Citation
Low-temperature response of asphalt concrete pavements

Material Information

Title:
Low-temperature response of asphalt concrete pavements
Creator:
Roque, Reynaldo
Publisher:
Reynaldo Roque
Publication Date:
Language:
English

Subjects

Subjects / Keywords:
Asphalt ( jstor )
Base courses ( jstor )
Bituminous concrete pavements ( jstor )
Cooling ( jstor )
Dynamic loads ( jstor )
Load tests ( jstor )
Low temperature ( jstor )
Pavements ( jstor )
Static loads ( jstor )
Structural deflection ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
030280416 ( alephbibnum )
16396428 ( oclc )

Downloads

This item has the following downloads:

oai.xml

UF00089974_00001.pdf

00067b.txt

00006.txt

UF00089974_00001_0050.txt

UF00089974_00001_0227.txt

00265.txt

00199.txt

UF00089974_00001_0189.txt

UF00089974_00001_0395.txt

00399.txt

UF00089974_00001_0362.txt

00409.txt

00206.txt

UF00089974_00001_0162.txt

00026.txt

UF00089974_00001_0327.txt

00047.txt

UF00089974_00001_0009.txt

00080.txt

00410.txt

UF00089974_00001_0375.txt

UF00089974_00001_0036.txt

00415.txt

UF00089974_00001_0250.txt

UF00089974_00001_0059.txt

UF00089974_00001_0301.txt

00288.txt

00058.txt

00339.txt

00372.txt

UF00089974_00001_0352.txt

UF00089974_00001_0190.txt

UF00089974_00001_0415.txt

UF00089974_00001_0033.txt

UF00089974_00001_0170.txt

UF00089974_00001_0084.txt

00105.txt

00060.txt

00054.txt

00092.txt

UF00089974_00001_0174.txt

UF00089974_00001_0270.txt

00282.txt

UF00089974_00001_0291.txt

UF00089974_00001_0023.txt

UF00089974_00001_0231.txt

UF00089974_00001_0220.txt

00233.txt

00430.txt

UF00089974_00001_0353.txt

UF00089974_00001_0284.txt

00280.txt

UF00089974_00001_0097.txt

UF00089974_00001_0002.txt

00051.txt

00269.txt

UF00089974_00001_0240.txt

00177.txt

UF00089974_00001_0058.txt

00380.txt

UF00089974_00001_0028.txt

00231.txt

UF00089974_00001_0180.txt

00263.txt

00416.txt

UF00089974_00001_0110.txt

00252.txt

00055.txt

00061.txt

00320.txt

00153.txt

00162.txt

00137.txt

UF00089974_00001_0271.txt

00205.txt

00253.txt

00392.txt

00296.txt

00183.txt

00067.txt

UF00089974_00001_0139.txt

UF00089974_00001_0054.txt

00142.txt

00181.txt

00237.txt

00037.txt

UF00089974_00001_0358.txt

UF00089974_00001_0006.txt

00326.txt

UF00089974_00001_0405.txt

00290.txt

00381.txt

UF00089974_00001_0267.txt

UF00089974_00001_0446.txt

00262.txt

00033.txt

UF00089974_00001_0307.txt

00440.txt

00215.txt

UF00089974_00001_0266.txt

UF00089974_00001_0428.txt

00100.txt

UF00089974_00001_0223.txt

00358.txt

00224.txt

00291.txt

UF00089974_00001_0168.txt

00096.txt

UF00089974_00001_0083.txt

00145.txt

00335.txt

UF00089974_00001_0172.txt

UF00089974_00001_0454.txt

UF00089974_00001_0276.txt

UF00089974_00001_0336.txt

UF00089974_00001_0111.txt

UF00089974_00001_0292.txt

00388.txt

00308.txt

UF00089974_00001_0232.txt

UF00089974_00001_0387.txt

UF00089974_00001_0420.txt

UF00089974_00001_0377.txt

UF00089974_00001_0090.txt

UF00089974_00001_0177.txt

UF00089974_00001_0215.txt

UF00089974_00001_0413.txt

00448.txt

00108.txt

UF00089974_00001_0453.txt

UF00089974_00001_0272.txt

UF00089974_00001_0262.txt

UF00089974_00001_0338.txt

UF00089974_00001_0071.txt

00316.txt

00338.txt

00333.txt

UF00089974_00001_0401.txt

UF00089974_00001_0078.txt

00174.txt

UF00089974_00001_0193.txt

UF00089974_00001_0403.txt

UF00089974_00001_0032.txt

UF00089974_00001_0414.txt

00317.txt

00062.txt

UF00089974_00001_0102.txt

00002.txt

00336.txt

UF00089974_00001_0101.txt

UF00089974_00001_0430.txt

UF00089974_00001_0029.txt

UF00089974_00001_0255.txt

UF00089974_00001_0450.txt

00112.txt

00146.txt

00243.txt

UF00089974_00001_0341.txt

00076.txt

UF00089974_00001_0386.txt

00057.txt

00378.txt

00293.txt

00433.txt

UF00089974_00001_0258.txt

00359.txt

00148.txt

00373.txt

UF00089974_00001_0005.txt

00182.txt

UF00089974_00001_0206.txt

UF00089974_00001_0337.txt

UF00089974_00001_0382.txt

UF00089974_00001_0238.txt

UF00089974_00001_0354.txt

UF00089974_00001_0339.txt

00158.txt

UF00089974_00001_0460.txt

UF00089974_00001_0452.txt

00087.txt

UF00089974_00001_0013.txt

UF00089974_00001_0061.txt

00371.txt

00066.txt

UF00089974_00001_0072.txt

00186.txt

00402.txt

00419.txt

00073.txt

00075.txt

UF00089974_00001_0095.txt

UF00089974_00001_0434.txt

UF00089974_00001_0425.txt

UF00089974_00001_0167.txt

UF00089974_00001_0330.txt

UF00089974_00001_0152.txt

UF00089974_00001_0026.txt

00267.txt

00279.txt

UF00089974_00001_0383.txt

00343.txt

00442.txt

UF00089974_00001_0027.txt

00367.txt

00194.txt

UF00089974_00001_0442.txt

UF00089974_00001_0145.txt

UF00089974_00001_0211.txt

UF00089974_00001_0379.txt

UF00089974_00001_0243.txt

00385.txt

UF00089974_00001_0165.txt

00007.txt

00127.txt

UF00089974_00001_0224.txt

00398.txt

00071a.txt

00235.txt

00027.txt

00404.txt

00063.txt

00387.txt

UF00089974_00001_0402.txt

00315.txt

UF00089974_00001_0221.txt

00270.txt

00352.txt

UF00089974_00001_0130.txt

UF00089974_00001_0155.txt

UF00089974_00001_0300.txt

00114.txt

UF00089974_00001_0288.txt

00221.txt

00091.txt

00071.txt

00120.txt

UF00089974_00001_0100.txt

00059.txt

00223.txt

UF00089974_00001_0105.txt

00136.txt

UF00089974_00001_0304.txt

UF00089974_00001_0265.txt

00439.txt

00259.txt

UF00089974_00001_0253.txt

00284.txt

00150.txt

00303.txt

UF00089974_00001_0324.txt

UF00089974_00001_0398.txt

00386.txt

00341.txt

UF00089974_00001_0053.txt

UF00089974_00001_0455.txt

UF00089974_00001_0022.txt

00444.txt

UF00089974_00001_0334.txt

00330.txt

UF00089974_00001_0062.txt

UF00089974_00001_0226.txt

00042.txt

UF00089974_00001_0329.txt

UF00089974_00001_0147.txt

UF00089974_00001_0429.txt

00012.txt

UF00089974_00001_0283.txt

UF00089974_00001_0183.txt

00201.txt

00360.txt

00445.txt

UF00089974_00001_0404.txt

00156.txt

00125.txt

UF00089974_00001_0010.txt

00023.txt

UF00089974_00001_0067.txt

00350.txt

00167.txt

UF00089974_00001_0419.txt

00039.txt

UF00089974_00001_0001.txt

UF00089974_00001_0051.txt

UF00089974_00001_0122.txt

00449.txt

00218.txt

UF00089974_00001_0217.txt

UF00089974_00001_0432.txt

UF00089974_00001_0186.txt

UF00089974_00001_0119.txt

00122.txt

UF00089974_00001_0384.txt

UF00089974_00001_0317.txt

00258.txt

00368.txt

00408.txt

UF00089974_00001_0146.txt

UF00089974_00001_0080.txt

00163.txt

00255.txt

00407.txt

00256.txt

UF00089974_00001_0075.txt

00133.txt

UF00089974_00001_0421.txt

UF00089974_00001_0316.txt

00210.txt

00072.txt

00426.txt

00081.txt

00382.txt

UF00089974_00001_0041.txt

00020.txt

UF00089974_00001_0219.txt

UF00089974_00001_0148.txt

UF00089974_00001_0309.txt

00318.txt

UF00089974_00001_pdf.txt

00274.txt

00038.txt

00322.txt

UF00089974_00001_0202.txt

UF00089974_00001_0007.txt

UF00089974_00001_0164.txt

UF00089974_00001_0229.txt

00268.txt

00309.txt

00213.txt

UF00089974_00001_0394.txt

00250.txt

00356.txt

00188.txt

00179.txt

UF00089974_00001_0287.txt

00403.txt

UF00089974_00001_0335.txt

UF00089974_00001_0082.txt

00379.txt

00425.txt

00193.txt

UF00089974_00001_0399.txt

UF00089974_00001_0248.txt

UF00089974_00001_0039.txt

00383.txt

UF00089974_00001_0123.txt

UF00089974_00001_0087.txt

UF00089974_00001_0312.txt

UF00089974_00001_0269.txt

00390.txt

UF00089974_00001_0369.txt

00151.txt

UF00089974_00001_0412.txt

UF00089974_00001_0088.txt

UF00089974_00001_0076.txt

00429.txt

00327.txt

00447.txt

UF00089974_00001_0203.txt

00101.txt

00011.txt

00238.txt

UF00089974_00001_0047.txt

00277.txt

00190.txt

00285.txt

UF00089974_00001_0086.txt

00160.txt

UF00089974_00001_0129.txt

UF00089974_00001_0342.txt

UF00089974_00001_0440.txt

UF00089974_00001_0351.txt

UF00089974_00001_0263.txt

UF00089974_00001_0370.txt

00034.txt

00010.txt

00083.txt

00377.txt

UF00089974_00001_0234.txt

00311.txt

00157.txt

00422.txt

00143.txt

00024.txt

UF00089974_00001_0345.txt

UF00089974_00001_0049.txt

00405.txt

UF00089974_00001_0348.txt

UF00089974_00001_0393.txt

UF00089974_00001_0372.txt

UF00089974_00001_0012.txt

UF00089974_00001_0163.txt

00110.txt

00093.txt

UF00089974_00001_0191.txt

UF00089974_00001_0116.txt

UF00089974_00001_0416.txt

UF00089974_00001_0166.txt

00354.txt

00423.txt

00117.txt

UF00089974_00001_0213.txt

UF00089974_00001_0381.txt

00247.txt

UF00089974_00001_0120.txt

00234.txt

UF00089974_00001_0364.txt

UF00089974_00001_0205.txt

00152.txt

00310.txt

UF00089974_00001_0319.txt

UF00089974_00001_0302.txt

00184.txt

UF00089974_00001_0132.txt

UF00089974_00001_0305.txt

00022.txt

UF00089974_00001_0118.txt

00204.txt

00119.txt

00189.txt

00168.txt

00328.txt

00111.txt

UF00089974_00001_0151.txt

00154.txt

UF00089974_00001_0144.txt

UF00089974_00001_0285.txt

UF00089974_00001_0197.txt

UF00089974_00001_0207.txt

00248.txt

00207.txt

UF00089974_00001_0371.txt

UF00089974_00001_0113.txt

UF00089974_00001_0290.txt

00019.txt

UF00089974_00001_0462.txt

00289.txt

00203.txt

UF00089974_00001_0435.txt

00251.txt

00126.txt

00135.txt

UF00089974_00001_0423.txt

UF00089974_00001_0245.txt

UF00089974_00001_0133.txt

UF00089974_00001_0204.txt

00283.txt

UF00089974_00001_0199.txt

UF00089974_00001_0332.txt

UF00089974_00001_0085.txt

00172.txt

00421.txt

00363.txt

00191.txt

00396.txt

UF00089974_00001_0417.txt

00170.txt

UF00089974_00001_0390.txt

UF00089974_00001_0138.txt

00220.txt

00246.txt

UF00089974_00001_0112.txt

UF00089974_00001_0233.txt

UF00089974_00001_0222.txt

00169.txt

UF00089974_00001_0136.txt

00299.txt

00070.txt

00032.txt

UF00089974_00001_0391.txt

00374.txt

UF00089974_00001_0365.txt

UF00089974_00001_0192.txt

UF00089974_00001_0459.txt

00337.txt

00411.txt

UF00089974_00001_0042.txt

00138.txt

00068.txt

UF00089974_00001_0142.txt

UF00089974_00001_0441.txt

00342.txt

UF00089974_00001_0448.txt

UF00089974_00001_0279.txt

00241.txt

00323.txt

UF00089974_00001_0015.txt

UF00089974_00001_0106.txt

00294.txt

UF00089974_00001_0438.txt

00437.txt

UF00089974_00001_0069.txt

UF00089974_00001_0055.txt

00107.txt

UF00089974_00001_0096.txt

00217.txt

UF00089974_00001_0160.txt

00346.txt

00428.txt

UF00089974_00001_0321.txt

00128.txt

00140.txt

UF00089974_00001_0385.txt

UF00089974_00001_0367.txt

00212.txt

UF00089974_00001_0360.txt

00355.txt

UF00089974_00001_0461.txt

00064.txt

00454.txt

UF00089974_00001_0038.txt

00008.txt

UF00089974_00001_0449.txt

UF00089974_00001_0117.txt

00035.txt

UF00089974_00001_0108.txt

UF00089974_00001_0056.txt

00095.txt

UF00089974_00001_0043.txt

00200.txt

00264.txt

00271.txt

00427.txt

00090.txt

00196.txt

00312.txt

UF00089974_00001_0242.txt

UF00089974_00001_0376.txt

00016.txt

UF00089974_00001_0431.txt

UF00089974_00001_0296.txt

UF00089974_00001_0368.txt

UF00089974_00001_0346.txt

UF00089974_00001_0297.txt

00222.txt

UF00089974_00001_0314.txt

UF00089974_00001_0356.txt

UF00089974_00001_0406.txt

00116.txt

UF00089974_00001_0218.txt

UF00089974_00001_0254.txt

UF00089974_00001_0175.txt

UF00089974_00001_0092.txt

00118.txt

UF00089974_00001_0239.txt

00005.txt

UF00089974_00001_0328.txt

UF00089974_00001_0093.txt

00103.txt

UF00089974_00001_0209.txt

00304.txt

00208.txt

UF00089974_00001_0184.txt

UF00089974_00001_0355.txt

UF00089974_00001_0396.txt

00166.txt

00394.txt

00301.txt

00197.txt

00063a.txt

UF00089974_00001_0407.txt

00017.txt

UF00089974_00001_0094.txt

00139.txt

UF00089974_00001_0244.txt

oai_xml.txt

00178.txt

00097.txt

UF00089974_00001_0359.txt

UF00089974_00001_0016.txt

00321.txt

UF00089974_00001_0310.txt

UF00089974_00001_0318.txt

00451.txt

UF00089974_00001_0157.txt

00050.txt

00397.txt

UF00089974_00001_0357.txt

00121.txt

00085.txt

00195.txt

00018.txt

00227.txt

00307.txt

00098.txt

UF00089974_00001_0331.txt

UF00089974_00001_0201.txt

Copyright.txt

UF00089974_00001_0126.txt

UF00089974_00001_0315.txt

UF00089974_00001_0456.txt

UF00089974_00001_0114.txt

00209.txt

00414.txt

UF00089974_00001_0282.txt

00113.txt

UF00089974_00001_0260.txt

00052.txt

UF00089974_00001_0128.txt

00375.txt

UF00089974_00001_0079.txt

00144.txt

UF00089974_00001_0185.txt

00434.txt

00084.txt

00347.txt

00452.txt

UF00089974_00001_0057.txt

UF00089974_00001_0159.txt

00069.txt

00245.txt

UF00089974_00001_0295.txt

UF00089974_00001_0030.txt

00134.txt

UF00089974_00001_0451.txt

00239.txt

UF00089974_00001_0323.txt

00004.txt

UF00089974_00001_0091.txt

UF00089974_00001_0065.txt

UF00089974_00001_0103.txt

UF00089974_00001_0278.txt

UF00089974_00001_0350.txt

00417.txt

UF00089974_00001_0286.txt

00088.txt

UF00089974_00001_0194.txt

00187.txt

UF00089974_00001_0274.txt

UF00089974_00001_0236.txt

00362.txt

00240.txt

00349.txt

00292.txt

00357.txt

00393.txt

00370.txt

UF00089974_00001_0182.txt

00286.txt

UF00089974_00001_0320.txt

00353.txt

UF00089974_00001_0161.txt

00287.txt

UF00089974_00001_0019.txt

00029.txt

00257.txt

00391.txt

UF00089974_00001_0021.txt

00175.txt

00226.txt

00272.txt

UF00089974_00001_0246.txt

00074.txt

UF00089974_00001_0373.txt

00254.txt

00432.txt

UF00089974_00001_0249.txt

UF00089974_00001_0210.txt

UF00089974_00001_0298.txt

00438.txt

UF00089974_00001_0044.txt

00249.txt

UF00089974_00001_0444.txt

00132.txt

00443.txt

UF00089974_00001_0034.txt

00060a.txt

UF00089974_00001_0003.txt

UF00089974_00001_0259.txt

UF00089974_00001_0303.txt

UF00089974_00001_0443.txt

UF00089974_00001_0343.txt

UF00089974_00001_0293.txt

UF00089974_00001_0322.txt

00077.txt

00300.txt

UF00089974_00001_0063.txt

UF00089974_00001_0294.txt

00219.txt

UF00089974_00001_0411.txt

00041.txt

00436.txt

UF00089974_00001_0256.txt

UF00089974_00001_0214.txt

00236.txt

00053.txt

00340.txt

00164.txt

UF00089974_00001_0181.txt

UF00089974_00001_0099.txt

00198.txt

UF00089974_00001_0374.txt

00229.txt

UF00089974_00001_0280.txt

00332.txt

00401.txt

00104.txt

UF00089974_00001_0208.txt

00453.txt

00185.txt

00067a.txt

UF00089974_00001_0178.txt

00115.txt

00078.txt

00149.txt

00141.txt

UF00089974_00001_0261.txt

UF00089974_00001_0268.txt

UF00089974_00001_0241.txt

UF00089974_00001_0173.txt

00324.txt

UF00089974_00001_0311.txt

UF00089974_00001_0017.txt

00131.txt

UF00089974_00001_0275.txt

00021.txt

UF00089974_00001_0004.txt

00424.txt

UF00089974_00001_0347.txt

UF00089974_00001_0237.txt

UF00089974_00001_0008.txt

UF00089974_00001_0408.txt

00028.txt

UF00089974_00001_0179.txt

00348.txt

UF00089974_00001_0361.txt

00216.txt

UF00089974_00001_0427.txt

UF00089974_00001_0388.txt

00275.txt

00331.txt

UF00089974_00001_0018.txt

00031.txt

00073a.txt

00009.txt

UF00089974_00001_0089.txt

00230.txt

00276.txt

00295.txt

00281.txt

00046.txt

00329.txt

UF00089974_00001_0378.txt

00298.txt

UF00089974_00001_0198.txt

00344.txt

UF00089974_00001_0014.txt

00278.txt

UF00089974_00001_0349.txt

00266.txt

UF00089974_00001_0235.txt

00366.txt

UF00089974_00001_0212.txt

00364.txt

00384.txt

00147.txt

00297.txt

00413.txt

00376.txt

UF00089974_00001_0252.txt

UF00089974_00001_0216.txt

00044.txt

UF00089974_00001_0073.txt

UF00089974_00001_0424.txt

00013.txt

UF00089974_00001_0400.txt

UF00089974_00001_0064.txt

00228.txt

00319.txt

00412.txt

UF00089974_00001_0070.txt

UF00089974_00001_0340.txt

00389.txt

00001.txt

00109.txt

UF00089974_00001_0308.txt

00225.txt

00099.txt

UF00089974_00001_0046.txt

UF00089974_00001_0195.txt

UF00089974_00001_0366.txt

00345.txt

UF00089974_00001_0426.txt

00102.txt

UF00089974_00001_0140.txt

UF00089974_00001_0127.txt

UF00089974_00001_0037.txt

00180.txt

UF00089974_00001_0251.txt

UF00089974_00001_0031.txt

00040.txt

00361.txt

UF00089974_00001_0156.txt

UF00089974_00001_0380.txt

00129.txt

00313.txt

UF00089974_00001_0389.txt

UF00089974_00001_0011.txt

00094.txt

00159.txt

00420.txt

00302.txt

00014.txt

UF00089974_00001_0277.txt

UF00089974_00001_0141.txt

UF00089974_00001_0060.txt

00086.txt

UF00089974_00001_0068.txt

UF00089974_00001_0107.txt

UF00089974_00001_0121.txt

UF00089974_00001_0115.txt

UF00089974_00001_0326.txt

UF00089974_00001_0131.txt

00242.txt

UF00089974_00001_0035.txt

00232.txt

UF00089974_00001_0257.txt

UF00089974_00001_0158.txt

UF00089974_00001_0437.txt

UF00089974_00001_0153.txt

UF00089974_00001_0457.txt

00305.txt

00130.txt

00049.txt

UF00089974_00001_0397.txt

UF00089974_00001_0149.txt

UF00089974_00001_0228.txt

00079.txt

UF00089974_00001_0436.txt

UF00089974_00001_0135.txt

UF00089974_00001_0200.txt

00048.txt

00165.txt

00306.txt

00071b.txt

00431.txt

UF00089974_00001_0025.txt

UF00089974_00001_0333.txt

00446.txt

00211.txt

00123.txt

UF00089974_00001_0124.txt

UF00089974_00001_0052.txt

00334.txt

UF00089974_00001_0306.txt

00065.txt

UF00089974_00001_0196.txt

00261.txt

00106.txt

UF00089974_00001_0281.txt

UF00089974_00001_0045.txt

00214.txt

00435.txt

00365.txt

UF00089974_00001_0109.txt

UF00089974_00001_0418.txt

UF00089974_00001_0409.txt

00369.txt

00015.txt

UF00089974_00001_0422.txt

UF00089974_00001_0439.txt

UF00089974_00001_0433.txt

UF00089974_00001_0299.txt

UF00089974_00001_0137.txt

UF00089974_00001_0392.txt

00314.txt

00056.txt

UF00089974_00001_0098.txt

00192.txt

00045.txt

UF00089974_00001_0171.txt

UF00089974_00001_0458.txt

00161.txt

00171.txt

00441.txt

00176.txt

UF00089974_00001_0230.txt

UF00089974_00001_0463.txt

UF00089974_00001_0154.txt

00173.txt

UF00089974_00001_0169.txt

00202.txt

UF00089974_00001_0344.txt

UF00089974_00001_0289.txt

UF00089974_00001_0143.txt

00418.txt

00351.txt

00030.txt

UF00089974_00001_0040.txt

00450.txt

UF00089974_00001_0077.txt

UF00089974_00001_0447.txt

UF00089974_00001_0363.txt

00325.txt

UF00089974_00001_0150.txt

UF00089974_00001_0024.txt

UF00089974_00001_0313.txt

00406.txt

00244.txt

UF00089974_00001_0264.txt

UF00089974_00001_0176.txt

00089.txt

UF00089974_00001_0074.txt

UF00089974_00001_0187.txt

00082.txt

UF00089974_00001_0445.txt

UF00089974_00001_0066.txt

UF00089974_00001_0081.txt

UF00089974_00001_0188.txt

UF00089974_00001_0104.txt

00155.txt

UF00089974_00001_0134.txt

00273.txt

00036.txt

UF00089974_00001_0225.txt

00124.txt

00068a.txt

UF00089974_00001_0273.txt

UF00089974_00001_0125.txt

UF00089974_00001_0247.txt

00260.txt

00400.txt

00043.txt

00395.txt

UF00089974_00001_0325.txt

00025.txt

UF00089974_00001_0410.txt

UF00089974_00001_0020.txt

UF00089974_00001_0048.txt

00003.txt


Full Text










LOW-TEMPERATURE RESPONSE
OF ASPHALT CONCRETE PAVEMENTS






By

REYNALDO ROQUE


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY


1986













ACKNOWLEDGMENTS


I would like to express my gratitude to Dr. Byron E. Ruth, Chairman

of my Graduate Supervisory Committee, for his guidance, encouragement,

and friendship. I would also like to thank Dr. F. C. Townsend,

Dr. J. L. Davidson, Professor W. H. Zimpfer, Dr. M. C. McVay,

Dr. D. L. Smith, and Dr. J. L. Eades for serving on my Graduate

Supervisory Committee. I consider myself fortunate to have had such a

distinguished committee.

A very special thanks goes to the Florida Department of

Transportation (FDOT) for providing the financial support, testing

facilities, materials, and personnel that made this research possible.

I would also like to thank the many individuals at the Bureau of

Materials and Research of the FDOT who contributed to this research

project by giving so generously of their time. I especially want to

thank the personnel in the Pavement Performance Division, Bituminous

Materials and Research Section, the Pavement Evaluation Section and the

Soil Materials and Research Section for their help and consideration.

I would also like to thank Candace Leggett for her

conscientiousness and diligence in typing this dissertation.

Finally, I would like to thank my wife Maria for encouraging me to

return to school, and for her encouragement and patience throughout my

Ph.D. program.














TABLE OF CONTENTS

Pane

ACKNOWLEDGMENTS................................................ ii

LIST OF TABLES............................................... ..... vi

LIST OF FIGURES ..................... ............................ x

ABSTRACT......................... .. ........... ... .... ....... xvi

CHAPTERS

I INTRODUCTION... .............. .............................. 1

II LITERATURE REVIEW .................................. ...... 5

2.1 Introduction........................................... 5
2.2 Distress In Asphalt Concrete Pavements.................. 6
2.2.1 Modes of Distress.................................. 6
2.2.2 Cracking Mechanisms................................. 9
2.3 Properties of Asphalt Cement and Asphalt Concrete
As Related to Low-Temperature Pavement Response
and Cracking........................................... 15
2.3.1 Asphalt Cement Properties.......................... 16
2.3.2 Asphalt Mixture Properties.......................... 20
2.4 Properties of Foundation Materials...................... 30
2.5 Prediction of Thermal- and Load-Induced Stresses,
Strains, and Failure In Asphalt Concrete Pavements...... 35

III EQUIPMENT AND FACILITIES..................................... 42

3.1 Description of Test Pit Facility....................... 42
3.2 System for Hot Mix Asphalt Distribution................. 43
3.3 Pavement Cooling System................................ 43
3.4 Measurement System for Pavement Response................ 47
3.4.1 Measuring Instruments............................ 49
3.4.2 Data Acquisition System............................ 55
3.5 Loading Systems: Rigid Plate Load vs. Flexible
Dual Wheels............................................. 62

IV EFFECT OF ENCLOSED CONCRETE TEST PIT ON PAVEMENT RESPONSE.... 67

4.1 Introduction........................................... 67
4.2 Preliminary Analysis .................................... 67
4.3 Effect of Test Pit Constraints.......................... 72
4.3.1 Analytical Model ..................... ............. 72














4.3.2 Effect of Constraints on Subgrade Response.......... 73
4.3.3 Effect of Constraints on Limerock Base Response..... 77
4.3.4 Effect of Constraints on Three-Layer System......... 85
4.4 Methodoloqy to Account for the Effect of Test Pit
Constraints on Pavement Response Prediction............. 91
4.4.1 Rigid Plate Loading on the Subgrade................. 91
4.4.2 Rigid Plate Loading on the Reinforcing Base
Layer............ ........ ................ ......... 92
4.4.3 Predicting Pavement System Response in the
Test Pit.................... .................. .. 104

V MATERIALS AND PLATE TESTING PROCEDURES....................... 113

5.1 Introduction............... ............................. 113
5.2 Laboratory Tests...................................... 113
5.2.1 Fairbanks Sand Subgrade............................. 113
5.2.2 Crushed Limerock Base............................... 113
5.2.3 Asphalt Cement and Asphalt Concrete................. 115
5.3 Material Placement and Compaction...................... 122
5.4 Material Properties In Situ ........................... 126
5.4.1 Plate Load Test Procedures......................... 126
5.4.2 Plate Tests Immediately After Placement............ 129
5.4.3 Plate Tests After Pavement Removal .................. 141

VI PROCEDURES.................................................. 150

6.1 Dynamic Plate Load Tests at Ambient Temperatures........ 150
6.2 Low-Temperature Pavement Response Tests.................. 151
6.2.1 Introduction...................................... 151
6.2.2 Pavement Cooling and Initial Dynamic Load Tests..... 153
6.2.3 Creep Test Procedures............................... 155

VII PAVEMENT RESPONSE AT AMBIENT TEMPERATURES.................... 157

7.1 Initial Plate Load Tests at Fast Loading Rate........... 159
7.1.1 Dynamic Load Test Results........................... 159
7.1.2 Elastic Layer Simulation and Evaluation of
Results................. .... .............. .. ........ 163
7.2 Initial Plate Load Tests at Slow Loading Rate........... 170
7.2.1 Dynamic Load Test Results........................... 170
7.2.2 Evaluation of Results............................... 170
7.3 Additional Plate Load Tests at Fast Loading Rate........ 182
7.3.1 Dynamic Load Test Results........................... 182
7.3.2 Elastic Layer Simulation and Evaluation of
Results................................... ... .. .... 186
7.4 Summary ... ......................... .................... 213














VIII LOW-TEMPERATURE PAVEMENT RESPONSE........................... 216

8.1 Preliminary Tests With the Rigid Plate.................. 216
8.2 Reinstrumentation for Tests With the Dual Wheels......... 221
8.3 Results of Tests With the Dual Wheel Loading System..... 221
8.3.1 Introduction............................ ...... 221
8.3.2 Pavement Response During Cooling.................. 223
8.3.3 Dynamic Load and Creep Response at Different
Temperatures........................................ 255
8.3.4 Combined Effect of Thermal and Load Response........ 320

IX RESPONSE PREDICTION AT LOW TEMPERATURES.................... 331

9.1 Dynamic Load Response......................... ........ 331
9.2 Thermal Resoonse...................................... 347
9.3 Creep Response........................................ 348

X CONCLUSIONS AND RECOM"ENOATIONS .............................. 350
10.1 Conclusions........................ ............... 350
10.1.1 Pavement Testinq and Evaluation Method............. 350
10.1.2 Thermal and Load Response of Asphalt Concrete
Pavements ....... ............... ............ ...... 356
10.2 Recommendations...................................... 356

APPENDICES

A RELATIONSHIPS BETWEEN ASPHALT CONCRETE PROPERTIES
AND ASPHALT CEMENT PROPERTIES............................... 359

B PAVEMENT TEMPERATURES DURING COOLING ........................ 361

C MEASURED THERMAL STRAINS DURING COOLING..................... 368

D DYNAMIC LOAD RESPONSE MEASUREMENTS .......................... 375

E CREEP TEST DATA............................................ 396

REFERENCES....................... ... ...... .......... .............. 433

BIOGRAPHICAL SKETCH.............................................. 443














LIST OF TABLES


Table Page

2.1 Primary Types and Causes of Distress In Asphalt
Concrete Pavements......................................... 7
2.2 Modes, Manifestations, and Mechanisms of Types of
Distress............ ..... .................... ............... 8

4.1 Sand Subgrade Modulus for Different Layer Depths
and Poisson's Ratio.................................... 75
4.2 Effect of Concrete Floor on Surface Deflections
for Different Base Stiffnesses .............................. 79
4.3 Predicted Deflections Using AXSYM........................... 83
4.4 Tabulated Deflection Basins to Show Effect of
Test Pit Floor on Pavements of Different Stiffness........... 87
4.5 Effect of Test Pit Walls on Surface Deflections
for Pavements of Different Stiffness......................... 90
4.6 Computer Runs to Determine Poisson's Ratio Effect............ 99
4.7 Measured and Predicted Surface Strains....................... 109

5.1 Laboratory Test Results: Fairbanks Sand.................... 114
5.2 Laboratory Test Results: Ocala Formation Limerock........... 116
5.3 Source of Materials and Job Mix Formula for Asphalt
Concrete.................................................. 117
5.4 Test Pit Asphalt Concrete Properties......................... 118
5.5 Rheology and Penetration of Asphalt Recovered From
Test Pit During Initial Placement: September, 1982.......... 120
5.6 Rheology and Penetration of Asphalt Recovered From
Test Pit After All Testing: September, 1985..................... 121
5.7 Load Increments Used for Plate Load Tests: Fairbanks
Sand Subgrade.............................................. 127
5.8 Load Increments Used for Plate Load Tests:
Limerock Base............................ .................... 128
5.9 Modulus Values Immediately After Placement:
Fairbanks Sand Suhgrade.................................... 134
5.10 Modulus Values Immediately After Placement:
Limerock Base.............................................. 139
5.11 Modulus Values Without Accounting for Test Pit
Constraints............................................ 142
5.12 Modulus Values After Pavement Removal:
Fairbanks Sand Subqrade.................................... 145
5.13 Modulus Values After Pavement Removal: Limerock Base........ 146

6.1 Summary of Order of Testing................................. 152

7.1 Summary of Dynamic Plate Load Tests at Ambient
Temperatures................................................. 158








Table Page

7.2 Surface Deflections: Fast Loadina Rate..................... 162
7.3 Surface Strains: Fast Loading Rate......................... 162
7.4 Surface Deflections: Slow Loading Rate...................... 173
7.5 Surface Strains: Slow Loading Rate.......................... 173
7.6 Measured Surface Deflections at 20.6 C (69 F):
Center Plate Loading Position............................... 184
7.7 Measured Surface Strains at 20.6 C (69 F):
Center Plate Loading Position .............................. 184
7.8 Measured Surface Deflection at 25.6 C (78 F):
South Plate Loading Position................................ 185
7.9 Measured Strains at 25.6 (78 F):
South Plate Loading Position................................ 185

8.1 Summary of Average Pavement Temperatures During Testing...... 259

B.1 Pavement Temperatures During Cooling: Test Position 1....... 362
B.2 Pavement Temperatures During Cooling: Test Position 2....... 364
B.3 Pavement Temperatures During Cooling: Test Position 3...... 366

C.1 Measured Thermal Strains During Cooling: Test Position 1.... 369
C.2 Measured Thermal Strains During Cooling: Test Position 2.... 371
C.3 Measured Thermal Strains During Cooling: Test Position 3.... 373

D.1 Measured Deflections, Test Position 1, 0 C (32 F)........... 376
D.2 Measured Strains, Test Position 1, 0 C (32 F)................ 377
0.3 Measured Deflections, Test Position 1, 6.7 C (44 F).......... 378
0.4 Measured Strains, Test Position 1, 6.7 C (44 F).............. 379
D.5 Measured Deflections, Test Position 1, 13.3 C (56 F)......... 380
D.6 Measured Strains, Test Position 1, 13.3 C (56 F)............. 381
D.7 Measured Deflections, Test Position 1, 0 C (32 F),
Repeat Test................................................. 382
D.8 Measured Strains, Test Position 1, 0 C (32 F),
Repeat Test................................... ....... 383
D.9 Measured Deflections, Test Position 2, 0 C (32 F)............ 384
D.10 Measured Strains, Test Position 2, 0 C (32 F).............. 385
D.11 Measured Deflections, Test Position 2, 6.7 C (44 F).......... 386
0.12 Measured Strains, Test Position 2, 6.7 C (44 F).............. 387
0.13 Measured Deflections, Test Position 2, 13.3 C (56 F)......... 388
D.14 Measured Strains, Test Position 2, 13.3 C (56 F) ........... 389
D.15 Measured Deflections, Test Position 3, 0 C (32 F)............ 390
0.16 Measured Strains, Test Position 3, 0 C (32 F)................ 391
D.17 Measured Deflections, Test Position 3, 6.7 C (44 F).......... 392
D.18 Measured Strains, Test Position 3, 6.7 C (44 F).............. 393
D.19 Measured Deflections, Test Position 3, 13.3 C (56 F)......... 394
D.20 Measured Strains, Test Position 3, 13.3 C (56 F)... .... .... 395

E.1 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 1,
13.3 C (56 F).............................. ....... ........... 397
E.2 Measured Creep Strains For Different Times of 10,000-1b.
Static Load Application, Position Number 1, 13.3 C (56 F).... 398








Table Page

E.3 Measured Dynamic Deflections at 10,000 lbs. After
Different Times of Static Load Application, Position
Number 1, 13.3 C (56 F)...................................... 399
E.4 Measured Dynamic Strains at 10,000 Ibs. After Different
Times of Static Load Application, Position Number 1,
13.3 C (56 F).............................................. 400
E.5 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 1,
6.7 C (44 F) ............... ......................... ...... .. 401
E.6 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position Number 1, 6.7 C (44 F)..... 402
E.7 Measured Dynamic Deflections at 10,000 Ibs. After
Different Times of Static Load Application, Position
Number 1, 6.7 C (44 F)....................................... 403
E.8 Measured Dynamic Strains at 10,000 lbs. After Different
Times of Static Load Application, Position Number 1,
6.7 C (44 F)........... ................ ........ .... 404
E.9 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 1,
0 C (32 F)..................... .............. 405
E.10 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position Number 1, 0 C (32 F)....... 406
E.11 Measured Dynamic Deflections at 10,000 lbs. After
Different Times of Static Load Application, Position
Number 1, 0 C (32 F)........................................ 407
E.12 Measured Dynamic Strains at 10,000 lbs. After Different
Times of Static Load Apolication, Positon Number 1,
0 C (32 F) .............................. .... ... .... ...... 408
E.13 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 2,
6.7 C (44 F)......................... .... .................... ...... 409
E.14 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position Number 2, 6.7 C (44 F)..... 410
E.15 Measured Dynamic Deflections at 10,000 Ibs. After
Different Times of Static Load Application, Position
Number 2, 6.7 C (44 F)............................ .......... 411
E.16 Measured Dynamic Strains at 10,000 Ibs. After Different
Times of Static Load Application, Position Number 2,
6.7 C (44 F) .................. ...... ......... .... .... .... 412
E.17 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 2,
0 C (32 F)................................................ 413
E.18 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position 2, 0 C (32 F).............. 414
E.19 Measured Dynamic Deflections at 10,000 Ibs. After
Different Times of Static Load Application, Position
Number 2, 0 C (32 F)... ........ ........................... 415
E.20 Measured Dynamic Strains at 10,000 Ibs. After Different
Times of Static Load Application, Position Number 2,
0 C (32 F)................................ .... ............. 416


viii








Table Page

E.21 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 2,
13.3 C (56 F)................................................ 417
E.22 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position Number 2, 13.3 C (56 F).... 418
E.23 Measured Dynamic Deflections at 10,000 Ibs. After
Different Times of Static Load Application, Position
Number 2, 13.3 C (56 F)...................................... 419
E.24 Measured Dynamic Strains at 10,000 Ibs. After Different
Times of Static Load Application, Position Number 2,
13.3 C (56 F)................................................ 420
E.25 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 3,
0 C (32 F)............................................ ..... 421
E.26 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position Number 3, 0 C (32 F)....... 422
E.27 Measured Dynamic Deflections at 10,000 Ibs. After
Different Times of Static Load Application, Position
Number 3, 0 C (32 F) .................................... 423
E.28 Measured Dynamic Strains at 10,000 Ibs. After Different
Times of Static Load Application, Position Number 3,
0 C (32 F)................................. .................. 424
E.29 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 3,
6.7 C (44 F)............................................ ... 425
E.30 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position Number 3, 6.7 C (44 F)..... 426
E.31 Measured Dynamic Deflections at 10,000 Ibs. After
Different Times of Static Load Application, Position
Number 3, 6.7 C (44 F)...................................... 427
E.32 Measured Dynamic Strains at 10,000 Ibs. After Different
Times of Static Load Application, Position Number 3,
6.7 C (44 F) .............. .................. ............ 428
E.33 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 3,
13.3 C (56 F) ....................................... 429
E.34 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position Number 3, 13.3 C (56 F).... 430
E.35 Measured Dynamic Deflections at 10,000 Ibs. After
Different Times of Static Load Application, Position
Number 3, 13.3 C (56 F).............................. ...... 431
E.36 Measured Dynamic Strains at 10,000 Ibs. After Different
Times of Static Load Application, Position Number 3,
13.3 C (56 F)............................ ................... 432













LIST OF FIGURES


Figure Page

3.1 Hopper for Asphalt Hot Mix Distribution.................... 44
3.2 Layout of Test Pit Cooling System.......................... 46
3.3 Insulated Test Pit Cover Completely Installed................ 48
3.4 Insulated Test Pit Cover With Panels Removed.............. 48
3.5 LVDT Support System for Plate Loading........................ 50
3.6 LVDT Support System for Dual Wheel Loading................... 51
3.7 LVDT Prepared for Tests at Low Temperatures................. 53
3.8 Test Pit Pavement Completely Instrumented and Ready
for Testing With Dual Wheels
a) Frontal View .......................................... 54
b) Diagonal View........................................... 54
3.9 Two-Inch Strain Gages Mounted on Asphalt Concrete........... 56
3.10 Schematic Diagram of Data Acquisition System................. 59
3.11 Data Acquisition System In Test Pit Facility................. 60
3.12 Typical Deflection Recording on Digital Oscilloscope......... 60
3.13 Typical Deflection Output on X-Y Plotter.................... 61
3.14 Rigid Plate Loading System........... ....................... 63
3.15 Flexible Dual Wheel Loading System......................... 64

4.1 Measured and Predicted Deflection Basins in the Test Pit..... 69
4.2 Effect of Concrete Floor at Different Depths on
Predicted Deflection Basins................................. 71
4.3 Effect of Different Base Layer Stiffness on
Predicted Deflection Basins...... ............................ 80
4.4 Effect of Test Pit Walls on Limerock Base Response.......... 82
4.5 Comparison of AXSYM and Elastic Layer Theory Solutions....... 84
4.6 Three-Layer Systems as Modeled for Analysis.................. 86
4.7 Pavement System Models to Determine Wall Effect.............. 89
4.8 Equivalent Systems Based on Maximum Plate Deflection
on Subgrade.............. ........... ...................... 94
4.9 Comparison of Response of Equivalent Systems Based
on Maximum Plate Deflection on Subgrade .................... 95
4.10 Comparison of Test Pit System and Burmister System........... 97
4.11 Stress Distribution Under Riqid Plate on
Semi-Infinite Mass: 50 psi Average Pressure................. 101
4.12 Measured vs. Predicted Deflection Basins at
18.3 C (65 F).................... ........................... 107

5.1 Viscosity Temperature Relationships for Asphalt
Recovered from the Test Pit................................. 123
5.2 Location of Plate Load Tests: Fairbanks Sand Subgrade....... 130
5.3 Location of Plate Load Tests: Limerock Base................. 130
5.4 Applied Stress vs. Deflection: 12-in. Plate on
Fairbanks Sand Subgrade .................................... 132








Figure Page

5.5 Applied Stress vs. Deflection: 30-in. Plate on
Fairbanks Sand Subgrade.................................... 133
5.6 Applied Stress vs. Deflection: 16-in. Plate on
Limerock Base..............................................136
5.7 Deflections Used to Calculate Limerock Moduli.............. 138
5.8 Location of Plate Load Tests: Fairbanks Sand Subgrade....... 144
5.9 Location of Plate Load Tests: Limerock Base................. 144

7.1 Test Pit Diagram: Elevation ............................... 160
7.2 Test Pit Diagram: Plan...................................... 161
7.3 Measured Deflection Basins: Fast Loading Rate............... 164
7.4 Measured Strain Distributions: Fast Loading Rate............ 165
7.5 Load-Deflection Relationships: Fast Loading Rate............ 166
7.6 Load-Strain Relationships: Fast Loading Rate................ 167
7.7 Test Pit Pavement System as Modeled for Elastic Layer
Analysis ..................................................... 169
7.8 Measured Deflection Basins: Slow Loading Rate............... 171
7.9 Measured Strain Distributions: Slow Loading Rate............ 172
7.10 Load-Deflection Relationships: Slow Loading Rate............ 175
7.11 Load-Strain Relationships: Slow Loading Rate............. 176
7.12 Deflection Basin Comparison for Fast and Slow Loading
Rates: 10,000 Ibs.............. ........... ................ 177
7.13 Deflection Basin Comparison for Fast and Slow Loading
Rates: 7,000 Ibs............................ .............. 178
7.14 Deflection Basin Comparison for Fast and Slow Loading
Rates: 4,000 Ibs............................................ 179
7.15 Deflection Basin Comparison for Fast and Slow Loading
Rates: 1,000 Ibs.............. ............................. 180
7.16 Location of Plate Loading Positions and Strain and
Deflection Measurements ......... ............ .............. 183
7.17 Measured Deflection Basins at 20.6 C (69 F).................. 187
7.18 Measured Deflection Basins at 25.6 C (78 F)............... 188
7.19 Measured Strain Distributions at 20.6 C (69 F).............. 189
7.20 Measured Strain Distributions at 25.6 C (69 F).............. 190
7.21 Deflection Basin Comparison at 10,000 Ibs.................. 191
7.22 Deflection Basin Comparison at 7,000 Ibs..................... 192
7.23 Deflection Basin Comparison at 4,000 Ibs.................... 193
7.24 Deflection Basin Comparison at 1,000 Ibs.................. 194
7.25 Strain Distribution Comparison at 10,000 Ibs................ 195
7.26 Strain Distribution Comparison at 7,000 Ibs................. 196
7.27 Strain Distribution Comparison at 4,000 Ibs................. 197
7.28 Strain Distribution Comparison at 1,000 Ibs................. 198
7.29 Measured vs. Predicted Deflections at 10,000 Ibs.:
E2 = 53,000 psi, 20.6 C (69 F)............................. 201
7.30 Measured vs. Predicted Strains at 10,000 Ibs.:
E2 = 53,000 psi, 20.6 C (69 F).............................. 202
7.31 Measured vs. Predicted Deflections at 10,000 Ibs.:
Eq = 75,000 psi, 20.6 C (69 F)............................. 203
7.32 Measured vs. Predicted Strains at 10,000 Ibs.:
E2 = 75,000 psi, 20.6 C (69 F)........................... 204
7.33 Measured Load-Deflection Relationships at 20.6 C (69 F)...... 209
7.34 Measured Load-Strain Relationships at 20.6 C (69 F).......... 210








Figure Page

7.35 Measured vs. Predicted Deflections at 4,000 Ibs.:
E2 = 40,000 psi, 20.6 C (69 F)............ ................... 211
7.36 Measured vs. Predicted Strains at 4,000 lbs.:
E2 = 40,000 psi, 20.6 C (69 F)............................... 212

8.1 Layout of Strain Gages and Cables in the Test Pit........... 222
8.2 Location of Test Positions in the Test Pit.................. 224
8.3 Thermocouple and Strain Gage Location During Cooling:
Test Position 1.... ....................... .......... 225
8.4 Thermocouple and Strain Gage Location During Cooling:
Test Position 2............................................ 226
8.5 Thermocouple and Strain Gage Location During Cooling:
Test Position 3........................................... ... 227
8.6 Measured Cooling Curves: Test Position 1.................... 228
8.7 Measured Cooling Curves: Test Position 2................... 229
8.8 Measured Cooling Curves: Test Position 3.................... 230
8.9 Change in Temperature Gradient During Cooling............... 232
8.10 Measured Longitudinal Strains vs. Temperature:
Test Position 1.............................. ...... ..... 233
8.11 Measured Longitudinal Strains vs. Temperature:
Test Position 2................ ............... .......... 234
8.12 Measured Longitudinal Strains vs. Temperature:
Test Position 3.................... ................... 235
8.13 Measured Transverse Strains vs. Temperature:
Test Position 1................................... ........... 236
8.14 Measured Transverse Strains vs. Temperature:
Test Position 2........................................... 237
8.15 Measured Transverse Strains vs. Temperature:
Test Position 3........ ............... ...... ......... 238
8.16 Longitudinal Strain Distributions During Cooling:
Test Position 1.............................................. 242
8.17 Longitudinal Strain Distributions During Cooling:
Test Position 2........................................... 243
8.18 Longitudinal Strain Distributions During Cooling:
Test Position 3........... ............................... 244
8.19 Transverse Strain Distributions During Cooling:
Test Position 1............................................. 245
8.20 Transverse Strain Distributions During Cooling:
Test Position 2............................. ....... ..... 246
8.21 Transverse Strain Distributions During Cooling:
Test Position 3............................................. 247
8.22 Comparison of Measured Thermal Strains for Different
Cooling Cycles: Six Feet from South Wall.................... 250
8.23 Comparison of Measured Thermal Strains for Different
Cooling Cycles: 8.33 Feet from South Wall ................. 251
8.24 Longitudinal Strain Distributions During Cooling............ 253
8.25 LVDT and Strain Gage Location During Load Tests:
Test Position 1..................................... ......... 256
8.26 LVDT and Strain Gage Location During Load Tests:
Test Position 2...................................... 257
8.27 LVDT and Strain Gage Location During Load Tests:
Test Position 3.............................................. 258








Figure Page

8.28 Load-Unload Times for Dynamic Loading with Dual Wheels...... 261
8.29 Measured Longitudinal Deflections at 0.0 C (32 F):
Test Position 3............................................. 263
8.30 Measured Longitudinal Strains at 0.0 C (32 F):
Test Position 3............... ........................... 264
8.31 Measured Transverse Deflections at 0.0 C (32 F):
Test Position 3.............................................. 265
8.32 Measured Transverse Strains at 0.0 C (32 F):
Test Position 3............ ........ ... ............... 266
8.33 Measured Longitudinal Deflections at 5.7 C (44 F):
Test Position 3............................................. 267
8.34 Measured Longitudinal Strains at 6.7 C (44 F):
Test Position 3............................................. 268
8.35 Measured Transverse Deflections at 6.7 C (44 F):
Test Position 3............................................. 269
8.36 Measured Transverse Strains at 6.7 C (44 F):
Test Position 3...................................... ........ 270
8.37 Measured Longitudinal Deflections at 13.3 C (56 F):
Test Position 3.................................... ..... 271
8.38 Measured Longitudinal Strains at 13.3 C (56 F):
Test Position 3............................................. 272
8.39 Measured Transverse Deflections at 13.3 C (56 F):
Test Position 3.............................................. 273
8.40 Measured Transverse Strains at 13.3 C (56 F):
Test Position 3.............................................. 274
8.41 Comparison of Measured Longitudinal Deflections at
Different Temperatures: Test Position 1, 10,000 Ibs......... 278
8.42 Comparison of Measured Longitudinal Strains at
Different Temperatures: Test Position 1, 10,000 Ibs......... 279
8.43 Comparison of Measured Transverse Deflections at
Different Temperatures: Test Position 1, 10,000 Ibs......... 280
8.44 Comparison of Measured Transverse Strains at
Different Temperatures: Test Position 1, 10,000 lbs......... 281
8.45 Comparison of Measured Longitudinal Deflections at
Different Temperatures: Test Position 2, 10,000 Ibs......... 282
8.46 Comparison of Measured Longitudinal Strains at
Different Temperatures: Test Position 2, 10,000 Ibs....... 283
8.47 Comparison of Measured Transverse Deflections at
Different Temperatures: Test Position 2, 10,000 Ibs......... 284
8.48 Comparison of Measured Transverse Strains at


Different Temperatures: Test Position 2, 10,000 Ibs.........
8.49 Comparison of Measured Longitudinal Deflections at
Different Temperatures: Test Position 3, 10,000 Ibs.........
8.50 Comparison of Measured Longitudinal Strains at
Different Temperatures: Test Position 3, 10,000 Ibs.........
8.51 Comparison of Measured Transverse Deflections at
Different Temperatures: Test Position 3, 10,000 Ibs.........
8.52 Comparison of Measured Transverse Strains at
Different Temperatures: Test Position 3, 10,000 Ibs.........
8.53 Comparison of Predicted Longitudinal Deflections at
Different Temperatures: 10,000 lbs.........................


286

287

288

289

290


xiii








Figure Page

8.54 Comparison of Predicted Longitudinal Strains at
Different Temperatures: 10,000 Ibs......................... 291
8.55 Permanent Longitudinal Deflections at 0.0 C (32 F):
Test Position 3.............................................. 294
8.56 Longitudinal Creep Strains at 0.0 C (32 F):
Test Position 3.................. ...................... 295
8.57 Permanent Transverse Deflections at 0.0 C (32 F):
Test Position 3........................................ 296
8.58 Transverse Creep Strains at 0.0 C (32 F):
Test Position 3............................................. 297
8.59 Comparison of Permanent Longitudinal Deflections at
Different Temperatures: Test Position 1...................... 300
8.60 Comparison of Longitudinal Creep Strains at Different
Temperatures: Test Position 1.............................. 301
8.61 Comparison of Permanent Longitudinal Deflections at
Different Temperatures: Test Position 2.................... 302
8.62 Comparison of Longitudinal Creep Strains at Different
Temperatures: Test Position 2............................... 303
8.63 Comparison of Permanent Longitudinal Deflections at
Different Temperatures: Test Position 3.................... 304
8.64 Comparison of Longitudinal Creep Strains at Different
Temperatures: Test Position 3............................. 305
8.65 Comparison of Dynamic Load Response Immediately Prior
to Creep Tests for Different Test Positions:
0.0 C (32 F)... ................................. ...... .. 308
8.66 Comparison of Permanent Longitudinal Deflections for
Different Test Positions: 0.0 C (32 F).................... 309
8.67 Comparison of Dynamic Load Response at Different Times:
0.0 C (32 F), Test Position 1............................... 310
8.68 Comparison of Dynamic Load Response at Different Times:
0.0 C (32 F), Test Position 2.............................. 311
8.69 Comparison of Dynamic Load Response at Different Times:
0.0 C (32 F), Test Position 3................................ 312
8.70 Comparison of Dynamic Load Response After Creep Tests
for Different Test Positions: 13.3 C (56 F)................ 317
8.71 Comparison of Dynamic Load Response After Creep Tests
for Different Test Positions: 6.7 C (44 F).................. 318
8.72 Comparison of Dynamic Load Response After Creep Tests
for Different Test Positions: 0.0 C (32 F).................. 319
8.73 Comparison of Load-Deflection Relationships for
Different Test Positions: 0.0 C (32 F)...................... 325
8.74 Comparison of Load-Deflection Relationships for
Different Test Positions: 6.7 C (44 F)...................... 326
8.75 Comparison of Load-Deflection Relationships for
Different Test Positions: 13.3 C (56 F)................... 327

9.1 Comparison of Measured and Predicted Longitudinal
Deflection Basins at 0.0 C (32 F)............................ 335
9.2 Comparison of Measured and Predicted Longitudinal
Strain Distributions at 0.0 C (32 F)........................ 336
9.3 Comparison of Measured and Predicted Transverse
Deflection Basins at 0.0 C (32 F)............................ 337


xiv








Figure Page

9.4 Comparison of Measured and Predicted Transverse
Strain Distributions at 0.0 C (32 F)....................... 338
9.5 Comparison of Measured and Predicted Longitudinal
Deflection Basins at 6.7 C (44 F).......................... 339
9.6 Comparison of Measured and Predicted Longitudinal
Strain Distributions at 6.7 C (44 F)......................... 340
9.7 Comparison of Measured and Predicted Transverse
Deflection Basins at 6.7 C (44 F)............................ 341
9.8 Comparison of Measured and Predicted Transverse
Strain Distributions at 6.7 C (44 F)......................... 342
9.9 Comparison of Measured and Predicted Longitudinal
Deflection Basins at 13.3 C (56 F)........................... 343
9.10 Comparison of Measured and Predicted Longitudinal
Strain Distributions at 13.3 C (56 F)....................... 344
9.11 Comparison of Measured and Predicted Transverse
Deflection Basins at 13.3 C (56 F)........................... 345
9.12 Comparison of Measured and Predicted Transverse
Strain Distributions at 13.3 C (56 F)....................... 346













Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy


LOW-TEMPERATURE RESPONSE
OF ASPHALT CONCRETE PAVEMENTS

By

REYNALDO ROQUE

May, 1986


Chairman: Byron E. Ruth
Major Department: Civil Engineering

The variable performance of asphalt concrete pavements indicates

that existing design procedures may be inadequate. However, before

improved design procedures can be developed, the mechanisms that lead to

pavement cracking must be fully understood. Therefore, this research

program was developed to monitor and define the response and failure

characteristics of pavements subjected to thermal and dynamic loading

conditions.

An asphalt concrete pavement was tested under controlled

temperature conditions in an enclosed concrete test pit. Thermal

contraction strains were measured as the pavement was cooled from room

temperature to temperatures below freezing. Load-induced deflections

and strains in the pavement were measured for dynamic load tests

performed at temperatures ranging from -6.7 C (20 F) to 25.6 C (78 F).

Finally, permanent deflections and creep strains were measured under

static loads at temperatures ranging from 0.0 C (32 F) to 13.3 C (56 F).








Load-induced deflections and strains measured at temperatures

ranging from 0.0 C (32 F) to 21.1 C (70 F) were accurately predicted

using linear elastic layer theory when suitable layer moduli were used

for input. Asphalt concrete moduli determined from correlations with

measured asphalt viscosity resulted in accurate prediction of measured

deflections and strains at all temperatures and load levels tested.

Suitable moduli for the subgrade and base layers were determined from

plate load tests performed on these materials in situ. The use of

proper analytical tools to evaluate the plate load test data was

critical in the determination of these moduli.

Thermal and load response measurement of the four-inch asphalt

concrete pavement indicated that temperature differentials produced by

rapid cooling caused the asphalt concrete layer to contract and bend in

such a way that it separated and uplifted from the base. The uplift

effect resulted in load-induced deflections, strains, and stresses at

0.0 C (32 F) that were in some cases more than double those expected for

pavements exhibiting elastic behavior. The exact mechanism that led to

the uplift phenomenon could not be determined from the measurements

obtained. Because of the uplift effect, measured creep strains and

failure could not be evaluated. Based on the findings from this

investigation, recommendations are presented for improved testing

procedures.


xvii














CHAPTER I
I INTRODUCTION


The Florida Department of Transportation (FDOOT) has had variable

performance with asphalt concrete pavements. Some pavements have

developed cracking within less than five years, while others have given

satisfactory performance after many years. In all cases, cracking was

observed with little or no distortion in the pavement layers and with no

apparent deficiencies in the asphalt concrete mixtures. The problem is

not unique to Florida. Pavement condition surveys of existing highways

and test roads around the United States indicate that traffic-associated

cracking is of major concern to highway engineers. Cracking is one of

the first indicators of distress observable in asphalt pavements and

often leads to other forms of distress.

The variable performance observed for pavements designed using

current design procedures indicates that these procedures are

deficient. Existing design procedures are empirically derived based on

correlations of certain material or pavement system parameters with

observed field performance. These procedures consider two forms of

cracking: low-temperature thermally-induced cracking and traffic-load-

induced fatigue cracking. The most commonly proposed approach to limit

thermal cracking is to limit the asphalt stiffness as measured for a

minimum design temperature, where the limiting stiffness is usually

obtained from correlations with observed field performance. Design

thickness requirements to provide adequate fatigue life are established








by attempting to limit pavement deflections or strains under a given

design load. These procedures neglect that pavement deflections,

stresses, and strains cover a wide spectrum of values dependent on

temperature and climatic fluctuations. The variable properties of

individual asphalts at low temperatures and the combined effect of

thermally- and load-induced stresses are not considered. In addition,

the two modes of cracking considered cannot explain certain types of

failures commonly encountered in practice.

Therefore, it is necessary to develop improved design procedures to

reduce the cracking potential of pavements. However, this will be

difficult to accomplish until the mechanisms that lead to cracking are

fully understood. Although the causes or factors involved with cracking

are known, the actual mechanisms that lead to cracking have not been

identified with definitive measurements on full-scale pavements.

Investigations sponsored by the FDOT led to the establishment of the

hypothesis that cracking of asphalt concrete pavements is a brittle

failure induced by short-term repetitive loads and thermal stresses that

occur during cool weather when the asphalt stiffness is high. This

suggests that asphalt concrete pavements should be designed for a

critical condition where stresses or creep strains induced in the

pavement are of sufficient magnitude to produce cracking. This critical

condition may be a result of the combined effects of asphalt age

hardening, base and subgrade support, asphalt concrete modulus,

vehicular loads, pavement cooling rate, and temperature.

The research work done in Florida, along with the variable

performance of existing pavements, indicated that a research program

should be developed to monitor and define the behavior of pavements








subjected to thermal and dynamic loading conditions. This resulted in

the formulation of a research program to test full-scale asphalt

concrete pavement systems under controlled conditions.

An FDOT test pit facility was developed for this purpose, since

there are obvious problems associated with trying to monitor this type

of behavior in the field. The test pit facility made it possible to

construct a layered system of materials to simulate a flexible pavement

system in the field. In order to provide temperature control, a cooling

system with an insulated cover was installed in the test pit.

Thus having developed the capability to simulate temperature and

loading conditions encountered in actual roadways, this research program

was initiated in an attempt to satisfy the following objectives:

1. To measure and evaluate the response of asphalt concrete

pavements to changes in temperature, and determine the effect

of this response on the dynamic load response and failure

characteristics of the pavement.

2. To measure and evaluate the dynamic load response of asphalt

concrete pavements at different temperatures and load levels.

3. To measure and evaluate load-induced creep strains and

permanent deflections induced in asphalt concrete pavements at

different temperatures.

4. To compare the response measurements listed in items one, two,

and three to values predicted by theoretical stress-strain

distributions and parameters obtained from laboratory tests.

In order to meet these objectives, a complete series of tests was

performed on a pavement section that was typical for Florida. A

measurement and data acquisition system was installed that was capable








of obtaining static and dynamic deflection and strain measurements at

ten different points in the pavement at any given time. Thermal strains

developed in the asphalt concrete layer during cooling were measured for

several cooling cycles. Dynamic load tests were performed at

temperatures ranging from -6.7 C (20 F) to 25.6 C (78 F), using both

rigid plate and flexible dual wheel loading systems. Finally, permanent

deflections and creep strains were measured for specified durations of

static loads at temperatures ranging from 0.0 C (32 F) to 13.3 C

(56 F). Dynamic load tests were also performed at different times

during creep tests to observe the effect of creep on the dynamic load

response of the pavement. Although the results of these tests were not

entirely definitive, they emphasized the need to consider the combined

effect of temperature and load in the analysis of asphalt concrete

pavements.














CHAPTER II
LITERATURE REVIEW


2.1 Introduction

The research presented in this document focuses on defining and

predicting low temperature response and failure of asphalt concrete

pavements. This includes the response of pavements to changes in

temperature and the effect these changes have on the load response and

failure limits of the asphalt concrete layer. Two elements of the

analysis system considered here make it unique: the use of measured

theological parameters of the asphalt at low temperatures to predict the

response and failure characteristics of the asphalt concrete; and the

fact that cracking is considered a short-term phenomenon that occurs

when the combined effect of temperature and traffic loads exceed the

failure limit of the asphalt concrete pavement. Although this approach

is totally different from traditional approaches, a review of the

literature will serve two purposes:

1) to establish the need and develop the rationale behind the

proposed method of analysis; and

2) to give an overview of existing knowledge of asphalt concrete

pavement response to temperature changes and traffic loads,

including an assessment of our ability to predict response and

failure.








2.2 Distress In Asphalt Concrete Pavements

2.2.1 Modes of Distress

The modes of distress in asphalt concrete pavements are well

recognized and the causes of distress, at least in general terms, are

also known. Tables 1 and 2 (1,2) are two examples of tables listing the

types and causes of distress in asphalt concrete pavements. These

tables show that failures can be grouped into three major categories:

cracking, rutting, and disintegration.

Pavement surveys around the country and the world indicate that of

these three categories, cracking is the major problem in terms of amount

and cost. Based on extensive observations by himself and others, Finn

(3) stated that traffic-associated cracking is the number one priority

item for improving and extending the performance of asphalt pavements.

Pedigo et al. (4) reviewed a great deal of work that has been done on

pavement distress and reached similar conclusions. Finn also stated

that traffic associated cracking is one of the first indicators of

distress observable in asphalt pavements, and that cracking is often

observed with little or no distortion. In reviewing the results of the

AASHO Road Test, he found that cracking led to other forms of distress

(such as rutting), and that more cracking occurred when the pavement was

cold than warm. However, there was a lack of information as to when and

where the first cracks occurred and how these cracks propagated.

Furthermore, asphalt properties were not measured at low temperatures.

Information of this type is lacking, even today. Measurements of

the environmental and loading conditions at the time of initial

cracking, along with relevant material properties, are crucial to the

development of damage criteria. Such information could not be found in








Table 2.1:


Primary Types and Causes of Distress in Asphalt Concrete
Pavements. After Ruth, 1985 (1)


Type Of Distress


Causes or Contributing Factors


- Consolidation
Shear failure
Low stability
Abrasion
Traffic
High temperatures


2. Thermal Cracking





3. Load Associated or
Fatigue cracking


4. Combined Thermal and
Induced Cracking

5. Heaving (Localized
or Extensive: Frost
Boils, Ice Lenses)




6. Settlement and Slope
Failures


- Thermal contraction
Shrinkage
Low temperatures
Fast rate of cooling
Excessively hard asphalts
Lack of snow cover (insulation)

- Traffic volume and loads
Deflection basin characteristics:

1. Layer moduli
2. Layer thickness
3. Asphalt viscosity

Climate Microclimate

1. Temperature
2. Drainage moisture variations

Material quality

- Combine factors in items 2 and 3


- Expansive soils
Frost susceptible soils
Drainage
Permeability
Capillarity
Depth and rate of frost
penetration

- Quality of in situ materials
Quality of construction
Drainage and moisture conditions
Mining activity
Karst terrain sinkholes and
cavity collapse


1. Rutting














Table 2.2:


Modes, Manifestations, and Mechanisms of Types of Distress.
After McCullough, 1971 (2)


Mode Manifestation Mechanism

Fracture Cracking Excessive loading
Repeated loading (i.e., fatigue)
Thermal changes
Moisture changes
Slippage (horizontal forces)
Shrinkage
Spalling Excessive loading
Repeated loading (i.e., fatigue)
Thermal changes
Moisture changes


E


distortionn Permanent deformation
Time-dependent deformation
(e.g., creep)
Densification (i.e., compa
Consolidation
Swelling
Faulting Excessive loading
Densification (i.e., compa
Consolidation
Swelling


Disintegration Stripping
Chemical reactivity
Abrasion by traffic
Raveling and scaling
Chemical reactivity
Abrasion by traffic
Degradation of aggregate
Durability of binder


Excessive loading


action)



:tion)


Adhesion (i.e., loss of bond)


Adhesion (i.e., loss of bond)








the literature. Thus, although the causes of cracking are well known,

the actual mechanisms that lead to cracking have not been verified with

definitive measurements of actual failures on full-scale pavements.

Several mechanisms have been proposed that cannot account for basic

material response and failure characteristics, variability of

environment, and loading conditions encountered in actual pavements.

These have led to empirical design procedures, which are valid only for

the conditions from which they were derived.



2.2.2 Cracking Mechanisms

Traditionally, cracking has been broken down into traffic-load

induced and thermally-induced, with little consideration for the

combined effects of the two mechanisms. Low-temperature transverse

cracking has been recognized as the most common non-traffic associated

failure mode and is a serious problem in Canada and parts of the United

States (5,6,7). This type of failure is generally considered a

temperature phenomenon caused by low temperatures. As the pavement

temperature decreases the asphalt concrete wants to contract, but

contraction is resisted by the friction between the asphalt concrete

layer and the base and by the length of the roadway in the longitudinal

direction. This resistance results in tensile stresses in the pavement,

which are greatest in the longitudinal direction.

Several researchers have postulated that cracking occurs when these

thermally induced tensile stresses exceed the tensile strength of the

asphalt concrete (8, 9, 10). This mechanism has been confirmed by

laboratory and field investigations (7-13), and provides the basis for

the hypotheses that have been presented for low temperature cracking.








The theological properties of the asphalt at low temperatures are

generally recognized as the most important factor in low-temperature

transverse cracking (5, 11, 14, 15, 16, and others). Many researchers

have associated low temperature cracking with properties such as asphalt

stiffness, viscosity, temperature susceptibility, and glass transition

temperature. These properties, of course, are all related to the

asphalt's ability to flow and thus relax stresses. All researchers have

found that the stiffer and more temperature susceptible the asphalt, the

greater the potential for cracking.

Probably the most commonly proposed approach to control thermal

cracking is to limit the asphalt stiffness as measured for a minimum

design temperature. McLeod (17) concluded that low temperature pavement

cracking is likely to occur whenever the stiffness of the pavement

attains a value of 6.9 E9 Pa (1.0 E6 psi) at a pavement depth of two

inches, at the minimum temperature encountered, and for a loading time

of 20,000 seconds. Fromm and Phang (18) proposed a value of 1.4 E8 Pa

(20,000 psi) at 10,000 seconds loading time. Gaw (19) reported that the

St. Anne test pavements cracked at an asphalt binder stiffness of 1.0 E9

Pa (145,000 psi) and a mixture stiffness of 2.0 E10 pa (2,900,000 psi)

at 1800 seconds loading time. Many researchers have found good

agreement between measured stiffness and observed cracking of pavements

in the field and confirmed that pavements using softer asphalts exhibit

less cracking (7, 12, 20, 21, 22).

Ruth (14) concluded that cracking would be reduced by using

asphalts with lower viscosities and improved theological behavior at low

temperature. Fabb (23) concluded that low viscosity and low temperature







susceptibility are conducive to reducing the temperature at which

fracture occurs.

The advantage of using a softer binder, particularly one with a low

temperature susceptibility, was demonstrated by Hills and Brien (8).

Fromm and Phang (16) also reported that less temperature susceptible

asphalts were associated with pavements exhibiting less cracking.

Schmidt (24) suggested that the glass transition temperature of the

asphalt might be a more definitive measure of non-load associated

cracking than measured viscosities, since at temperatures lower than the

glass transition temperature the asphalt behaves elastically, while at

higher temperatures it exhibits viscoelastic response. Thus, below the

glass transition temperature there is almost no potential for stress

relaxation.

Other factors have been found to influence low temperature

transverse cracking, but to a lesser degree than asphalt properties.

Tuckett et al. (25) found that higher asphalt contents reduced thermal

cracking. Fabb (23) reported that increasing binder content reduced

thermal fracture, but only slightly. He also concluded that the

properties and grading of the aggregate had little or no effect on the

resistance of the asphalt concrete to thermal cracking. Cooling rate

was found to have little effect on the failure temperature by Fabb (23)

and Fromm and Phang (16). However, they only compared relatively high

cooling rates. Finally, results of the St. Anne Test Road indicated

that only half the frequency of low temperature cracking occurred in

10-inch pavements than did in 4-inch pavements (26).

The concept of fatigue is probably the most recognized concept that

has been suggested for use in the evaluation of traffic-load associated








failure (27-31). Fatigue distress is the phenomenon of fracture under

repeated stresses which are less than the tensile strength of the

material. Fatigue characterization of materials has been studied

extensively and there are innumerable references on this topic (e.g. 31-

39).

The philosophy behind the approach to the analysis and design of

asphalt concrete pavements considered in this thesis is totally

different from conventional approaches based on fatigue. In fact, the

fatigue concept is considered erroneous, and will not be covered in much

detail. Design procedures based on fatigue assume that there is some

average pavement condition for which an equivalent amount of damage will

be incurred under each passing wheel load. These procedures neglect

that deflections, strains, and stresses cover a wide spectrum of values

dependent on temperature and climatic fluctuations. They cannot

properly account for the variable properties of individual asphalts at

low temperatures.

Several researchers have proposed modifications to fatigue life

predictions based on temperature, recognizing that the fatigue life of

materials tested in the laboratory is dependent on temperature.

However, the basic concept of fatigue damage has remained unchanged.

Rauhut and Kennedy (40) proposed one such modification and discuss

modifications proposed by other researchers. They also recognize that

fatigue damage is difficult to evaluate since there is limited knowledge

as to fatigue life relations for real pavements, reliable test data

exists for only a limited number of mixtures, and there is insufficient

information to define how fatigue life varies with temperature and








mixture characteristics. Furthermore, they point out that no laboratory

fatigue test comes close to simulating actual field conditions.

One very significant point is that fatigue life is highly dependent

on the type of fatigue test performed. While illustrating the effect of

temperature on fatigue life, Pell and Cooper (34) showed that as the

temperature is lowered, fatigue life increases under stress controlled

tests, but decreases under strain controlled tests.

Recently, investigators have found that rest periods markedly

increase the fatigue life of bituminous mixtures (41). This seems to

indicate that asphalt concrete has the potential to heal, or that the

actual failure mode is not a true fatigue phenomenon. Both of these

ideas negate the validity of conventional fatigue approaches.

Ruth and Maxfield (42) found the concept of fatigue did not apply

to specimens from test roads in Florida. They found that the failure

strains for in-service cores were the same as for fabricated cores.

They concluded that fracture of asphalt concrete is related to a process

of cumulative creep strain and that fracture strain is primarily depen-

dent on asphalt properties and loading conditions. Ruth et al. (43)

pointed out that during warm weather, temperatures are high enough to

eliminate stress or strain accumulation that leads to fatigue failure.

Pavements designed using conventional fatigue approaches have given

marginal performance. Ruth et al. (43) has stated that for similarly

designed asphalt concrete pavements in Florida some have developed

cracking within less than five years, while others give satisfactory

performance after many years. Roberts et al. (44) stated that very few

highways have served without maintenance even for five or ten years, and

in many cases roads with low traffic volumes have experienced premature








failure. Rauhut and Kennedy (40) stated that the occurence of fatigue

cracking in the field is quite variable, even for apparently identical

sections.

In addition, traditional approaches are unable to explain certain

types of failures observed in the field. It is well recognized that

immediate and disastrous failures may occur with weakened subsoil

conditions after just a few passes of a heavy vehicle (45). Molenaar

(46) has pointed out that traditional approaches cannot explain

longitudinal cracks observed to occur at the pavement surface. This

type of cracking is very common in practice.

Several studies indicate that the combined effects of thermally and

load induced stresses may cause cracking. Haas and Topper (13)

indicated that even if the thermal stresses are not sufficient to cause

cracking the addition of load associated stresses may result in pavement

failure. From and Phang (16) reported a case where heavily loaded

trucks were carried during the winter months in one direction only.

They found that there was a greater incidence of transverse cracking on

the heavily loaded side, thereby illustrating the combined effect of

thermal and load stresses. Ruth et al. (43) hypothesized that this type

of mechanism may be the cause of some early pavement failures in Florida

and elsewhere. However, almost all studies presented in the literature

consider only the load effect or the thermal effect.

Ruth et al. (43) were the first to present an approach that

combines the effect of thermal and dynamic stresses as the main cause of

failure. They considered pavement cracking to be caused by brittle

failure induced by short term repetitive loads and thermal stresses that

occur during cool weather when the asphalt stiffness gets very high.







Their idea is to design the pavement for a critical condition based on

material properties, loads, and environment. They developed a pavement

analysis model that considers cracking as a result of asphalt properties

(including age hardening), vehicular loads, pavement cooling rate and

temperature. The analysis program was used to evaluate the effect of

different asphalt viscosities, cooling rates, and pavement thicknesses

on pavement performance. Predictions of cracking temperatures for a

Pennsylvania DOT test road were obtained which identified the two

cracked sections in the test road. Analysis of typical highways in

Florida indicated that some pavements may give marginal performance,

which was indirectly substantiated by observed early cracking of pave-

ments, particularly those located in northern Florida.



2.3 Properties of Asphalt Cement and Asphalt Concrete As
Related To Low-Temperature Pavement Response and Cracking

The response and failure of asphalt concrete pavements have been

shown to be highly dependent on the properties of the asphalt cement.

Thus, proper characterization of asphaltic materials is extremely

important. The characterization of bituminous materials for use in

conventional design methods is based mostly on empirical procedures

which rely on correlations of their results with field performance. The

Marshall and Hveem Stabilometer tests are most commonly used for this

purpose (27). These tests are performed at high temperatures and relate

mainly to the problems of stability, workability, and durability. Fun-

damental properties cannot be obtained directly from these tests.

Several researchers have attempted to correlate Marshall results with

fundamental properties (47), but it will be pointed out later that such








an approach can lead to serious error, particularly when predicting

properties at lower temperatures.

As explained earlier, the analysis method used in this dissertation

considers cracking to be caused by brittle failure induced by short term

repetitive loads and thermal stresses that occur when the asphalt

stiffness is high. Therefore, the emphasis here is placed on the behavior

at relatively low temperatures, roughly in the range from -10 C (14 F), or

approximately the glass transition temperature, to 25 C (77 F). This

temperature range is referred to as the near transition region (48).


2.3.1 Asphalt Cement Properties

The response of asphalt to an applied stress is time dependent,

where the strain increases at a given rate with time. At lower temper-

atures many asphalts are also shear susceptible, with the change in

creep strain rate not being proportional to the change in applied

stress. Finally, the behavior of all asphalts is highly dependent on

temperature.

In general, as the temperature is lowered, asphalts become more /

viscous and eventually exhibit glassiness, where different elasto- /

viscous behavior is observed, their coefficients of expansion change,\

and brittle fracture may develop (48). Jongepier and Kuilman (49)

explained the behavior of asphalt as a viscoelastic liquid. At low

temperatures, asphalt behaves like an elastic solid, while at high

temperatures its behavior is comparable to a viscous liquid. At inter-

mediate temperatures the behavior is influenced by both viscous and

elastic components. Asphalt cements show a characteristic common to

other amorphous materials; the glass transition phenomenon. Schweyer








and Burns (50) found that the glass transition temperature for a wide

variety of asphalts is between -10 C and 5 C (14 and 41 F).

The viscoelastic response of asphalts and asphalt mixtures is often

approached using mechanical models that combine Hookean springs and

Newtonian dashpots in various combinations (51). As mentioned above,

asphalt behavior at lower temperatures is often non-Newtonian (shear

susceptible), making these models unsuitable for complete description.

In any case, asphalt behavior is commonly described in terms of

theological parameters, and at low temperatures, asphalts can be

described in terms of three theological parameters: consistency or

viscosity, shear susceptibility, and temperature susceptibility (52).

However, the measurement of viscosity at low temperatures is rather

problematic because the asphalt behaves closer to an elastic material

with relatively low creep deformation rates. This means that creep or

viscosity tests will require an extremely long time of loading to obtain

measurable deformations at low stress levels. High stress levels are

usually necessary to obtain measurements within reasonable time

intervals, but if the material is shear susceptible the results may not

be representative of the material's behavior within the range of

interest.

Schweyer presented a pictorial review of an extensive number of

devices that have been used over the years to measure theological

properties (53). A more concise literature survey of the different

methods to measure low temperature rheology of asphalts is presented in

reference 48. In general, the traditional transient rheometers are not

directly adaptable to low temperature work (54). However, several

special testing devices have been used to conduct investigations of








asphalt properties at low temperatures. These have led to improved

understanding of low temperature asphalt behavior.

For example several investigators have concluded that low

temperature asphalt properties cannot be predicted from properties

measured at higher temperatures. Schweyer et al. (55) reported that

different asphalts demonstrate very different low temperature

theological properties. They emphasized that temperature susceptibility

in the near transition region can and should be evaluated by absolute

viscosity measurements rather than by empirical tests. They also stated

that temperature susceptibility cannot be predicted from behavior

exhibited at higher temperatures. In a comprehensive study of different

asphalts at the Asphalt Institute, Puzinauskas (56) reached the

following conclusions:

generally, viscosity at low temperature is affected more by

heating than viscosity at high temperature;

the low temperature viscosity of asphalt cements was found to

vary extremely and the variation increases with decreasing

temperature; and

shear effects become more pronounced with increasing viscosity or

with decreasing temperature.

Schmidt (57) investigated the reliability of standard ASTM tests to

predict low temperature stiffness of mixtures made with a wide variety

of asphalts. He concluded that low temperature thermally induced

cracking should not be implied from high temperature viscosity measure-

ments on diverse types of asphalts. Although many researchers have

found reasonable correlation between measured stiffness and observed

field cracking, relatively noor agreement has been obtained by







researchers estimating low temperature stiffness by means of tests at

higher temperatures. Pink et al. (54) and Keyser and Ruth (58) also

emphasized the importance of experimental measurements rather than the

use of empirical extrapolations to determine low temperature properties.

Probably the most significant advancement to the understanding of

low temperature response and failure properties of asphalts and asphalt

mixtures was the development in the 1970's of the Schweyer constant

stress rheometer (59). This device has the capability of measuring

theological properties at -10 C (14 F) and lower. Furthermore, Schweyer

established theological concepts that led to a definitive theological

model and methods to evaluate important parameters that relate to low

temperature behavior, including shear susceptibility.

The proposed theological model is the Burns-Schweyer model (55).

The model is a Burgers model with a modified dashpot to incorporate a

self-generating feedback system to regulate the rate of viscous flow.

Thus, the model accounts for viscous behavior for both Newtonian and

shear susceptible materials, as well as for elastic, and delayed elastic

behavior. Details pertaining to the measurement and evaluation of

theological parameters using the Schweyer rheometer may be found in

references 50, 55, 59, and 60. Ruth and Schweyer (61) showed that the

Burns-Schweyer model gives accurate prediction of the theological

properties of asphalts, including those that are very shear

susceptible. Keyser and Ruth (58) concluded that the Schweyer rheometer

is an excellent device for low temperature measurements of asphalt

properties and that the concepts developed by Schweyer provide values

more closely related to shear and strain rates encountered in the

laboratory and in actual pavements.








Therefore, the Schweyer rheometer has the unique advantage of

direct measurement of low temperature asphalt properties at any given

temperature. This instrument and the theological concepts presented by

Schweyer were key elements in the development of the analysis method

considered in this dissertation.


2.3.2 Asphalt Mixture Properties

Tests on asphalt mixtures are usually conducted to determine their

failure characteristics. However, the increased trend toward

mechanistic approaches and the application of elastic theory to pavement

evaluation and design, has initiated a concerted effort to define the

stress-strain response of bituminous mixtures (62). The variability of

materials in asphalt mixtures and the nature of pavement structures are

unlimited, making it nearly impossible to uniquely characterize their

stress-strain properties (63). Furthermore, a compromise between a

rigorous design solution and practicality is necessary. Material

characterization should be based on conditions that are believed to be

critical with respect to pavement response and failure.

As mentioned earlier, the approach to pavement failures in this

dissertation considers pavement cracking to be caused by brittle

fracture induced by short term repetitive loads and the thermal stresses

that occur during cool weather when the asphalt stiffness is high.

Using this approach it is necessary to predict the following: the

strains and deflections induced by applied dynamic stresses (wheel

loads); the short term creep strains induced by these dynamic stresses;

and the stresses and strains induced by temperature changes. References

cited later will show that by modelling the asphalt concrete layer as an








elastic continuum, researchers have obtained fairly accurate predictions

of measured strains and deflections in asphalt concrete pavements under

dynamic wheel loads, particularly at low ambient temperatures. Thus,

for a given set of temperature and loading conditions, it appears that

the stress-strain response of an asphalt mixture can be characterized

using an elastic modulus or E-value. The obvious point should be made

that asphalt mixtures are viscoelastic and describing their stress-

strain behavior by using an E-value is an idealization. The idealized

E-value will depend on how it is defined and the test method used to

measure it. Therefore, it is.necessary to identify laboratory

procedures that yield E-values that are suitable for proper pavement

response prediction. Suitable parameters for creep response, thermal

expansion and contraction, and failure limits are also necessary.

A variety of test methods has been used to test asphalt concrete

mixtures for characterization including compression (unconfined and

triaxial), bending (flexure), tension (direct and indirect), and shear

tests. Some of the different laboratory procedures and the different

idealized stiffness values they yield are described in reference 27.

There are certain advantages and disadvantages for each test, but these

are beyond the scope of this review. One problem with all the tests is

the effect of time or rate of loading, which makes suspect the elastic

equations used to analyze the test data. The effect of creep on stress

redistribution is almost always ignored because of the complexities

introduced into the analysis of test data. Many researchers (64, 65, 66

and others) have recommended using the indirect tensile test as the most

suitable for routine characterization in terms of practicality,

simulation of actual conditions, economy, and ease of testing.








Many researchers have concentrated on defining the relative effects

of different variables on the response characteristics of asphalt

mixtures. These studies are usually limited to those variables that are

considered to have a significant effect on the material behavior within

the researchers' scope of interest.

Deacon (38) summarized the major variables affecting the stiffness

and range of linear response for asphalt concrete mixtures and divided

them into three major categories: 1) loading; 2) mixture; and 3) envi-

ronmental related variables. He stated that four mixture related

variables have a considerable effect on the stiffness of asphalt paving

mixtures: air void content, asphalt content, asphalt viscosity, and

filler content. Temperature, mainly through its effect on asphalt

viscosity, was recognized as the major .factor in the behavior of

bituminous mixtures. Deacon also stated that the range of linear

response increases with increasing load frequency, decreasing

temperature and void content, and increasing asphalt content, asphalt

viscosity or filler content.

Bazin and Saunier (37) recognized the difficulty in changing any

one parameter without changing another, so they studied variation in

modulus for very different mixtures. They found that for correct binder

dosages and normal voids (4 to 8 percent), all other parameters had

little influence on modulus when compared to the effect of variation in

binder type, temperature, and time of loading. They suggest a linear

relation between log of modulus and void content of mix and proposed the

time-temperature superposition principle. This principle suggests that

there is an equivalency between time of loading and temperature.








Several researchers have investigated the stress dependence of

dynamic modulus. Cragg and Pell (67) reported stress dependence in

dynamic modulus but indicated that the change was small when compared

with temperature induced changes in modulus. Gonzalez et al. (64)

concluded that instantaneous modulus of elasticity decreased with

increasing temperatures and increasing number of applications but was

not affected by magnitude of applied stress. Kallas and Riley (68)

reported stress independent moduli for stresses between 17 and 70 psi

and temperatures from 4.4 C to 38 C (40 F to 100 F). Monismith et al.

(39) reported stress independent moduli over the range of 100 to 125 psi

using repeated flexure tests. Constant dynamic moduli were reported by

Pell and Taylor (33) for stresses below 125 psi at a temperature of 10 C

(50 F). They also indicated that low voids, low temperatures or high

loading rates, and adequate quantities of binder and filler were

conducive to linear response.

The effect of loading frequency on modulus has also been the

subject of several investigations. Yeager and Wood (69) described the

behavior of test specimens subjected to different frequencies. For

faster frequencies there is little time for flow and the mixture

behavior is more elastic. Slower load rates result in larger total

strains and lower calculated moduli. They reported a seven-fold

increase in modulus at 4.4 C (40 F) and a stress of 50 psi as the rate

of loading was increased from 1 to 12 cycles per second. Barksdale (70)

studied stress pulses applied by moving wheel loads and developed

relationships of stress pulse times as a function of vehicle velocity

and depth beneath the pavement surface. He developed charts that have







been subsequently recommended for determining loading times for dynamic

laboratory tests (71).

Investigations have also been conducted to evaluate different test

methods and differences in tensile and compressive properties. Kallas

(72) compared dynamic moduli of dense graded mixtures with normal air

voids at 4.4, 21.1, and 37.8 C (40, 70, and 100 F) and frequencies of 1,

4, and 16 cycles per second in tension, tension-compression, and

compression. He reported small differences in dynamic modulus in

tension, tension-compression, and compression for temperatures between

4.4 and 21.1 C (40 and 70 F) and frequencies between 1 and 16 cycles per

second. However, at 1 cycle per second and temperatures between 21.1

and 37.8 C (70 and 100 F), the dynamic tension or tension-compression

moduli averaged 1/2 to 2/3 of moduli in compression. The viscous

component of response was found to be considerably greater in tension

than in compression for frequencies of 1 to 16 cycles per second and

temperatures from 21.1 to 37.8 C (40 to 100 F). Wallace and Monismith

(73) compared moduli results from diametial indirect tension and

triaxial tests and analyzed the effect of anisotropy on both testing

procedures. They estimated that a more relevant assessment of modulus

could be obtained from diametral tests.

Although direct characterization is one approach to obtain mixture

parameters, the cost is prohibitive for most agencies (27), particularly

considering the variable properties of asphalts with temperature and

stress, and the fact that these properties can vary significantly for

different asphalts. Therefore, several investigators have established

relationships to predict mixture response parameters based on

temperature, time of loading, and material characteristics from







conventional tests. Miller et al. (62) reviewed some of the methods

that have been presented for predicting modulus from physical and

mechanical properties of the mixture, which usually take the form of

nomographs or master equations. Two of the better known methods are:

1) the different versions of the Shell Nomographs, and 2) the Asphalt

Institute Bituminous Mix Modulus Predictive Equation. The Shell

Nomographs were developed from Van der Poel's original stiffness charts

for asphalt cements for a given load rate and temperature. Relation-

ships were developed to translate bitumen stiffness to mixture stiffness

and these have been modified by several investigators. Modulus

prediction is based on asphalt stiffness as derived from softening point

and penetration, temperature, and frequency of loading. The asphalt

stiffness is then modified for mixture characteristics to obtain the

modulus of the mixture. The Asphalt Institute Equation is based on

tests performed on different mixtures at varying temperatures and fre-

quencies. Modulus prediction is based on load frequency, void content,

filler content, asphalt viscosity at 21.1 C (70 F), temperature, and

asphalt content.

Note that these methods cannot account for the variable properties

of individual asphalts at low temperature. Asphalt properties at

different temperature are inferred from consistency measurements at high

temperatures, and in the case of the Shell Nomographs, these measure-

ments are entirely empirical in nature. The temperature and shear

susceptibility of individual asphalts is ignored. As pointed out

earlier in this review, several researchers have shown that these

properties are a major factor in the response of the mixture, and they

can vary significantly from asphalt to asphalt.








Ruth et al. (43) were the first to present modulus relationships

based on a direct evaluation of measured asphalt viscosities at

different temperatures. The dynamic modulus relationship was developed

for dense-graded mixtures and a loading time of 0.1 seconds. It

requires the asphalt viscosity at a specific temperature as determined

from measurements with the Schweyer rheometer. Viscosity measurements

are made at several temperatures and stress levels (comparable to those

induced in laboratory tests on mixtures), in order to develop asphalt

viscosity-temperature relations for input. Therefore, this prediction

method properly accounts for the effects of temperature and shear

susceptibility on mixture response, and thus shows the most promise for

accurate modulus determination. Furthermore, the tests required are

simple and inexpensive.

The evaluation of these relationships is one goal of the current

investigation. It has been recognized that there is much variability in

characterization parameters reported in the literature (63). Part of

the problem is that resilient modulus tests have not been standardized

and certain aspects of the test that have an important effect on results

have not been defined (66). One reason for the lack of standardization

is that little if any work has been done to identify test procedures and

data interpretation methods that yield modulus values which give

accurate prediction of measured response on full-scale asphalt concrete

pavements under varying temperature and loading conditions. In other

words, although the relationships between asphalt viscosities and

laboratory measured dynamic moduli seem adequate, it has not been

verified that laboratory generated moduli can be used to predict

response on full-scale pavements, particularly at low temperatures.








Ruth and Maxfield (42) also developed relationships between asphalt

viscosity and what he defined to be a static modulus. The static

modulus was calculated from results of constant stress creep tests from

which he also calculated a pseudo mix viscosity for creep strain

prediction. Ruth found good correlation between asphalt viscosity and

pseudo mix viscosity, which also correlated well with static modulus.

Thus he developed relationships for both static modulus and creep

strains that are based on the low temperature viscosity of the

asphalt. The creep strain prediction model is a function of the asphalt

viscosity (function of temperature), the applied stress and the failure

stress. The static modulus is representative of the material's long

term response, as for the case of thermally induced stresses.

It should be noted that creep prediction models presented by other

researchers were almost exclusively developed to predict rutting. Their

results will not be discussed because they have not considered low

temperature creep response.

With the emphasis on fatigue, only a limited amount of work has

been done to define the failure limits of asphalt concrete mixtures in

terms of stresses or strains. Even for low temperature transverse

cracking, failure parameters are usually presented in terms of a

limiting asphalt stiffness or a fracture temperature, determined from

correlations with observed cracking in the field.

Several researchers have reported that tensile strength and failure

strain of asphalt mixtures are dependent on rate of loading and

temperature, and all have determined that the tensile strength increases

as the temperature decreases and the rate of loading increases (9, 12,

36, 74, 75, 76). Ruth (14) reported that as the temperature decreases








the failure stress increases but remains constant below some transition

temperature which is dependent on asphalt properties. Heukelom (74)

presented evidence that the tensile strengths of mixtures are related to

asphalt properties. Various researchers have reported failure stresses

for conditions of low temperatures and fast loading rates. Finn (77)

reported fracture strengths of 290 to 580 psi for asphalts in bulk under

low temperature conditions and rapid loading. For asphalt mixtures

under the same conditions he reported strengths generally ranging from

400 to 700 psi. Ruth and Olson (78) reported failure stresses between

380 and 440 psi and chose a value of 400 psi as typical.

Ruth and Olson (78) reported that as the temperature decreases the

tensile strain at failure decreases. Tons and Krokosky (76) reported

that increasing asphalt content within practical limits had little

effect on strain at ultimate strength. They also stated that rate of

loading had little effect at low temperatures. Epps and Monismith (36)

showed that the strain at failure is related to the stiffness of the

mixture, the strain at failure decreasing with increasing mixture

stiffness. Pavlovich and Goetz (79) computed limiting strains from

axial deformations in direct tension tests and determined that

temperature is the most significant factor affecting limiting strains.

Strain rate has some effect hut not as much as temperature. They found

the limiting strain at 60 C (140 F) was 300 to 500 times greater than

that at -27.5 C (-17.5 F). Salam and Monismith (80) presented an

equation to determine the strain at failure for asphalt mixtures based

on the strain at failure of the asphalt, the asphalt stiffness, and the

mixture stiffness. The asphalt properties used are penetration and ring








and ball softening point, which are empirical in nature and performed at

high temperatures.

Ruth and Maxfield (42) tested different mixtures and determined

that measured failure strains are primarily a function of asphalt

viscosity. Ruth and Potts (81) found that the energy required to

fracture a specimen decreased with increasing viscosity. These findings

led to the development of relationships between asphalt viscosities and

strain and energy at failure. The relationships are unique, since they

can directly account for the properties of individual asphalts at

different temperatures.

Investigations on the thermal expansion and contraction charac-

teristics of asphalt concrete mixtures have led to the following

conclusions (82, 83):

1. Different asphalts produce different amounts of expansion and

contraction.

2. The amount of shrinkage during cooling is more than the amount

of expansion during heating.

3. There are two different coefficients of expansion between -10 F

and 140 F, and the transition temperature between the two was

found to vary between 70 F and 86 F. Values in the low and

hiqh range have been called the solid and fluid thermal

coefficients, respectively.

4. The greater the asphalt content, the greater the thermal

expansion and contraction.

5. In the fluid state, the amount of expansion (contraction)

depends on the degree of restraint, while in the solid state

the expansion is the same for free and friction conditions.








6. Jones et al. (83) developed the following equation for

predicting the cubic coefficient of expansion in the solid

state:


mix


= ac Bac + Vagg Bagg
Vix
mix


where, Bmix


Bac


Bagg


Vmix =

Vac =

Vaaa
aa =


cubic thermal coefficient of expansion for
asphalt concrete

cubic thermal coefficient of expansion for
asphalt (glassy state)

cubic thermal coefficient of expansion for
aggregate

volume of. asphalt plus aggregate

volume of asphalt

volume of aggregate.


Assuming isotropic properties the linear coefficient of thermal

contraction (a) is:

mix
mix ~


2.4 Properties of Foundation Materials

The characterization of soils and granular base and subbase

materials, for both analysis and performance, is required data for all

structural pavement design methods, and is often treated with

considerable simplification. Geotechnical engineers often feel that

structural engineers have little interest in those parts of their work

below the ground level, and their feelings are certainly justified in

the case of pavements (84). Of the 185 papers presented to the past

five International Conferences on the Structural Design of Asphalt

Pavements only 15 have been concerned in any detail with the mechanical







properties of soils and granular materials, and much of this work has

concentrated on the prediction of rutting.

Soil behavior depends on many factors including water content, dry

density, stress level, stress states, stress path, structure, stress

history, and soil moisture tension (85). Although the relative effect

of these variables can be investigated in the laboratory, the cost of

even a limited number of tests may be prohibitive for most highway

projects.

Reference 27 describes tests that are typically used to

characterize soils and granular materials for pavement analysis,

including CBR, plate load tests, and triaxial testing. These tests are

usually performed for a limited set of conditions that in all proba-

bility do not encompass the variable conditions encountered during the

life of the pavement. Furthermore, the parameters obtained from such

tests are dependent on sample preparation, testing procedures, and data

interpretation methods, all of which have not been standardized.

Several researchers have presented relationships of resilient
modulus as a function of dry density and moisture content for specified

soils, based on laboratory testing. However, even these types of rela-

tionships are of limited value since they may ignore certain variables

that may have a significant effect on response. Also there is little

evidence that response parameters determined from conventional tests, or

any other test, provide for accurate prediction of response of full-

scale pavements.

A detailed review of work that has been done to define the relative

effects of different factors on soil response, or the relationships

developed to predict response parameters measured in the laboratory,








will not be presented here. This information is beyond the scope of the

work presented in the dissertation. However, some of the more recent

work done will be reviewed in an attempt to present a general view of

the state of knowledge in the area of characterizing soils for pavement

response prediction.

References cited later show that stresses and strains within the

asphalt concrete layer can be predicted fairly accurately by using an

effective modulus for the soil in an elastic layer analysis. They also

state that it is unlikely that stresses within the soil layers can be

predicted using a similar approach. Even for the simpler case of

predicting stresses in the asphalt concrete layer, the problem of deter-

mining a suitable effective modulus remains.

It has become evident in recent years that it is difficult or even

impossible to predict the behavior of pavements solely from laboratory

test data (86). Therefore, there has been much more emphasis on full-

scale pavement testing. A great deal of work is being done to determine

response parameters from nondestructive testing devices such as

Dynaflect and Falling Weight Deflectometer, as evidenced by the many

papers presented on this subject in the latest International Conference

on the Bearing Capacity of Roads and Airfields (87), and the 1985

sessions of the Transportation Research Board. This approach involves

the calculation of effective layer moduli based on measured surface

deflections. The approach has advantages and disadvantages. One

advantage is that many tests can he performed quickly and easily over

several miles of roadway under in situ conditions. Therefore, modulus

relationships can be developed for widely varying conditions. However,

surface deflections alone may not yield unique solutions and the








peculiarities of the different testing devices may result in loading

conditions that are not directly comparable to moving wheel truck

loads. In addition, the method is only suitable for existing pavements

and the results obtained will represent only the behavior during the

particular time tested.

Maree et al. (86) presented an approach to determine layer moduli

based on a device they developed to measure deflections at different

depths within the pavement structure. They suggested that effective

moduli for use in elastic layer theory can be determined from correction

factors or shift factors established from field measurements using their

device at different times of the year and under different conditions.

They determined effective moduli from tests performed at different load

levels and under different environmental conditions using their

multidepth deflectometer. They found that moisture condition has a

significant effect on response and its effect can be as important as

stress state. They also demonstrated that the modulus of individual

materials depends on the modulus of the underlying layers. The stress

dependence of several pavement materials was also demonstrated from the

field data. They found that most granular materials behave in a stress-

stiffening way, while most subgrade materials (e.g. weathered shale)

behave in a stress-softening way. Comparison of field and laboratory

data showed that although the trends apparent in the field were also

apparent in the laboratory, the constant-confining pressure triaxial

tests overestimated the modulus of the base materials tested.

Accurate prediction of stresses and strains within the foundation

layers is necessary for the prediction of rutting and stability failures

within these layers. Luhr and McCullough (88) state that moduli of








unbound materials should vary both horizontally and vertically in the

pavement structure according to equations equating modulus with state of

stress (i.e. MR = Aod for fine-grained materials and MR = kik2 for

granular materials). They stated that although this variation can be

satisfactorily represented by finite element models (F.E.M.), these

models have the following disadvantages: 1) they usually require large

amounts of computer time; 2) the variability in pavement performance

data may make their precision superfluous; and 3) F.E.M.'s are usually

too complex and consume too much time to be used routinely in pavement

management systems. They concluded that elastic layer theory with an

equivalent effective modulus gives reasonable response prediction.

Brown and Pappin (89) seem to disagree. They developed a contour model

to predict non-linear behavior and lack of tensile strength in soils and

incorporated this model into the finite element program SENOL. Develop-

ment and limited verification of their model involved both laboratory

tests and full-scale testing of pavements under controlled conditions.

Based on a parametric study and limited measurements they determined the

following:

deflections and strains in the asphalt layer can be reasonably

predicted using an equivalent effective modulus in elastic layer

theory;

it is unlikely that elastic layer theory can be used to predict

stresses and strains within the unbound layers;

they suggest that the K-o approach is less satisfactory than a

linear elastic solution; and

the simplest approach for design calculations is the use of a

linear elastic system provided adeaqate equivalent moduli are used.








They suggest that detailed nonlinear analysis using the SENOL model

would lead to adequate selection of these moduli.

It seems clear that more work is needed to determine suitable

parameters for soil response and to develop models for more accurate

prediction of stresses and strains within the soil layers. Finite

element models should lead to more satisfactory results but the stress-

strain relationships still have to be improved. However, accurate

measurements on full-scale pavements to develop such models are lacking.



2.5 Prediction of Thermal- and Load-Induced Stresses,
Strains, and Failure In Asphalt Concrete Pavements

Several investigators have developed models to predict thermal

stresses and strains in asphalt concrete. Hills and Brien (8) developed

a simple calculation procedure to predict thermally induced stresses.

Their solution was based on a restrained, infinitely long pseudoelastic

beam exposed to a uniform temperature drop. Lateral restraint was

neglected. Limited experimental work showed that the method gave

reasonable predictions. Haas and Topper (13) used the basic equation

presented by Hills and Brien to develop a procedure to calculate thermal

stresses that recognizes the temperature and stiffness gradients that

exist in actual pavements. Shahin (90) modified the Hills and Brien

equation to accommodate coefficients of contraction not constant with

temperature.

Christison et al. (7) used five different analyses for thermal

stress computation and compared predicted and observed values. He

concluded that the pseudoelastic beam yields reasonable results but

points out certain difficulties with this analysis: 1) the predicted

stress depends on the time interval used in the calculation; and 2) the








method does not allow for stress relaxation subsequent to the time

interval in which stresses are computed.

Monismith et al. (9) developed a thermal stress equation for an

infinite viscoelastic slab and complete restraint. Stress is calculated

as a function of depth, time, and temperature. A relaxation modulus is

required from uniaxial creep tests.

Ruth et al. (43) presented a stress equation that considers rate of

cooling, creep rate, and variation in modulus with temperature. The

model requires the asphalt viscosity-temperature relationship from which

all calculations are made. The method is unique in that it can account

for individual asphalt properties.

The prediction of thermal cracking in asphalt pavements is usually

based on comparing the accumulated thermal stress with the tensile

breaking strength of the asphalt concrete. Several investigators have

compared stresses as predicted by the various models with observed

cracking.

Burgess et al. (12) reported that the method of Hills and Brien

correlated well with cracking observed in the St. Anne Test Road.

Christison (7) compared results from pseudoelastic beam, viscoelastic

beam, and viscoelastic slab analyses to cracking observed at St. Anne.

He determined that thermal cracking could be predicted by using the

computed stresses from either analysis at 1/2-inch depth.

Haas and Topper (13) indicated that Monismith's method appeared to

predict unusually high stresses, which may be due to his assumption of

infinite lateral extent.

Models to predict thermal stress cracking have also been

presented. The computer program COLD (91) was developed based on the








Hills and Brien equation. It uses a stress criteria of 200 psi. Shahin

and McCullough (92) developed a model to predict the amount of thermal

cracking. The model predicts temperatures, thermal stresses, low-

temperature cracking, and thermal fatigue cracking. They indicated that

comparisons of predicted cracking with measurements at the Ontario and

St. Anne Test Roads were reasonable. Lytton and Shanmugham (93)

developed a mechanistic model based on fracture mechanics to predict

thermal cracking of asphalt concrete pavements. The model assumes

cracking is initiated at the surface of the pavement and propagates

downward as temperature cycling occurs. The prediction of transverse

cracking and cracking temperature can also be attempted with the Haas

model (94) and the Asphalt Institute Procedure (95). Keyser and Ruth

(58) found absolutely no correlation between actual cracking in 6- to 9-

year old pavements in Quebec, and cracking predicted using the latter

two models.

Ruth et al. (43) developed a thermal cracking model based on their

stress equation. The model computes stresses, creep strains, and

applied creep energy, which are all used as failure parameters.

Predictions of cracking temperatures for a Pennsylvania D.O.T. test road

were obtained which identified the two cracked sections in the test

road.

It should be noted that although indirect comparisons have been

made with observed field cracking, little if any measurements exist of

actual contraction and deformation of asphalt concrete pavements during

cool ing.

Over the years, various solutions have been presented to predict

stresses, strains, and deflections within the pavement system due to








applied wheel loads. In 1885, Boussinesq presented the mathematical

solution for a concentrated load on a boundary of a semi-infinite

body. Love extended this solution to solve for a distributed load on a

circular area. Burmister (96) was the first to present a solution for a

multi-layered elastic system, and developed solutions for specific two-

and three-layered systems. Schiffmann (97) extended Burmister's

solutions to include shear stresses at the surface. Elastic solutions

for layered systems have been presented in the form of tables, graphs,

and equations that include a wide range of parameters (98). Also,

several computer programs have been developed for elastic layer

analysis.

With the advent of the finite element method, more sophisticated

models have been introduced, including viscoelastic analysis (99), Brown

and Pappin's contour model (89), and Yandell's mechano-lattice analysis

(100). Many of the finite element models have been developed for the

problem of rut prediction, and more recently to predict response within

the soil layers more accurately.

For the purpose of predicting stresses, strains and deflections

within the asphalt concrete layer, there seems to he general agreement

that elastic layer analysis is suitable. Although there is some

question of its ability to predict response in the underlying layers,

there has been considerable verification of its ability to predict

response in bound layers.

Avital (101) discusses a series of computer programs available for

the analysis of multi-layered systems. Barksdale and Hicks (32) present

a general description of multi-layered systems and finite element

approaches, along with extensive references on their detailed








development. They recommend the use of elastic layered systems for

pavement analysis since only two variables are needed (modulus and

Poisson's ratio).

The moderators for the last International Conference on the

Structural Design of Asphalt Pavements (84) indicated that use of linear

elastic theory for determining stresses, strains, and deflections is

reasonable as long as the time-dependent and nonlinear response of the

paving materials are recognized. They noted that the papers presented

at the conference confirm that multilayer elastic models generally yield

good results in layers containing binders. It is interesting to note

that of the 16 papers relating to analyzing pavements, 14 used

multilayer elastic theory and the two that used viscoelastic procedures

reduced their viscoelastic idealization to equivalent elastic layered

systems.

Ros et al. (102) measured strains and pressures at different levels

in a variety of trial sections subjected to standard loads at varying

speeds and temperatures. They found good correspondence between

measured values and values computed with an elastic layer program

(BISAR, Ref. 103), using properties determined in situ and in the

laboratory. Correlation was especially good at hich asphalt

stiffness. Halim et al. (104) performed tests on reinforced and

unreinforced flexible pavements in a test pit and found that elastic

layer theory (BISAR) provides a reliable tool to predict flexible

pavement response. They suggested that use of a calibration factor for

stress-dependent materials is more efficient and less time consuming

than more sophisticated models. Waterhouse (105) measured axial

stresses vertically below the central axis of a circular load and








determined that elastic theory gave reasonable prediction of stress

distribution. However, as pointed out in section 2.4 by Brown and

Pappin (89), it is unlikely that elastic theory can predict stresses

within the soil layers, although they did find good correspondence

within the asphalt layer and at the surface of the subgrade using

elastic theory. Several other researchers have also found good

correspondence with measured results using elastic layer theory (e.g.

106, 107).

As mentioned earlier, the approach to pavement failure considered

in this dissertation is based on the idea that pavement cracking is

caused by brittle failure induced by short term repetitive loads and

thermal stresses that occur during cool weather when the asphalt

stiffness is high. Therefore, pavement design and analysis methods

should be based on the theological properties of the asphalt which are

used to predict thermal and dynamic load stresses and strains at thermal

conditions typical of the lowest temperatures expected. The predicted

values can then be compared to the failure limits of the material for a

direct evaluation of failure. This method was proposed by Ruth et al.

(15).

Avital (101) implemented a computer program (CRACK) to handle this

interaction of load and temperature induced stresses. The program

combines an elastic layer computer program for response prediction under

dynamic load and a thermal program for thermal stress and creep strain

prediction. The program requires temperature conditions, traffic

volumes, theological properties of the asphalt, and the pavement struc-

ture characteristics (layer thickness, modulus, and poisson's ratio).





41

This approach is totally different from conventional fatigue

approaches. Therefore, the different distress prediction models that

have been developed will not be presented here. These models usually

attempt to predict pavement life by using different empirically based

sub-models that predict fatigue cracking, rutting, and thermal cracking

separately. The most sophisticated of these are the different versions

of the VESYS model presented by Kenis et al. (108).













CHAPTER III
EQUIPMENT AND FACILITIES


3.1 Description of Test Pit Facility

A test pit facility located at the Office of Materials and Research

of the Florida Department of Transportation (FDOT) was used in this

research project. The facility included a 15 ft. long, 13'-4" wide, 6'-

2" deep concrete pit with a test area of 8 ft. by 12 ft. The test pit

made it possible to construct a layered system of materials to simulate

a flexible pavement system in the field. It had the following features:

control of water level in the test pit;

a hydraulic loading system that could apply both static and

dynamic loads anywhere within the 8 ft. by 12 ft. test area;

two linear displacement transducers for deflection measurements;

a 20,000 lb. capacity load cell; and

a four-channel continuous plotter used for both the transducers

and the load cell.

A detailed description of this facility is given in Research Report S-1-

63, Civil Engineering Department, Engineering and Industrial Experiment

Station, University of Florida. The facility was formerly used for the

evaluation of Florida base course materials using a variation of the

plate bearing test. However, the following modifications were necessary

to make it suitable for testing complete flexible pavement systems:

a proper method for distributing hot asphalt concrete mix within

a reasonable time period to allow for proper compaction and to

avoid premature cooling of the mix;








a cooling system with proper insulation to provide the capability

of testing at different temperatures; and

a proper measurement and data acquisition system for deflection,

strain, and temperature measurements.

Therefore, a good deal of work was done to design, procure materials,

and construct equipment to overcome these deficiencies.



3.2 System for Hot Mix Asphalt Distribution

A triangular shaped hopper that spans the 8 ft. width of the test

pit and distributes material through an adjustable opening at its bottom

was designed to distribute the hot mix. A picture of the hopper is

shown as Figure 3.1. The hopper was made of steel angle and plate and

stands approximately 2'-6" high. It had a level capacity of 33 cf which

made it possible to place a four-inch lift in the test pit in only one

pass. The hopper was designed for completely manual operation and ran

on steel angle rails that were installed in the test pit. A concrete

pad was placed immediately north of the hopper to allow use of a dump

truck for asphalt concrete distribution.



3.3 Pavement Cooling System

Different alternatives were evaluated to select a pavement cooling

system for installation in the test pit area. An air cooling unit was

chosen as the most suitable system. A system of this type is clean,

essentially maintenance free, and provides for reasonably accurate

temperature control. A local mechanical engineer was hired to design

and prepare equipment specifications for a suitable air cooling system.
















7 -/- '. .-


Figure 3.1: Hopper for Asphalt Hot Mix Distribution


,,
~pph~
-


' ; nIR~fFPh29i
~a


,-c- .-
6-' ~c~~--. .
,*ra :~








The system was designed to cool six-inch thick pavements to a

temperature of -10 C (15 F) at a rate of 3.3 C (5 F)/hour, as measured

at a depth of 1/4-inch from the surface of the pavement. Temperature

control was achieved by manually controlling the cooling unit. A fully

automatic system with greater cooling capacity was originally

considered, but its cost was prohibitive. In any case, automatic

controls are of limited value, since temperature gradients would be

present in the pavement as long as the unit was running.

The cooling system consists of a direct expansion, low temperature

refrigeration system. The evaporator (Larkin ELT-300) was located

directly in the test pit and cooled the pavement by recirculating cold

air across the test pit surface. Temperature control was achieved by

lowering the pavement temperature below the required test temperature,

stopping the refrigeration system, and allowing the test pit temperature

to drift upward. The condensing unit (Larkin CS 0750L1), which housed a

7.5 HP compressor, sat outside the building's north wall. Refrigerant

hoses and electrical cables from the condensing unit to the evaporator

unit were connected through holes drilled in the north wall of the test

pit. Drainage was accomplished with a heated drain pipe, which was

passed through a hole drilled in the test pit wall. A layout of the

test pit cooling system is shown in Figure 3.2.

An insulated cover for the test pit area was designed and

constructed. The cover had a solid wood frame which enclosed the test

pit area. The top of the cover consisted of five removable wood panels

that spanned the 8-ft. width of the test area and were supported by a

ledger on the cover's frame. One panel had a one-ft. diameter hole to

allow for placement of the loading ram. The panel dimensions are such













Condensing
Unit


North Wall of
Building

Evaporator
Location
During
Operation


Bay Door


-Flexible
Refrigerant
Hose


i i Evaporator
S__Location
-- When not
L.J Testing
Test Pit
Reaction
Beam -/


NOTH


Insulated Test
Pit Cover


Figure 3.2: Layout of Test Pit Cooling System








that the panel with the hole can be moved to three different positions,

thereby providing for three different loading positions during cooling.

All sections of the cover, including the frame and the panels, were

insulated with six inches of polystyrene, as recommended by the

contractor. Once the frame and panels were in place, all the joints

were sealed with clay and tape to reduce moisture migration and

infiltration into the test pit during operation. A nylon sheet was

placed over the entire cover to further reduce infiltration. A picture

of the cover, completely installed, is shown in Figure 3.3. Figure 3.4

shows the cover removed for access to the test pit.


3.4 Measurement System for Pavement Response

The test pit facility was formerly used exclusively to evaluate

Florida base course materials using a variation of the plate bearing

test. These tests required a loading system, capability for two

deflection measurements, and a recording device for the load and two

displacements. Considerably more extensive measurements were required

for complete evaluation of asphalt concrete pavements at different

temperatures. Static and dynamic deflection and strain measurements at

different points in the pavement were required to define the pavement

response during loading, and the contraction of the pavement during

cooling. Temperature measurements were also required. Therefore, a

measurement and data acquisition system was designed and installed for

this purpose.
































Insulated Test Pit Cover Completely Installed


Insulated Test Pit Cover With Panels Removed


Figure 3.3:


Figure 3.4:








3.4.1 Measuring Instruments

Linear variable differential transformers (LVDT's) were purchased

to obtain static and dynamic deflection measurements at different points

on the pavement surface. Schaevitz model DCD-200 LVDT's, with a range

of 0.20 in. (0.5 cm) and an output of 50 V/in. (19.7 V/cm) were

used. Two dual-output power supplies were purchased to operate these

units. All LVDT's were individually calibrated using a micrometer and a

voltmeter.

Two LVDT support systems were designed and constructed: one for

use with the plate loading system, and the other for use with a dual

wheel loading system designed for use in the test pit. Figure 3.5 shows

a plan view of the test pit area with the LVDT support system used for

plate testing. The system consisted of wood LVDT mounts supported by

1.5 in. diameter pipes that spanned the eight-foot width of the test

pit. The mounts could be positioned to obtain deflection measurements

at any specified distance from the load, and the entire system could be

moved for testing at different positions. The LVDT's were spring-loaded

and could be adjusted vertically by way of an adjustment screw on the

wood mount.

Figure 3.6 shows the LVDT support system used for loading with the

dual wheels. The longitudinal support was a laminated two-by-four-inch

beam which was located underneath the axle of the dual wheel system.

The entire length of the beam was grooved to accept the vertically

adjustable LVDT mounts shown in Section B-B (Figure 3.6). Thus the

mounts could be positioned at any specified distance from the load.

The same mounts used for plate testing (Figure 3.5) were used to obtain

transverse deflections with the dual wheel system. These mounts were





50









PLAN VIEW


Wood LVDT
Mounts



Loading Plate -


A
Ledger to
Support Pipes


~~1~~~~

=




-5---
=
=


11/2"Pipe



-N


A
_4

----------


Section A-A
Hole for LVDT Wires
0i o Hold-Down

jo,
01 1 I [j_- Clamp


Adjustment-.
Screw

SHold-Down
Si* Clamp


Spring-I
LVDT'S


Figure 3.5: LVDT Support System for Plate Loading














PLAN VIEW


Note: See Figure 3.5
for Section A-A.


SECTION B-B
Vertically Adjustable
LVDT Mount





Groovi
for
=-== Mountir
pring-Loaded
LVDT


Figure 3.6: LVDT Support System for Dual Wheel Loading







supported on 1 1/2-inch diameter pipes that rested on longitudinal beams

as shown in Figure 3.6. The supports could all be moved for testing at

different positions.

During initial cooling trials, the LVDT's malfunctioned at low

temperatures, even when covered with heavy insulation. The low

operating temperature of the cooling unit would eventually penetrate the

insulation and cause these units to malfunction. Therefore, a heating

system was installed to maintain the LVDT's at fairly constant temper-

ature during cooling. Variable output heater wire (0 to 4 watts/ft. of

wire) was wrapped around the individual LVDT's, which were then covered

with a 1/2-inch layer of rubber foam insulation. A voltage regulator

was used to control the output of the heater wire, which was adjusted as

necessary to maintain the LVDT's at constant temperature of about 25 C

(77 F) during cooling. A picture illustrating how the LVDT's were

prepared for testing is shown in Figure 3.7. A picture of the entire

LVDT support system with the dual wheel loading system in place and

ready for testing is shown in Figure 3.8.

Strain measurements were obtained with two-inch bonded wire strain

gages (Micro-Measurements EA-06-20CBW-120). Two methods were used to

position the gages at a given location. For surface strain measure-

ments, the gages were mounted at specified points on the pavement after

it was placed and compacted. Several gages were also installed for

measurements at the bottom of the asphalt concrete. These gages were

first mounted on asphalt concrete cores (4-inch diameter and 2 1/2-

inches high) and then were positioned at specified locations on the

compacted base material. The cores were prepared in the laboratory,

using an asphalt concrete mixture similar to the one used for the rest

of the pavement.




















wI


Figure 3.7:


LVDT Prepared for Tests at Low Temperatures







































a) Frontal View


C~~-~- ;r
~~

i.
*-I,


b) Diagonal View


Figure 3.8: Test Pit Pavement Completely Instrumented and Ready
for Testing With Dual Wheels


~t rx








The strain gages were mounted with epoxy directly on the asphalt

concrete surface. The following procedure was used to prepare the

surface and mount the gages. The asphalt concrete surface was prepared

by sanding; first with a belt sander, and then with progressively finer

sandpaper until the surface was "glass-smooth". Clear tape was then

used to lift off all loose particles from the surface. Cleaning with

the tape was repeated until the tape was completely clean when lifted

off the surface. A thin layer of epoxy was then applied to the clean

surface and the gage was positioned, taking care to remove any air

bubbles trapped underneath the gage. A thin layer of epoxy was also

applied to the surface of the gage, for protection and to aid in

bonding. A sheet of cellophane was placed on top of the gage and clear

tape was used to hold the gage in position until the epoxy set. The

cellophane was used to prevent possible damage from the tape adhering

directly to the gage. Once the epoxy set, the tape was removed and the

strain gage wires were soldered to the gage. The completely installed

gage was covered with a piece of masking tape followed by a piece of

duct tape for protection. A picture of the two-inch strain gages

mounted on asphalt concrete cores is shown in Figure 3.9.


3.4.2 Data Acquisition System

A data acquisition system was designed and installed which was

capable of monitoring and recording ten dynamic deflection measurements,

ten dynamic strain measurements, 20 temperature measurements, and load

magnitude and time of loading. As mentioned earlier, only one recording

device, a Gould model 2400 strip chart recorder, was available in the

test pit facility, since this was all that was needed to evaluate base






































Figure 3.9:


Two-Inch Strain Gages Mounted on Asphalt Concrete








course materials. This high speed recorder had four channels, but only

three channels with amplifiers were available for use. Two channels

were used in conjunction with LVDT's to obtain deflection measurements,

and the third channel was used with the load cell to monitor and control

load magnitude and time of loading.

Five dual-beam digital oscilloscopes were purchased to record

additional'deflection and strain measurements. The oscilloscopes used

were Nicolet Explorer Series 2090 with model 201 plug-in units. These

instruments had the capability of monitoring and recording displacements

or strains continuously with time. All five oscilloscopes had a

temporary recording system for indefinite storage of measurements taken

within a specified time interval (i.e. one sweep of the oscilloscope).

In addition, three of the five oscilloscopes were equipped with a floppy

disc recording system for permanent data storage. Permanent storage of

data obtained with the other two oscilloscopes was accomplished by using

an X-Y plotter (Hewlett-Packard Model 7046B). Once data were temporar-

ily stored for a given series of loading cycles, they were immediately

output to a calibrated X-Y plotter for permanent recording.

Four oscilloscopes, two with floppy disc recording systems and two

without, were used in conjunction with eight LVDT's to obtain deflection

measurements. These eight measurements, plus the two on the strip chart

recorder provided for ten simultaneous deflection measurements at

different points on the pavement.

The following equipment was used in conjunction with the strain

gages to obtain strain measurements:

a Vishay/Ellis (V/E) 21 AK switch, balance, and calibration

module;








a V/E 20 AJMLH strain gage indicator; and

a digital oscilloscope with floppy disc recording system.

These units could handle 10 strain gages, in either a 1/4-, 1/2-, or

full-bridge arrangement. However, only one gage could be monitored at

any given time with the strain gage indicator. Output from the

indicator was sent to the digital oscilloscope for continuous recording

with time.

Deflection and strain measurements stored on floppy discs, were

later output to calibrated X-Y plotters. The plotters were calibrated

for an average LVDT output, since each LVDT had a slightly different

calibration. The measurements, as determined from the X-Y plotter

output, were then adjusted for the calibration factor of the particular

LVDT used. All cables going to the LVDT's and the strain gages were

passed through an access hole drilled through the side of the insulated

test pit cover.

A schematic diagram of the data acquisition system is shown in

Figure 3.10. A picture of the system is shown in Figure 3.11. Figure

3.12 is a picture of a typical recording of deflections on a digital

oscilloscope and Figure 3.13 shows a typical output recorded on an X-Y

plotter.

Temperature measurements were obtained with a Fluke Model 2240A

Datalogger. This unit used thermocouple wires to receive and record

temperature measurements for up to 20 different positions at one time.

It could record temperatures at specified time intervals or could be

triggered to record at any given time. The unit automatically records

the date and time of the readings.













Record on
Floppy Discs


J To and From
Load Cell
To and From
LVDT'S 9&10


Remote Control
for simultaneous
Operation of
Digital Oscilloscopes


Input Voltage
to LVDT'S


Output Voltage from LVDT'S
to Digital Oscilloscopes


Figure 3.10: Schematic Diagram of Data Acquisition System
















. ......
_.....~... 1*-


I i I


Figure 3.11:


Data Acquisition System in Test Pit Facility


Figure 3.12: Typical Deflection Recording on Digital Oscilloscope


- ---------------------L-~*lwr*uurr~llll I~-






































Figure 3.13:


Typical Deflection Output on X-Y Plotter







3.5 Loading Systems: Rigid Plate Load vs. Flexible Dual Wheels

Two loading devices were used during the course of this research:

a rigid plate loading and flexible dual wheel loading. Rigid plate

loading was accomplished with a 12-inch diameter steel plate. A cage

was used in conjunction with the plate to evenly distribute the load

over the plate's area, and to increase the plate's rigidity. The plate

was always set on a thin layer of hydrocal (plaster) for levelling and

to evenly distribute the load. A diagram of the rigid plate loading

system is shown in Figure 3.14.

A set of small jet aircraft wheels (Piper Aircraft 31T) were used

for flexible dual wheel loading. The wheels were purchased from a

second-hand dealer for use in the test pit. An axle, which was

compatible with the existing loading system, was designed and machined

for the wheels. The wheels carried a pressure of 100 psi and were

designed to operate at 3,750 Ibs. However, the wheels easily carried

5,000 Ibs. each for a total of 10,000 Ibs. on the dual wheel system.

The actual loaded area was determined from wheel imprints made in the

laboratory at different load levels. A picture of the dual wheel system

is shown in Figure 3.15. Note that it was necessary to attach cables to

the wheels to prevent them from rotating about their vertical axis.

A hydraulic loading system, which could apply static and dynamic

loads, was used with both rigid plate and dual wheel loading. Loading

cycles could be preset for any combination of loading time and rest

period. The time required for the load to be fully applied could not be

controlled, and was dependent on the distance the loading ram had to

travel. Therefore, the load was applied faster with the rigid plate

than with the flexible wheels.


















,Swivel Plate


Loading Cage to
Distribute Load -



Hydrocal to
Level Plate


12" Diameter
Steel Plate


Surface


Figure 3.14: Rigid Plate Loading System






































Figure 3.15:


Flexible Dual Wheel Loading System








After extensive experience with the rigid plate and the dual wheel

devices, several advantages and disadvantages were observed for each.

These are as follows:

I. Rigid Plate Loading.

A. Advantages

loading time could be controlled very accurately, since the

load came on almost instantaneously;

the loading area was circular and constant; and

the position of the load was always known because it was

difficult for the plate to move during loading; and

there was less wear and tear on the loading system since very

little ram movement was required for loading.

B. Disadvantages

very high shear stresses were induced at the edge of the

plate, causing it to sink and forcing a plane of failure;

analysis of a rigid plate on a multi-layer system was a major

problem, since there was no computer program available that

accurately predicted stresses and strains under a rigid

load. An analytical procedure was developed using an elastic

layer analysis, but it proved to be extremely tedious.
II. Flexible Dual Wheel Loading.

A. Advantages

this loading was close to flexible type loading (constant

pressure) and could be modeled more easily with existing

computer programs;

dual wheel loading was more representative of actual truck

loads in the field; and








the wheels were easier to position than the plate, since they

did not require setting with hydrocal.

B. Disadvantages

the load did not come on instantaneously since the wheels had

to deform before a load was applied. Furthermore, the time

required for the load to come on, depended on load

magnitude. Therefore, it was difficult to set the loading

time and to evaluate creep strain accumulation for dynamic

loading conditions. For this reason, all creep tests were

performed using static loads.

up to 2 1/2 inches of ram movement was required for loading,

which caused greater wear and tear in the loading system;

the loading area varied with load and was not circular (more

difficult to model); and

the wheels tended to roll during loading so that the exact

position of load was not known.



It was evident that neither of the loading devices was perfect, but

the disadvantages associated with the plate were overwhelming. The

loading was not representative of wheel loads and the analysis procedure

required tremendous amounts of time. Therefore, the dual wheel loading

system was used for the majority of tests performed.














CHAPTER IV
EFFECT OF ENCLOSED CONCRETE
TEST PIT ON PAVEMENT RESPONSE


4.1 Introduction
The test pit used in this research was made up of 8-inch concrete

walls and a 12-inch concrete slab, which enclosed a volume of 8 ft. by

12 ft. by 6 ft. deep. The layered pavement system was placed and tested

within this volume. Most analytical solutions and computer programs,

consider the layered system (or soil mass) to be infinite in lateral

extent and semi-infinite in depth. Three-dimensional finite element

programs could model the test pit, but as discussed later, these pro-

grams were found to be either too expensive or inaccurate. Therefore, a

study was undertaken to evaluate the effects of the test pit floor and

walls on the response of the layered system to an applied load, and to

establish a methodology to account for these effects.



4.2 Preliminary Analysis

An initial attempt was made to predict the measured response of the

asphalt concrete pavement with an elastic layer computer program. The

pavement deflection and strain measurements used for this analysis were

obtained at a temperature of 18.3 C (65 F), using the plate loading

system at 10,000 Ibs. and 0.1 sec. loading time (see Section 7.1).

Parameters for the sand subgrade and limerock base were determined from

plate load tests performed in the test pit. Conventional analytical

solutions, which consider the pavement lavers to be infinite in lateral







extent and semi-infinite in depth, were used to obtain these

parameters. A sand subgrade modulus of 14,000 psi was calculated using

Boussinesq's theory for a rigid circular load on a semi-infinite mass.

Using Burmister's two-layer theory for similar conditions, a limerock

modulus of 75,000 psi resulted (see Table 5.11 and accompanying

discussion). This procedure is commonly used to obtain modulus values

from plate bearing tests performed in the field. The modulus values

obtained were essentially equivalent moduli, which represent the

behavior of all the materials below the tested surface. These values

are sometimes used to predict the response of the entire pavement system

using elastic layer analysis.

An asphalt concrete modulus of 145,000 psi was calculated for a

temperature of 18.3 C (65 F), using previously established correlations

with measured asphalt viscosity (see Appendix A). Therefore, the

following moduli, Poisson's ratios, and layer thicknesses were used in

an elastic layer computer program (BISAR) to predict pavement response:

Layer Poisson's
Material Thickness (in.) Modulus (psi) Ratio

Asphalt Concrete 4 1/8 145,000 0.35

Limerock Base 6 3/4 75,000 0.40

Sand Subgrade semi-infinite 14,000 0.40

Figure 4.1 shows the measured deflections and the deflections

predicted for the system above (Predicted 1). Although the shapes of

the deflection basins matched reasonably well outside the loaded area,

the measured deflections were grossly overpredicted. It appeared that

the effect of the floor was not properly accounted for by simply using

an effective layer modulus for the sand subgrade. A second computer run










DISTANCE FROM LOAD CENTER (INS.)
6 12 18 24 30 36


Measured


"-'ePredicted C


Predicted (


_~/


EAC = 145,000 psi d = 41/8" V= 0.35
ELR = 75,000 psi d =63/4" V= 0.40
ESAND = 14,000 psi d =oo = 0.40


SEAC = 145,000 psi d = 41/" V= 0.35
ELR = 75,000 psi d = 63/4" V= 0.40
ESAND = 14,000 psi d =48" V= 0.40
ECONC = 3,500,000 d = V= 0.20


20-

Ij


/


Figure 4.1: Measured and Predicted Deflection Basins in the Test Pit


/








was made with a semi-infinite concrete embankment (E = 3,500,000 psi

and P = 0.20) underneath the sand layer. The sand layer was assigned a

finite depth of 48 in. and all other parameters were unchanged. The

predicted deflection basin for this case is also shown in Figure 4.1

(Predicted 2), and shows that the concrete foundation had a very

significant effect on the pavement's response. However, even with the

concrete embankment at a depth of 48 in., the measured response was

overpredicted, which made it clear that there were other factors

affecting the pavement response that were not being accounted for in the

analysis.

A series of program runs was made to determine the depth at which

the concrete floor had no effect on response. The depth of the sand

layer was varied from 48 in. to infinity, while maintaining all other

parameters constant. The results of this analysis are shown in Figure

4.2. Note that even at a depth of 120 in. (10 ft.) the concrete floor

had a significant effect on the predicted pavement response.

The following conclusions were drawn from this preliminary

analysis:

the effect of the concrete floor must be directly accounted for

in the analysis procedure. It cannot be accounted for by simply

using an equivalent layer modulus determined from plate load

tests;

the walls may have an effect on the response of the pavement.

The measured deflections were overpredicted, even when a concrete

embankment was introduced. Therefore, the wall effect needed to

be evaluated and accounted for in the analysis; and











DISTANCE FROM LOAD CENTER (INS.)
12 18 24


(No Concrete)






EAC = 145,000 psi d = 41/8" V = 0.35
ELR = 75,000 psi d = 63/4" V = 0.40
ESAND 17,000 psi d = Variable V= 0.30
ECONC = 3,500,000 psi d =o V= 0.20


Figure 4.2: Effect of Concrete Floor at Different Depths on Predicted Deflection Basins


Measured







the effect of the floor and walls must also be accounted for when

analyzing plate test data. Layer moduli determined from

analytical solutions for systems of semi-infinite depth, are not

suitable for pavement response prediction.

Therefore, a study was undertaken to evaluate the effect of the test pit

floor and walls on the response of the subgrade, the limerock base, and

the complete pavement system.



4.3 Effect of Test Pit Constraints

4.3.1 Analytical Model

Although the elastic layer theory computer program can model a

floor, it cannot model walls. The program considers all materials

infinite in the lateral direction. In addition, the program can only

handle flexible loads and the plate loading system is rigid. Therefore,

several available finite element computer programs were considered to

evaluate the effect of the test pit constraints on pavement perfor-

mance. The AXSYM computer program was chosen for this purpose.

AXSYM is a three-dimensional finite element program written by

E. L. Wilson at the University of California at Berkeley. The program

is for solution of axisymmetric stress-deformation problems using

nonlinear stress-strain characteristics. It is specifically designed

for analysis of vertically loaded circular footings resting on or

beneath the surface of a soil mass. Only linear elastic stress-strain

characteristics were used in conjunction with the program. The basic

difference between the AXSYM model and the test pit is that AXSYM models

the walls as a circular tank, whereas the test pit is rectangular. This







was not considered a major problem and the effects could be defined and

approximated using this model.

Two other programs were also investigated; YBFE1 and SAPIV. YBFE1

is a two-dimensional (plane strain) finite element program developed for

soil-structure interaction problems. Preliminary program runs using

YBFE1 revealed that this program was unsuitable for predicting flexible

pavement response. The error was probably introduced by the plane-

strain nature of the model and the type of finite element used. SAPIV

is a three-dimensional finite element program developed for structural

dynamics but can be used to model homogeneous masses by means of a brick

or plate element. This program would have been most accurate in

modeling the test pit, but preliminary attempts at running the program

showed that an excessive amount of computer space was needed. This

space was not available on the current version of SAPIV at the

University of Florida. In addition, the cost of running the program was

prohibitive for the purposes of this project.



4.3.2 Effect of Constraints on Subgrade Response

Closed form solutions are available for the vertical displacement

of a rigid circle on both a semi-infinite mass and on a finite layer.

The following equations may be used to calculate these displacements:

Semi-infinite: Pz = (1-2) ava

Finite Layer: Pz I Pavg(a)

where z vertical displacement of rigid circle (ins.)

u Poisson's ratio

Pavg average pressure on the rigid circle (psi)








a radius of circle (ins.)

E Young's modulus (psi)

IP influence coefficient: function of P and depth of
finite layer.

Subgrade moduli were calculated with these equations to show the effect

of assuming different finite layer depths and different Poisson's

ratios. Plate deflections measured in the test pit were used in the

calculations. The modulus values calculated are shown in Table 4.1.

As expected, the modulus increases with decreasing Poisson's ratio

and increases with increasing layer depth. However, the main purpose of

this comparison is to show that by assuming an infinite layer as opposed

to a finite layer, errors in excess of 20 percent may result in subgrade

modulus calculations. Similarly, errors in excess of 20 percent may

result in the modulus if a Poisson's ratio of 0.5 is assumed as opposed

to 0.3. Therefore, when calculating modulus for pavement response

prediction using plate load data, it is necessary to use the finite

layer solution. Poisson's ratio values are difficult to determine, but

values of 0.3 to 0.4 are usually considered reasonable for granular

materials. A Poisson's ratio of 0.3 was assumed for the sand subgrade

in the test pit, since laboratory tests by other researchers indicated

that this was a typical value for the Fairbanks Sand.

AXSYM was used to determine the effect of the test pit walls on the

response of the sand subgrade. The sand subgrade was assumed to be a

finite layer of 48-inch thickness. A subgrade modulus of 15,420 psi was

calculated for an assumed Poisson's ratio of 0.3 (see Table 4.1). The

following computer runs were made with these parameters to determine the

wall effect:




















Table 4.1: Sand Subgrade Modulus for Different Layer Depths and
Poisson's Ratios


Modulus Values: Sand Subgrade

Poisson's Depth of Finite Layer (in.)

Ratio Semi-Infinite 24 36 48

0.2 17,890 14,590 16,020 16,610

0.3 16,960 13,410 14,530 15,420

0.4 15,650 12,100 13,290 13,940

0.5 13,980 10,260 11,510 12,160


Note:
(a) Calculated using average 12-in. plate deflection at 15
loading cycle.


psi on 5th







wall at 15 ft., frictionless;

wall at 8 ft., frictionless;

wall at 4 ft., frictionless; and

wall at 4 ft., full friction.

It should be noted that rigid plate loading and finite element grids of

similar geometry were used in all runs.

The plate deflections as well as the deflection basins predicted by

the program, were identical for all cases, indicating that the wall had

absolutely no effect on the response. However, the deflections

predicted by the AXSYM program were about 14 percent less than predicted

by closed form solution (6.55 E-3 vs. 7.59 E-3 in.). It seemed like the

finite element grid used in the AXSYM runs was not fine enough.

Therefore, the number of elements was doubled and the program was

rerun. Although the program solution was closer, it underpredicted

deflections by about ten percent (6.83 E-3 vs. 7.59 E-3 in.). However,

one interesting point is that deflections remained unchanged away from

the loaded plate for the increased element grid.

Based on these results it seemed apparent that the type of finite

element used by AXSYM could not properly handle the high stress

concentrations at the edge of the rigid circle. Therefore, one cannot

put reliance on the rigid plate deflections predicted by AXSYM, except

on a relative basis. It also seems that the error introduced by the

high stress concentrations on these elements does not affect the

deflections away from the loaded area.

Several runs were made with the elastic layer computer program to

verify the accuracy of the AXSYM Drogram away from the rigid loaded

area. A finite layer of 48 in. was used with a modulus of 17,000 psi








and a Poisson's ratio of 0.3. For the AXSYM program, a flexible load

was used and the walls were placed at 8 ft. and assumed frictionless.

The elastic layer program predicted deflections that were identical to

the deflections determined by closed form solution. The AXSYM solution

was identical to the elastic layer solution outside the loaded area and

was within 3 percent of the elastic layer solution underneath the load.

One additional AXSYM run was made to insure that the rigid and

flexible plate AXSYM solutions gave the same results away from the

loaded area, since the comparison of AXSYM and the elastic layer

solution was done for a flexible load. This comparison showed that the

AXSYM rigid and flexible plate solution predict identical deflections

beyond 2 in. of the loaded area.

The following conclusions were made after having verified the

deflections predicted by AXSYM:

the walls have absolutely no effect on the response of the sand

subgrade to a load applied at its surface; and

the floor has a definite effect on the sand subgrade response. A

finite layer closed form solution should he used to determine the

subgrade modulus from plate bearing test data in the test pit.


4.3.3 Effect of Constraints on Limerock Base Response

The effect of the concrete floor on the response of the limerock

base was investigated by a series of runs with the elastic layer

solution computer program. Two systems were analyzed: a 6.75 in.

limerock base over a semi-infinite subgrade; and a 6.75 in. limerock

base over a finite subgrade of 48 in. on a concrete embankment. Three

limerock base moduli were used for each system: 30,000, 60,000, and








100,000 psi. All other material properties were the same for all runs

and are given in Table 4.2.

The predicted deflection basins for each system are presented in

Table 4.2. The deflection differences for systems with and without a

concrete embankment (or floor) are also shown in Table 4.2. These

differences indicate that the effect of the concrete floor was to reduce

the deflections by an amount that is relatively independent of the

stiffness of the limerock base (approximately 2.5 E-3 in.).

A comparison of the deflection basins for the three cases studied

is presented in Figure 4.3. Clearly, the effect of the concrete floor

is considerable and must be accounted for in the analysis.

The following AXSYM runs were made to determine the effect of the

walls on the limerock base response:

wall at 7 ft., no friction, rigid plate;

wall at 7 ft., no friction, flexible plate;

wall at 4 ft., no friction, rigid plate; and

wall at 4 ft., full friction, rigid plate.

The following pavement system was used in the analysis:

Modulus (psi) Poisson's Ratio Thickness (in.)

Limerock: 90,000 0.40 6.75

Sand: 14,530 0.30 36

This system was underlain by a rigid base. A relatively high limerock

modulus was chosen for the analysis, since this stiffer system would be

affected to a greater degree by wall friction. A pressure of 50 psi on

a 12-inch diameter area was applied in all cases.













Table 4.2: Effect of Concrete Floor on Surface Deflections for Different Base Stiffnesses


Surface Deflections (in.)

Modull Poisson's Thickness Distance From Center of the Plate (in.)
f of Layers (psi) Ratios (in.) 0 4 6 9 12 18 24 30 36 48

2 30,000 0.4 6.75 2.43E-2 2.17E-2 1.70E-2 1.13E-2 8.74E-3 5.89E-3 4.39E-3 3.50E-3 2.91E-3 2.19E-3
15,420 0.3 SM-INF

3 30,000 0.4 6.75
15,420 0.3 48.0 2.18E-2 1.92E-2 1.45E-2 8.83E-3 6.27E-3 3.48E-3 2.06E-3 1.26E-3 7.79E-4 2.84E-4
3,500,000 0.2 SM-INF
Difference 2.5E-3 2.5E-3 2.5E-3 2.47E-3 2.47E-3 2.41E-3 2.33E-3 2.24E-3 2.24E-3 1.91E-3

2 60,000 0.4 6.75 1.89E-2 1.72E-2 1.44E-2 1.08E-2 8.68E-3 6.00E-3 4.46E-3 3.52E-3 2.90E-3 2.17E-3
15,420 0.3 SM-INF

3 60,000 0.4 6.75
15,420 0.3 48.0 1.64E-2 1.47E-2 1.20E-2 8.33E-3 6.27E-3 3.64E-3 2.17E-3 1.32E-3 8.10E-4 2.92E-4
3,500,000 0.2 SM-INF
Difference 2.5E-3 2.5E-3 2.4E-3 2.47E-3 2.41E-3 2.36E-3 2.29E-3 2.20E-3 2.09E-3 1.88E-3

2 100,000 0.4 6.75 1.59E-2 1.47E-2 1.28E-2 1.02E-2 8.49E-3 6.05E-3 4.53E-3 3.55E-3 2.92E-3 2.16E-3
15,420 0.3 SM-INF

3 100,000 0.4 6.75
15,420 0.3 48.0 1.35E-2 1.23E-2 1.04E-2 7.81E-3 6.12E-3 3.74E-3 2.28E-3 1.39E-3 8.52E-4 3.03E-4
3,500,000 0.2 SM-INF
Difference 2.4E-3 2.4E-3 2.4E-3 2.39E-3 2.37E-3 2.31E-3 2.25E-3 2.16E-3 2.07E-3 1.86E-3











DISTANCE FROM CENTER OF PLATE, IN.
12 18 24


5E-3





10E-3
z
0
U
-J
u.
o 15E-3


// 48" Subgrade Concrete Slab
// ----Semi-Infinite Subgrade


i/ Base Modulus:
'/ 0 30,000 psi
0 60,000 psi
A 100,000 psi










of Different Base Layer Stiffness on Predicted Deflection Basins


/ 1'
/O








Figure 4.4 shows a comparison between deflection basins for the

wall at 7 ft. with no friction and the wall at 4 ft. with full

friction. This comparison gives a direct indication of the effect of

having the wall at 4 ft. as opposed to having no wall. As shown in the

figure, the effect of the wall was to shift the deflection basin upward

by a small amount. The actual deflections for each case, given in Table

4.3, show that the deflection difference between the two basins is about

0.23 E-3 in., or about 2.2 percent of the plate deflection. This effect

is relatively insignificant, especially considering that the accuracy of

our measurements was somewhere in this range.

An elastic layer program run was made to evaluate the accuracy of

the AXSYM solution for this two-layer case. The elastic layer program

run gave a deflection basin identical to the AXSYM flexible plate run

with wall at 7 ft. The basin from the AXSYM rigid plate run with wall

at 7 ft. was also identical to these basins outside of the loaded

area. Figure 4.5 shows the deflection basins plotted for these three

runs. Note that the rigid plate deflections seem low relative to the

flexible plate, again showing AXSYM's inability to handle the high

stress gradients induced at the edge of the plate. There is also some

discrepancy under the load between the AXSYM flexible plate solution and

the elastic layer solution, but this is small. In general, it seemed

that the AXSYM solution was accurate and could be used to evaluate the

wall effects on a relative basis.

Therefore, the following conclusions were made concerning the

effect of the test pit constraints on the limerock base response:

the floor effect is significant and must be accounted for, but

the effect is independent of limerock base modulus; and

the wall effect is insignificant and can be ignored.






















6 1


DISTANCE FROM LOAD CENTER (INS.)

18 24 30


AXSYM., Rigid Plate, Wall @ 4 ft., Full Friction

AXSYM., Rigid Plate, Wall @ 7 ft., No Friction



Limerock: Modulus = 90,000 psi, d = 63/4", V= 0.40

Sand: Modulus = 14,530 psi, d = 36", V= 0.30

Underlain by Rigid Layer


Figure 4.4: Effect of Test Pit Walls on Limerock Base Response


LI
z

z 5E.
0

0
w
Uj
-j
u.
u 10E.
0
w

B
15
0 5E
0r
iSE














Table 4.3: Predicted Deflections Using AXSYM


Deflections (E-3 in.)
Position Distance From Load Center

of Wall 0.0 6.0 7.5 9.0 12.0 16.0 20.0 24.0 30.0 36.0 42.0 48.0

4 ft. (NF)* 10.19 10.17 8.06 7.09 5.46 3.82 2.61 1.74 0.90 0.41 0.17 0.10

A-4 ft. (FF)* 10.16 10.14 8.03 7.06 5.43 3.78 2.56 1.68 0.82 0.32 0.08 0.0

B-7 ft. (NF)* 10.38 10.36 8.25 7.27 5.65 4.01 2.79 1.92 1.06 0.54 0.25 0.08

B A 0.22 0.22 0.22 0.23 0.22 0.23 0.23 0.24 0.24 0.22 0.18 0.08

* F No Friction
FF Full Friction




Full Text
33
peculiarities of the different testing devices may result in loading
conditions that are not directly comparable to moving wheel truck
loads. In addition, the method is only suitable for existing pavements
and the results obtained will represent only the behavior during the
particular time tested.
Maree et al (86) presented an approach to determine layer moduli
based on a device they developed to measure deflections at different
depths within the pavement structure. They suggested that effective
moduli for use in elastic layer theory can be determined from correction
factors or shift factors established from field measurements using their
device at different times of the year and under different conditions.
They determined effective moduli from tests performed at different load
levels and under different environmental conditions using their
multidepth deflectmeter. They found that moisture condition has a
significant effect on response and its effect can he as important as
stress state. They also demonstrated that the modulus of individual
materials depends on the modulus of the underlying layers. The stress
dependence of several pavement materials was also demonstrated from the
field data. They found that most granular materials behave in a stress-
stiffening way, while most subgrade materials (e.g. weathered shale)
behave in a stress-softening way. Comparison of field and 1aboratory
data showed that although the trends apparent in the field were also
apparent in the laboratory, the constant-confining pressure triaxial
tests overestimated the modulus of the base materials tested.
Accurate prediction of stresses and strains within the foundation
layers is necessary for the prediction of rutting and stability failures
within these layers. Luhr and McCullough (88) state that moduli of


DEFLECTION (E-3 ins)
DISTANCE FROM LOAD CENTER (ins)
6 12 18 24 30 36
Figure 7.35: Measured vs. Predicted Deflections at 4,000 lbs.: E2 = 40,000 psi, 20.6 C (69 F)


173
Table 7.4: Surface Deflections: Slow Loading Rate
Surface Deflection (x 10"3 in.)
Load
Distance
From Center of Load (in.)
(lbs.)
0 in.
8 5/16 in.
12 1/16 in. 18 5/16 in.
28 1/4 in.
10,000
17.6
6.8
4.2 4.35
1.7
7,000
14.0
5.3
2.9

4,000
9.0
5.0
3.4
0.6
1,000
5.4
2.6
1.7 1.0
0.0
Notes:
(a) Readings shown are average of two LVDT's
(b) Average pavement temperature: 21.7 C (71 F)
Table 7.5: Surface Strains: Slow Loading Rate
Surface Strains
(micro-strain)
Load
Distance From
Center of Load (in.)
(lbs.)
8 3/32
in. 12 1/16
in. 17 15/16 in.
28 5/16 in.
10,000
47.6
87.9
67.6
41.0
7,000
39.6
76.0
54.2
32.0
4,000
37.8
51.4
36.8
21.9
1,000
25.9
33.4
21.5
14.0
Notes:
(a) Readings shown are average of two strain gages
(b) Average pavement temperature: 21.7 C (71 F)


Table D.4: Measured Strains, Test Position 1, 6.7 C (44 F)
Load
Longitudinal
Strains (micro-strain)^
Transverse Strains
(micro-strain)
(kips)
Distance from Load Center
(in.):
N-S(b)
Distance
from Load Center (in.
): E-W(c)
0BT{d) 0BL{e)
8{N) 12(S)
24 (S)
36 (N)
12(E)
18 (W)
24(E)
36 (W)
10
-52 246
31 66
38
8
18
96
45
17
7
-32 191
22 47
29
6
16
66
32
13
4
-13 112
10 28
20
4
9
40
20
8
1
-2 30
2 7
6
1
2
10
5
2
(a) Tension is positive.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.
(d) This gage located at the center of load, at the bottom of pavement, in the transverse direction.
(e) This gage located at the center of load, at the bottom of pavement in the longitudinal direction.


Tensile Strain (E-6ln/ln)
i oo
Distance from
load center (Ins.)
Bottom of pavement
x Measured
ISO
O Predicted
'J
E, =320,000psl
E2= 25, OOOpsI
zoo
E3 = 25, OOOpsI
2B0
Compressive
Strain (E-6ln/ln)
Fiqure 9.12: Comparison of Measured and Predicted Transverse Strain Distributions at 13.3 C (56 F)


146
Moduli for the limerock base were calculated based on resilient
plate deflections, using the procedure outlined in Section 4.4.2.1, with
an assumed Poisson's ratio of 0.4. A subgrade modulus of 18,730 psi was
used for the plate tests performed at locations where pavement response
tests had been performed (test numbers one, two, and three). This value
was the average of the two moduli determined from plate tests on the
subgrade at locations where pavement response tests had been
performed. For plate tests at locations where no load tests were
performed on the pavement, a subgrade modulus of 15,420 psi was used,
which is the value calculated from the original plate tests on the sand
(Section 5.4.2).
The calculated limerock moduli are given in Table 5.13. These
values are siginficantly lower than the limerock moduli determined when
the limerock was initially placed and compacted (see Table 5.10). This
reduction seems to be directly related to the difference in water
content and dry density in the limerock. The original water content was
2.0% lower and the original dry density was 4.0 pcf higher than when the
pavement was removed. The reason for these changes is difficult to
determine with available data. However, it appears that loading had
little effect on the observed changes in limerock properties. The
modulus values were about equal for both loaded and unloaded areas, as
were the measured water contents and dry densities. Therefore, it seems
like some other phenomenon was responsible for these changes.
There are several possibilities for the observed increase in
moisture. Capillary rise may have occurred which transferred moisture
from the sand subgrade to the limerock base. This seems reasonable,
since there is a significant amount of fines in the crushed limerock


Distance From Center of Test Position (ins.)
South
Figure 8.68: Comparison of Dynamic Load Response at Different Times: 0.0 C (32 F), Test Position 2


Table
Page
E.21 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 2,
13.3C (56 F) 417
E.22 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position Number 2, 13.3 C (56 F).... 418
E.23 Measured Dynamic Deflections at 10,000 lbs. After
Different Times of Static Load Application, Position
Humber 2, 13.3 C (56 F) 419
E.24 Measured Dynamic Strains at 10,000 lbs. After Different
Times of Static Load Application, Position Number 2,
13.3C (56 F) 420
E.25 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 3,
0 C (32 F) 421
E.26 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position Number 3, 0 C (32 F) 422
E.27 Measured Dynamic Deflections at 10,000 lbs. After
Different Times of Static Load Application, Position
Number 3, 0 C (32 F) 423
E.28 Measured Dynamic Strains at 10,000 lbs. After Different
Times of Static Load Application, Position Number 3,
0 C (32 F) 424
E.29 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 3,
6.7 C (44 F) 425
E.30 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position Number 3, 6.7 C (44 F) 426
E.31 Measured Dynamic Deflections at 10,000 lbs. After
Different Times of Static Load Application, Position
Number 3, 6.7 C (44 F) 427
E.32 Measured Dynamic Strains at 10,000 lbs. After Different
Times of Static Load Application, Position Number 3,
6.7 C (44 F) 428
E.33 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 3,
13.3C (56 F) 429
E.34 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position Number 3, 13.3 C (56 F).... 430
E.35 Measured Dynamic Deflections at 10,000 lbs. After
Different Times of Static Load Application, Position
Number 3, 13.3 C (56 F) 431
E.36 Measured Dynamic Strains at 10,000 lbs. After Different
Times of Static Load Application, Position Number 3,
13.3C (56 F) 432
ix


APPENDIX A
RELATIONSHIPS BETWEEN ASPHALT CONCRETE PROPERTIES
AND ASPHALT CEMENT PROPERTIES
A.l Dynamic Modulus
Ruth et al. (43) presented relationships between measured asphalt
viscosity and the dynamic modulus of the asphalt concrete mixture at
different temperatures. These relationships are as follows:
For < 9.19 E8 Pa*s
log Eq j = 7.18659 + 0.30677 log n^g (Equation A.2)
For n^gg > 9.19 E8 Pa*s
log Eg j = 9.51354 + 0.04716 log n10g (Equation A.l)
where, Eg ^ dynamc modulus 0f the asphalt concrete
n100 aPParent viscosity of the asphalt at constant
power of 100 watts/m3
The relationships were developed for dense-graded mixtures tested
in indirect tension for a loading time of 0.1 seconds. They require the
constant power viscosity of the asphalt at a specific temperature as
determined from measurements with the Schweyer rheometer. Viscosity-
temperature relationships for the asphalt can be developed by obtaining
viscosity measurements at several temperatures and stress levels.
A.2 Creep Strain Rate
Ruth et al. (43) also presented relationships between the creep
strain rate of the mixture at a given stress level and the
359


19
researchers estimating low temperature stiffness by means of tests at
higher temperatures. Pink et al. (54) and Keyser and Ruth (58) also
emphasized the importance of experimental measurements rather than the
use of empirical extrapolations to determine low temperature properties.
Probably the most significant advancement to the understanding of
low temperature response and failure properties of asphalts and asphalt
mixtures was the development in the 1970's of the Schweyer constant
stress rheometer (59). This device has the capability of measuring
rheological properties at -10 C (14 F) and lower. Furthermore, Schweyer
established rheological concepts that led to a definitive rheological
model and methods to evaluate important parameters that relate to low
temperature behavior, including shear susceptibility.
The proposed rheological model is the Burns-Schweyer model (55).
The model is a Burgers model with a modified dashpot to incorporate a
self-generating feedback system to regulate the rate of viscous flow.
Thus, the model accounts for viscous behavior for both Newtonian and
shear susceptible materials, as well as for elastic, and delayed elastic
behavior. Details pertaining to the measurement and evaluation of
rheological parameters using the Schweyer rheometer may be found in
references 50, 55, 59, and 60. Ruth and Schweyer (61) showed that the
Burns-Schweyer model gives accurate prediction of the rheological
properties of asphalts, including those that are very shear
susceptible. Keyser and Ruth (58) concluded that the Schweyer rheometer
is an excellent device for low temperature measurements of asphalt
properties and that the concepts developed by Schweyer provide values
more closely related to shear and strain rates encountered in the
laboratory and in actual pavements.


234
Figure 8.11: Measured Longitudinal Strains vs. Temperature:
Test Position 2


CHAPTER III
EQUIPMENT AMD FACILITIES
3.1 Description of Test Pit Facility
A test pit facility located at the Office of Materials and Research
of the Florida Department of Transportation (FDOT) was used in this
research project. The facility included a 15 ft. long, 13'-4" wide, 6'-
2" deep concrete pit with a test area of 8 ft. by 12 ft. The test pit
made it possible to construct a layered system of materials to simulate
a flexible pavement system in the field. It had the following features:
- control of water level in the test pit;
- a hydraulic loading system that could apoly both static and
dynamic loads anywhere within the 8 ft. by 12 ft. test area;
- two linear displacement transducers for deflection measurements;
- a 20,000 lb. capacity load cell; and
- a four-channel continuous plotter used for both the transducers
and the load cell.
A detailed description of this facility is given in Research Report S-l-
63, Civil Engineering Department, Engineering and Industrial Experiment
Station, University of Florida. The facility was formerly used for the
evaluation of Florida base course materials using a variation of the
plate bearing test. However, the following modifications were necessary
to make it suitable for testing complete flexible pavement systems:
- a proper method for distributing hot asphalt concrete mix within
a reasonable time period to allow for proper compaction and to
avoid premature cooling of the mix;
42


West
42
ro
co
cn
Figure 8.48: Comparison of Measured Transverse Strains at Different Temperatures:
Test Position 2, 10,000 lbs.


Tensile Strain (E 6In/In)
South
Figure 9.2: Comparison of Measured and Predicted Longitudinal Strain Distributions at 0.0 C (32 F)


174
proportionately, while the deflections next to the plate remained
essentially constant. This effect is more clearly illustrated by the
load-deflection relationships for the different measurement positions.
These are shown in Figure 7.10. The load-deflection relationships for
the positions next to the plate (8- and 12-inches from load center) were
almost vertical for loads greater than 4,000 lbs., while the
relationship for the plate had a much more gradual slope. Apparently,
the high shear stresses at the edge of the rigid plate, which are
theoretically infinite, combined with the increased loading time, caused
the plate to sink into the asphalt concrete layer.
This phenomenon appeared to have two effects on the pavement system
response: 1) it prevented the asphalt concrete layer from acting as a
structural unit in flexure, since yielding was taking place on a
vertical plane at the edge of the plate; and 2) it caused a volume of
asphalt concrete to be transferred from underneath the plate to the
surrounding areas. Load-strain relationships for the different
measurement positions are shown in Figure 7.11. This figure shows that
the strain response at a point two in. from the edge of the plate (eight
in. from load center), was totally different from the strain response
further from the plate. As the load increased, the rate of increase in
strain at this position decreased dramatically as compared to the other
positions, indicating that strains, and thus stresses, were not
increasing in proportion to the applied load. This suggests that the
behavior of the asphalt concrete layer in the vicinity of the plate was
discontinuous in flexure.
Figures 7.12 through 7.15 are comparisons of deflection basins for
the fast and slow loading rates for the different load levels used. The


Table E.3: Measured Dynamic Deflections at 10,000 lbs. After Different Times of Static Load Application,
Position Number 1, 13.3 C (56 F)
Load^
Duration
Longitudinal Deflections 1
[E-3 in.)(b>
Transverse Deflections (E-3
in.)
Distance
from Load
Center (in.): N-
s(c)
Distance
from Load
Center (in.
): E-W(d)
(Seconds)
0
8(N)
12 (S)
18 (S)
24 (N)
36 (N)
13 (W)
18(E)
24 (W)
36(E)
0
12.58
7.10
6.75
3.04
0.68
0.0
9.46
4.37
3.20
0.3
50
12.09
7.43
5.88
2.94
0.77
0.0
8.65
3.97
2.96
0.1
100
12.19
7.05
6.03
2.94
0.72
0.0
9.08
3.87
2.86
0.1
500
11.90
6.77
5.83
2.94
0.82
0.0
8.69
3.72
2.86
0.1
1000
11.60
6.67
5.64
2.90
0.92
0.0
8.50
3.47
2.71
0.2
(a) A rest period equal to four times the load duration was allowed before dynamic testing.
(b) Positive is downward.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.
399


16
an approach can lead to serious error, particularly when predicting
properties at lower temperatures.
As explained earlier, the analysis method used in this dissertation
considers cracking to be caused by brittle failure induced by short term
repetitive loads and thermal stresses that occur when the asphalt
stiffness is high. Therefore, the emphasis here is placed on the behavior
at relatively low temperatures, roughly in the range from -10 C (14 F), or
approximately the glass transition temperature, to 25 C (77 F). This
temperature range is referred to as the near transition region (48).
2.3.1 Asphalt Cement Properties
The response of asphalt to an applied stress is time dependent,
where the strain increases at a given rate with time. At lower temper
atures many asphalts are also shear susceptible, with the change in
creep strain rate not being proportional to the change in applied
stress. Finally, the behavior of all asphalts is highly dependent on
temperature.
In general, as the temperature is lowered, asphalts become more ^
viscous and eventually exhibit glassiness, where different elasto- /
viscous behavior is observed, their coefficients of expansion change, \
y
and brittle fracture may develop (48). Jongepier and Kuilman (49)
explained the behavior of asphalt as a viscoelastic liquid. At low
temperatures, asphalt behaves like an elastic solid, while at high
temperatures its behavior is comparable to a viscous liquid. At inter
mediate temperatures the behavior is influenced by both viscous and
elastic components. Asphalt cements show a characteristic common to
other amorphous materials; the glass transition phenomenon. Schweyer


154
4. The test pit cover was then sealed for cooling.
5. The strain gages were balanced and zeroed and an initial
temperature reading was taken prior to starting the cooling
unit.
6. After starting the cooling unit, temperature and strain
measurements were taken at specified time intervals. The time
intervals used are shown in the tables of pavement temperatures
during cooling presented in Appendix B.
7. The pavement was cooled until the temperature at the bottom of
the asphalt concrete layer was at 0.0 C (32 F). At this time,
the strains in the pavement were recorded and the cooling unit
was turned off.
8. Since the top of the pavement was colder than the bottom, the
pavement temperature was allowed to equalize so that the
properties of the asphalt concrete layer were uniform during
dynamic load tests. During this initial equalization period,
the temperature of the bottom of the pavement remained close to
0.0 C (32 F).
9. Once the top and bottom of the pavement were within about 1.5 C
(2.7 F), a final strain reading was made and dynamic load tests
were begun.
10.The strain gage dummy was changed from the gage located outside
the test pit to one located on the pavement surface. The dummy
gage was located at the north end of the pavement, perpendicu
lar to the axis of loading, so that it would not be affected by
applied loads. Each gage was then rebalanced with the new
dummy.


CHAPTER IV
EFFECT OF ENCLOSED CONCRETE
TEST PIT ON PAVEMENT RESPONSE
4.1 Introduction
The test pit used in this research was made up of 8-inch concrete
walls and a 12-inch concrete slab, which enclosed a volume of 8 ft. by
12 ft. by 6 ft. deep. The layered pavement system was placed and tested
within this volume. Most analytical solutions and computer programs,
consider the layered system (or soil mass) to be infinite in lateral
extent and semi-infinite in depth. Three-dimensional finite element
programs could model the test pit, but as discussed later, these pro
grams were found to be either too expensive or inaccurate. Therefore, a
study was undertaken to evaluate the effects of the test pit floor and
walls on the response of the layered system to an applied load, and to
establish a methodology to account for these effects.
4.2 Preliminary Analysis
An initial attempt was made to predict the measured response of the
asphalt concrete pavement with an elastic layer computer program. The
pavement deflection and strain measurements used for this analysis were
obtained at a temperature of 18.3 C (65 F), using the plate loading
system at 10,000 lbs. and 0.1 sec. loading time (see Section 7.1).
Parameters for the sand subgrade and 1imerock base were determined from
plate load tests performed in the test pit. Conventional analytical
solutions, which consider the pavement layers to be infinite in lateral
67


158
Table 7.1: Summary of Dynamic Plate Load Tests at Ambient Temperatures
Test
Series No.
Pavement
Temperature
Load Rate
(sec. on/sec. off)
Load Level
(kips)
Test
Position
Date
1
18.3 C/65 F
0.1/0.4
1,4,7 & 10
Center^
3/83
2
21.7 C/71 F
1.0/4.0
1,4,7 & 10
Center^3)
3/83
3
20.6 C/69 F
0.1/0.4
1,4,7 & 10
Center^
4/84
4
25.6 C/78 F
0.1/0.4
1,4,7 & 10
South^
4/84
(a) Plate was set at the center of the test pit.
(b) Plate was set three feet south of center.


254
an even sharper decrease in compressive strains occurred to the north of
this section where the surface of the pavement actually went into
tension at about nine feet from the south wall of the test pit.
As mentioned earlier, the most interesting thing about these
changes in response, was that there was almost no change in pavement
temperature during the time that they occurred. The cooling rate was
negligible so that the thermal stresses accumulated during this time
were probably also negligible. The small temperature reduction that did
occur should have resulted in either no change in strain or a small
uniform increase in compressive strains. Therefore, the change in
thermal response observed was truly unusual. But so was the
differential strain distribution observed before these changes
occurred. It is possible that the two were related.
There are two ways, aside from changes in moisture, for a change in
strain to occur in any structure: with a change in temperature or with
an application of load. As mentioned before, the change in temperature
was very small and could not explain the observed response. This
implies that the strain changes occurred as a result of an applied
load. The only load on the pavement during cooling was the weight of
the asphalt concrete itself, and the only way that the weight of the
asphalt concrete could have caused the observed changes in strain, was
if the pavement was uplifted from the base prior to the time the changes
were observed. Therefore, it seems like the effect of cooling, prior to
the time this change in strain response was observed, was to uplift the
asphalt concrete layer from the base.
It will be shown in Section 8.3.3 that highly uncommon load
response was observed in the pavement, which seemed to be related to the


275
(see Figures 8.25, 8.26, and 8.27), each deflection basin and strain
distribution was defined by twice that number of data points (see
Figures 8.29 to 8.40). This was because the deflections and strains
were assumed to be symmetrical about the center of loading, and
measurements taken at different distances on either side of the load
were used to define the measured distribution on the opposite side.
Initial analyses of dynamic load response data indicated that the
wheels moved during dynamic load tests. Therefore, the exact position
of the wheels during loading was not known and had to be interpreted for
each set of deflection and strain measurements. The figures presented
for test position 3 (Figures 8.29 through 8.40) show that for almost all
tests performed, the measured strains and deflections were not
symmetrical about the center of the test position, where the wheels were
originally placed. However, the figures also show, that as expected,
the measurements were in all cases symmetrical. The point of symmetry,
which was determined by trial and error for each set of measurements
(i.e. for each load level and temperature), was interpreted as the
center of loading. It is interesting to note that the center of loading
was determined from deflection and strain measurements independently,
and in all cases the tv/o centers corresponded exactly. This indicated
three things: that the method used to establish the center of loading
was valid; that the precision of the measurements was good; and that
there was good correspondence between measured deflections and measured
strains.
It seemed clear from these results that there were deficiencies in
the loading system that caused the wheels to move longitudinally with
repeated loading, and in some cases, caused the wheels to rotate as the


6
2.2 Distress In Asphalt Concrete Pavements
2.2.1 Modes of Distress
The modes of distress in asphalt concrete pavements are well
recognized and the causes of distress, at least in general terms, are
also known. Tables 1 and 2 (1,2) are two examples of tables listing the
types and causes of distress in asphalt concrete pavements. These
tables show that failures can be grouped into three major categories:
cracking, rutting, and disintegration.
Pavement surveys around the country and the world indicate that of
these three categories, cracking is the major problem in terms of amount
and cost. Based on extensive observations by himself and others, Finn
(3) stated that traffic-associated cracking is the number one priority
item for improving and extending the performance of asphalt pavements.
Pedigo et al. (4) reviewed a great deal of work that has been done on
pavement distress and reached similar conclusions. Finn also stated
that traffic associated cracking is one of the first indicators of
distress observable in asphalt pavements, and that cracking is often
observed with little or no distortion. In reviewing the results of the
AASHO Road Test, he found that cracking led to other forms of distress
(such as rutting), and that more cracking occurred when the pavement was
cold than warm. However, there was a lack of information as to when and
where the first cracks occurred and how these cracks propagated.
Furthermore, asphalt properties were not measured at low temperatures.
Information of this type is lacking, even today. Measurements of
the environmental and loading conditions at the time of initial
cracking, along with relevant material properties, are crucial to the
development of damage criteria. Such information could not be found in


215
this was developed using an elastic layer analysis program, but was
found to be very tedious. Therefore, if at all possible, rigid plate
loading systems should not be used to evaluate asphalt concrete
pavements. After these plate tests were completed, a flexible dual
wheel loading system was designed and constructed for use in the test
pit.


SURFACE STRAINS (p.e) TENSION
Figure 7.32: Measured vs. Predicted Strains at 10,000 lbs.: E2 = 75,000 psi, 20.6 C (69 F)


Dislance From Load Center (ins)
West
Figure 9.3: Comparison of Measured and Predicted Transverse Deflection Basins at 0.0 C (32 F)


South
Figure 8.34: Measured Longitudinal Strains at 6.7 C (44 F): Test Position 3
268


DEFLECTION, IN.
DISTANCE FROM CENTER OF PLATE, IN.


ACKNOWLEDGMENTS
I would like to express my gratitude to Dr. Byron E. Ruth, Chairman
of my Graduate Supervisory Committee, for his guidance, encouragement,
and friendship. I would also like to thank Dr. F. C. Townsend,
Dr. J. L. Davidson, Professor W. H. Zimpfer, Dr. M. C. McVay,
Dr. D. L. Smith, and Dr. J. L. Eades for serving on my Graduate
Supervisory Committee. I consider myself fortunate to have had such a
distinguished committee.
A very special thanks goes to the Florida Department of
Transportation (FDOT) for providing the financial support, testing
facilities, materials, and personnel that made this research possible.
I would also like to thank the many individuals at the Bureau of
Materials and Research of the FDOT who contributed to this research
project by giving so generously of their time. I especially want to
thank the personnel in the Pavement Performance Division, Bituminous
Materials and Research Section, the Pavement Evaluation Section and the
Soil Materials and Research Section for their help and consideration.
I would also like to thank Candace Leggett for her
conscientiousness and diligence in typing this dissertation.
Finally, I would like to thank my wife Maria for encouraging me to
return to school, and for her encouragement and patience throughout my
Ph.D. program.
ii


2'0" 2l0"
( ) ( ) ( 1
C ) CZ3 CUD
POSITION I POSITION 2 POSITION 3
5'.0"
Figure 8.2: Location of Test Positions in the Test Pit


41
This approach is totally different from conventional fatigue
approaches. Therefore, the different distress prediction models that
have been developed will not be presented here. These models usually
attempt to predict pavement life by using different empirically based
sub-models that predict fatigue cracking, rutting, and thermal cracking
separately. The most sophisticated of these are the different versions
of the VESYS model presented by Kenis et al. (108).


11
susceptibility are conducive to reducing the temperature at which
fracture occurs.
The advantage of using a softer binder, particularly one with a low
temperature susceptibility, was demonstrated by Hills and Brien (8).
Fromm and Phang (16) also reported that less temperature susceptible
asphalts were associated with pavements exhibiting less cracking.
Schmidt (24) suggested that the glass transition temperature of the
asphalt might be a more definitive measure of non-load associated
cracking than measured viscosities, since at temperatures lower than the
glass transition temperature the asphalt behaves elastically, while at
higher temperatures it exhibits viscoelastic response. Thus, below the
glass transition temperature there is almost no potential for stress
relaxation.
Other factors have been found to influence low temperature
transverse cracking, but to a lesser degree than asphalt properties.
Tuckett et al. (25) found that higher asphalt contents reduced thermal
cracking. Fabb (23) reported that increasing binder content reduced
thermal fracture, but only slightly. He also concluded that the
properties and grading of the aggregate had little or no effect on the
resistance of the asphalt concrete to thermal cracking. Cooling rate
was found to have little effect on the failure temperature by Fabb (23)
and Fromm and Phang (16). However, they only compared relatively high
cooling rates. Finally, results of the St. Anne Test Road indicated
that only half the frequency of low temperature cracking occurred in
10-inch pavements than did in 4-inch pavements (26).
The concept of fatigue is probably the most recognized concept that
has been suggested for use in the evaluation of traffic-load associated


DEFLECTION (x 103 INS.)
DISTANCE FROM CENTER OF LOAD (INS.)
Figure 7.3: Measured Deflection Basins: Fast Loading Rate
164


93
not properly account for the effect of the test pit floor. Modulus
values determined from 12-inch plate tests were 25 percent less than
moduli from 16-inch plate tests.
The major source of error in this approach is the assumption of an
equivalent subgrade modulus based on the semi-infinite layer solution.
An equivalent modulus based on plate test deflections is only equivalent
for the same loading configuration and layer geometry that existed
during testing. That is, the equivalent modulus for the case of a 12-
inch plate on the sand subgrade will only give equivalent plate
deflections for similar conditions. Once the limerock base layer is
introduced, these equivalent moduli will give neither equivalent
deflection basins nor equivalent maximum deflections.
Two elastic layer theory computer runs were made to illustrate this
point. Figure 4.8 shows the layer thicknesses and material properties
used for the two runs. Equivalent subgrade moduli were calculated for
these runs based on 12-inch plate deflections measured in the test
pit. For the semi-infinite case, a modulus of 16,960 psi was obtained,
and for the finite layer case, a modulus of 14,530 psi was calculated
(See Table 4.1). The limerock modulus was chosen as 90,000 psi.
Figure 4.9 shows the predicted deflection basins for each of these
systems. As shown in the figure, neither the deflection basin nor the
maximum deflection are the same. This clearly shows that using an
equivalent subgrade modulus and Burmister's theory to determine the base
layer modulus is inadequate, since two totally different base moduli
would result for systems with and without a floor.
The computer program AXSYM models the floor and the walls of the
test pit, as well as rigid plate loading, but earlier analyses showed


255
unusual thermal response just described. Extremely high deflections
under load were observed at 0.0 C (32 F) at test positions 1 and 3 (see
Figure 8.24). In section 8.3.4, it will be shown that the high
deflections could only be attributed to an uplifted pavement, which
substantiates the observations made above. It is interesting to note
that relatively higher compressive strains were observed during cooling
at both test positions 1 and 3, until the time the strain pattern
changed. These observations did not help to better define the mechanism
that led to the observed thermal response, but it did reinforce the
concept that there was some correspondence between the thermal response
and the load response of the pavement.
8.3.3 Dynamic Load and Creep Response at Different Temperatures
For all dynamic load tests and creep tests performed, deflection
and strain measurements were taken at ten different points in the
pavement during load tests. Figures 8.25, 8.26, and 8.27 show the
positions of the LVDT's and strain gages monitored at test positions 1,
2, and 3, respectively. These instruments were positioned so that the
pavement deflection and strain distributions were adequately defined in
both the longitudinal and transverse directions.
Table 6.1 summarized the order in which dynamic load tests and
creep tests were performed. For the sake of clarity, the order of
testing was given in terms of nominal test temperatures in this table.
A summary of the measured pavement temperatures during testing is given
in Table 8.1. As shown in Table 6.1, dynamic load tests were performed
at all three test temperatures before any creep tests were performed.
This was done to eliminate any effect that creep might have had on the


Viscosity (Pa.Sec.)
123
Temperature
Figure 5.1: Viscosity Temperature Relationships for Asphalt
Recovered from the Test Pit


37
Hills and Brien equation. It uses a stress criteria of 200 psi. Shahin
and McCullough (92) developed a model to predict the amount of thermal
cracking. The model predicts temperatures, thermal stresses, low-
temperature cracking, and thermal fatigue cracking. They indicated that
comparisons of predicted cracking with measurements at the Ontario and
St. Anne Test Roads were reasonable. Lytton and Shanmugham (93)
developed a mechanistic model based on fracture mechanics to predict
thermal cracking of asphalt concrete pavements. The model assumes
cracking is initiated at the surface of the pavement and propagates
downward as temperature cycling occurs. The prediction of transverse
cracking and cracking temperature can also be attempted with the Haas
model (94) and the Asphalt Institute Procedure (95). Keyser and Ruth
(58) found absolutely no correlation between actual cracking in 6- to 9-
year old pavements in Quebec, and cracking predicted using the latter
two models.
Ruth et al. (43) developed a thermal cracking model based on their
stress equation. The model computes stresses, creep strains, and
applied creep energy, which are all used as failure parameters.
Predictions of cracking temperatures for a Pennsylvania D.Q.T. test road
were obtained which identified the two cracked sections in the test
road.
It should be noted that although indirect comparisons have been
made with observed field cracking, little if any measurements exist of
actual contraction and deformation of asphalt concrete pavements during
cooling.
Over the years, various solutions have been presented to predict
stresses, strains, and deflections within the pavement system due to


Tensile Strain (E-6ln/ln)
West
too
Bol tom of
pavement
100
Distance from
load center (Ins.)
x Measured
|B0 O Predicted
E, =490,000psi
E2= 25,000psi
E, = 25,000psl
200
280
Compressive Strain (E-6ln/in)
Figure 9.8: Comparison of Measured and Predicted Transverse Strain Distributions at 6.7 C (44 F)


LIST OF TABLES
Table Page
2.1 Primary Types and Causes of Distress In Asphalt
Concrete Pavements 7
2.2 Modes, Manifestations, and Mechanisms of Types of
Distress 8
4.1 Sand Subgrade Modulus for Different Layer Depths
and Poisson's Ratio 75
4.2 Effect of Concrete Floor on Surface Deflections
for Different Rase Stiffnesses 79
4.3 Predicted Deflections Using AXSYM 83
4.4 Tabulated Deflection Basins to Show Effect of
Test Pit Floor on Pavements of Different Stiffness 87
4.5 Effect of Test Pit Walls on Surface Deflections
for Pavements of Different Stiffness 90
4.6 Computer Runs to Determine Poisson's Ratio Effect 99
4.7 Measured and Predicted Surface Strains 109
5.1 Laboratory Test Results: Fairbanks Sand 114
5.2 Laboratory Test Results: Ocala Formation Limerock 116
5.3 Source of Materials and Job Mix Formula for Asphalt
Concrete 117
5.4 Test Pit Asphalt Concrete Properties 118
5.5 Rheology and Penetration of Asphalt Recovered From
Test Pit During Initial Placement: September, 1982 120
5.6 Rheology and Penetration of Asphalt Recovered From
Test Pit After All Testing: September, 1985 121
5.7 Load Increments Used for Plate Load Tests: Fairbanks
Sand Subgrade 127
5.8 Load Increments Used for Plate Load Tests:
Limerock Base 128
5.9 Modulus Values Immediately After Placement:
Fairbanks Sand Subgrade 134
5.10 Modulus Values Immediately After Placement:
Limerock Base 139
5.11 Modulus Values Without Accounting for Test Pit
Constraints 142
5.12 Modulus Values After Pavement Removal:
Fairbanks Sand Subgrade 145
5.13 Modulus Values After Pavement Removal: Limerock Base 146
6.1 Summary of Order of Testing 152
7.1 Summary of Dynamic Plate Load Tests at Ambient
Temperatures 158
vi


Table D.14: Measured Strains, Test Position 2, 13.3 C (56 F)
Load
Longitudinal
Strains (micro-strain)(a)
Transverse Strains
(micro-strain)
(kips)
Distance from Load
Center
(in.):
N-S(b>
Distance
from Load Center (in.
): E-W(c)
0
8(S)
12 (N)
16 (S)
24 (N!)
36 (S)
12 (W)
18(E)
24 (W)
36(E)
10
-274
-6
56
67
32
10
30
86
49
15
7
-216
-5
43
49
24
8
24
62
37
12
4
-136
-5
26
30
15
5
14
37
23
7
1
-36
-2
7
8
4
1
2
9
6
2
(a) Tension is positive.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.
389


MEASURED STRAINS DURING COOLING (E*6 IN./IN.)
I l l I I I' 9
251
Figure 8.23: Comparison of Measured Thermal Strains for Different
Cooling Cycles: 8.33 Feet from South Wall


Table E.34: Measured Creep Strains For Different Times of 10,000-lb. Static Load Application,
Position Number 3, 13.3 C (56 F)
Load^
Longitudinal
Strains
(micro-
strain)
Transverse Strains
(micro-strain)
Duration
Distance from
Load Center (in
.): N-S{ci
Distance
from Load Center (in.): E-W^)
(Seconds)
0
8(N)
12(S)
16(N)
24(S) 40(S)
12(W)
18(E)
24(W) 36(E)
50
8
-6
8
-2
-4
0
4
12
0 -2
50
11
-6
0 -
0
-3
0
0
5
0 -1
400
33
-2
-10
-2
-6
-2
0
11
0 -2
500
29
-2
-12
-3
-6
-2
2
10
0 -3
(a) Loads were applied sequentially with a rest period of four times the load duration between loads.
(b) Measured creep strains are the residual strains after the rest period. Tension is positive.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.


Tensile Strain (E-6In/In)
100
South
x 0.0 C(32F), E, = 750,000psi
O 6.7 C (44F), E, = 490,000psi
I3.3C(56F), E, = 320,000psl
NOTE! For all temperatures,
E2= 25,000 psi
E3= 25, OOOpsI
280
Compressive Strain (E-6ln/ln)
Figure 8.54: Comparison of Predicted Longitudinal Strains at Different Temperatures: 10,000 lbs.
291


250
Figure 8.22: Comparison of Measured Thermal Strains for Different
Cooling Cycles: Six Feet from South Wall


Table E.16: Measured Dynamic Strains at 10,000 lbs. After Different Times of Static Load Application,
Position Number 2, 6.7 C (44 F)
Load^
Longitudinal
Strains
(micro-
strain)^
Transverse Strains
(micro-strain)
Duration
Distance from
Load Center (in
.): N-S(c)
Distance
from Load Center (in.
): E-W(d)
(Seconds)
0
8 {S)
12 (N)
16 (S}
24(M) 36(S)
12(W)
18(E)
24 (W)
36(E)
0
-206
-17
38
46
29
12
15
56
37
13
50
-206
-16
39
47
30
11
17
57
36
12
100
-208
-15
38
48
28
12
17
57
37
13
500
-312
-13
37
49
29
12
19
59
38
14
1000
-232
-11
40
50
30
11
21
62
39
13
(a) A rest
period equal to four
times
the load
duration
was
allowed before dynamic
testing.
(b) Tension is positive.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.


207
expected. This behavior indicates that stresses were not effectively
transmitted away from the plate by the asphalt concrete layer.
A BISAR comDUter run was made with an asphalt concrete modulus of
90,000 psi, determined from Equation 5.1 and A.l for a temperature of
25.6 C (78 F) and the sand and limerock moduli found to give the best
prediction of measured response in previous analyses (15,420 psi and
53,000 psi, respectively). As expected, the measured strains were
grossly overpredicted by the program, as was the measured deflection
basin. The shapes of the predicted and measured deflection basins were
not even close. It was obvious that the measured response could not be
predicted using elastic layer theory, so additional attempts were not
made.
A major finding from these tests was to set an upper temperature
limit for pavement response evaluation with the rigid plate.
Apparently, for temperatures greater than about 21 C (70 F), excessive
plate identation can occur as a result of the high shear stresses at the
edge of the rigid plate, even at the fast loading rate. Therefore, the
plate should not be used at temperatures higher than 21 C (70 F).
Furthermore, rigid plate loading is not recommended for pavement
evaluation at any temperature because the analytical procedures required
to model rigid plate loading are extremely tedious and time consuming.
It should be noted that strain measurements were obtained at the
bottom of the asphalt concrete for the test performed at 25.6 C
(78 F). Unfortunately, the measurements obtained from these gages could
not be evaluated because of the uncharacteristic response that was
observed. However, the measurements seemed very reasonable and the
gages seemed to function well.


152
Table 6.1: Summary of Order of Testing
Test
Series
Order of
Testing
Position
1
Position
2
Position
3
Dynamic
Load
1
0
C
(32 F)
0
C (32 F)
0
C (32 F)
2
6.7
C
(44 F)
6.7
C (44 F)
6.7
C (44 F)
Tests
3
13.3
c
(56 F)
13.3
C (56 F)
13.3
C (56 F)
4
0
c
(32 F)


Creep
1
13.3
c
(56 F)
6.7
C (44 F)
0
C (32 F)
Tests
2
6.7
c
(44 F)
0
C (32 F)
6.7
C (44 F)
3
0
c
(32 F)
13.3
C (56 F)
13.3
C (56 F)
Notes:
(a) Temperatures shown are nominal test temperatures. The measured
averaae pavement temperatures for each test series are given in
Table 8.2.
(b) All dynamic load tests and creep test were completed at one
position before going to the next. The positions were tested in
the following order: Position 3, Position 2, Position 1.


66
- the wheels were easier to position than the plate, since they
did not require setting with hydrocal.
B. Disadvantages
- the load did not come on instantaneously since the wheels had
to deform before a load was applied. Furthermore, the time
required for the load to come on, depended on load
magnitude. Therefore, it was difficult to set the loading
time and to evaluate creep strain accumulation for dynamic
loading conditions. For this reason, all creep tests were
performed using static loads.
- up to 2 1/2 inches of ram movement was required for loading,
which caused greater wear and tear in the loading system;
- the loading area varied with load and was not circular (more
difficult to model); and
- the wheels tended to roll during loading so that the exact
position of load was not known.
It v/as evident that neither of the loading devices was perfect, but
the disadvantages associated with the plate were overwhelming. The
loading was not representative of wheel loads and the analysis procedure
required tremendous amounts of time. Therefore, the dual wheel loading
system was used for the majority of tests performed.


156
until the bottom of the pavement was at or slightly below the test
temperature. Then the cooling unit was turned off and the temperature
of the top and bottom of the asphalt concrete layer were allowed to
stabilize. To warm up to a given test temperature from a cooler
temperature, the pavement temperature was simply allowed to increase
slowly with the insulated cover in place.
Once the pavement was at a specified test temperature, creep tests
were performed. Immediately prior to performing any creep tests,
dynamic load tests at 10,000 lbs. were performed, and deflections and
strains were recorded for several successive repetitions of dynamic
load. These initial measurements could then be compared to dynamic load
measurements after different durations of static load applications to
determine the effect of creep on the dynamic load response of the
pavement. In addition, these initial measurements provided a reference
from which permanent deflections and creep strains could be monitored.
Creep tests were then performed by apnl.ying different durations of
10,000-lb. static load. Load durations of 50, 50, 400, and 500 seconds
were applied sequentially with a rest period between loads equal to four
times the load duration. Dynamic load tests at 10,000 lbs. were
performed after each rest period, and deflections and strains were
recorded for several repetitions of dynamic load for each series of
tests. Permanent deflections and creep strains could then be determined
for any given duration of static load application by comparing the
deflections and strains recorded during dynamic load tests performed at
different times. The difference between the unloaded values of
deflections or strains measured at different times was the permanent
deflection or creep strain accumulated during that time.


438
57. Schmidt, R. J., "Use of ASTM Tests to Predict Low Temperature
Stiffness of Asphalt Mixes," Transportation Research Record
Mo. 544. Transportation Research Board, 19/5, pp. 35-45.
58. Keyser, J. H., and Ruth, B. E., "Comparison of the Sensitivity of
Asphalt Concrete Mixture Strength Tests to Changes in Asphalt
Binder Properties," Proceedings, Association of Asphalt
Paving Technologists, Vol. 53, pp. 583-617 (1984).
59. Schweyer, H. E., Smith, L. L., and Fish, G. W., "A Constant Stress
Rheometer for Asphalt Cements Rheological Background,"
Special Technical Publication Mo. 628, American Society for
Testing and Materials, 1977, pp. 5-42.
60. Schweyer, H. E., and Kafka, F. Y., "Constant Stress Rheology of
Asphalt Cements," Reprint from Industrial and Engineering
Chemical Fundamentals, Vol. 15, pp. 138-144 (1976).
61. Ruth, B. E., and Schweyer, H. E., "Asphalt and Asphalt Mixture
Rheology as Related to Cracking of Pavements," Proceedings,
Eighteenth Paving Conference, Civil Engineering Department,
The University of New Mexico, 1981.
62. Miller, J. S., Uzan, J., and Witczak, M. W., "Modification of the
Asphalt Institute Bituminous Mix Modulus Predictive
Equation," Transportation Research Record No. 911,
Transportation Research Board, 1983, pp. 27-36.
63. "Structural Design of Asphalt Concrete Pavements," Special Report
No. 126, Highway Research Board, 1971.
64. Gonzalez, G., Kennedy, T. W., and Angnos, J. N., "Evaluation of
the Resilient Elastic Characteristics of Asphalt Mixtures
Using the Indirect Tensile Test," Research Report Mo. 183-6,
Center for Highway Research, The University of Texas at
Austin, 1975, pp. 1-71.
65. Kennedy, T. W., "Characterization of Asphalt Pavement Materials
Using the Indirect Tensile Test," Proceedings, Association of
Asphalt Paving Technologists, Vol.46, pp. 132-150 (1977).
66. Puyana, E., "Characterization of Asphalt Concrete Pavement
Materials," Unpublished Master's Thesis, Department of Civil
Engineering, University of Florida, 1983.
67. Cragg, R. and Pell, P. S., "The Dynamic Stiffness of Bituminous
Road Materials," Proceedings, Association of Asphalt Paving
Material s, Vol. 40, pp. 126-147 (1971).
68. Kallas, B. F., and Riley, J. C., "Mechanical Properties of Asphalt
Pavement Materials," Proceedings, Second International
Conference on the Structural Design of Asphalt Pavements,
1967, pp. 931-952.


260
dynamic response of the pavement. Dynamic load tests were performed
sequentially at 0.0 C (32 F), 6.7 C (44 F), and 13.3 C (56 F) at all
three test positions, mainly as a matter of convenience. By performing
tests in this order, the pavement was cooled only once to 0.0 C (32 F)
and tests at the higher temperatures were performed as the pavement
warmed up. Creep tests were performed in a different sequence at each
test position in order to observe the effect of creep at one temperature
on the dynamic load response and creep response at another temperature.
Deflections and strains measured during dynamic load tests at each
of the three test positions are given in Appendix D. At all test
temperatures, load tests were performed at levels of 1000, 4000, 7000,
and 10,000 lbs. The actual load-unload times used for dynamic testing
are shown in Figure 8.28. For all load levels, the maximum load was
left on for 0.1 sec., hut the time to reach the maximum load increased
as the maximum load increased. This was because a greater amount of
wheel deformation was required to reach higher loads. The rest period
was set at four times the total load time at the 10,000-lb. load level,
since Ruth and Maxfield (42) found that in laboratory samples all
delayed elastic response was recovered after a rest period equal to four
times the loading time.
Data collected during creep tests at each test position are
presented in Appendix E. For all test temperatures, permanent
deflections and creep strains were measured after different durations of
10,000-lb. static load applications. Load durations of 50, 50, 400, and
500 seconds were applied sequentially with a rest period between loads,
equal to four times the load duration. The permanent deflections and
creep strains measured were the residual deflections and strains after


320
(32 F), respectively. The deflection basins shown were measured after
all creep tests were performed at the respective temperatures. Although
the initial response of the pavement was very different for different
positions, Figures 8.70 and 8.71 show that the response after creep was
almost identical at all test positions for the two higher tests
temperatures. Figure 8.72 shows there was some discrepancy at 0.0 C
(32 F) but the response was much closer than it was for initial dynamic
load tests at this temperature. The reason for the discrepancy at this
lower temperature is that permanent deflections and creep strains
induced at 0.0 C (32 F) were not enough to seat the pavement completely
at test position 3, where creep tests were performed at 0.0 C (32 F)
first.
It is also interesting to note that the deflection basin approached
at all test positions, had the expected pattern with respect to
temperature. That is, the measured deflections decreased slightly as
the temperature decreased. The same patterns were also observed for the
measured strain distributions.
8.3.4 Combined Effect of Thermal and Load Response
The measured thermal response, dynamic load response, and creep
response of the pavement described earlier in this chapter appeared to
be related. All three sets of measurements showed unusual response
characteristics. The thermal response of the pavement, as determined by
the strains measured during initial cooling cycles, was much different
from the response expected for a pavement under normal conditions.
Observed changes in strain distributions with time indicated that
continued cooling of the pavement resulted in some combination of


SEMI-INFINITE
(a)
Figure 4.8: Equivalent Systems Based on Maximum Plate Deflection on Subgrade


276
load was increased. Apparently, the clamping system could not prevent
the reaction beam from moving longitudinally under repeated 10,000-lb.
loads. Also, the reaction beam was not stiff enough in torsion to
prevent wheel rotation. Figure 8.29 shows the center of loading for
each load level and how this center migrated southward as the load was
increased. This figure clearly illustrates that the wheels were rolling
as the reaction beam rotated in torsion with increasing load. Figure
8.31 shows that for this case, the wheels also moved slightly to the
east as the load was increased.
These problems with the loading system introduced additional
uncertainty in the data, since the exact center of loading was not
known. This uncertainty resulted in very tedious data interpretation
procedures. Therefore, if at all possible, the loading system should be
improved by either stiffening the existing system or by installing a new
system with a dual reaction beam. Also, the clamping system should be
improved so that the loading ram can be positioned more easily.
Because the wheels moved during dynamic load tests, the center of
loading was different for different sets of deflection and strain
measurements (see Figures 8.29 to 8.40). This made it difficult to
compare measurements taken at different temperatures, positions, etc.
Therefore, after the center of loadings were determined for each set of
data, the deflection basins and strain distributions were shifted so
that the center of loading coincided with the center of the test
position. These shifted drawings could then be easily compared to each
other. All deflection basins and strain distributions presented after
Figure 8.40 were shifted in this way.


CHAPTER VIII
LOW-TEMPERATURE
PAVEMENT RESPONSE
8.1 Preliminary Tests With the Rigid Plate
Once the cooling system was installed and operational, the test pit
pavement was subjected to several cooling cycles to evaluate the
performance of the cooling unit and the instrumentation system at low
temperatures. In addition, dynamic load tests were performed in an
attempt to evaluate pavement system response at low temperatures. The
rigid plate loading system was used because the dual wheels were not yet
available. These initial tests led to several findings concerning the
cooling unit capability, problems and deficiencies with the
instrumentation and loading systems at cold temperatures, and a low-
temperature limit for pavement evaluation.
The cooling system capability was determined by allowing the
cooling unit to run until the pavement temperature would not go any
lower. A temperature of -6.7 C (20.0 F) was achieved at the bottom of
the 4 1/8-inch asphalt concrete layer with a pavement surface temper
ature of -17.8 C (0.0 F). Measured cooling rates are presented in
conjunction with pavement response measurements in a later section.
During the first cooling cycle, an attempt was made to measure the
vertical movement at different points on the pavement surface. The
LVDT's were individually insulated with a loosely-fitting, cylindrically
shaped piece of 1/2-inch rubber foam insulation and the space between
the LVDT and the rubber foam was filled with expanding foam insulation.
216


371
Table C.
2: Measured Thermal
Strains During
Cooling:
Test
Position
2
Gage Humber
Time From Start
of Cooling (Hours/Minutes)
Thermocouple
0
1
2
1
4
Humber
0
30
0
30
0 30
0
30
0
0
Reading
Temperature (F)
0
61.5
-30
54.0
-40
45.3
-47
40.6
-61 -74
37.6 35.1
-8
32.5
-mo
30.7
-115
28.9
Adjusted Strain
0
-25
-29
-32
-43 -53
-63
-/3
-86
1
Reading
Temperature (F)
0
61.3
-36
48.1
-58
39.3
-75
34.5
-96 -115
31.5 29.1
-132
26.9
-145
25.3
-156
24.4
Adjusted Strain
0
-27
-41
-54
-70 -86
-100
-no
-120
2
Reading
Temperature (F)
0
61.5
-33
54.0
-65
45.3
-83
40.6
-103 -118
37.6 35.1
-131
32.5
-143
30.7
-153
28.9
Adjusted Strain
0
-28
-54
-68
-85 -97
-106
-116
-124
3
Reading
Temperature (F)
0
61.2
-48
45.9
-105
33.3
-134
28.4
-166 -190
25.3 23.2
-213
21.2
-230
19.9
-245
19.9
Adjusted Strain
0
-38
-81
-104
-131 -152
-172
-186
-201
4
Reading
Temperature (F)
0
63.3
-24
54.7
-51
45.9
-75
41.5
-101 -120
38.5 36.1
-133
31.8
-152
31.6
-166
.29.8
Adjusted Strain
0
-20
-41
-61
-84 -100
-115
-126
-138
5
Reading
Temperature (F)
0
61.2
-12
45.9
-34
33.3
-61
28.4
-87 -108
25.3 23.2
-12'
21.2
19.9
-153
19.9
Adjusted Strain
0
-2
-10
-31
-52 -70
-85
-96
-109
6
Reading
Temperature (F)
0
62.0
029
54.5
-51
45.4
-71
40.5
-95 -114
37.0 34.5
-132
29.6
-146
30.2
-159
28.5
Adjusted Strain
0
-24
-40
-56
-76 -92
-104
-118
-129
7
Reading
Temperature (F)
0
62.1
-50
52.1
-105
41.8
-146
37.0
-184 -214
33.7 31.3
-239
29.0
-261
27.3
-277
26.0
Adjusted Strain
0
-44
-91
-127
-161 -188
-210
-229
-243
8
Reading
Temperature (F)
0
62.4
-34
55.0
-88
45.5
-126
40.3
-162 -189
36.5 34.0
-213
31.5
-233
29.7
-250
28.0
Adjusted Strain
0
-30
-77
-110
-142 -166
-187
-205
-219
9
Reading
Temperature (F)
0
62.6
-71
50.2
-221
38.3
-320
33.3
-396 -448
29.8 27.5
-490
25.5
-523
23.9
-547
23.0
Adjusted Strain
0
-64
-203
-296
-368 -417
-456
-486
-509
Motes:
(a) Strains 1n micro-strain. Compression Is negative.
(b) See Figure 8.4 for gage and thermocouple location.
(c) Strain adjusted using temperature-Induced apparent strain relationship furnished by strain gage
manufacturer.
(d) Defrost cycle came on just before 6 hours, therefore, these readings are not typical.


Table E.8: Measured Dynamic Strains at 10,000 lbs. After Different Times of Static Load Application,
Position Number 1, 6.7 C (44 F)
Load^
Longitudinal
Strains
(micro-
strain)^)
Transverse Strains
(micro-strain)
Duration
Distance from
Load Center (in
.): N-S{c)
Distance
from Load Center (in.
): E-W(d)
(Seconds)
0BT(e) 0BL(f)
8(N)
12 (S)
24(S) 36(N)
12(E)
18(H)
24(E)
36 (W)
0
-82 144
11
30
25
8
14
54
28
12
50
-80 142
11
30
26
8
15
53
28
13
100
-82 134
10
28
26
9
16
50
28
13
500
-84 138
12
29
26
8
16
53
28
12
1000
-90
13
31
27
9
19
54
27
14
(a) A rest
period equal to four
times
the load
duration
was
allowed before dynamic
testing.
(b) Tension
is positive.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.
(e) This gage located at the center of load, at the bottom of pavement, in the transverse direction.
(f) This gage located at the center of load, at the bottom of pavement in the longitudinal direction.


APPENDIX B
PAVEMENT TEMPERATURES DURING COOLING


73
was not considered a major problem and the effects could be defined and
approximated using this model.
Two other programs were also investigated; YBFE1 and SAPIV. YBFE1
is a two-dimensional (plane strain) finite element program developed for
soil-structure interaction problems. Preliminary program runs using
YBFE1 revealed that this program was unsuitable for predicting flexible
pavement response. The error was probably introduced by the plane-
strain nature of the model and the type of finite element used. SAPIV
is a three-dimensional finite element program developed for structural
dynamics but can be used to model homogeneous masses by means of a brick
or plate element. This program would have been most accurate in
modeling the test pit, but preliminary attempts at running the program
showed that an excessive amount of computer space was needed. This
space was not available on the current version of SAPIV at the
University of Florida. In addition, the cost of running the program was
prohibitive for the purposes of this project.
4.3.2 Effect of Constraints on Subgrade Response
Closed form solutions are available for the vertical displacement
of a rigid circle on both a semi-infinite mass and on a finite layer.
The following equations may be used to calculate these displacements:
Semi-infinite:
Ip Pavg(a)
pz = U-2) Pavgjal
Finite Layer:
where
' z F
p vertical displacement of rigid circle (ins.)
y Poisson's ratio
Pavg average pressure on the rigid circle (psi)


161
.
8'
>
AXIS OF SYMMETRY
>'
Figure 7.2: Test Pit Diagram: Plan


199
The deflection basin comparisons show that at all load levels, the
deflection basin at 20.6 C (69 F) was very similar in shape to the one
at 18.3 C (65 F), but was shifted slightly upward (i.e., slightly lower
deflections were measured at the higher temperature). Comparisons of
the strain distributions at the same two temperatures show that the
measured strains were significantly higher at 20.6 C (69 F) than at
18.3 C (65 F) for all load levels. Assuming that foundation conditions
were the same for both tests, the observed difference in strains was as
expected, while the observed deflection difference was opposite of what
was expected. However, a period of one year elapsed between the two
tests and the limerock modulus may have changed with time or moisture
changes. Furthermore, this position was loaded extensively at fast and
slow loading rates prior to testing at 20.6 C (69 F) which may have
densified one or all three of the pavement layers.
BISAR was used to predict the measured deflections and strains at
20.6 C (69 F) and thereby deduce changes in the individual pavement
layer properties which would explain the observed differences in
response between the initial tests at 18.3 C (65 F) and the tests
performed one year later at 20.6 C (69 F). This analysis would also
serve to evaluate the asphalt concrete dynamic modulus prediction
equations (Appendix A) at a higher temperature.
The procedure outlined in Section A.4.3.1 was used along with the
rigid plate approximation procedure developed and described in Section
4.4.2 to predict pavement response. The pavement system was again
modeled as shown in Figure 7.7. An asphalt concrete modulus of
124,400 psi was calculated for use in the program from Equations 5.1 and
A.l for a temperture of 20.6 C (69 F). Different values of limerock


Table E.l: Measured Permanent Deflections For Different Times of 10,000-lb. Static Load Application,
Position Number 1, 13.3 C (56 F)
Load^
Longitudinal Deflections
(E-3 in.)(b>
Transverse Deflections (E-3 in.)
Duration
Distance
from Load
Center
(in.): N-S^
Distance
from Load
Center (in.): E-W^)
(Seconds)
0
8(M)
12 (S)
18 (S)
24(N) 36(M)
13(H)
18(E)
24(W) 36(E)
50
3.67
0.71
0.1
-0.24
-0.24 0.0
3.58
1.79
0.86 0.1
50
2.11
0.38
0.0
-0.39
-0.24 0.0
1.91
0.25
0.07 0.1
400
1.66
-0.28
0.25
0.14
0.0 0.0
1.96
0.30
0.30 0.1
500
1.76
0.0
0.43
-0.48
-0.58 0.0
1.53
0.Q4
0.25 0.2
(a) Loads were applied sequentially with a rest period of four times the load duration between loads.
(b) Measured permanent deflections are the residual deflections after the rest period. Positive is
downward.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.


437
45. Johnson, T. C., Cole, D. M., and Irwin, L. H., "Characterization
of Freeze/Thaw-Affected Granular Soils For Pavement
Evaluation," Proceedings, Fifth International Conference on
the Structural Design of Asphalt Pavements, 1982, pp. 805-
817:
46. Molenaar, A. A. A., "Fatigue and Reflection Cracking Due to
Traffic Loads," Proceedings, Association of Asphalt Paving
Techno!ogists, Vol. 53, pp. 440-474 (1984).
47. Tia, M.j "Characterization of Cold-Recycled Asphalt Mixtures,"
Joint Highway Research Project, Engineering Experiment
Station, Purdue University, 1982.
48. Schweyer, H. E., "Asphalt Rheology in the Mear Transition
Temperature Range," Highway Research Record No. 628, Highway
Research Board, 1973, pp. 1-15.
49. Jongepier, R., and Kuilman, B., "Characteristics of the Rheology
of Bitumens," Proceedings, Association of Asphalt Paving
Technologists, 'V'ol. 38, pp.~9'8-l'22 (1969) "
50. Schweyer, H. E., and Burns, A. M., "Low Temperature Rheology of
Asphalt Cements, III. Generalized Stiffness-Temperature
Relations of Different Asphalts," Proceedings, Association of
Asphalt Paving Technologists, Vol. 47, pp. 1-18 (1978).
51. Monismith, C. L., "Viscoelastic Behavior of Asphalt Concrete
Pavements," Proceedings, First International Conference on
the Structural Design of Asphalt Pavements, 1962.
52. Duthie, J. L., "Proposed Bitumen Specifications Derived from
Fundamental Parameters," Proceedings, Association of Asphalt
Paving Technologists, Vol. 41, pp. 70-117 (1972).
53. Schweyer, H. E., "A Pictorial Review of Asphalt (Bitumen)
Rheology," Proceedings, Association of Asphalt Paving
Technologists, Vol. 43A, pp. 121-157 (1974).
54. Pink, H. S., Mery, R. E., and Bosniack, D. S., "Determination of
Dynamic Modul i at Low Temperatures," Proceedings, Association
of Asphalt Paving Technologists, Vol. 49, pp. 64-94 (1980).
55. Schweyer, H. E., Baxley, R. L., and Burns, A. M., "Low-Temperature
Rheology of Asphalt Cements-Rheological Background," Special
Technical Publication Mo. 628, American Society for Testing
and Materials, 1976, pp. 5-42.
56. Puzinauskas, V. P., "Properties of Asphalt Cements," Proceedings,
Association of Asphalt Paving Technologists, Vol. 48,
pioram


36"
24" O
18"
O
24" 18"
24"
PAVEMENT: ONE
LONGITUDINAL AMD
ONE TRANSVERSE
3'. O"
36" O

X LVDT AND STRAIN GAGE
O LVDT ONLY
STRAIN GAGE ONLY
Jfahi
SCALE:
NOTE: DIMENSION SHOWN
INDICATES DISTANCE
FROM CENTER CF
TEST POSITION
Figure 8.25: LVDT and Strain Gage Location During Load Tests: Test Position 1


MEASURED STRAINS DURING COOLING (E-6 IN./IN.)
i l 1
246
Figure 8.20
Transverse Strain Distributions During Cooling:
Test Position 2


322
stresses in the pavement, temperature changes were having another,
possibly more significant effect on the pavement. Furthermore, the
effect was detrimental, since it resulted in a weaker pavement system,
as evidenced by the high deflections and strains observed under load.
However, it was still unclear what changes occurred in the pavement
system to cause the observed response. Therefore, the measured thermal
response and load response of the pavement were used to formulate a
hypothesis to explain this phenomenon.
There are two possible ways in which a reduction in temperature
resulted in a weaker pavement system: 1) by weakening the base or
subgrade; or 2) by causing the asphalt concrete layer to deform in such
a way that it separated from the base. Of course, it is also possible
that both these effects occurred at the same time. However, based on
the observed response it seems like the second possibility occurred in
the test pi t.
Although it is possible that a temperature reduction at the surface
of the pavement could weaken or reduce the modulus of the foundation
materials, the measured response indicated that this was not the case.
There are two ways in which a reduction in the modulus can occur in the
foundation materials: 1) an increase in moisture in the limerock base;
or 2) a decrease in capillary tension in either the sand subqrade or the
limerock base. Either of these effects could occur as a result of
upward moisture migration due to the thermal gradient created in the
pavement system during cooling. However, even if these changes did
occur, their effect would probably not be enough to explain the
magnitude of the deflections and strains measured at lower
temperatures. These changes would have occurred only in the top two


54
a) Frontal View
b) Diagonal View
Figure 3.8: Test Pit Pavement Completely Instrumented and Ready
for Testing With Dual Wheels


Table D.10: Measured Strains, Test Position 2, 0 C (32 F)
Load
(kips)
Longitudinal
Strains (micro-strain)
(a)
Transverse Strains
(micro-
strain)
Distance from Load
Center i
(in.): N-
Sib)
Distance
from Load Center (in.): E-W^0^
0
8(S)
12(N)
16 (S)
24 (N)
36 (S)
12(W)
18(E)
24 (W)
36(E)
10
-185
-25
5
48
20
14
-3
44
34
11
7
-138
-21
8
35
15
10
-3
35
24
9
4
-84
-14
5
20
9
7
-2
11
15
5
1
-20
-4
.5
4
2
2
-1
5
4
2
(a) Tension is positive.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.


Figure 3.13: Typical Deflection Output on X-Y Plotter


DISTANCE FROM CENTER OF LOAD (INS.)
Figure 7.12: Deflection Basin Comparison for Fast and Slow Loading Rates: 10,000 lbs.


Table D.12: Measured Strains, Test Position 2, 6.7 C (44 F)
Load
Longitudinal
Strains (micro-strain)
(a)
Transverse Strains
(micro-
strain)
(kips)
Distance from Load
Center
(in.): N-
s(b)
Distance
from Load Center (in.): E-W^
0
8(S)
12 (N)
16 (S)
24 (N)
36 (S)
12(W)
18(E)
24(W)
36(E)
10
-231
-22
51
55
31
14
16
65
43
15
7
-182
-18
38
40
25
12
14
50
34
11
4
-110
-14
23
24
14
6
7
28
20
6
1
-26
-4
5
5
4
1
.5
7
5
2
(a) Tension is positive.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.


15
Their idea is to design the pavement for a critical condition based on
material properties, loads, and environment. They developed a pavement
analysis model that considers cracking as a result of asphalt properties
(including age hardening), vehicular loads, pavement cooling rate and
temperature. The analysis program was used to evaluate the effect of
different asphalt viscosities, cooling rates, and pavement thicknesses
on pavement performance. Predictions of cracking temperatures for a
Pennsylvania DOT test road were obtained which identified the two
cracked sections in the test road. Analysis of typical highways in
Florida indicated that some pavements may give marginal performance,
which was indirectly substantiated by observed early cracking of pave
ments, particularly those located in northern Florida.
2.3 Properties of Asphalt Cement and Asphalt Concrete As
Related To Low-Temperature Pavement Response and Cracking
The response and failure of asphalt concrete pavements have been
shown to be highly dependent on the properties of the asphalt cement.
Thus, proper characterization of asphaltic materials is extremely
important. The characterization of bituminous materials for use in
conventional design methods is based mostly on empirical procedures
which rely on correlations of their results with field performance. The
Marshall and Hveem Stabilometer tests are most commonly used for this
purpose (27). These tests are performed at high temperatures and relate
mainly to the problems of stability, workability, and durability. Fun
damental properties cannot be obtained directly from these tests.
Several researchers have attempted to correlate Marshal 1 results with
fundamental properties (47), but it will be pointed out later that such


Table E.2: Measured Creep Strains For Different Times of 10,000-lb. Static Load Application,
Position Number 1, 13.3 C (56 F)
Load^
Longitudinal
Strains
(micro-
strain)^)
Transverse Strains
(micro-strain)
Duration
Distance from
Load Center (in
.): N-S(c)
Distance
from Load Center (in.
): E-W(d)
(Seconds)
0BT(e)
0BL(f)
8 (N)
12 {S)
24(S) 36(N)
12(E)
18 (W)
24(E)
36 (W)
50
-212
148
48
60
-6 -5
42
47
-2
-2
50
-141
119
32
35
-6 -4
38
28
-4
-2
400
-342
254
-2
-11
-55 -41
39
-18
-61
-44
500
-104
122
9
1
-14 -3
46
-6
-23
-7
(a) Loads were applied sequentially with a rest period of four times the load duration between loads.
(b) Measured creep strains are the residual strains after the rest period. Tension is positive.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.
(e) This gage located at the center of load, at the bottom of pavement, in the transverse direction.
(f) This gage located at the center of load, at the bottom of pavement in the longitudinal direction.


85
4.3.4 Effect of Constraints on Three-Layer System
A series of runs was made with the elastic layer computer program
to evaluate the effect of the concrete floor on the pavement system
response. Three asphalt concrete moduli were used: 145,000 psi (room
temperature), 1,500,000 psi (very cold pavement), and 3,500,000 psi
(Portland cement concrete). Two runs were made for each of these
pavement systems: one with the floor at a depth of 48 in. and the other
with a semi-infinite subgrade layer. Figure 4.6 shows the layer
thicknesses and properties used for each system.
The predicted deflection basins for each of these pavement systems
are tabulated in Table 4.4. The deflection differences resulting from
the presence of the floor are also listed in the table. For all cases,
the effect of the floor is to shift the deflection basin upward (i.e.
the deflections decrease by a uniform amount). In addition, the
decrease in deflection is relatively independent of the stiffness of the
asphalt concrete layer. The magnitude of the decrease is approximately
3.5 E-3 in. for the pavement at room temperature and 3.2 E-3 in. for the
Portland cement concrete pavement. This decrease, of course, is for a
particular set of support conditions and will be different for another
set. The magnitude of the decrease would also change with magnitude of
load and load configuration (e.g. 16-inch plate vs. 12-inch plate or
dual vs. single load). The main point is that the effect of the floor
in the test pit is significant and must be accounted for, whether by
estimating the magnitude or preferably by modeling the floor with
appropriate boundary conditions.
The test pit wall effect on the three-layer system was determined
using the AXSYM computer program. Again three pavement modulus values


PREDICTED DEFLECTIONS
5E-:
10E:
15E-:
DISTANCE FROM LOAD CENTER (INS.)
oo
-p>
Figure 4.5: Comparison of AXSYM and Elastic Layer Theory Solutions


Table E.18:
Measured Creep Strains
Position Humber 2, 0 C
For Different Times of 10
(32 F)
,000-lb. Static
Load Application,
Load^3^
Longitudinal
Strains (micro-strain)^
Transverse
Strains
(micro-strain)
Duration
Distance from
Load
Center (in.): H-S^
Distance from
Load Center (in.):
E-W(c0
(Seconds)
0
8 (S)
12(H)
16(S) 24(N) 36(S)
12(W) 18(E)
24(W)
36(E)
50
-18
-22
-11
23 -0.4 12
23.5
4
14
9.5
50
8
-19
2
14 -3 8
14.5
6
10.5
9
400
35
-41
30
32 -17 28
42
48
27.5
-29
500
40
-26
-46
20.5 -23 13.5
22
32
13

(a) Loads were applied sequentially with a rest period of four times the load duration between loads.
(b) Measured creep strains are the residual strains after the rest period. Tension is positive.
(c) H is for north, S is for south of load center.
(d) E is for east, W is for west of load center.


12
failure (27-31). Fatigue distress is the phenomenon of fracture under
repeated stresses which are less than the tensile strength of the
material. Fatigue characterization of materials has been studied
extensively and there are innumerable references on this topic (e.g. 31-
39).
The philosophy behind the approach to the analysis and design of
asphalt concrete pavements considered in this thesis is totally
different from conventional approaches based on fatigue. In fact, the
fatigue concept is considered erroneous, and will not be covered in much
detail. Design procedures based on fatigue assume that there is some
average pavement condition for which an equivalent amount of damage will
be incurred under each passing wheel load. These procedures neglect
that deflections, strains, and stresses cover a wide spectrum of values
dependent on temperature and climatic fluctuations. They cannot
properly account for the variable properties of individual asphalts at
low temperatures.
Several researchers have proposed modifications to fatigue life
predictions based on temperature, recognizing that the fatigue life of
materials tested in the laboratory is dependent on temperature.
However, the basic concept of fatigue damage has remained unchanged.
Pxauhut and Kennedy (40) proposed one such modification and discuss
modifications proposed by other researchers. They also recognize that
fatigue damage is difficult to evaluate since there is limited knowledge
as to fatigue life relations for real pavements, reliable test data
exists for only a limited number of mixtures, and there is insufficient
information to define how fatigue life varies with temperature and


239
A close examination of the strain-temperature plots show that there
were three distinct changes in the thermal response characteristics of
the pavement during cooling. Initially, the rate of contraction with
respect to temperature was very low for all surface gages. The reason
for this was that the thermal stresses developed in the asphalt concrete
were not yet high enough to overcome the frictional resistance between
the asphalt concrete layer and the underlying base. This was mainly
because the asphalt viscosity near room temperature is relatively low,
so that the asphalt concrete has a high capacity to relax stresses
induced thermally or otherwise.
An increase in contraction rate occurred for all surface gages
after about two hours of cooling. This increase in rate appears as a
distinct change in slope in the strain-temperature plots (Figures 8.10
to 8.15). This break was observed in both the longitudinal and
transverse directions. It is interesting to note that the break
occurred at the same time, but at different temperatures, for gages
located at different positions on the pavement. Therefore, it seems
like the change in thermal response was related more to the overall
behavior of the pavement than to the response characteristics of the
asphalt concrete itself. Apparently, after about two hours of cooling,
the thermal stresses induced in the pavement were enough to overcome
frictional resistance and thus initiate a significant amount of movement
or contraction in the pavement. Measured cooling curves (Figures 8.6 to
8.8) and plots of thermal gradient with time (Figure 8.9) indicated that
this change occurred when the average pavement temperature was about
eight to ten degrees C (46.4 to 50.0 F) and when the temperature
gradient reached a maximum.


434
11. Anderson, K. 0., and Hahn, W. P., "Design and Evaluation of
Asphalt Concrete With ResDect to Thermal Cracking,"
Proceedings, Association of Asphalt Paving Technologists,
V17 3/, pp. 1-31 (1968).
12. Burgess, R. A., Kopvillem, 0., and Young, F. D., "Ste. Anne Test
Road-Relationships Between Predicted Fracture Temperatures
and Low Temperature Field Performance," Proceedings,
Association of Asphalt Paving Technologists, Vol. 40,
ppl-148-170 (1571).
13. Haas, R. C. G., and Topper, T. H., "Thermal Fracture Phenomena in
Bituminous Surfaces," Special Report Mo. 101, Highway
Research Board, 1969, pp. 136-153.
14. Ruth, B. E., "Prediction of Low-Temperature Creep and Thermal
Strain in Asphalt Concrete Pavements," Special Technical
Publication No. 623, American Society for Testing and
Materials, 1977, pp. 68-83.
15. Ruth, B. E., Schweyer, H. E., Davis, A. S., and Maxfield, J. 0.,
"Asphalt Viscosity: An Indicator of Low Temperature Fracture
Strain in Asphalt Mixtures," Proceedings, Association of
Asphalt Paving Technologists, Vol. 48, pp. 221-23? (1979).
16. Fromm, H. J., and Phang, W. A., "A Study of Transverse Cracking of
Bituminous Pavements," Proceedings, Association of Asphalt
Paving Technologists, Vol. 41, pp. 383-418 (1972).
17. McLeod, M. W., "A Four-Year Survey of Low-Temperature Transverse
Pavement Cracking on Three Ontario Tests Roads," Proceedings,
Association of Asphalt Paving Technologists, Vol. 41,
pp": 424-468 (1972).
18. Fromm, H. J., and Phang, W. A., "Temperature Susceptibility
Control in Asphalt Cement Specifications," Highway Research
Record Mo. 350, Highway Research Board, 1971, pp. 30-45.
19. Gaw, W. J., "Measurement and Prediction of Asphalt Stiffnesses and
Their Use in Developing Specifications to Control Low-
Temperature Pavement Transverse Cracking," Special Technical
Pub!ication Mo. 628, American Society for Testing and
Materials, 1977, pp. 57-67.
20. Kandhal P. S., "Low Temperature Shrinkage Cracking of Pavements
in Pennsylvania," Proceedings, Association of Asphalt Paving
Technologists, Vol. 47, pp. 73-98 (1978).
21. Hignell, E. T., Hajik, 0. J., and Haas, R. C. J., "Modifications
of Temperature Susceptibilities of Certain Asphalt
Concretes," Proceedings, Association of Asphalt Paving
Technologists, Vol. 4l, pp. 524-553 (19/2). ~


325
Figure 8.73: Comparison of Load-Deflection Relationships for
Different Test Positions: 0.0 C (32 F)


370
Table
C.l-
extended
5
6
8
li
17
20
0
0
30
0
0
0
-279
47.8
-306
45.0
-350
40.3
-380
36.3
-383
32.0
-204
33.1
-270
-295
-334
-360
-358
-180
-225
47.8
-354
45.0
-397
40.3
-428
36.3
-461
32.0
-289
33.1
-216
-343
-381
-408
-436
-265
-281
29.7
-319
27.4
-373
23.3
-420
19.1
-449
15.5
-297
33.3
-253
-287
-335
-375
-397
-273
-141
28.2
-166
26.1
-201
22.8
-227
18.9
-289
14.9
-177
32.4
-111
-132
-162
-181
-236
-152
-171
31.1
-197
28.6
-253
23.7
-283
19.4
-281
16.2
-162
34.2
-145
-167
-216
-238
-230
-140
413
31.1
-446
28.6
-500
23.7
-536
19.4
-544
16.2
-284
34.2
-387
-416
-463
491
493
-242
-146
23.1
-175
20.7
-219
18.1
-257
15.0
-403
11.7
-353
31.2
-103
-133
-172
-204
-343
-327
-292
26.3
-328
23.8
-384
20.0
-422
16.1
-396
13.3
-178
33.0
-259
-291
-340
-371
-340
-154
-352
21.4
-389
19.0
-443
16.2
-478
12.7
-449
10.4
-141
31.8
-311
-344
-392
-420
-387
-116
-365
29.1
-407
26.2
-460
24.6
-500
20.1
-503
14.9
-272
33.1
-336
-374
-424
-457
-450
-248


MEASURED STRAINS DURING COOLING (E-6 IN./IN.)
242
Figure 8.16: Longitudinal Strain Distributions During Cooling:
Test Position 1


VIII LOW-TEMPERATURE PAVEMENT RESPONSE 216
8.1 Preliminary Tests With the Rigid Plate 216
8.2 Reinstrumentation for Tests With the Dual Wheels 221
8.3 Results of Tests With the Dual Wheel Loading System 221
8.3.1 Introduction 221
8.3.2 Pavement Response During Cooling 223
8.3.3 Dynamic Load and Creep Response at Different
Temperatures 255
8.3.4 Combined Effect of Thermal and Load Response 320
IX RESPONSE PREDICTION AT LOW TEMPERATURES 331
9.1 Dynamic Load Response 331
9.2 Thermal Response 347
9.3 Creep Response 348
X CONCLUSIONS A RECOMMENDATIONS 350
10.1 Conclusions 350
10.1.1 Pavement Testinq and Evaluation Method 350
10.1.2 Thermal and Load Response of Asphalt Concrete
Pavements 356
10.2 Recommendations 356
APPENDICES
A RELATIONSHIPS BETWEEN ASPHALT CONCRETE PROPERTIES
AND ASPHALT CEMENT PROPERTIES 359
B PAVEMENT TEMPERATURES DURING COOLING 361
C MEASURED THERMAL STRAINS DURING COOLING 368
D DYNAMIC LOAD RESPONSE MEASUREMENTS 375
E CREEP TEST DATA 396
REFERENCES 433
BIOGRAPHICAL SKETCH 443
v


Figure 7.20: Measured Strain Distributions at 25.6 C (69 F)


321
contraction and bending which may have caused the asphalt concrete layer
to separate and uplift from the base at one or more points.
The initial dynamic load response of the pavement at different
temperatures was totally contrary to the response expected for pavements
under normal conditions. Deflections and strains measured under load at
0.0 C (32 F), were in some cases more than double those measured at
higher temperatures. Furthermore, the dynamic load response varied
significantly for different test positions, even though the pavement
section was uniform.
The creep test results substantiated the results observed during
initial dynamic load tests. Comparitively high permanent deflections
and creep strains were measured for cases where high deflections and
strains were measured under dynamic loads. Dynamic load tests performed
during creep tests showed that there was a settling or stiffening effect
with sustained applications of static load. In addition, the dynamic
load response of the pavement after all creep tests were performed, was
found to be very similar for all test positions. This latter response,
which was approached with increased settlement at all test positions,
showed the expected changes in response with respect to temperature.
Dynamic load tests performed during creep tests showed that the
effect of temperature on load response depended on the method of cooling
used to arrive at a given temperature. These tests also showed that
response at a given temperature may have been affected by the amount of
permanent deflection and creep strain accumulated at a different
temperature. It is possible that these two effects were interrelated.
These observations left little doubt that aside from the expected
changes in asphalt concrete stiffness and possibly inducing thermal


366
Table B.3: Pavement Temperatures During Cooling: Test Position 3
Time
(Hours/Minutes)
Thermocouple
0
1
2
3
4
Number
0
30
0
30
0
30
0
30
0
1
C
22.8
18.5
12.3
9.4
CO
r-
6.9
6.2
4.7
3.3
F
73.0
65.3
54.1
48.9
46.0
44.4
13.2
40.5
37.9
2
C
22.7
16.3
8.3
5.0
3.2
2.8
2.6
1.1
1
o
F
72.9
61.3
46.9
41.0
37.8
37.0
36.7
34.0
30.9
3
C
22.6
19.2
13.3
10.1
8.2
7.5
7.0
5.8
4.3
F
72.7
66.6
55.9
50.2
46.8
45.5
44.6
42.4
39.7
4
C
22.5
18.0
11.9
9.2
7.8
7.1
6.4
4.9
3.3
F
72.5
64.4
53.4
48.6
18.0
44.8
43.5
40.8
37.9
5
C
22.2
13.3
3.4
0.2
-1.2
0.8
-0.6
-1.5
-2.8
F
72.0
55.9
38.1
32.4
29.8
30.6
30.9
29.3
27.0
C
22.4
17.3
9.8
6.6
5.0
4.4
4.0
2.7
1.2
F
72.3
63.1
49.6
43.9
41.0
39.9
39.2
36.9
34.0
C
22.4
13.0
2.4
-1.3
-2.6
-1.8
-1.4
CO
CM
1
-4.5
F
72.3
55.4
36.3
29.7
27.3
28.8
29.5
27.0
23.9
8
C
22.6
14.4
5.0
1.3
1
o
a*
-0.8
-0.5
-1.9
-3.5
F
72.7
57.9
41.0
34.3
30.9
30.6
31.1
28.6
25.7
C
22.5
22.0
16.9
13.4
10.7
9.1
8.9
7.3
5.9
F
72.5
71.6
62.4
56.1
51.3
48.4
46.9
45.1
42.6
10
C
22.5
22.4
19.5
16.3
13.8
12.0
10.9
9.9
8.7
F
72.5
72.3
67.1
61.3
56.8
53.6
51.6
49.8
47.7
11
C
22.4
22.5
21.4
19.1
17.1
15.4
14.1
13.1
12.1
F
72.3
72.5
70.5
66.4
62.8
59.7
57.4
55.6
53.8
12
C
22.4
22.4
22.0
20.7
19.2
17.8
16.5
15.5
14.6
F
72.3
72.3
71.6
69.3
66.6
64.0
61.7
59.9
58.3
13
C
22.3
21.9
16.7
12.5
9.8
8.3
7.4
6.6
5.5
F
72.1
71.4
62.1
54.5
49.6
46.9
15.3
43.9
41.9
14
C
22.2
22.1
19.0
15.3
12.5
10.7
9.6
8.7
7.7
F
72.0
71.8
66.2
59.5
54.5
51.3
19.3
47.7
45.9
15
C
22.2
22.2
21.1
18.7
16.4
14.5
13.2
12.2
11.3
F
72.0
72.0
70.0
65.7
61.5
58.1
55.8
54.0
52.3
16
C
22.1
22.2
21.8
20.4
18.7
17.0
15.7
14.7
13.8
F
71.8
72.0
71.2
68.7
65.7
62.6
60.3
58.5
56.8
Notes: (a) See Figure 8.5 for thermocouple location.
(b) Defrost cycle came on just before 6 hours. Therefore, temperatures are higher than normal.


STRAIN GAGE NUMBER
CABLE NUMBER
VIII
16x
xx xx
U U.
I II
VII
2 Ox
x24
15x
X
19x
x23
XII
X X
X
X
XXX
X
26
X
5J5
_ 8^
9 10 11 12
.25 x .
III
Tv
V
VI
XlT
13x
17x
x21
14x
IX
XI
18x
x22
Scale:
12"
NOTE: DIMENSIONS OMMITED FOR CLARITY.
A. TWO GAGES LOCATED AT PAVEMENT/LIME ROCK INTERFACE:
ONE TRANSVERSE AND ONE LONGITUDINAL
Figure 8.1: Layout of Strain Gages and Cables in the Test Pit


Tensile Strain (E-6ln/ln)
Wesi
loo
Distance from
load center (ins.)
too
x Measured
O Predicted
E, = 750,000 psl
E2 25,000 psl
E3= 25,000psl
200
Compressive Strain (E-6in/in)
East
Figure 9.4: Comparison of Measured and Predicted Transverse Strain Distributions at 0.0 C (32 F)


323
feet or so of the foundation materials, so that drastic reductions in
moduli would be necessary to effect the observed changes in load
response.
Several other observations indicated that the properties of the
foundation materials did not change during cooling. First, the changes
in response were not observed at all test positions, and it seems
unlikely that weakening of the foundation materials occurred in
localized areas. The moisture content and density of the sand subgrade
and limerock base were measured before the asphalt concrete layer was
placed and after it was removed. In both cases, these properties were
uniform throughout the pavement area (see Chapter V). Second, very
significant changes in pavement response were observed within a period
of a few hours, and for relatively small increases in temperature. As
shown in Figure 8.49, there was a drastic reduction (60%) in measured
deflections under load, when the pavement warmed up from 0.0 C (32 F) to
6.7 C (44 F). This change in temperature occurred within a few hours.
A reduction in base layer moisture could not have occurred during this
time, and no other effect related to changes in foundation material
properties could explain the observed increase in pavement stiffness.
Third, the fact that the load response after creep tests was very
similar for all three test positions, indicated that the same foundation
conditions existed under each test position. Finally, creep tests were
performed using the same level of static load for different time
durations, and permanent deflections continued to accumulate with
increased time of loading. It is difficult to believe that these time-
dependent permanent deflections were a result of densification of the
granular subgrade and base materials under a static load. There were


444
In 1985, he was registered as a professional engineer in the state
of Florida. He is a member of the American Society of Civil Engineers,
the Association of Asphalt Paving Technologists, and the Transportation
Research Board.
Reynaldo was awarded a postdoctoral fellowship by the Royal
Norwegian Council for Scientific and Industrial Research, to work for
one year with scientists at the Norwegian Geotechnical Institute in
Trondheim, Norway. He will begin this appointment in March, 1986.


436
34. Pell, P. S., and Cooper, K. E., "The Effect of Testing and Mix
Variables on the Fatigue Performance of Bituminous
Materials," Proceedings, Association of Asphalt Paving
Technologists, Vol. 44, pp. 1-31 (1975).
35. Moni smith, C. L., and Epps, J. A., "Influence of Mixture Variables
on the Flexural Fatigue Properties of Asphalt Concrete,"
Proceedings, Association of Asphalt Paving Technologists,
Vol. 38, pp. 4'23-4'64""QM5i:
36. Epps, J. A., and Monismith, C. L., "Influence of Mixture Variables
on the Direct Tensile Properties of Asphalt Concrete,"
Proceedings, Association of Asphalt Paving Technologists,
Vol. 39, pp. 207-236 (1970).
37. Bazin, P., and Saunier, J., "Deformability, Fatigue, and Healing
Properties of Asphalt Mixes," Proceedings, Second Inter
national Conference on the Structural Design of Asphalt
Pavements, 1967, pp. 109-140.
38. Deacon, J. A., "Materials Characterization Experimental
Behavior," Special Report Mo. 126, Highway Research Board,
1971, pp. m^iw.
39. Monismith, C. L., Seed, H. B., Mitry, F. G., and Chan, C. K.,
"Prediction of Pavement Deflections from Laboratory Tests,"
Proceedings, Second International Conference on the
StructuraHDesign of Asphalt Pavements, 1967, pp. 109-140.
40. Rauhut, J. B., and Kennedy, T. W., "Characterizing Fatigue Life
for Asphalt Concrete Pavements," Transportation Research
Record Mo. 888, Transportation Research Board, 1982, pp. 47-
ss:
41. Francken, L., "Fatigue Performance of a Bituminous Road Mix Under
Realistic Test Conditions," Transportation Research Record
No. 712, Transportation Research Board, 1979, pp. 30-36.
42. Ruth, B. E., and Maxfield, J. D., "Fatigue of Asphalt Concrete,"
Final Report, Project 245-D54, Department of Civil
Engineering, University of Florida, 1977.
43. Ruth, B. E., Bloy, L. A. K., and Avital A. A., "Low Temperature
Asphalt Rheology as Related to Thermal and Dynamic Behavior
of Asphalt Pavements," Final Report, Project 245-U20,
Department of Civil Engineering, University of Florida, 1981.
44. Roberts, F. L., Von Quintus, H., and Hudson, W. R., "Design
Procedure for Premium Flexible Pavements," Proceedings, Fifth
International Conference on the Structural Design of Asphalt
Pavements, 1982.


Figure
Page
8.28 Load-Unload Times for Dynamic Loading with Dual Wheels 261
8.29 Measured Longitudinal Deflections at 0.0 C (32 F):
Test Position 3 263
8.30 Measured Longitudinal Strains at 0.0 C (32 F):
Test Position 3 264
8.31 Measured Transverse Deflections at 0.0 C (32 F):
Test Position 3 265
8.32 Measured Transverse Strains at 0.0 C (32 F):
Test Position 3 266
8.33 Measured Longitudinal Deflections at 5.7 C (44 F):
Test Position 3 267
8.34 Measured Longitudinal Strains at 6.7 C (44 F):
Test Position 3 268
8.35 Measured Transverse Deflections at 6.7 C (44 F):
Test Position 3 269
8.36 Measured Transverse Strains at 6.7 C (44 F):
Test Position 3 270
8.37 Measured Longitudinal Deflections at 13.3 C (56 F):
Test Position 3 271
8.38 Measured Longitudinal Strains at 13.3 C (56 F):
Test Position 3 272
8.39 Measured Transverse Deflections at 13.3 C (56 F):
Test Position 3 273
8.40 Measured Transverse Strains at 13.3 C (56 F):
Test Position 3 274
8.41 Comparison of Measured Longitudinal Deflections at
Different Temperatures: Test Position 1, 10,000 lbs 278
8.42 Comparison of Measured Longitudinal Strains at
Different Temperatures: Test Position 1, 10,000 lbs. 279
8.43 Comparison of Measured Transverse Deflections at
Different Temperatures: Test Position 1, 10,000 lbs 280
8.44 Comparison of Measured Transverse Strains at
Different Temperatures: Test Position 1, 10,000 lbs 281
8.45 Comparison of Measured Longitudinal Deflections at
Different Temperatures: Test Position 2, 10,000 lbs 282
8.46 Comparison of Measured Longitudinal Strains at
Different Temperatures: Test Position 2, 10,000 lbs 283
8.47 Comparison of Measured Transverse Deflections at
Different Temperatures: Test Position 2, 10,000 lbs 284
8.48 Comparison of Measured Transverse Strains at
Different Temperatures: Test Position 2, 10,000 lbs 285
8.49 Comparison of Measured Longitudinal Deflections at
Different Temperatures: Test Position 3, 10,000 lbs 286
8.50 Comparison of Measured Longitudinal Strains at
Different Temperatures: Test Position 3, 10,000 lbs 287
8.51 Comparison of Measured Transverse Deflections at
Different Temperatures: Test Position 3, 10,000 lbs 288
8.52 Comparison of Measured Transverse Strains at
Different Temperatures: Test Position 3, 10,000 lbs 289
8.53 Comparison of Predicted Longitudinal Deflections at
Different Temperatures: 10,000 lbs 290
xi i i


Figure 3.1: Hopper for Asphalt Hot Mix Distribution


55
The strain gages were mounted with epoxy directly on the asphalt
concrete surface. The following procedure was used to prepare the
surface and mount the gages. The asphalt concrete surface was prepared
by sanding; first with a belt sander, and then with progressively finer
sandpaper until the surface was "glass-smooth". Clear tape was then
used to lift off all loose particles from the surface. Cleaning with
the tape was repeated until the tape was completely clean when lifted
off the surface. A thin layer of epoxy was then applied to the clean
surface and the gage was positioned, taking care to remove any air
bubbles trapped underneath the gage. A thin layer of epoxy was also
applied to the surface of the gage, for protection and to aid in
bonding. A sheet of cellophane was placed on top of the gage and clear
tape was used to hold the gage in position until the epoxy set. The
cellophane was used to prevent possible damage from the tape adhering
directly to the gage. Once the epoxy set, the tape was removed and the
strain gage wires were soldered to the gage. The completely installed
gage was covered with a piece of masking tape followed by a piece of
duct tape for protection. A picture of the two-inch strain gages
mounted on asphalt concrete cores is shown in Figure 3.9.
3.4.2 Data Acquisition System
A data acquisition system was designed and installed which was
capable of monitoring and recording ten dynamic deflection measurements,
ten dynamic strain measurements, 20 temperature measurements, and load
magnitude and time of loading. As mentioned earlier, only one recording
device, a Gould model 2400 strip chart recorder, was available in the
test pit facility, since this was all that was needed to evaluate base


78
100,000 psi. All other material properties were the same for all runs
and are given in Table 4.2.
The predicted deflection basins for each system are presented in
Table 4.2. The deflection differences for systems with and without a
concrete embankment (or floor) are also shown in Table 4.2. These
differences indicate that the effect of the concrete floor was to reduce
the deflections by an amount that is relatively independent of the
stiffness of the 1imerock base (approximately 2.5 E-3 in.).
A comparison of the deflection basins for the three cases studied
is presented in Figure 4.3. Clearly, the effect of the concrete floor
is considerable and must be accounted for in the analysis.
The following AXSYM runs were made to determine the effect of the
walls on the 1imerock base response:
- wall at 7 ft., no friction, rigid plate;
- wall at 7 ft., no friction, flexible plate;
- wall at 4 ft., no friction, rigid plate; and
- wall at 4 ft., full friction, rigid plate.
The following pavement system was used in the analysis:
Modulus (psi)
Poisson's Ratio
Thickness (in.)
Limerock:
90,000
0.40
6.75
Sand:
14,530
0.30
36
This system was underlain by a rigid base. A relatively high 1imerock
modulus was chosen for the analysis, since this stiffer system would be
affected to a greater degree by wall friction. A pressure of 50 psi on
a 12-inch diameter area was applied in all cases.


Table E.22: Measured Creep Strains For Different Times of 10,000-lb. Static Load Application,
Position Number 2, 13.3 C (56 F)
Load^
Longitudinal
Strains
(micro-
strain)^)
Transverse Strains
(micro-strain)
Duration
Distance from
Load Center (in
.): N-
Sic)
Distance
from Load Center (in.
): E-W(d>
(Seconds)
0
8(S)
12(N)
16 (S)
24(N)
36 (S)
12(W)
18(E)
24(W)
36(E)
50
7
32
O
6
-23
-2
38
35
2.4
0
50
12
15
-7
3
-10
0
25
16
2.4
-2
400
73
23
-45
-8
-20
-3
39
24.4
-12
-6
500
53
38
-38
-3
-22
-5
43
23
-8
-3
(a) Loads were applied sequentially with a rest period of four times the load duration between loads.
(b) Measured creep strains are the residual strains after the rest period. Tension is positive.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.


Table E.13: Measured Permanent Deflections For Different Times of 10,000-lb. Static Load Application,
Position Number 2, 6.7 C (44 F)
Load^a)
Longitudinal Deflections
(E-3 in
.)(b)
Transverse Deflections (E-3
in.)
Duration
Distance
from Load
Center
(in.):
N-S(c)
Distance
from Load
Center (in
.): E-W{di
(Seconds)
0
8(S)
12(N)
16 (S)
24(N)
36 (S)
13 (W)
18(E)
24 (W)
36(E)
50
0.98
-0.24
0.72
-0.19
-0.24
0.2
0.53
0.25
-0.25
-0.1
50
1.27
-0.24
0.63
-0.24
0.12
0.0
0.96
0.37
0.0
0.0
400
1.10
0.90
0.60
-0.12
-0.48
-0.1
0.81
0.35
-0.39
-0.1
500
0.61
-0.19
0.36
0.0
-.05
-0.1
0.76
0.0
0.12
-0.3
(a) Loads were applied sequentially with a rest period of four times the load duration between loads.
(b) Measured permanent deflections are the residual deflections after the rest period. Positive is
downward.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.


151
4. Fifty dynamic load repetitions at 10,000-lbs. were applied as a
seating load.
5. Dynamic load response measurements were made at load levels of
1,000, 4,000, 7,000, and 10,000 lbs., beginning with the
greatest load first.
6. For each load level, deflection and strain measurements were
recorded with the digital oscilloscopes for several successive
repetitions of dynamic load. Since the data acquisition system
could only monitor one strain gage at a time, dynamic loads
were applied until recordings were made for all ten strain
gages being monitored. Therefore, each strain gage was
monitored at a slightly different time.
6.2 Low-Temperature Pavement Response Tests
6.2.1 Introduction
These procedures correspond to the results presented in Chapter
VIII. The tests performed can be broken down as follows: thermal
response of the pavement during cooling; dynamic load tests at different
pavement temperatures; creep tests to determine the permanent
deflections and creep strains accumulated under static loads; and
dynamic load tests performed during creep tests to determine the effect
of permanent deformations on the dynamic load response of the
pavement. These tests were performed at three different positions in
the test pit pavement, and except for the temperature order of testing,
the procedures used were the same at all three tests positions.
Table 6.1 summarizes the temperature order in which these tests
were performed. As shown in this table, dynamic load tests were


314
However, the fact remained that this effect resulted in very
unusual pavement response. The measured response varied considerably
from position to position, and the highest dynamic deflections and creep
response were observed at the lowest test temperature. Further analysis
of the data provided additional clues to explain the phenomenon that
resulted in the observed pavement behavior.
Creep test results pointed to other factors that had an effect on
the dynamic load response and creep response of the pavement. It was
found that the order, with respect to temperature, in which creep tests
were performed, had a significant effect on the response of the
pavement. As mentioned earlier although initial dynamic load tests were
performed in the same order of temperature at all test positions, creep
tests were performed in a different order at each test position. As
shown in Table 8.1, the order of creep tests at test position 1 was the
reverse order of the creep tests performed at test position 3.
Therefore, a comparison of the measured resDonse for these two positions
will help demonstrate the effects of the order of temperature on the
measured response of the pavement.
The initial dynamic response at 0.0 C (32 F) at test positions 1
and 3 was very similar and unusually high (see Figures 8.67 and 8.69).
However, the dynamic response immediately prior to creep tests was
totally different at the same two positions at the same temperature
(again see Figures 8.67 and 8.69). Figure 8.66 shows that the permanent
deflections recorded at these two positions at 0.0 C (32 F) were also
totally different, where much higher permanent deflections were recorded
at test position 3. A similar, but opposite effect was observed for
these two positions at 13.3 C (56 F). As shown in Figure 8.59, a


Figure 5.6: Applied Stress vs. Deflection: 16-in. Plate on Limerock Base


9
the literature. Thus, although the causes of cracking are well known,
the actual mechanisms that lead to cracking have not been verified with
definitive measurements of actual failures on full-scale pavements.
Several mechanisms have been proposed that cannot account for basic
material response and failure characteristics, variability of
environment, and loading conditions encountered in actual pavements.
These have led to empirical design procedures, which are valid only for
the conditions from which they were derived.
2.2.2 Cracking Mechanisms
Traditionally, cracking has been broken down into traffic-load
induced and thermally-induced, with little consideration for the
combined effects of the two mechanisms. Low-temperature transverse
cracking has been recognized as the most common non-traffic associated
failure mode and is a serious problem in Canada and parts of the United
States (5,6,7). This type of failure is generally considered a
temperature phenomenon caused by low temperatures. As the pavement
temperature decreases the asphalt concrete wants to contract, but
contraction is resisted by the friction between the asphalt concrete
layer and the base and by the length of the roadway in the longitudinal
direction. This resistance results in tensile stresses in the pavement,
which are greatest in the longitudinal direction.
Several researchers have postulated that cracking occurs when these
thermally induced tensile stresses exceed the tensile strength of the
asphalt concrete (8, 9, 10). This mechanism has been confirmed by
laboratory and field investigations (7-13), and provides the basis for
the hypotheses that have been presented for low temperature cracking.


367
Table B.3extended
5
6
8
10
12
15
18
21
24
payJ^-^2
0
0
0
0
0
0
0
0
0
0
1.1
0.1
-1.0
-2.4
-3.6
-5.0
-5.8
-6.0
-6.7
1.0
34.0
32.2
30.2
27.7
25.5
23.0
21.6
21.2
19.9
33.8
-3.1
-4.6
-6.1
-7.9
-9.1
-9.7
-10.9
-10.5
-10.7
-0.7
26.4
23.7
21.0
17.8
15.6
11.5
12.4
13.1
12.7
30.7
1.4
-0.6
-1.8
-4.3
-6.1
-6.4
-8.1
-7.7
-8.3
-0.1
34.5
30.9
28.8
24.3
21.0
20.5
17.4
18.1
17.1
31.8
1.0
-0.3
-1.2
-3.3
-4.8
-5.3
-6.7
-6.5
-7.5
-0.1
33.8
31.5
29.8
26.1
23.4
22.5
19.9
20.3
18.5
32.2
-4.8
-6.3
-9.6
-11.2
-12.0
-12.8
-13.9
-13.7
-13.8
-2.9
23.4
20.7
14.7
11.8
10.4
9.0
7.0
7.3
7.2
26.8
-1.4
-3.1
-5.3
-7.0
-8.5
-9.8
-10.8
-10.7
-11.0
-2.4
28.5
26.4
22.5
19.1
16.7
14.4
12.6
12.7
12.2
27.7
-6.9
-8.4
-11.2
-12.7
-13.8
-14.3
-15.0
-14.4
-2.9
19.6
16.9
11.8
9.1
7.2
6.3
5.0
6.1
5.0
26.8
-5.8
-7.2
-9.5
-11.5
-12.6
-13.1
-14.6
-13.7
-14.8
-3.0
21.6
19.0
14.9
11.3
9.3
8.4
5.7
7.3
5.4
26.6
3.4
1.6
-0.3
-2.3
-3.0
-4.7
-6.0
-6.3
-6.5
-0.9
38.1
34.9
31.5
27.9
26.6
23.5
21.2
20.7
20.3
30.4
6.3
4.4
2.3
0.1
-0.5
-25
-4.0
-4.4
-4.5
-0.7
43.3
39.9
36.1
32.2
31.5
27.5
24.8
24.1
23.7
30.7
9.9
8.0
5.7
3.1
2.1
0.7
-1.0
-1.6
-2.0
0.0
49.8
46.4
42.3
38.1
35.8
33.3
30.2
29.1
28.4
32.0
12.6
10.8
8.4
6.1
4.6
2.9
1.4
0.7
0.2
0.7
54.7
51.4
47.1
43.0
40.3
37.2
34.5
33.3
32.4
33.3
3.5
1.8
-1.3
-3.3
-5.6
-6.1
-7.4
-7.4
-8.2
-2.4
38.3
35.2
29.7
26.1
21.9
21.0
18.7
18.7
17.2
27.7
5.7
3.9
0.9
-0.8
-3.5
-4.4
-5.6
-5.7
-6.5
-2.0
42.3
39.0
33.6
30.6
25.7
24.1
21.9
21.7
70.3
28.4
9.4
7.6
4.7
2.4
0.5
-1.0
-2.2
-2.5
-3.2
-0.9
48.9
45.7
40.5
36.3
32.9
30.2
28.0
27.5
26.2
30.4
. 12.0
10.4
7.6
5.1
3.0
1.5
0.4
-0.1
-0.6
0.1
53.6
50.7
45.7
41.2
37.4
34.7
32.7
31.8
30.9
32.2


10
The rheological properties of the asphalt at low temperatures are
generally recognized as the most important factor in low-temperature
transverse cracking (5, 11, 14, 15, 16, and others). Many researchers
have associated low temperature cracking with properties such as asphalt
stiffness, viscosity, temperature susceptibility, and glass transition
temperature. These properties, of course, are all related to the
asphalt's ability to flow and thus relax stresses. All researchers have
found that the stiffer and more temperature susceptible the asphalt, the
greater the potential for cracking.
Probably the most commonly proposed approach to control thermal
cracking is to limit the asphalt stiffness as measured for a minimum
design temperature. McLeod (17) concluded that low temperature pavement
cracking is likely to occur whenever the stiffness of the pavement
attains a value of 6.9 E9 Pa (1.0 E6 psi) at a pavement depth of two
inches, at the minimum temperature encountered, and for a loading time
of 20,000 seconds. Fromm and Phang (18) proposed a value of 1.4 E8 Pa
(20,000 psi) at 10,000 seconds loading time. Gaw (19) reported that the
St. Anne test pavements cracked at an asphalt binder stiffness of 1.0 E9
Pa (145,000 psi) and a mixture stiffness of 2.0 E10 pa (2,900,000 psi)
at 1800 seconds loading time. Many researchers have found good
agreement between measured stiffness and observed cracking of pavements
in the field and confirmed that pavements using softer asphalts exhibit
less cracking (7, 12, 20, 21, 22).
Ruth (14) concluded that cracking would be reduced by using
asphalts with lower viscosities and improved rheological behavior at low
temperature. Fabb (23) concluded that low viscosity and low temperature


Table E.30: Measured Creep Strains For Different Times of 10,000-lb. Static Load Application,
Position Number 3, 6.7 C (44 F)
Load^
Longitudinal
Strains
(micro-
strain)^)
Transverse Strains
(micro-strain)
Duration
Distance from
Load Center (in
.): N-S(c)
Distance
from Load Center (in.
): E-W{d>
(Seconds)
0
8(N)
12 (S)
16 (N)
24(S) 40(S)
12(W)
18(E)
24 (W)
36(E)
50
-32
-38
39
-10
-6 -2
-8
12
-4
0
50
-22
-16
21
-4
-4 -2
0
10
-3
0
400
-6

26
-12
-18 -6
-10
16
-3
0
500
-52
12
0
-9 -4
-5
18
0
0
(a) Loads were applied sequentially with a rest period of four times the load duration between loads.
(b) Measured creep strains are the residual strains after the rest period. Tension is positive.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.


129
the seating load, all instrumentation was zeroed prior to
testing. Each increment of static loading was maintained
until there was essentially no increase in deflection before
proceeding to the next load increment. This was determined
visually by continuously monitoring the deflection readings on
the chart recorder. After the maximum load was applied for a
particular cycle, it was released completely before initiating
the next loading cycle. Tables 5.7 and 5.8 give the
incremental loads used for each material and plate size. In
all cases the first increment of load was 800 lbs. This was
inevitable, since it was the minimum load that could be
applied once the hydraulic loading system was activated.
5) This incremental loading sequence was repeated for a total of
five cycles.
6) After the last static load cycle was applied, 100 dynamic load
repetitions were applied. The stress levels used for dynamic
loading on each material are also given in Tables 5.7 and 5.8.
7) Once all testing was complete a nuclear density test was
performed and a soil sample was taken for moisture content
determination.
5.4.2 Plate Tests Immediately After Placement
Three 12-inch and three 16-inch diameter plate load tests were
performed on both the sand subgrade and 1imerock base shortly after they
were placed and compacted. In addition, one 30-inch diameter plate was
performed on the sand. Figure 5.2 shows the positions of each plate
load test performed on the Fairbanks sand and Figure 5.3 shows the


SURFACE STRAINS (jjie) TENSION
Figure 7.25: Strain Distribution Comparison at 10,000 lbs.


363
Table B.lextended
5
6
8
11
17
20
0
0
30
0
0
0
-0.5
-1.9
-4.6
-7.0
-8.8
1.2
31.1
28.6
23.7
19.4
16.2
34.2
-5.9
-7.2
-8.8
-10.7
-12.0
-.1
21.4
19.0
16.2
12.7
10.4
31.8
-1.6
-3.2
-4.1
-6.6
-8.6
0.6
29.1
26.2
24.6
20.1
16.5
33.1
-2.1
-3.3
-5.1
-7.3
-9.5
0.2
28.2
26.1
22.8
18.9
14.9
32.4
-7.8
-9.3
-10.4
-11.8
-13.1
-1.1
18.0
15.3
13.3
10.8
8.4
30.0
-4.5
-6.6
-7.7
-9.6
-11.3
-0.9
23.9
21.2
18.1
14.7
11.7
30.4
-10,1
-11.3
-12.5
-14.0
-14.6
-1.1
13.8
11.7
9.5
6.8
5.7
30.0
-9.4
-10.9
-11.6
-13.6
-14.8
-1.2
15.1
12.4
11.1
7.5
5.4
29.8
0.2
-1.4
-2.0
-5.6
-7.5
-0.3
32.4
29.5
28.4
21.9
18.5
31.5
2.8
1.2
+0.3
-3.2
-5.5
-0.2
37.0
34.2
32.5
26.2
22.1
31.6
6.1
4.5
2.3
0.2
-2.5
0.2
43.0
40.1
36.1
32.4
27.5
32.4
8.8
7.2
4.6
2.4
0.0
0.6
47.8
45.0
40.3
36.3
32.0
33.1
0.5
-0.4
-3.9
-6.3
-7.9
-0.8
31.1
31.3
25.0
20.7
17.8
30.6
1.8
0.9
-1.8
-4.4
-6.2
-0.6
35.2
33.6
28.8
24.1
20.8
30.9
5.6
3.9
1.5
-0.7
-2.9
-0.1
42.1
39.0
34.7
30.7
26.8
31.8
8.4
6.6
3.9
1.8
-0.3
0.5
47.1
43.9
39.0
35.2
31.5
32.9


x Surface Thermocouples
O* Strain Gages
*
5
x O
8
1
x
4
O
3
x O
4
1
O
6
O
0
o
2
O
x
2&
9-12(a>
7
O
x
6
9
O
- N
12
i 1
(a) FOUR THERMOCOUPLES LOCATED AT DEPTHS OF 1 3/8, 2 1/8, 3 1/8 & 4 1/8 INS.
ro
ro
-'~l
Figure 8.5: Thermocouple and Strain Gage Location During Cooling: Test Position 3


149
clear from these latter plate test results: 1) the 1 imerock modulus
decreased; and 2) the subgrade modulus increased relative to when they
were initially placed and compacted. It will be shown in Chapter IX
that the same conclusions were reached based on pavement response
measurements taken close to the time the pavement was removed.


?.08
Load-deflection and load-strain relationships were plotted for the
pavement response measurements taken at 20.6 C (69 F). These plots are
shown for the different measurement positions, in Figures 7.33 and 7.34,
respectively. Both the deflection and strain relationships indicate
that the pavement response was nonlinear in a stress-stiffening way.
This is consistent with the response observed for the initial tests
performed at 18.3 C (65 F) and with the nonlinear response observed on
the plate tests performed on the limerock base (see Chapter V).
BISAR was used to predict the pavement response at 4,000 lbs. to
see if the observed nonlinearity could be accounted for by changing only
the limerock base modulus. The procedure outlined in Section 4.4.3.1
was used along with the rigid plate approximation procedure to predict
response using different limerock moduli until the best correspondence
was achieved between measured and predicted response at 4,000 lbs. The
best prediction was obtained with the following parameters.
Asphalt Concrete:
E =
124,400
psi
u = 0.35
Limerock Base:
E =
40,000
psi
v = 0.40
Sand Subgrade:
E =
15,420
psi
y = 0.30
Note that the modulus of the sand and the asphalt concrete are the same
as those used to predict response at the 10,000-lb. load level. Only
the limerock modulus was changed.
Comparisons of measured and predicted deflections and strains are
shov/n in Figures 7.35 and 7.36, respectively. As shown in these
figures, these parameters resulted in excellent predictions of both
deflections and strains, indicating that the observed nonlinearity was
mainly caused by the limerock base. Plate load tests performed on the
limerock also indicated that this material responded non! inearly in a


Table D.ll: Measured Deflections, Test Position 2, 6.7 C (44 F)
Load
Longitudinal Deflections
(E-3 in
.) (a)
Transverse Deflections (E-3 in.)
(kips)
Distance
from Load
Center
(in.):
N-S(b)
Distance
from Load
Center (in.):
: E-W(c)
0 8(S)
12(N)
16 (S)
24(N)
36 (S)
13 (VI)
18(E)
24(W)
36(E)
10
10.53 4.26
5.78
-0.24
3.02
0.0
8.56
4.30
2.48
.3
7
7.94 3.08
5.31
-0.25
2.17
0.0
5.97
3.22
1.97
0.0
4
4.53 1.66
3.01
-0.24
1.1
0.0
3.34
1.49
0.2
0.0
1
.73 .12
.24
0.0
0.0
0.0
.48
0.0
0.0
0.0
(a) Positive is downward deflection.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.


205
good deflection and strain prediction for the earlier tests performed at
18.3 C (65 F). Therefore, it seems like the flexural properties of the
asphalt concrete have not changed.
One logical explanation for the discrepancy between measured and
predicted deflections at 20.6 C (69 F) is that the asphalt concrete may
have developed anistropic properties as a result of previous loading at
this position. The initial tests performed at this position,
particularly at the slow loading rate, may have compressed the asphalt
concrete layer and decreased its vertical compressibility under the
plate. A 12-inch diameter circular depression was clearly visible after
these initial plate tests were performed. Therefore, the asphalt
concrete responded more stiffly vertically than radially. Since BISAR
cannot account for anisotropic properties, it overpredicted the measured
deflections, even though it predicted the measured radial strains almost
exactly.
These results indicate that more reliance should be placed on
measured strains than on measured deflections for the evaluation of
pavement response. Errors in interpreting deflection measurements can
result from uncertain foundation conditions, anisotropy, or the volume
transfer phenomenon observed earlier for the slow loading rate.
Although these may have some effect on measured strains, the effect is
less. In addition, strains are a direct measure of the response of the
material, making their interpretation much less dependent on the model
used to predict response. Thus a more direct evaluation of response can
be made. It is interesting to note that most researchers rely mainly on
deflection data alone to evaluate pavement response, since deflections
are much simpler to obtain.


114
Table 5.1: Laboratory Test Results: Fairbanks Sand
Limerock Bearing Ratio (LBR) 31
Optimum Moisture Content 12.3
Dry Density of Optimum Moisture 108.9
Sieve Analysis
Sieve Size Percent Passing
# 10 100
# 40 90
# 60 59
#200 5


Table 5.10 Modulus Values Immediately After Placement: Limerock Base
12" Plate
16" Plate
F(a)
r-res
F(b)
ttani
F(c)
Ltanf
Water
Dry Unit
E(a)
tres
p(b)
ttani
p(c)
Ltanf
Water
Dry Unit
Test
Test
Content
Weight
Content
Weight
No.
Type
(psi)
(psi)
(psi)
w-%
(pcf)
(psi)
(psi)
(psi)
v-%
(pcf)
Static^
48500
34850
63500
50250
24500
77100
1
Dynamic(
43800


9.5
117.4
63350


9.7
116.7
Static
55700
37450
89300
56800
27650
83250
2
Dynamic
51900


9.2
117.7
47100


10.5
117.2
Static
53450
40400
73250
56000
24500
77100
3
Dynamic
51000


10.4
116.7
47500


9.3
117.6
Static
52550
37550
75350
54350
25550
79150
Average
Dynamic
48900
9.7
117.3
52650
9.8
117.2
(a) Resilient Modulus
(b) Initial Tangent Modulus (Low Stresses)
(c) Final Tangent Modulus (High Stresses)
(d) Determined on fifth loading cycle
(e) Determined on 100th repetition of load


Distance From Load Center (ins)
Figure 8.47: Comparison of Measured Transverse Deflections at Different Temperatures:
Test Position 2, 10,000 lbs.
284


South
FO
ro
Figure 8.38: Measured Longitudinal Strains at 13.3 C (56 F): Test Position 3


Table 4.3: Predicted Deflections Using AXSYM
Deflections (E-3 in.)
Position
Distance
From
Load Center
of Wall
0.0
6.0
7.5
9.0
12.0
16.0
20.0
24.0
30.0
36.0
42.0
48.0
4 ft. (NF)*
10.19
10.17
8.06
7.09
5.46
3.82
2.61
1.74
0.90
0.41
0.17
0.10
A-4 ft. (FF)*
10.16
10.14
8.03
7.06
5.43
3.78
2.56
1.68
0.82
0.32
0.08
0.0
B-7 ft. (NF)*
10.38
10.36
8.25
7.27
5.65
4.01
2.79
1.92
1.06
0.54
0.25
0.08
B A
0.22
0.22
0.22
0.23
0.22
0.23
0.23
0.24
0.24
0.22
0.18
0.08
* NF Ho Friction
FF Full Friction


88
were used: 145,000 psi, 1,500,000 psi, and 3,500,000 psi. For each
pavement system, the following AXSYM runs were made: frictionless wall
at 7 ft., frictionless wall at 4 ft., and full friction wall at 4 ft.
Figure 4.7 shows the thicknesses and properties used for each layer and
illustrates the systems as modeled. Mote that for all cases the sand
was modeled as a 48-inch layer underlain by a rigid base.
The predicted deflection basins for each of these runs are tabu
lated in Table 4.5. These results show that for all pavement stiff
nesses, the effect of the wall was to shift the deflection basin upward
by a small amount. Note that the wall effect was considered to be the
difference in deflection between the case of the frictionless wall at 7
ft. and the case of the full friction wall at 4 ft. The effect of the
wall increased with increasing pavement stiffness and was relatively
small for all cases. For the range of modulus values used in the test
pit (100,000 psi to 1,500,000 psi), it seems that the wall effect will
be essentially constant for a given stress level and support condi
tions. A shift factor of 1.0 E-3 in. could be used for a load of 10,000
lbs. and the support conditions used. It may be that this value is
adequate for most support conditions used in the test pit, but one
program run should be made to check this value once the subgrade and
limerock moduli are determined.
The following conclusions were drawn concerning the effect of the
test pit constraints on the three-layer system:
- the floor effect must be considered in the analysis of measured
deflection basins, but the effect is independent of the asphalt
concrete modulus for a given set of foundation conditions;


Distance From Load Center (ins)
West
Figure 8.51: Comparison of Measured Transverse Deflections at Different Temperatures:
Test Position 3, 10,000 lbs.
288


249
response occurred. This indicates that the mechanism may be related to
the thermal gradient in the asphalt concrete layer, since the thermal
gradient reached a maximum during the first two hours of cooling (Figure
8.9). The thermal gradient decreased comparatively slowly after two
hours, and no change in the shape of the thermal strain distribution
occurred for these conditions under continued cooling.
The next change in thermal response occurred anywhere from 10 to 12
hrs. of cooling, depending on the cooling cycle. This change is evident
by comparing the second and third longitudinal strain distributions for
each cooling cycle. The most significant changes are shown in Figures
8.17 and 8.18, which show the greatest changes in the shape of the
longitudinal strain distribution. The section to the left of the center
of the test pit (six feet from south wall) went into compression, while
the area to the right of center decreased in compression, and at one
point, the surface of the pavement actually went into tension.
The most interesting thing about the observed change in thermal
response was that there were very small changes in pavement temperature
during the time the strain changes were observed. The respective
cooling curves shown in Figures 8.6 to 8.8, show that the cooling rate
after about 11 hrs. was almost negligible (about 0.3 C/0.6 F per hr.).
Yet even with these small temperature reductions, relatively large and
unusual changes in strains were observed in the pavement.
The repeatability of the measurements indicated that the
measurements were good. Several gages were monitored during different
cooling cycles and their response during all cycles was almost
identical. Figures 8.22 and 8.23 show that the strain-temperature
relationship for two gages that were monitored during different cooling


MEASURED STRAINS DURING COOLING (E-6 IN./IN.)
I t i i i i '
237
Figure 8.14: Measured Transverse Strains vs. Temperature:
Test Position 2


Distance From Center of Test Position (ins,)
Figure 8.65: Comparison of Dynamic Load Response Immediately Prior to Creep Tests
for Different Test Positions: 0.0 C (32 F)


Table D.7: Measured Deflections, Test Position 1, 0 C (32 F), Repeat Test
Load
Longitudinal Deflections
(E-3 in
.)(a)
Transverse Deflections (E-3 in.)
(kips)
Distance
from Load
Center
(in.):
N-S{b)
Distance
from Load
Center (in.):
: E-W(c)
0
8( M)
12 (S)
18 (S)
24 (N)
36 (N)
13 (W)
18(E)
24(W)
36(E)
10
22.0
14.6
14.0
9.0
3.1
-.1
17.1
5.5
6.2
0
7
16.6
11.5
10.8
6.6
2.5
0
13.2
4.0
4.9
0
4
10.4
7.4
6.8
4.2
1.5
0
8.4
2.2
3.2
0
1
2.3
1.4
1.3
.8
.1
0
1.5
.7
0
0
(a) Positive is downward deflection.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.


36
method does not allow for stress relaxation subsequent to the time
interval in which stresses are computed.
Monismith et al. (9) developed a thermal stress equation for an
infinite viscoelastic slab and complete restraint. Stress is calculated
as a function of depth, time, and temperature. A relaxation modulus is
required from uniaxial creep tests.
Ruth et al. (43) presented a stress equation that considers rate of
cooling, creep rate, and variation in modulus with temperature. The
model requires the asphalt viscosity-temperature relationship from which
all calculations are made. The method is unique in that it can account
for individual asphalt properties.
The prediction of thermal cracking in asphalt pavements is usually
based on comparing the accumulated thermal stress with the tensile
breaking strength of the asphalt concrete. Several investigators have
compared stresses as predicted by the various models with observed
cracking.
Burgess et al. (12) reported that the method of Hills and Brien
correlated well with cracking observed in the St. Anne Test Road.
Christison (7) compared results from pseudoelastic beam, viscoelastic
beam, and viscoelastic slab analyses to cracking observed at St. Anne.
He determined that thermal cracking could be predicted by using the
computed stresses from either analysis at 1/2-inch depth.
Haas and Topper (13) indicated that Moni smith's method appeared to
predict unusually high stresses, which may be due to his assumption of
infinite lateral extent.
Models to predict thermal stress cracking have also been
presented. The computer program COLO (91) was developed based on the


A39
69. Yeager, L. L., and Wood, L. E., "Recommended Procedure for
Determining the Dynamic Modulus of Asphalt Mixtures,"
Transportation Research Record No. 549, Transportation
Research Board, 1975, pp. 26-38.
70. Barksdale, R. G., "Compressive Stress Pulse Times in Flexible
Pavements for Use in Dynamic Testing," Highway Research
Record No. 345, Highway Research Board, 1971, pp. 32-34.
71. "Test Procedures For Characterizing Dynamic Stress-Strain
Properties of Pavement Materials," Special Report No. 162,
Transportation Research Board, 1975T
72. Kallas, B. F., "Dynamic Modulus of Asphalt Concrete in Tension and
Tension-Compression," Proceedings, Association of Asphalt
Paving Technologists, Vol. 39, pp. 1-23 (1970).
73. Wallace, K., and Monismith, C. L., "Diametral Modulus Testing on
Nonlinear Pavement Materials," Proceedings, Association of
Asphalt Paving Technologists, Vol. 49, pp. 633-652 (1980).
74. Heukelom, W., "Observations on the Rheology and Fracture of
Bitumens and Asphalt Mixes," Proceedings, Association of
Asphalt Paving Technologists, Vol. 35, pp. 358-396 (1966).
75. Majidzadeh, K., and Herrin, M., "Modes of Failure and Strength of
Asphalt Films Subjected to Tensile Stress," Highway Research
Record No. 67, Highway Research Board, 1965, pp. 98-121.
76. Tons, E., and Krokosky, E. M., "Tensile Properties of Dense Graded
Bituminous Concrete," Proceedings, Association of Asphalt
Paving Technologists, Vol. 32, pp. 497-524 (1963).
77. Finn, F. N., "Factors Involved in the Design of Asphaltic Pavement
Surfaces," National Cooperative Highway Research Program
Report No. 39, 1967.
78. Ruth, B. E., and Olson, G. K., "Creep Effects on Fatigue Testing
of Asphalt Concrete," Proceedings, Association of Asphalt
Paving Technologists, Vol. 46, pp. 176-192 (1977)'.
79. Pavlovich, R. D., and Goetz, W. H., "Direct Tension Results for
Some Asphalt Concretes," Proceedings, Association of Asphalt
Paving Technologists, Vol.'45, pp. 400-424 (1976).
80. Sal am, Y. M., and Monismith, C. L., "Fracture Characteristics of
Asphalt Concrete," Proceedings, Association of Asphalt Paving
Techno!ogists, Vol. 41, pp. 215-253 (1972).
81. Ruth, B. E., and Potts, C. F., "Changes in Asphalt Concrete
Mixture Properties as Affected by Absorption, Hardening, and
Temperature," Transportation Research Record No. 515,
Transportation Research Board, 1974, pp. 55-66.


CHAPTER II
LITERATURE REVIEW
2.1 Introduction
The research presented in this document focuses on defining and
predicting low temperature response and failure of asphalt concrete
pavements. This includes the response of pavements to changes in
temperature and the effect these changes have on the load response and
failure limits of the asphalt concrete layer. Two elements of the
analysis system considered here make it uni cue: the use of measured
rheological parameters of the asphalt at low temperatures to predict the
response and failure characteristics of the asphalt concrete; and the
fact that cracking is considered a short-term phenomenon that occurs
when the combined effect of temperature and traffic loads exceed the
failure limit of the asphalt concrete pavement. Although this approach
is totally different from traditional approaches, a review of the
literature will serve two purposes:
1) to establish the need and develop the rationale behind the
proposed method of analysis; and
2) to give an overview of existing knowledge of asphalt concrete
pavement response to temperature changes and traffic loads,
including an assessment of our ability to predict response and
failure.
5


Distance From Center of Test Position (ins.)
Figure 8.71: Comparison of Dynamic Load Response After Creep Tests for
Different Test Positions: 6.7 C (44 F)


45
The system was designed to cool six-inch thick pavements to a
temperature of -10 C (15 F) at a rate of 3.3 C (5 F)/hour, as measured
at a depth of 1/4-inch from the surface of the pavement. Temperature
control was achieved by manually controlling the cooling unit. A fully
automatic system with greater cooling capacity was originally
considered, but its cost was prohibitive. In any case, automatic
controls are of limited value, since temperature gradients would be
present in the pavement as long as the unit was running.
The cooling system consists of a direct expansion, low temperature
refrigeration system. The evaporator (Larkin ELT-300) was located
directly in the test pit and cooled the pavement by recirculating cold
air across the test pit surface. Temperature control was achieved by
lowering the pavement temperature below the required test temperature,
stopping the refrigeration system, and allowing the test pit temperature
to drift upward. The condensing unit (Larkin CS 0750L1), which housed a
7.5 HP compressor, sat outside the building's north wall. Refrigerant
hoses and electrical cables from the condensing unit to the evaporator
unit were connected through holes drilled in the north wall of the test
pit. Drainage was accomplished with a heated drain pipe, which was
passed through a hole drilled in the test pit wall. A layout of the
test pit cooling system is shown in Figure 3.2.
An insulated cover for the test pit area was designed and
constructed. The cover had a solid wood frame which enclosed the test
pit area. The top of the cover consisted of five removable wood panels
that spanned the 8-ft. width of the test area and were supported by a
ledger on the cover's frame. One panel had a one-ft. diameter hole to
allow for placement of the loading ram. The panel dimensions are such


210
Figure 7.34: Measured Load-Strain Relationships at 20.6 C (69 F)


313
after creep tests. The initial dynamic load test results will be
temporarily ignored. These figures show that for positions 2 and 3,
where permanent deflections were higher than for position 1, there was
also a greater change in dynamic load response, from before to after
creep tests, than at position 1. It seemed that as permanent
deflections and creep strains were accumulated, one or more of the
pavement layers settled and caused the pavement to respond more
stiffly. Conversely, the magnitude of permanent deflections and creep
strains were dependent on the capacity of one or more of the pavement
layers to settle.
These observations were not unusual, and were in fact, logical.
One would expect higher permanent deflections and creep strains for a
pavement system that responds less stiffly. Less stiff response implies
that higher stresses are being induced in the asphalt concrete for any
given load, and higher stresses imply higher creep rates. It also makes
sense that the pavement responded more stiffly as more permanent
deflections and creep strains were accumulated, since this implies that
either settlement, consolidation, or both were taking place, and these
are normally associated with an increase in stiffness.
These observations resulted in increased confidence in all the
measurements obtained. Therefore, it appeared that the creep test
results substantiated the observation made from initial dynamic load
test results: that aside from changing the stiffness of the asphalt
concrete and possibly inducing thermal stresses in the pavement,
temperature changes were having another, possibly more significant
effect, on the pavement's behavior.


131
positions of the tests performed on the limerock base. The moisture
content (w) and dry density (y^) measured at each test position is also
shown on these figures.
Stress-deflection curves were plotted for each plate load test
performed. Typical examples of these plots for tests performed on the
sand subgrade are shown in Figure 5.4 and 5.5 (12- and 30-inch plate,
respectively). Both the 12- and 16-inch plate tests on the sand
resulted in a linear stress-deflection relationship after the first
loading cycle, indicating that this material was stress independent
within the range of stresses applied. However, a curvilinear
relationship was observed for the 30-inch plate on the sand, even though
the same stress levels were used. The apparent reason for the nonlinear
response observed with the larger plate is discussed later in this
section.
Static and dynamic moduli for Fairbanks sand subgrade were
calculated based on resilient plate deflections, using the procedure
outlined in Section 4.4.1.1, with an assumed Poisson's ratio of 0.3.
The results of these calculations are given in Table 5.9. The
calculated moduli were very consistent for both the 12-and 16-inch plate
tests, except for test number one with the 16-inch plate, where a higher
water content and lower dry density were measured. Almost no difference
(0.5 percent) was observed between the average static and dynamic
modulus for the 12-inch plate tests, and only a small difference (4.6
percent) was observed for the 16-inch plate tests.
Since the stress-deflection relationship for the 30-inch plate test
was curvilinear, two modulus values were calculated: a resilient modulus
and a tangent modulus. Although the resilient modulus was unreasonably


Table E.17: Measured Permanent Deflections For Different Times of 10,000-lb. Static Load Application,
Position Number 2, 0 C (32 F)
Load^
Longitudinal Deflections
(E-3 in.)
(b)
Transverse Deflections (E-3 in.)
Duration
Distance
from Load
Center
(in.): N-
S(c)
Distance
from Load
Center (in.):
: E-W{di
(Seconds)
0
8(S)
12(N)
16 (S)
24(N)
36 (S)
13(W)
18(E)
24 (W)
36(E)
50
3.55
4.26
2.17
3.38
1.21
0.0
2.39
3.22
1.48
8
50
1.35
1.42
0.36
1.69
0.0
0.0
.48
2.98
0.49
6.4
400
0.49
2.84
-0.12
5.80
-0.60
0.0
1.07
7.19
0.99
13.6
500
0.24
2.72
-0.12
8.46
0.12
-0.2
1.31
1.98
1.49
2.5
(a) Loads were applied sequentially with a rest period of four times the load duration between loads.
(b) Measured permanent deflections are the residual deflections after the rest period. Positive is
downward.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.


Distance From Center of Test Position (ins )
Figure 8.33: Measured Longitudinal Deflections at 6.7 C (44 F): Test Position 3


167
Figure 7.6: Load-Strain Relationships: Fast Loading Rate


Table D.13: Measured Deflections, Test Position 2, 13.3 C (56 F)
Load
Longitudinal Deflections
(E-3 in
.) (a)
Transverse Deflections (E-3
in.)
(kips)
Distance
from Load
Center
(in.):
N-S(b)
Distance
from Load
Center (in
.): E-W(c)
0 8(S)
12 (M)
16 {S)
24 (N)
36 (S)
13 (W)
18(E)
24 (W)
36(E)
10
9.55 3.08
5.78
-.24
1.21
0
8.12
3.97
2.22
.25
CO
7
7.10 2.13
4.10
-.20
.72
0
5.97
2.48
1.48
0
00
4
4.16 1.18
2.41
-.20
.24
0
3.34
1.24
.49
0
1
.49 0
.24
0
0
0
.72
-.12
0
0
(a) Positive is downward deflection.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.


LIST OF FIGURES
Figure Page
3.1 Hopper for Asphalt Hot Mix Distribution 44
3.2 Layout of Test Pit Cooling System 46
3.3 Insulated Test Pit Cover Completely Installed 48
3.4 Insulated Test Pit Cover With Panels Removed 48
3.5 LVDT Support System for Plate Loading 50
3.6 LVDT Support System for Dual Wheel Loading 51
3.7 LVDT Prepared for Tests at Low Temperatures 53
3.8 Test Pit Pavement Completely Instrumented and Ready
for Testing With Dual Wheels
a) Frontal View 54
b) Diagonal View 54
3.9 Two-Inch Strain Gages Mounted on Asphalt Concrete 56
3.10 Schematic Diagram of Data Acquisition System 59
3.11 Data Acquisition System In Test Pit Facility 60
3.12 Typical Deflection Recording on Digital Oscilloscope 60
3.13 Typical Deflection Output on X-Y Plotter 61
3.14 Rigid Plate Loading System 63
3.15 Flexible Dual Wheel Loading System 64
4.1 Measured and Predicted Deflection Basins in the Test Pit 69
4.2 Effect of Concrete Floor at Different Depths on
Predicted Deflection Basins 71
4.3 Effect of Different Base Layer Stiffness on
Predicted Deflection Basins 80
4.4 Effect of Test Pit Walls on Limerock Base Response 82
4.5 Comparison of AXSYM and Elastic Layer Theory Solutions 84
4.6 Three-Layer Systems as Modeled for Analysis 86
4.7 Pavement System Models to Determine Wall Effect 89
4.8 Equivalent Systems Based on Maximum Plate Deflection
on Subgrade 94
4.9 Comparison of Response of Equivalent Systems Based
on Maximum Plate Deflection on Subgrade 95
4.10 Comparison of Test Pit System and Burmister System 97
4.11 Stress Distribution Under Rigid Plate on
Semi-Infinite Mass: 50 psi Average Pressure 101
4.12 Measured vs. Predicted Deflection Basins at
18.3C (65 F) 107
5.1 Viscosity Temperature Relationships for Asphalt
Recovered from the Test Pit 123
5.2 Location of Plate Load Tests: Fairbanks Sand Subgrade 130
5.3 Location of Plate Load Tests: Limerock Base 130
5.4 Applied Stress vs. Deflection: 12-in. Plate on
Fairbanks Sand Subgrade 132
x


50
PLAN VIEW
Section A-A
Figure 3.5: LYDT Support System for Plate Loading


Table E.7: Measured Dynamic Deflections at 10,000 lbs. After Different Times of Static Load Application,
Position Number 1, 6.7 C (44 F)
Load^
Longitudinal Deflections
(E-3 in.)
(b)
Transverse Deflections (E-3
in.)
Duration
Distance
from Load
Center
(in.): N-
s(c)
Distance
from Load
Center (in.
): E-W
(Seconds)
0
8(N)
12 (S)
18 (S)
24(N)
36 (N)
13 (W)
18(E)
24(W)
36(E)
0
10.87
6.63
5.78
3.33
0.97
0
7.64
3.97
2.87
.2
50
10.18
6.58
5.78
3.28
1.11
0
7.52
3.97
2.96
.3
100
10.67
6.48
5.74
3.33
1.06
0
7.64
3.72
2.96
.3
500
10.67
6.53
5.74
3.26
1.11
0
7.55
3.57
2.96
.3
1000
10.87
6.67
5.74
3.33
1.01
0
7.74
3.87
2.86
.3
(a) A rest period equal to four times the load duration was allowed before dynamic testing.
(b) Positive is downward.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.


LOW-TEMPERATURE RESPONSE
OF ASPHALT CONCRETE PAVEMENTS
By
REYNALDO ROQUE
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
1986


34
unbound materials should vary both horizontally and vertically in the
pavement structure according to equations equating modulus with state of
stress (i.e. Mr = Ao^8 for fine-qrained materials and = k1k2 for
granular materials). They stated that although this variation can be
satisfactorily represented by finite element models (F.E.M.), these
models have the following disadvantages: 1) they usually require large
amounts of computer time; 2) the variability in pavement performance
data may make their precision superfluous; and 3) F.E.M.'s are usually
too complex and consume too much time to be used routinely in pavement
management systems. They concluded that elastic layer theory with an
equivalent effective modulus gives reasonable response prediction.
Brown and Pappin (89) seem to disagree. They developed a contour model
to predict non-linear behavior and lack of tensile strength in soils and
incorporated this model into the finite element program SENOL. Develop
ment and limited verification of their model involved both laboratory
tests and full-scale testing of pavements under controlled conditions.
Based on a parametric study and limited measurements they determined the
following:
- deflections and strains in the asphalt layer can be reasonably
predicted using an equivalent effective modulus in elastic layer
theory;
- it is unlikely that elastic layer theory can be used to predict
stresses and strains within the unbound layers;
- they suggest that the K-0 approach is less satisfactory than a
linear elastic solution; and
- the simplest approach for design calculations is the use of a
linear elastic system provided adequate equivalent moduli are used.


106
0.1 seconds loading time. Previously determined subgrade and limerock
base moduli were used. The pavement system was modeled as follows:
Layer Poisson's
Material Thickness (in.) Modulus Ratio
Asphalt Concrete
4 1/8
(a)
0.35
Limerock Base
6 3/4
53,000
0.40
Sand Subgrade
48
15,420
0.30
Concrete Embankment
semi-infinite
3,500,000
0.20
(a) Modulus of asphalt concrete to be determined by trial and error.
The modulus of the asphalt concrete layer was adjusted by trial and error
until the predicted plate deflection matched the measured plate
deflection, adjusted for wall effect. The wall effect for this system
was determined to be 1.0 E-3 in. in Section 4.3.4.
This procedure resulted in a predicted asphalt concrete modulus of
170,000 psi, which is within 12 percent of the modulus of 145,000 psi
determined from correlations with measured asphalt viscosity. This
prediction is very reasonable, and it should be emphasized that these
results were obtained from the rational evaluation of direct
measurements. Therefore, it seems that the analysis methods used to
evaluate the measurements are reasonable.
Deflections and strains were measured at several points throughout
the pavement and these were used to further evaluate the analysis
methods. Figure 4.12 shows a comparison of the measured and predicted
deflection basins. The plate deflections match exactly, since the
asphalt concrete modulus was adjusted to match the measured plate
deflection. In addition, there is good correspondence for the entire
deflection basin.


DEFLECTION (E-3 ins)
DISTANCE FROM LOAD CENTER (ins)
Figure 7.29: Measured vs. Predicted Deflections at 10,000 lbs.: E2 = 53,000 psi, 20.6 C (69 F)


Table E.20: Measured Dynamic Strains at 10,000 lbs. After Different Times of Static Load Application,
Position Number 2, 0 C (32 F)
Load^
Longitudinal
Strains
(micro-
strain) ^)
Transverse Strains
(micro-strain)
Duration
Distance from
Load Center (in
.): N-S(c)
Distance
from Load Center (in.
): E-W(d>
(Seconds)
0
8{S)
12(N)
16 (S)
24(N) 36(S)
12(W)
18(E)
24 (W)
36(E)
0
-154
12
-17
45
20
13
6
12
25
13
50
-146
-7
-20
40
16
12
6
16
25
12
100
-146
-9
-13
38
16
12
6
24
25
11
500
-150
-12
3
37
15
12
6
40
26
10
1000
-165
-16
14
37
18
12
10
45
28
11
(a) A rest period equal to four times the load duration was allowed before dynamic testing.
(b) Tension is positive.
(c) N is for north, S is for south of load center.
(d) E is for east, Vi is for west of load center.


170
described in Section 4.4.2 was used. The response predicted by the
program was compared to the measured pavement response in the test pit.
A comparison of measured and predicted deflection basins for a load
of 10,000 lbs. is shown in Figure 4.12. Measured and predicted strains
are compared in Table 4.5. These comparisons show that both deflections
and strains were accurately predicted by the program with the parameters
determined above.
7.2 Initial Plate Load Tests at Slow Loading Rate
Test, series number 2, listed in Table 7.1, were performed a few
days after test series number 1. The tests were performed at the same
loading position. The instrumentation layout used for these tests was
the same as for the fast loading rate, which was shown in Figures 7.2.
7.2.1 Dynamic Load Test Results
The measured deflections and strains for this series of tests are
given in Tables 7.4 and 7.5. The measurements shown are the average of
two readings taken on either side of the load. As for the earlier
tests, the readings were repeatable and nearly identical for gages
located at equal distances, but on opposite sides of the load. Measured
deflection basins and surface strain distributions are shown in
Figure 7.8 and 7.9, respectively.
7.2.2 Evaluation of Results
The deflection basins shown in Figure 7.8 clearly show that
excessive plate indentation was occurring for the slow loading rate. As
the load was increased, the plate deflections seemed to increase


103
distribution under the plate by trial and error. Therefore, the method
was considered impractical.
4.4.2.1 Procedure to Determine the Base Layer Modulus in the Test Pit
Two procedures were established to determine the base layer modulus
from plate deflections measured in the test pit. The first method
involved adjusting the measured deflections so that Burmister's theory
could be used. The second method involved approximating the stresses
under a rigid plate using an elastic layer computer program and
determining the base modulus by trial and error until the measured
deflections were matched. Modulus values predicted by both methods were
almost identical. However, the second method was very tedious and was
considered impractical. Therefore, the first method is recommended to
determine the base layer modulus.
The following procedure should be followed:
1. Calculate the subgrade modulus as per Section 4.4.1.1.
2. Determine the effect of the floor on deflections as follows:
a. Assume a reasonable base layer modulus and make two runs
with the elastic layer computer program: one for a two-
layer semi-infinite system, and one for a two-layer system
underlain by a rigid base. The rigid base can modeled using
a concrete modulus (3,500,000 psi) or higher.
b. Calculate the difference between the maximum deflections
predicted by the program for these two systems. This
difference is the floor effect.
3. If the Poisson's ratio for the suhgrade or base layers are
assumed different than 0.5, their effect on deflections relative


APPENDIX C
MEASURED THERMAL STRAINS DURING COOLING


South
NOTE:
Tensile Strain (E- 6in/in)
400
200
s
1 1 42 36 30 24 ,/|8 12
zf
Tensile strains at
bottom of pavement
~i - North
42
*1 A
200
400
Distance from center
of test position (ins)
CREEP STRAINS SHOWN ARE
RESIDUAL STRAINS AFTER A
1000-sec. DURATION OF 10,000-lb.
STATIC LOAD FOLLOWED BY A
4000-sec. REST PERIOD.
600
eoo
x 0.0 C (32 F)
O 6.7 C (44F)
13.3 C(32F)
IOOO
f Compressive Strain (E-6 in/ln)
Figure 8.60: Comparison of Longitudinal Creep Strains at Different Temperatures: Test Position 1


130
Figure 5.2: Location of Plate Load Tests: Fairbanks Sand Subgrade
Figure 5.3: Location of Plate Load Tests: Limerock Base


Figure 3.14: Rigid Plate Loading System


58
- a V/E 20 AJMLH strain gage indicator; and
- a digital oscilloscope with floppy disc recording system.
These units could handle 10 strain gages, in either a 1/4-, 1/2-, or
full-bridge arrangement. However, only one gage could be monitored at
any given time with the strain gage indicator. Output from the
indicator was sent to the digital oscilloscope for continuous recording
with time.
Deflection and strain measurements stored on floppy discs, were
later output to calibrated X-Y plotters. The plotters were calibrated
for an average LVDT output, since each LVDT had a slightly different
calibration. The measurements, as determined from the X-Y plotter
output, were then adjusted for the calibration factor of the particular
LVDT used. All cables going to the LVDT's and the strain gages were
passed through an access hole drilled through the side of the insulated
test pit cover.
A schematic diagram of the data acquisition system is shown in
Figure 3.10. A picture of the system is shown in Figure 3.11. Figure
3.12 is a picture of a typical recording of deflections on a digital
oscilloscope and Figure 3.13 shows a typical output recorded on an X-Y
plotter.
Temperature measurements were obtained with a Fluke Model 2240A
Datalogger. This unit used thermocouple wires to receive and record
temperature measurements for up to 20 different positions at one time.
It could record temperatures at specified, time intervals or could be
triggered to record at any given time. The unit automatically records
the date and time of the readings.


Table E.9: Measured Permanent Deflections For Different Times of 10,000-lb. Static Load Application,
Position Number 1, 0 C (32 F)
Load^
Longitudinal Deflections
(E-3 in.)(b)
Transverse Deflections (E-3
in.)
Duration
Distance from Load
Center
(in.): N-S^
Distance
from Load
Center (in.
): E-W(di
(Seconds)
0
8(N) 12(S)
18 (S)
24(N) 36 (N)
13 (W)
18(E)
24 (W)
36(E)
50
......
_
0.91
2.28
0.10
1.7
50
0.24
-0.05 -0.10
-0.24
-0.24 0.4
0.24
1.24
0.0
0.7
400
-0.15
0.05 -0.43
-0.53
-0.48 1.1
-0.19
0.84
-0.39
0.2
500
-0.20
-0.71 -0.63
-0.82
-1.11 -.5
0.24
0.60
-0.25
0.0
(a) Loads were applied sequentially with a rest period of four times the load duration between loads.
(b) Measured permanent deflections are the residual deflections after the rest period. Positive is
downward.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.


South-*
NOTE
GJ
O
o
Figure 8.59: Comparison of Permanent Longitudinal Deflections at Different Temperatures: Test Position 1


24
been subsequently recommended for determining loading times for dynamic
laboratory tests (71).
Investigations have also been conducted to evaluate different test
methods and differences in tensile and compressive properties. Kallas
(72) compared dynamic moduli of dense graded mixtures with normal air
voids at 4.4, 21.1, and 37.8 C (40, 70, and 100 F) and frequencies of 1,
4, and 16 cycles per second in tension, tension-compression, and
compression. He reported small differences in dynamic modulus in
tension, tension-compression, and compression for temperatures between
4.4 and 21.1 C (40 and 70 F) and frequencies between 1 and 16 cycles per
second. However, at 1 cycle per second and temperatures between 21.1
and 37.8 C (70 and 100 F), the dynamic tension or tension-compression
moduli averaged 1/2 to 2/3 of moduli in compression. The viscous
component of response was found to be considerably greater in tension
than in compression for frequencies of 1 to 16 cycles per second and
temperatures from 21.1 to 37.8 C (40 to 100 F). Wallace and Monismith
(73) compared moduli results from diametial indirect tension and
triaxial tests and analyzed the effect of anisotropy on both testing
procedures. They estimated that a more relevant assessment of modulus
could be obtained from diametral tests.
Although direct characterization is one approach to obtain mixture
parameters, the cost is prohibitive for most agencies (27), particularly
considering the variable properties of asphalts with temperature and
stress, and the fact that these properties can vary significantly for
different asphalts. Therefore, several investigators have established
relationships to predict mixture response parameters, based on
temperature, time of loading, and material characteristics from


DISTANCE FROM LOAD CENTER (ins)
Figure 7.31: Measured vs. Predicted Deflections at 10,000 lbs.: E¡> = 75,000 psi, 20.6 C (69 F)


6
\Q
o^'*
,ied Stress vs. _
rbanks ^and Subgrade
f\ec
Xo*
*.*
Applleu _
Fai "!/'i Sand


293
observations that the measurements obtained were a true representation
of the dynamic load response of the pavement.
The uncommon response observed at different temperatures and
different positions in the test pit, indicated that changes in
temperature were having a significant effect on the pavement's behavior
that could not be explained by the expected changes in the stiffness of
the asphalt concrete layer. In other words, aside from increasing the
pavement stiffness and possibly inducing thermal stresses in the
pavement, the temperature reduction apparently had another effect on the
pavement in the test pit. This other effect resulted in deflections and
strains, and thus stresses, at lower temperatures, that were more than
double those expected and normally used for design. This is evident by
comparing the response measured at 0.0 C (32 F) at test position 3
(Figures 8.49 and 8.50) with the response predicted by elastic layer
theory for this temperature (Figures 8.53 and 8.54).
The high stresses resulting from this temperature effect combined
with the fact that the asphalt concrete is more brittle at lower
temperatures, may create a very critical condition in the pavement. If
this phenomenoni occurs in the field, it could explain the cause of
certain early pavement failures as well as the source of longitudinal
wheel-path cracking. Therefore, this observation was very significant,
and it is important to determine the mechanism that led to the observed
response. However, before specul ating on the cause of this phenomenon,
the creep test results will be presented, since these results provided
additional insight into the observed behavior.
Longitudinal and transverse permanent deflection and creep strain
distributions were plotted for all creep tests performed at each


LOW-TEMPERATURE RESPONSE
OF ASPHALT CONCRETE PAVEMENTS
By
REYNALDO ROQUE
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
1986

ACKNOWLEDGMENTS
I would like to express my gratitude to Dr. Byron E. Ruth, Chairman
of my Graduate Supervisory Committee, for his guidance, encouragement,
and friendship. I would also like to thank Dr. F. C. Townsend,
Dr. J. L. Davidson, Professor W. H. Zimpfer, Dr. M. C. McVay,
Dr. D. L. Smith, and Dr. J. L. Eades for serving on my Graduate
Supervisory Committee. I consider myself fortunate to have had such a
distinguished committee.
A very special thanks goes to the Florida Department of
Transportation (FDOT) for providing the financial support, testing
facilities, materials, and personnel that made this research possible.
I would also like to thank the many individuals at the Bureau of
Materials and Research of the FDOT who contributed to this research
project by giving so generously of their time. I especially want to
thank the personnel in the Pavement Performance Division, Bituminous
Materials and Research Section, the Pavement Evaluation Section and the
Soil Materials and Research Section for their help and consideration.
I would also like to thank Candace Leggett for her
conscientiousness and diligence in typing this dissertation.
Finally, I would like to thank my wife Maria for encouraging me to
return to school, and for her encouragement and patience throughout my
Ph.D. program.
ii

TABLE OF CONTENTS
Page
ACKNOWLEDGMENTS ii
LIST OF TABLES vi
LIST OF FIGURES x
ABSTRACT xv 1
CHAPTERS
I INTRODUCTION 1
II LITERATURE REVIEW 5
2.1 Introduction 5
2.2 Distress In Asphalt Concrete Pavements 6
2.2.1 Modes of Distress 6
2.2.2 Cracking Mechanisms 9
2.3 Properties of Asphalt Cement and Asphalt Concrete
As Related to Low-Temperature Pavement Response
and Cracking 15
2.3.1 Asphalt Cement Properties 16
2.3.2 Asphalt Mixture Properties 20
2.4 Properties of Foundation Materials 30
2.5 Prediction of Thermal- and Load-Induced Stresses,
Strains, and Failure In Asphalt Concrete Pavements 35
III EQUIPMENT AND FACILITIES 42
3.1 Description of Test Pit Facility 42
3.2 System for Hot Mix Asphalt Distribution 43
3.3 Pavement Cooling System 43
3.4 Measurement System for Pavement Response 47
3.4.1 Measuring Instruments 49
3.4.2 Data Acquisition System 55
3.5 Loading Systems: Rigid Plate Load vs. Flexible
Dual Wheels 62
IV EFFECT OF ENCLOSED CONCRETE TEST PIT ON PAVEMENT RESPONSE.... 67
4.1 Introduction 67
4.2 Preliminary Analysis 67
4.3 Effect of Test Pit Constraints 72
4.3.1 Analytical Model 72
i i i

4.3.2 Effect of Constraints on Subgrade Response 73
4.3.3 Effect of Constraints on Limerock Base Response 77
4.3.4 Effect of Constraints on Three-Layer System 85
4.4Methodology to Account for the Effect of Test Pit
Constraints on Pavement Response Prediction 91
4.4.1 Rigid Plate Loading on the Subgrade 91
4.4.2 Rigid Plate Loading on the Reinforcing Base
Layer 92
4.4.3 Predicting Pavement System Response in the
Test Pit 104
VMATERIALS AND PLATE TESTING PROCEDURES 113
5.1 Introduction 113
5.2 Laboratory Tests 113
5.2.1 Fairbanks Sand Subgrade 113
5.2.2 Crushed Limerock Base 113
5.2.3 Asphalt Cement and Asphalt Concrete 115
5.3 Material Placement and Compaction 122
5.4 Material Properties In Situ 126
5.4.1 Plate Load Test Procedures 126
5.4.2 Plate Tests Immediately After Placement 129
5.4.3 Plate Tests After Pavement Removal 141
VIPROCEDURES 150
6.1 Dynamic Plate Load Tests at Ambient Temperatures 150
6.2 Low-Temperature Pavement Response Tests 151
6.2.1 Introduction 151
6.2.2 Pavement Cooling and Initial Dynamic Load Tests 153
6.2.3 Creep Test Procedures 155
VIIPAVEMENT RESPONSE AT AMBIENT TEMPERATURES 157
7.1 Initial Plate Load Tests at Fast Loading Rate 159
7.1.1 Dynamic Load Test Results 159
7.1.2 Elastic Layer Simulation and Evaluation of
Results 163
7.2 Initial Plate Load Tests at Slow Loading Rate 170
7.2.1 Dynamic Load Test Results 170
7.2.2 Evaluation of Results 170
7.3 Additional Plate Load Tests at Fast Loading Rate 182
7.3.1 Dynamic Load Test Results 182
7.3.2 Elastic Layer Simulation and Evaluation of
Results 186
7.4 Summary 213

VIII LOW-TEMPERATURE PAVEMENT RESPONSE 216
8.1 Preliminary Tests With the Rigid Plate 216
8.2 Reinstrumentation for Tests With the Dual Wheels 221
8.3 Results of Tests With the Dual Wheel Loading System 221
8.3.1 Introduction 221
8.3.2 Pavement Response During Cooling 223
8.3.3 Dynamic Load and Creep Response at Different
Temperatures 255
8.3.4 Combined Effect of Thermal and Load Response 320
IX RESPONSE PREDICTION AT LOW TEMPERATURES 331
9.1 Dynamic Load Response 331
9.2 Thermal Response 347
9.3 Creep Response 348
X CONCLUSIONS A RECOMMENDATIONS 350
10.1 Conclusions 350
10.1.1 Pavement Testinq and Evaluation Method 350
10.1.2 Thermal and Load Response of Asphalt Concrete
Pavements 356
10.2 Recommendations 356
APPENDICES
A RELATIONSHIPS BETWEEN ASPHALT CONCRETE PROPERTIES
AND ASPHALT CEMENT PROPERTIES 359
B PAVEMENT TEMPERATURES DURING COOLING 361
C MEASURED THERMAL STRAINS DURING COOLING 368
D DYNAMIC LOAD RESPONSE MEASUREMENTS 375
E CREEP TEST DATA 396
REFERENCES 433
BIOGRAPHICAL SKETCH 443
v

LIST OF TABLES
Table Page
2.1 Primary Types and Causes of Distress In Asphalt
Concrete Pavements 7
2.2 Modes, Manifestations, and Mechanisms of Types of
Distress 8
4.1 Sand Subgrade Modulus for Different Layer Depths
and Poisson's Ratio 75
4.2 Effect of Concrete Floor on Surface Deflections
for Different Rase Stiffnesses 79
4.3 Predicted Deflections Using AXSYM 83
4.4 Tabulated Deflection Basins to Show Effect of
Test Pit Floor on Pavements of Different Stiffness 87
4.5 Effect of Test Pit Walls on Surface Deflections
for Pavements of Different Stiffness 90
4.6 Computer Runs to Determine Poisson's Ratio Effect 99
4.7 Measured and Predicted Surface Strains 109
5.1 Laboratory Test Results: Fairbanks Sand 114
5.2 Laboratory Test Results: Ocala Formation Limerock 116
5.3 Source of Materials and Job Mix Formula for Asphalt
Concrete 117
5.4 Test Pit Asphalt Concrete Properties 118
5.5 Rheology and Penetration of Asphalt Recovered From
Test Pit During Initial Placement: September, 1982 120
5.6 Rheology and Penetration of Asphalt Recovered From
Test Pit After All Testing: September, 1985 121
5.7 Load Increments Used for Plate Load Tests: Fairbanks
Sand Subgrade 127
5.8 Load Increments Used for Plate Load Tests:
Limerock Base 128
5.9 Modulus Values Immediately After Placement:
Fairbanks Sand Subgrade 134
5.10 Modulus Values Immediately After Placement:
Limerock Base 139
5.11 Modulus Values Without Accounting for Test Pit
Constraints 142
5.12 Modulus Values After Pavement Removal:
Fairbanks Sand Subgrade 145
5.13 Modulus Values After Pavement Removal: Limerock Base 146
6.1 Summary of Order of Testing 152
7.1 Summary of Dynamic Plate Load Tests at Ambient
Temperatures 158
vi

Tab! e
Page
7.2 Surface Deflections: Fast Loading Rate 162
7.3 Surface Strains: Fast Loading Rate 162
7.4 Surface Deflections: Slow Loading Rate 173
7.5 Surface Strains: Slow Loading Rate 173
7.6 Measured Surface Deflections at 20.6 C (69 F):
Center Plate Loading Position 184
7.7 Measured Surface Strains at 20.6 C (69 F):
Center Plate Loading Position 184
7.8 Measured Surface Deflection at 25.6 C (78 F):
South Plate Loading Position 185
7.9 Measured Strains at 25.6 (78 F):
South Plate Loading Position 185
8.1 Summary of Average Pavement Temperatures During Testing 259
B.l Pavement Temperatures During Cooling: Test Position 1 362
B.2 Pavement Temperatures During Cooling: Test Position 2 364
B.3 Pavement Temperatures During Cooling: Test Position 3 366
C.l Measured Thermal Strains During Cooling: Test Position 1.... 369
C.2 Measured Thermal Strains During Cooling: Test Position^.... 371
C.3 Measured Thermal Strains During Cooling: Test Position 3.... 373
D.l Measured Deflections, Test Position 1, 0 C (32 F) 376
D.2 Measured Strains, Test Position 1, 0 C (32 F) 377
D.3 Measured Deflections, Test Position 1, 6.7 C (44 F) 378
D.4 Measured Strains, Test Position 1, 6.7 C (44 F) 379
D.5 Measured Deflections, Test Position 1, 13.3 C (56 F) 380
D.6 Measured Strains, Test Position 1, 13.3 C (56 F) 381
D.7 Measured Deflections, Test Position 1, 0 C (32 F),
Repeat Test 382
D.8 Measured Strains, Test Position 1, 0 C (32 F),
Repeat Test 383
D.9 Measured Deflections, Test Position 2, 0 C (32 F) 384
D.10 Measured Strains, Test Position 2, 0 C (32 F) 385
D.ll Measured Deflections, Test Position 2, 6.7 C (44 F) 386
D.12 Measured Strains, Test Position 2, 6.7 C (44 F) 387
D.13 Measured Deflections, Test Position 2, 13.3 C (56 F) 388
D.14 Measured Strains, Test Position 2, 13.3 C (56 F) 389
D.15 Measured Deflections, Test Position 3, 0 C (32 F) 390
D.16 Measured Strains, Test Position 3, 0 C (32 F) 391
D.17 Measured Deflections, Test Position 3, 6.7 C (44 F) 392
D.18 Measured Strains, Test Position 3, 6.7 C (44 F) 393
D.19 Measured Deflections, Test Position 3, 13.3 C (56 F) 394
D.20 Measured Strains, Test Position 3, 13.3 C (56 F) 395
E.l Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 1,
13.3 C (56 F) 397
E.2 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position Number 1, 13.3 C (56 F).... 398
vii

Tab! e
Page
E.3 Measured Dynamic Deflections at 10,000 lbs. After
Different Times of Static Load Application, Position
Number 1, 13.3 C (56 F) 399
E.4 Measured Dynamic Strains at 10,000 lbs. After Different
Times of Static Load Application, Position Number 1,
13.3 C (56 F) 400
E.5 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 1,
6.7C (44 F) 401
E.6 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position Number 1, 6.7 C (44 F) 402
E.7 Measured Dynamic Deflections at 10,000 lbs. After
Different Times of Static Load Application, Position
Number 1, 6.7 C (44 F) 403
E.8 Measured Dynamic Strains at 10,000 lbs. After Different
Times of Static Load Application, Position Number 1,
6.7C (44 F) 404
E.9 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 1,
0 C (32 F) 405
E.10 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position Number 1, 0 C (32 F) 406
E.ll Measured Dynamic Deflections at 10,000 lbs. After
Different Times of Static Load Application, Position
Number 1, 0 C (32 F) 407
E.12 Measured Dynamic Strains at 10,000 lbs. After Different
Times of Static Load AoDlication, Positon Number 1,
0 C (32 F) 408
E.13 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 2,
6.7C (44 F) 409
E.14 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position Number 2, 6.7 C (44 F) 410
E.15 Measured Dynamic Deflections at 10,000 lbs. After
Different Times of Static Load Application, Position
Number 2, 6.7 C (44 F) 411
E.16 Measured Dynamic Strains at 10,000 Tbs. After Different
Times of Static Load Application, Position Number 2,
6.7C (44 F) 412
E.17 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 2,
0 C (32 F) 413
E.18 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position 2, 0 C (32 F) 414
E.19 Measured Dynamic Deflections at 10,000 lbs. After
Different Times of Static Load Application, Position
Number 2, 0 C (32 F) 415
E.20 Measured Dynamic Strains at 10,000 lbs. After Different
Times of Static Load Application, Position Number 2,
0 C (32 F) 416
viii

Table
Page
E.21 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 2,
13.3C (56 F) 417
E.22 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position Number 2, 13.3 C (56 F).... 418
E.23 Measured Dynamic Deflections at 10,000 lbs. After
Different Times of Static Load Application, Position
Humber 2, 13.3 C (56 F) 419
E.24 Measured Dynamic Strains at 10,000 lbs. After Different
Times of Static Load Application, Position Number 2,
13.3C (56 F) 420
E.25 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 3,
0 C (32 F) 421
E.26 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position Number 3, 0 C (32 F) 422
E.27 Measured Dynamic Deflections at 10,000 lbs. After
Different Times of Static Load Application, Position
Number 3, 0 C (32 F) 423
E.28 Measured Dynamic Strains at 10,000 lbs. After Different
Times of Static Load Application, Position Number 3,
0 C (32 F) 424
E.29 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 3,
6.7 C (44 F) 425
E.30 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position Number 3, 6.7 C (44 F) 426
E.31 Measured Dynamic Deflections at 10,000 lbs. After
Different Times of Static Load Application, Position
Number 3, 6.7 C (44 F) 427
E.32 Measured Dynamic Strains at 10,000 lbs. After Different
Times of Static Load Application, Position Number 3,
6.7 C (44 F) 428
E.33 Measured Permanent Deflections For Different Times of
10,000-lb. Static Load Application, Position Number 3,
13.3C (56 F) 429
E.34 Measured Creep Strains For Different Times of 10,000-lb.
Static Load Application, Position Number 3, 13.3 C (56 F).... 430
E.35 Measured Dynamic Deflections at 10,000 lbs. After
Different Times of Static Load Application, Position
Number 3, 13.3 C (56 F) 431
E.36 Measured Dynamic Strains at 10,000 lbs. After Different
Times of Static Load Application, Position Number 3,
13.3C (56 F) 432
ix

LIST OF FIGURES
Figure Page
3.1 Hopper for Asphalt Hot Mix Distribution 44
3.2 Layout of Test Pit Cooling System 46
3.3 Insulated Test Pit Cover Completely Installed 48
3.4 Insulated Test Pit Cover With Panels Removed 48
3.5 LVDT Support System for Plate Loading 50
3.6 LVDT Support System for Dual Wheel Loading 51
3.7 LVDT Prepared for Tests at Low Temperatures 53
3.8 Test Pit Pavement Completely Instrumented and Ready
for Testing With Dual Wheels
a) Frontal View 54
b) Diagonal View 54
3.9 Two-Inch Strain Gages Mounted on Asphalt Concrete 56
3.10 Schematic Diagram of Data Acquisition System 59
3.11 Data Acquisition System In Test Pit Facility 60
3.12 Typical Deflection Recording on Digital Oscilloscope 60
3.13 Typical Deflection Output on X-Y Plotter 61
3.14 Rigid Plate Loading System 63
3.15 Flexible Dual Wheel Loading System 64
4.1 Measured and Predicted Deflection Basins in the Test Pit 69
4.2 Effect of Concrete Floor at Different Depths on
Predicted Deflection Basins 71
4.3 Effect of Different Base Layer Stiffness on
Predicted Deflection Basins 80
4.4 Effect of Test Pit Walls on Limerock Base Response 82
4.5 Comparison of AXSYM and Elastic Layer Theory Solutions 84
4.6 Three-Layer Systems as Modeled for Analysis 86
4.7 Pavement System Models to Determine Wall Effect 89
4.8 Equivalent Systems Based on Maximum Plate Deflection
on Subgrade 94
4.9 Comparison of Response of Equivalent Systems Based
on Maximum Plate Deflection on Subgrade 95
4.10 Comparison of Test Pit System and Burmister System 97
4.11 Stress Distribution Under Rigid Plate on
Semi-Infinite Mass: 50 psi Average Pressure 101
4.12 Measured vs. Predicted Deflection Basins at
18.3C (65 F) 107
5.1 Viscosity Temperature Relationships for Asphalt
Recovered from the Test Pit 123
5.2 Location of Plate Load Tests: Fairbanks Sand Subgrade 130
5.3 Location of Plate Load Tests: Limerock Base 130
5.4 Applied Stress vs. Deflection: 12-in. Plate on
Fairbanks Sand Subgrade 132
x

Figure
Page
5.5 Applied Stress vs. Deflection: 30-in. Plate on
Fairbanks Sand Subgrade 133
5.6 Applied Stress vs. Deflection: 16-in. Plate on
Limerock Base 136
5.7 Deflections Used to Calculate Limerock Moduli 138
5.8 Location of Plate Load Tests: Fairbanks Sand Subgrade 144
5.9 Location of Plate Load Tests: Limerock Base 144
7.1 Test Pit Diagram: Elevation 160
7.2 Test Pit Diagram: Plan 161
7.3 Measured Deflection Basins: Fast Loading Rate 164
7.4 Measured Strain Distributions: Fast Loading Rate 165
7.5 Load-Deflection Relationships: Fast Loading Rate 166
7.6 Load-Strain Relationships: Fast Loading Rate 167
7.7 Test Pit Pavement System as Modeled for Elastic Layer
Analysis 169
7.8 Measured Deflection Basins: Slow Loading Rate 171
7.9 Measured Strain Distributions: Slow Loading Rate 172
7.10 Load-Deflection Relationships: Slow Loading Rate 175
7.11 Load-Strain Relationships: Slow Loading Rate... 176
7.12 Deflection Basin Comparison for Fast and Slow Loading
Rates: 10,000 lbs 177
7.13 Deflection Basin Comparison for Fast and Slow Loading
Rates: 7,000 lbs 178
7.14 Deflection Basin Comparison for Fast and Slow Loading
Rates: 4,000 lbs 179
7.15 Deflection Basin Comparison for Fast and Slow Loading
Rates: 1,000 lbs 180
7.16 Location of Plate Loading Positions and Strain and
Deflection Measurements 183
7.17 Measured Deflection Basins at 20.6 C (69 F) 187
7.18 Measured Deflection Basins at 25.6 C (78 F) 188
7.19 Measured Strain Distributions at 20.6 C (69 F) 189
7.20 Measured Strain Distributions at 25.6 C (69 F) 190
7.21 Deflection Basin Comparison at 10,000 lbs 191
7.22 Deflection Basin Comparison at 7,000 lbs 192
7.23 Deflection Basin Comparison at 4,000 lbs 193
7.24 Deflection Basin Comparison at 1,000 lbs 194
7.25 Strain Distribution Comparison at 10,000 lbs 195
7.26 Strain Distribution Comparison at 7,000 lbs 196
7.27 Strain Distribution Comparison at 4,000 lbs 197
7.28 Strain Distribution Comparison at 1,000 lbs 198
7.29 Measured vs. Predicted Deflections at 10,000 lbs.:
E2 = 53,000 psi, 20.6 C (69 F) 201
7.30 Measured vs. Predicted Strains at 10,000 lbs.:
E2 = 53,000 psi, 20.6 C (69 F) 202
7.31 Measured vs. Predicted Deflections at 10,000 lbs.:
E? = 75,000 psi, 20.6 C (69 F) 203
7.32 Measured vs. Predicted Strains at 10,000 lbs.:
E2 = 75,000 psi, 20.6 C (69 F) 204
7.33 Measured Load-Deflection Relationships at 20.6 C (69 F) 209
7.34 Measured Load-Strain Relationships at 20.6 C (69 F) 210
xi

Figure
Page
7.35 Measured vs. Predicted Deflections at 4,000 lbs.:
E2 = 40,000 psi, 20.6 C (69 F) 211
7.36 Measured vs. Predicted Strains at 4,000 lbs.:
E2 = 40,000 psi, 20.6 C (69 F) 212
8.1 Layout of Strain Gages and Cables in the Test Pit 222
8.2 Location of Test Positions in the Test Pit 224
8.3 Thermocouple and Strain Gage Location During Cooling:
Test Position 1 225
8.4 Thermocouple and Strain Gage Location During Cooling:
Test Position 2 226
8.5 Thermocouple and Strain Gage Location During Cooling:
Test Position 3 227
8.6 Measured Cooling Curves: Test Position 1 228
8.7 Measured Cooling Curves: Test Position 2 229
8.8 Measured Cooling Curves: Test Position 3 230
8.9 Change in Temperature Gradient During Cooling 232
8.10 Measured Longitudinal Strains vs. Temperature:
Test Position 1. 233
8.11 Measured Longitudinal Strains vs. Temperature:
Test Position 2 234
8.12 Measured Longitudinal Strains vs. Temperature:
Test Position 3 235
8.13 Measured Transverse Strains vs. Temperature:
Test Position 1 236
8.14 Measured Transverse Strains vs. Temperature:
Test Position 2 237
8.15 Measured Transverse Strains vs. Temperature:
Test Position 3 238
8.16 Longitudinal Strain Distributions During Cooling:
Test Position 1 242
8.17 Longitudinal Strain Distributions During Cooling:
Test Position 2 243
8.18 Longitudinal Strain Distributions During Cooling:
Test Position 3 244
8.19 Transverse Strain Distributions During Cooling:
Test Position 1 245
8.20 Transverse Strain Distributions During Cooling:
Test Position 2 246
8.21 Transverse Strain Distributions During Cooling:
Test Position 3 247
8.22 Comparison of Measured Thermal Strains for Different
Cooling Cycles: Six Feet from South Wall 250
8.23 Comparison of Measured Thermal Strains for Different
Cooling Cycles: 8.33 Feet from South Wall 251
8.24 Longitudinal Strain Distributions During Cooling 253
8.25 LVDT and Strain Gage Location During Load Tests:
Test Position 1 256
8.26 LVDT and Strain Gage Location During Load Tests:
Test Position 2 257
8.27 LVDT and Strain Gage Location During Load Tests:
Test Position 3 258
xii

Figure
Page
8.28 Load-Unload Times for Dynamic Loading with Dual Wheels 261
8.29 Measured Longitudinal Deflections at 0.0 C (32 F):
Test Position 3 263
8.30 Measured Longitudinal Strains at 0.0 C (32 F):
Test Position 3 264
8.31 Measured Transverse Deflections at 0.0 C (32 F):
Test Position 3 265
8.32 Measured Transverse Strains at 0.0 C (32 F):
Test Position 3 266
8.33 Measured Longitudinal Deflections at 5.7 C (44 F):
Test Position 3 267
8.34 Measured Longitudinal Strains at 6.7 C (44 F):
Test Position 3 268
8.35 Measured Transverse Deflections at 6.7 C (44 F):
Test Position 3 269
8.36 Measured Transverse Strains at 6.7 C (44 F):
Test Position 3 270
8.37 Measured Longitudinal Deflections at 13.3 C (56 F):
Test Position 3 271
8.38 Measured Longitudinal Strains at 13.3 C (56 F):
Test Position 3 272
8.39 Measured Transverse Deflections at 13.3 C (56 F):
Test Position 3 273
8.40 Measured Transverse Strains at 13.3 C (56 F):
Test Position 3 274
8.41 Comparison of Measured Longitudinal Deflections at
Different Temperatures: Test Position 1, 10,000 lbs 278
8.42 Comparison of Measured Longitudinal Strains at
Different Temperatures: Test Position 1, 10,000 lbs. 279
8.43 Comparison of Measured Transverse Deflections at
Different Temperatures: Test Position 1, 10,000 lbs 280
8.44 Comparison of Measured Transverse Strains at
Different Temperatures: Test Position 1, 10,000 lbs 281
8.45 Comparison of Measured Longitudinal Deflections at
Different Temperatures: Test Position 2, 10,000 lbs 282
8.46 Comparison of Measured Longitudinal Strains at
Different Temperatures: Test Position 2, 10,000 lbs 283
8.47 Comparison of Measured Transverse Deflections at
Different Temperatures: Test Position 2, 10,000 lbs 284
8.48 Comparison of Measured Transverse Strains at
Different Temperatures: Test Position 2, 10,000 lbs 285
8.49 Comparison of Measured Longitudinal Deflections at
Different Temperatures: Test Position 3, 10,000 lbs 286
8.50 Comparison of Measured Longitudinal Strains at
Different Temperatures: Test Position 3, 10,000 lbs 287
8.51 Comparison of Measured Transverse Deflections at
Different Temperatures: Test Position 3, 10,000 lbs 288
8.52 Comparison of Measured Transverse Strains at
Different Temperatures: Test Position 3, 10,000 lbs 289
8.53 Comparison of Predicted Longitudinal Deflections at
Different Temperatures: 10,000 lbs 290
xi i i

Figure
Page
8.54 Comparison of Predicted Longitudinal Strains at
Different Temperatures: 10,000 lbs 291
8.55 Permanent Longitudinal Deflections at 0.0 C (32 F):
Test Position 3 294
8.56 Longitudinal Creep Strains at 0.0 C (32 F):
Test Position 3 295
8.57 Permanent Transverse Deflections at 0.0 C (32 F):
Test Position 3 296
8.58 Transverse Creep Strains at 0.0 C (32 F):
Test Position 3 297
8.59 Comparison of Permanent Longitudinal Deflections at
Different Temperatures: Test Position 1 300
8.60 Comparison of Longitudinal Creep Strains at Different
Temperatures: Test Position 1 301
8.61 Comparison of Permanent Longitudinal Deflections at
Different Temperatures: Test Position 2 302
8.62 Comparison of Longitudinal Creep Strains at Different
Temperatures: Test Position 2 303
8.63 Comparison of Permanent Longitudinal Deflections at
Different Temperatures: Test Position 3 304
8.64 Comparison of Longitudinal Creep Strains at Different
Temperatures: Test Position 3 305
8.65 Comparison of Dynamic Load Response Immediately Prior
to Creep Tests for Different Test Positions:
0.0 C (32 F) 308
8.66 Comparison of Permanent Longitudinal Deflections for
Different Test Positions: .0 C (32 F) 309
8.67 Comparison of Dynamic Load Response at Different Times:
0.0 C (32 F), Test Position 1 310
8.68 Comparison of Dynamic Load Response at Different Times:
0.0 C (32 F), Test Position 2 311
8.69 Comparison of Dynamic Load Response at Different Times:
0.0 C (32 F), Test Position 3 312
8.70 Comparison of Dynamic Load Response After Creep Tests
for Different Test Positions: 13.3 C (56 F) 317
8.71 Comparison of Dynamic Load Response After Creep Tests
for Different Test Positions: 6.7 C (44 F) 318
8.72 Comparison of Dynamic Load Response After Creep Tests
for Different Test Positions: 0.0 C (32 F) 319
8.73 Comparison of Load-Deflection Relationships for
Different Test Positions: 0.0 C (32 F) 325
8.74 Comparison of Load-Deflection Relationships for
Different Test Positions: 6.7 C (44 F) 326
8.75 Comparison of Load-Deflection Relationships for
Different Test Positions: 13.3 C (56 F) 327
9.1 Comparison of Measured and Predicted Longitudinal
Deflection Basins at 0.0 C (32 F) 335
9.2 Comparison of Measured and Predicted Longitudinal
Strain Distributions at 0.0 C (32 F) 336
9.3 Comparison of Measured and Predicted Transverse
Deflection Basins at 0.0 C (32 F) 337
xiv

Figure Page
9.4 Comparison of Measured and Predicted Transverse
Strain Distributions at 0.0 C (32 F) 338
9.5 Comparison of Measured and Predicted Longitudinal
Deflection Basins at 6.7 C (44 F) 339
9.6 Comparison of Measured and Predicted Longitudinal
Strain Distributions at 6.7 C (44 F) 340
9.7 Comparison of Measured and Predicted Transverse
Deflection Basins at 6.7 C (44 F) 341
9.8 Comparison of Measured and Predicted Transverse
Strain Distributions at 6.7 C (44 F) 342
9.9 Comparison of Measured and Predicted Longitudinal
Deflection Basins at 13.3 C (56 F) 343
9.10 Comparison of Measured and Predicted Longitudinal
Strain Distributions at 13.3 C (56 F) 344
9.11 Comparison of Measured and Predicted Transverse
Deflection Basins at 13.3 C (56 F) 345
9.12 Comparison of Measured and Predicted Transverse
Strain Distributions at 13.3 C (56 F) 346
xv

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
LOW-TEMPERATURE RESPONSE
OF ASPHALT CONCRETE PAVEMENTS
By
REYNALDO ROQUE
May, 1986
Chairman: Byron E. Ruth
Major Department: Civil Engineering
The variable performance of asphalt concrete pavements indicates
that existing design procedures may be inadequate. However, before
improved design procedures can be developed, the mechanisms that lead to
pavement cracking must be fully understood. Therefore, this research
program was developed to monitor and define the response and failure
characteristics of pavements subjected to thermal and dynamic loading
conditions.
An asphalt concrete pavement was tested under controlled
temperature conditions in an enclosed concrete test pit. Thermal
contraction strains were measured as the pavement was cooled from room
temperature to temperatures below freezing. Load-induced deflections
and strains in the pavement were measured for dynamic load tests
performed at temperatures ranging from -6.7 C (20 F) to 25.6 C (78 F).
Finally, permanent deflections and creep strains were measured under
static loads at temperatures ranging from 0.0 C (32 F) to 13.3 C (56 F).
xv i

Load-induced deflections and strains measured at temperatures
ranging from 0.0 C (32 F) to 21.1 C (70 F) were accurately predicted
using linear elastic layer theory when suitable layer moduli were used
for input. Asphalt concrete moduli determined from correlations with
measured asphalt viscosity resulted in accurate prediction of measured
deflections and strains at all temperatures and load levels tested.
Suitable moduli for the subgrade and base layers were determined from
plate load tests performed on these materials in situ. The use of
proper analytical tools to evaluate the plate load test data was
critical in the determination of these moduli.
Thermal and load response measurement of the four-inch asphalt
concrete pavement indicated that temperature differentials produced by
rapid cooling caused the asphalt concrete layer to contract and bend in
such a way that it separated and uplifted from the base. The uplift
effect resulted in load-induced deflections, strains, and stresses at
0.0 C (32 F) that were in some cases more than double those expected for
pavements exhibiting elastic behavior. The exact mechanism that led to
the uplift phenomenon could not be determined from the measurements
obtained. Because of the uplift effect, measured creep strains and
failure could not be evaluated. Based on the findings from this
investigation, recommendations are presented for improved testing
procedures.

CHAPTER I
INTRODUCTION
The Florida Department of Transportation (FDOT) has had variable
performance with asnhalt concrete pavements. Some pavements have
developed cracking within less than five years, while others have given
satisfactory performance after many years. In all cases, cracking was
observed with Tittle or no distortion in the pavement layers and with no
apparent deficiencies in the asphalt concrete mixtures. The problem is
not unique to Florida. Pavement condition surveys of existing highways
and test roads around the United States indicate that traffic-associated
cracking is of major concern to highway engineers. Cracking is one of
the first indicators of distress observable in asphalt pavements and
often leads to other forms of distress.
The variable oerformance observed for pavements designed using
current design procedures indicates that these procedures are
deficient. Existing design procedures are emoirically derived based on
correlations of certain material or pavement system parameters with
observed field performance. These procedures consider two forms of
cracking: low-temperature thermally-induced cracking and traffic-load-
induced fatigue cracking. The most commonly proposed approach to limit
thermal cracking is to limit the asphalt stiffness as measured for a
minimum design temperature, where the limiting stiffness is usually
obtained from correlations with observed field performance. Design
thickness requirements to provide adequate fatigue life are established
1

2
by attempting to limit pavement deflections or strains under a given
design load. These procedures neglect that pavement deflections,
stresses, and strains cover a wide spectrum of values dependent on
temperature and climatic fluctuations. The variable properties of
individual asphalts at low temperatures and the combined effect of
thermally- and load-induced stresses are not considered. In addition,
the two modes of cracking considered cannot explain certain types of
failures commonly encountered in practice.
Therefore, it is necessary to develop improved design procedures to
reduce the cracking potential of pavements. However, this will be
difficult to accomplish until the mechanisms that lead to cracking are
fully understood. Although the causes or factors involved with cracking
are known, the actual mechanisms that lead to cracking have not been
identified with definitive measurements on full-scale pavements.
Investigations sponsored by the FDOT led to the establishment of the
hypothesis that cracking of asphalt concrete pavements is a brittle
failure induced by short-term repetitive loads and thermal stresses that
occur during cool weather when the asphalt stiffness is high. This
suggests that asphalt concrete pavements should be designed for a
critical condition where stresses or creep strains induced in the
pavement are of sufficient magnitude to produce cracking. This critical
condition may be a result of the combined effects of asphalt age
hardening, base and subgrade support, asphalt concrete modulus,
vehicular loads, pavement cooling rate, and temperature.
The research work done in Florida, along with the variable
performance of existing pavements, indicated that a research program
should be developed to monitor and define the behavior of pavements

3
subjected to thermal and dynamic loading conditions. This resulted in
the formulation of a research program to test full-scale asphalt
concrete pavement systems under controlled conditions.
An FOOT test nit facility was developed for this purpose, since
there are obvious problems associated with trying to monitor this type
of behavior in the field. The test pit facility made it possible to
construct a layered system of materials to simulate a flexible pavement
system in the field. In order to provide temperature control, a cooling
system with an insulated cover was installed in the test pit.
Thus having developed the capability to simulate temperature and
loading conditions encountered in actual roadways, this research program
was initiated in an attempt to satisfy the following objectives:
1. To measure and evaluate the response of asphalt concrete
pavements to changes in temperature, and determine the effect
of this response on the dynamic load response and failure
characteristics of the pavement.
2. To measure and evaluate the dynamic load response of asphalt
concrete pavements at different temperatures and load levels.
3. To measure and evaluate load-induced creep strains and
permanent deflections induced in asphalt concrete pavements at
different temperatures.
4. To compare the response measurements listed in items one, two,
and three to values predicted by theoretical stress-strain
distributions and parameters obtained from laboratory tests.
In order to meet these objectives, a complete series of tests was
performed on a pavement section that was typical for Florida. A
measurement and data acquisition system was installed that was capable

4
of obtaining static and dynamic deflection and strain measurements at
ten different points in the pavement at any given time. Thermal strains
developed in the asphalt concrete layer during cooling were measured for
several cooling cycles. Dynamic load tests were performed at
temperatures ranging from -6.7 C (20 F) to 25.6 C (78 F), using both
rigid plate and flexible dual wheel loading systems. Finally, permanent
deflections and creep strains were measured for specified durations of
static loads at temperatures ranging from 0.0 C (32 F) to 13.3 C
(56 F). Dynamic load tests were also performed at different times
during creep tests to observe the effect of creep on the dynamic load
response of the pavement. Although the results of these tests were not
entirely definitive, they emphasized the need to consider the combined
effect of temperature and load in the analysis of asphalt concrete
pavements.

CHAPTER II
LITERATURE REVIEW
2.1 Introduction
The research presented in this document focuses on defining and
predicting low temperature response and failure of asphalt concrete
pavements. This includes the response of pavements to changes in
temperature and the effect these changes have on the load response and
failure limits of the asphalt concrete layer. Two elements of the
analysis system considered here make it uni cue: the use of measured
rheological parameters of the asphalt at low temperatures to predict the
response and failure characteristics of the asphalt concrete; and the
fact that cracking is considered a short-term phenomenon that occurs
when the combined effect of temperature and traffic loads exceed the
failure limit of the asphalt concrete pavement. Although this approach
is totally different from traditional approaches, a review of the
literature will serve two purposes:
1) to establish the need and develop the rationale behind the
proposed method of analysis; and
2) to give an overview of existing knowledge of asphalt concrete
pavement response to temperature changes and traffic loads,
including an assessment of our ability to predict response and
failure.
5

6
2.2 Distress In Asphalt Concrete Pavements
2.2.1 Modes of Distress
The modes of distress in asphalt concrete pavements are well
recognized and the causes of distress, at least in general terms, are
also known. Tables 1 and 2 (1,2) are two examples of tables listing the
types and causes of distress in asphalt concrete pavements. These
tables show that failures can be grouped into three major categories:
cracking, rutting, and disintegration.
Pavement surveys around the country and the world indicate that of
these three categories, cracking is the major problem in terms of amount
and cost. Based on extensive observations by himself and others, Finn
(3) stated that traffic-associated cracking is the number one priority
item for improving and extending the performance of asphalt pavements.
Pedigo et al. (4) reviewed a great deal of work that has been done on
pavement distress and reached similar conclusions. Finn also stated
that traffic associated cracking is one of the first indicators of
distress observable in asphalt pavements, and that cracking is often
observed with little or no distortion. In reviewing the results of the
AASHO Road Test, he found that cracking led to other forms of distress
(such as rutting), and that more cracking occurred when the pavement was
cold than warm. However, there was a lack of information as to when and
where the first cracks occurred and how these cracks propagated.
Furthermore, asphalt properties were not measured at low temperatures.
Information of this type is lacking, even today. Measurements of
the environmental and loading conditions at the time of initial
cracking, along with relevant material properties, are crucial to the
development of damage criteria. Such information could not be found in

7
Table 2.1: Primary Types and Causes of Distress in Asphalt Concrete
Pavements. After Ruth, 1985 (1)
Type Of Distress
1. Rutting
2. Thermal Cracking
3. Load Associated or
Fatigue cracking
4. Combined Thermal and
Induced Cracking
5. Heaving (Localized
or Extensive: Frost
Boils, Ice Lenses)
6. Settlement and Slope
Failures
Causes or Contributing Factors
- Consolidation
Shear failure
Low stability
Abrasion
Traffic
High temperatures
- Thermal contraction
Shrinkage
Low temperatures
Fast rate of cooling
Excessively hard asphalts
Lack of snow cover (insulation)
- Traffic volume and loads
Deflection basin characteristics:
1. Layer moduli
2. Layer thickness
3. Asphalt viscosity
Climate Microclimate
1. Temperature
2. Drainage moisture variations
Material quality
- Combine factors in items 2 and 3
- Expansive soils
Frost susceptible soils
Drainage
Permeability
Capillarity
Depth and rate of frost
penetration
- Quality of in situ materials
Quality of construction
Drainage and moisture conditions
Mining activity
Karst terrain sinkholes and
cavity collapse

8
Table 2.2: Modes, Manifestations, and Mechanisms of Types of Distress.
After McCullough, 1971 (2)
Mode Manifestation
Mechanism
Fracture Cracking Excessive loading
Repeated loading (i.e., fatigue)
Thermal changes
Moisture changes
Slippage (horizontal forces)
Shrinkage
Spalling Excessive loading
Repeated loading (i.e., fatigue)
Thermal changes
Moisture changes
Distortion Permanent deformation Excessive loading
Time-dependent deformation
(e.g., creep)
Densification (i.e., compaction)
Consol idation
Swell ing
Faulting Excessive loading
Densification (i.e., compaction)
Consolidation
Swelling
Disintegration Stripping Adhesion (i.e., loss of bond)
Chemical reactivity
Abrasion by traffic
Raveling and scaling Adhesion (i.e., loss of bond)
Chemical reactivity
Abrasion by traffic
Degradation of aggregate
Durability of binder

9
the literature. Thus, although the causes of cracking are well known,
the actual mechanisms that lead to cracking have not been verified with
definitive measurements of actual failures on full-scale pavements.
Several mechanisms have been proposed that cannot account for basic
material response and failure characteristics, variability of
environment, and loading conditions encountered in actual pavements.
These have led to empirical design procedures, which are valid only for
the conditions from which they were derived.
2.2.2 Cracking Mechanisms
Traditionally, cracking has been broken down into traffic-load
induced and thermally-induced, with little consideration for the
combined effects of the two mechanisms. Low-temperature transverse
cracking has been recognized as the most common non-traffic associated
failure mode and is a serious problem in Canada and parts of the United
States (5,6,7). This type of failure is generally considered a
temperature phenomenon caused by low temperatures. As the pavement
temperature decreases the asphalt concrete wants to contract, but
contraction is resisted by the friction between the asphalt concrete
layer and the base and by the length of the roadway in the longitudinal
direction. This resistance results in tensile stresses in the pavement,
which are greatest in the longitudinal direction.
Several researchers have postulated that cracking occurs when these
thermally induced tensile stresses exceed the tensile strength of the
asphalt concrete (8, 9, 10). This mechanism has been confirmed by
laboratory and field investigations (7-13), and provides the basis for
the hypotheses that have been presented for low temperature cracking.

10
The rheological properties of the asphalt at low temperatures are
generally recognized as the most important factor in low-temperature
transverse cracking (5, 11, 14, 15, 16, and others). Many researchers
have associated low temperature cracking with properties such as asphalt
stiffness, viscosity, temperature susceptibility, and glass transition
temperature. These properties, of course, are all related to the
asphalt's ability to flow and thus relax stresses. All researchers have
found that the stiffer and more temperature susceptible the asphalt, the
greater the potential for cracking.
Probably the most commonly proposed approach to control thermal
cracking is to limit the asphalt stiffness as measured for a minimum
design temperature. McLeod (17) concluded that low temperature pavement
cracking is likely to occur whenever the stiffness of the pavement
attains a value of 6.9 E9 Pa (1.0 E6 psi) at a pavement depth of two
inches, at the minimum temperature encountered, and for a loading time
of 20,000 seconds. Fromm and Phang (18) proposed a value of 1.4 E8 Pa
(20,000 psi) at 10,000 seconds loading time. Gaw (19) reported that the
St. Anne test pavements cracked at an asphalt binder stiffness of 1.0 E9
Pa (145,000 psi) and a mixture stiffness of 2.0 E10 pa (2,900,000 psi)
at 1800 seconds loading time. Many researchers have found good
agreement between measured stiffness and observed cracking of pavements
in the field and confirmed that pavements using softer asphalts exhibit
less cracking (7, 12, 20, 21, 22).
Ruth (14) concluded that cracking would be reduced by using
asphalts with lower viscosities and improved rheological behavior at low
temperature. Fabb (23) concluded that low viscosity and low temperature

11
susceptibility are conducive to reducing the temperature at which
fracture occurs.
The advantage of using a softer binder, particularly one with a low
temperature susceptibility, was demonstrated by Hills and Brien (8).
Fromm and Phang (16) also reported that less temperature susceptible
asphalts were associated with pavements exhibiting less cracking.
Schmidt (24) suggested that the glass transition temperature of the
asphalt might be a more definitive measure of non-load associated
cracking than measured viscosities, since at temperatures lower than the
glass transition temperature the asphalt behaves elastically, while at
higher temperatures it exhibits viscoelastic response. Thus, below the
glass transition temperature there is almost no potential for stress
relaxation.
Other factors have been found to influence low temperature
transverse cracking, but to a lesser degree than asphalt properties.
Tuckett et al. (25) found that higher asphalt contents reduced thermal
cracking. Fabb (23) reported that increasing binder content reduced
thermal fracture, but only slightly. He also concluded that the
properties and grading of the aggregate had little or no effect on the
resistance of the asphalt concrete to thermal cracking. Cooling rate
was found to have little effect on the failure temperature by Fabb (23)
and Fromm and Phang (16). However, they only compared relatively high
cooling rates. Finally, results of the St. Anne Test Road indicated
that only half the frequency of low temperature cracking occurred in
10-inch pavements than did in 4-inch pavements (26).
The concept of fatigue is probably the most recognized concept that
has been suggested for use in the evaluation of traffic-load associated

12
failure (27-31). Fatigue distress is the phenomenon of fracture under
repeated stresses which are less than the tensile strength of the
material. Fatigue characterization of materials has been studied
extensively and there are innumerable references on this topic (e.g. 31-
39).
The philosophy behind the approach to the analysis and design of
asphalt concrete pavements considered in this thesis is totally
different from conventional approaches based on fatigue. In fact, the
fatigue concept is considered erroneous, and will not be covered in much
detail. Design procedures based on fatigue assume that there is some
average pavement condition for which an equivalent amount of damage will
be incurred under each passing wheel load. These procedures neglect
that deflections, strains, and stresses cover a wide spectrum of values
dependent on temperature and climatic fluctuations. They cannot
properly account for the variable properties of individual asphalts at
low temperatures.
Several researchers have proposed modifications to fatigue life
predictions based on temperature, recognizing that the fatigue life of
materials tested in the laboratory is dependent on temperature.
However, the basic concept of fatigue damage has remained unchanged.
Pxauhut and Kennedy (40) proposed one such modification and discuss
modifications proposed by other researchers. They also recognize that
fatigue damage is difficult to evaluate since there is limited knowledge
as to fatigue life relations for real pavements, reliable test data
exists for only a limited number of mixtures, and there is insufficient
information to define how fatigue life varies with temperature and

13
mixture characteristics. Furthermore, they point out that no laboratory
fatigue test comes close to simulating actual field conditions.
One very significant point is that fatigue life is highly dependent
on the type of fatigue test performed. While illustrating the effect of
temperature on fatigue life, Pell and Cooper (34) showed that as the
temperature is lowered, fatiaue life increases under stress controlled
tests, but decreases under strain controlled tests.
Recently, investigators have found that rest periods markedly
increase the fatigue life of bituminous mixtures (41). This seems to
indicate that asphalt concrete has the potential to heal, or that the
actual failure mode is not a true fatigue phenomenon. Both of these
ideas negate the validity of conventional fatigue approaches.
Ruth and Maxfield (42) found the concept of fatigue did not apply
to specimens from test roads in Florida. They found that the failure
strains for in-service cores were the same as for fabricated cores.
They concluded that fracture of asphalt concrete is related to a process
of cumulative creep strain and that fracture strain is primarily depen
dent on asphalt properties and loading conditions. Ruth et al. (43)
pointed out that during warm weather, temperatures are high enough to
eliminate stress or strain accumulation that leads to fatigue failure.
Pavements designed using conventional fatigue approaches have given
marginal performance. Ruth et al. (43) has stated that for similarly
designed asphalt concrete pavements in Florida some have developed
cracking within less than five years, while others give satisfactory
performance after many years. Roberts et al. (44) stated that very few
highways have served without maintenance even for five or ten years, and
in many cases roads with low traffic volumes have experienced premature

14
failure. Rauhut and Kennedy (40) stated that the occurence of fatigue
cracking in the field is quite variable, even for apparently identical
sections.
In addition, traditional approaches are unable to explain certain
types of failures observed in the field. It is well recognized that
immediate and disastrous failures may occur with weakened subsoil
conditions after just a few passes of a heavy vehicle (45). Molenaar
(46) has pointed out that traditional approaches cannot explain
longitudinal cracks observed to occur at the pavement surface. This
type of cracking is very common in practice.
Several studies indicate that the combined effects of thermally and
load induced stresses may cause cracking. Haas and Topper (13)
indicated that even if the thermal stresses are not sufficient to cause
cracking the addition of load associated stresses may result in pavement
failure. Fromm and Phang (16) reported a case where heavily loaded
trucks were carried during the winter months in one direction only.
They found that there was a greater incidence of transverse cracking on
the heavily loaded side, thereby illustrating the combined effect of
thermal and load stresses. Ruth et al. (43) hypothesized that this type
of mechanism may be the cause of some early pavement failures in Florida
and elsewhere. However, almost all studies presented in the literature
consider only the load effect or the thermal effect.
Ruth et al. (43) were the first to present an approach that
combines the effect of thermal and dynamic stresses as the main cause of
failure. They considered pavement cracking to be caused by brittle
failure induced by short term repetitive loads and thermal stresses that
occur during cool weather when the asphalt stiffness gets very high.

15
Their idea is to design the pavement for a critical condition based on
material properties, loads, and environment. They developed a pavement
analysis model that considers cracking as a result of asphalt properties
(including age hardening), vehicular loads, pavement cooling rate and
temperature. The analysis program was used to evaluate the effect of
different asphalt viscosities, cooling rates, and pavement thicknesses
on pavement performance. Predictions of cracking temperatures for a
Pennsylvania DOT test road were obtained which identified the two
cracked sections in the test road. Analysis of typical highways in
Florida indicated that some pavements may give marginal performance,
which was indirectly substantiated by observed early cracking of pave
ments, particularly those located in northern Florida.
2.3 Properties of Asphalt Cement and Asphalt Concrete As
Related To Low-Temperature Pavement Response and Cracking
The response and failure of asphalt concrete pavements have been
shown to be highly dependent on the properties of the asphalt cement.
Thus, proper characterization of asphaltic materials is extremely
important. The characterization of bituminous materials for use in
conventional design methods is based mostly on empirical procedures
which rely on correlations of their results with field performance. The
Marshall and Hveem Stabilometer tests are most commonly used for this
purpose (27). These tests are performed at high temperatures and relate
mainly to the problems of stability, workability, and durability. Fun
damental properties cannot be obtained directly from these tests.
Several researchers have attempted to correlate Marshal 1 results with
fundamental properties (47), but it will be pointed out later that such

16
an approach can lead to serious error, particularly when predicting
properties at lower temperatures.
As explained earlier, the analysis method used in this dissertation
considers cracking to be caused by brittle failure induced by short term
repetitive loads and thermal stresses that occur when the asphalt
stiffness is high. Therefore, the emphasis here is placed on the behavior
at relatively low temperatures, roughly in the range from -10 C (14 F), or
approximately the glass transition temperature, to 25 C (77 F). This
temperature range is referred to as the near transition region (48).
2.3.1 Asphalt Cement Properties
The response of asphalt to an applied stress is time dependent,
where the strain increases at a given rate with time. At lower temper
atures many asphalts are also shear susceptible, with the change in
creep strain rate not being proportional to the change in applied
stress. Finally, the behavior of all asphalts is highly dependent on
temperature.
In general, as the temperature is lowered, asphalts become more ^
viscous and eventually exhibit glassiness, where different elasto- /
viscous behavior is observed, their coefficients of expansion change, \
y
and brittle fracture may develop (48). Jongepier and Kuilman (49)
explained the behavior of asphalt as a viscoelastic liquid. At low
temperatures, asphalt behaves like an elastic solid, while at high
temperatures its behavior is comparable to a viscous liquid. At inter
mediate temperatures the behavior is influenced by both viscous and
elastic components. Asphalt cements show a characteristic common to
other amorphous materials; the glass transition phenomenon. Schweyer

17
and Burns (50) found that the glass transition temperature for a wide
variety of asphalts is between -10 C and 5 C (14 and 41 F).
The viscoelastic response of asphalts and asphalt mixtures is often
approached using mechanical models that combine Hookean springs and
Newtonian dashpots in various combinations (51). As mentioned above,
asphalt behavior at lower temperatures is often non-Newtonian (shear
susceptible), making these models unsuitable for complete description.
In any case, asphalt behavior is commonly described in terms of
rheological parameters, and at low temperatures, asphalts can be
described in terms of three rheological parameters: consistency or
viscosity, shear susceptibility, and temperature susceptibility (52).
However, the measurement of viscosity at low temperatures is rather
problematic because the asphalt behaves closer to an elastic material
with relatively low creep deformation rates. This means that creep or
viscosity tests will require an extremely long time of loading to obtain
measurable deformations at low stress levels. High stress levels are
usually necessary to obtain measurements within reasonable time
intervals, but if the material is shear susceptible the results may not
be representative of the material's behavior within the range of
interest.
Schweyer presented a pictorial review of an extensive number of
devices that have been used over the years to measure rheological
properties (53). A more concise literature survey of the different
methods to measure low temperature rheology of asphalts is presented in
reference 48. In general, the traditional transient rheometers are not
directly adaptable to low temperature work (54). However, several
special testing devices have been used to conduct investigations of

18
asphalt properties at low temperatures. These have led to improved
understanding of low temperature asphalt behavior.
For example several investigators have concluded that low
temperature asphalt properties cannot be predicted from properties
measured at higher temperatures. Schweyer et al. (55) reported that
different asphalts demonstrate very different low temperature
rheological properties. They emphasized that temperature susceptibility
in the near transition region can and should be evaluated by absolute
viscosity measurements rather than by empirical tests. They also stated
that temperature susceptibility cannot be predicted from behavior
exhibited at higher temperatures. In a comprehensive study of different
asphalts at the Asphalt Institute, Puzinauskas (56) reached the
following conclusions:
- generally, viscosity at low temperature is affected more by
heating than viscosity at high temperature;
- the low temperature viscosity of asphalt cements was found to
vary extremely and the variation increases with decreasing
temperature; and
- shear effects become more pronounced with increasing viscosity or
with decreasing temperature.
Schmidt (57) investigated the reliability of standard ASTM tests to
predict low temperature stiffness of mixtures made with a wide variety
of asphalts. He concluded that low temperature thermally induced
cracking should not be implied from high temperature viscosity measure
ments on diverse types of asphalts. Although many researchers have
found reasonable correlation between measured stiffness and observed
field cracking, relatively Door agreement has been obtained by

19
researchers estimating low temperature stiffness by means of tests at
higher temperatures. Pink et al. (54) and Keyser and Ruth (58) also
emphasized the importance of experimental measurements rather than the
use of empirical extrapolations to determine low temperature properties.
Probably the most significant advancement to the understanding of
low temperature response and failure properties of asphalts and asphalt
mixtures was the development in the 1970's of the Schweyer constant
stress rheometer (59). This device has the capability of measuring
rheological properties at -10 C (14 F) and lower. Furthermore, Schweyer
established rheological concepts that led to a definitive rheological
model and methods to evaluate important parameters that relate to low
temperature behavior, including shear susceptibility.
The proposed rheological model is the Burns-Schweyer model (55).
The model is a Burgers model with a modified dashpot to incorporate a
self-generating feedback system to regulate the rate of viscous flow.
Thus, the model accounts for viscous behavior for both Newtonian and
shear susceptible materials, as well as for elastic, and delayed elastic
behavior. Details pertaining to the measurement and evaluation of
rheological parameters using the Schweyer rheometer may be found in
references 50, 55, 59, and 60. Ruth and Schweyer (61) showed that the
Burns-Schweyer model gives accurate prediction of the rheological
properties of asphalts, including those that are very shear
susceptible. Keyser and Ruth (58) concluded that the Schweyer rheometer
is an excellent device for low temperature measurements of asphalt
properties and that the concepts developed by Schweyer provide values
more closely related to shear and strain rates encountered in the
laboratory and in actual pavements.

20
Therefore, the Schweyer rheometer has the unique advantage of
direct measurement of low temperature asphalt properties at any given
temperature. This instrument and the rheological concepts presented by
Schweyer were key elements in the development of the analysis method
considered in this dissertation.
2.3.2 Asphalt Mixture Properties
Tests on asphalt mixtures are usually conducted to determine their
failure characteristics. However, the increased trend toward
mechanistic approaches and the application of elastic theory to pavement
evaluation and design, has initiated a concerted effort to define the
stress-strain response of bituminous mixtures (62). The variability of
materials in asphalt mixtures and the nature of pavement structures are
unlimited, making it nearly impossible to uniquely characterize their
stress-strain properties (63). Furthermore, a compromise between a
rigorous design solution and practicality is necessary. Material
characterization should be based on conditions that are believed to be
critical with respect to pavement response and failure.
As mentioned earlier, the approach to pavement failures in this
dissertation considers pavement cracking to be caused by brittle
fracture induced by short term repetitive loads and the thermal stresses
that occur during cool weather when the asphalt stiffness is high. v-
Using this approach it is necessary to predict the following: the
strains and deflections induced by applied dynamic stresses (wheel
loads); the short term creep strains induced by these dynamic stresses;
and the stresses and strains induced by temperature changes. References
cited later will show that by modelling the asphalt concrete layer as an

21
elastic continuum, researchers have obtained fairly accurate predictions
of measured strains and deflections in asphalt concrete pavements under
dynamic wheel loads, particularly at low ambient temperatures. Thus,
for a given set of temperature and loading conditions, it appears that
the stress-strain response of an asphalt mixture can be characterized
using an elastic modulus or E-value. The obvious point should be made
that asphalt mixtures are viscoelastic and describing their stress-
strain behavior by using an E-value is an idealization. The idealized
E-value will depend on how it is defined and the test method used to
measure it. Therefore, it is.necessary to identify laboratory
procedures that yield E-values that are suitable for proper pavement
response prediction. Suitable parameters for creep response, thermal
expansion and contraction, and failure limits are also necessary.
A variety of test methods has been used to test asphalt concrete
mixtures for characterization including compression (unconfined and
triaxial), bending (flexure), tension (direct and indirect), and shear
tests. Some of the different laboratory procedures and the different
idealized stiffness values they yield are described in reference 27.
There are certain advantages and disadvantages for each test, but these
are beyond the scope of this review. One problem with all the tests is
the effect of time or rate of loading, which makes suspect the elastic
equations used to analyze the test data. The effect of creep on stress
redistribution is almost always ignored because of the complexities
introduced into the analysis of test data. Many researchers (64, 65, 66
and others) have recommended using the indirect tensile test as the most
suitable for routine characterization in terms of practicality,
simulation of actual conditions, economy, and ease of testing.

22
Many researchers have concentrated on defining the relative effects
of different variables on the response characteristics of asphalt
mixtures. These studies are usually limited to those variables that are
considered to have a significant effect on the material behavior within
the researchers' scope of interest.
Deacon (38) summarized the major variables affecting the stiffness
and range of linear response for asphalt concrete mixtures and divided
them into three major categores: 1) loading; 2) mixture; and 3) envi
ronmental related variables. He stated that four mixture related
variables have a considerable effect on the stiffness of asphalt paving
mixtures: air void content, asphalt content, asphalt viscosity, and
filler content. Temperature, mainly through its effect on asphalt
vi scosity, was recognized as the major factor....in the behavior of j
bituminous mixtures. Deacon also stated that the range of linear
response increases with increasing load frequency, decreasing
temperature and void content, and increasing asphalt content, asphalt
viscosity or filler content.
Bazin and Saunier (37) recognized the difficulty in changing any
one parameter without changing another, so they studied variation in
modulus for very different mixtures. They found that for correct binder
dosages and normal voids (4 to 8 percent), all other parameters had
little influence on modulus when compared to the effect of variation in
binder type, temperature, and time of loading. They suggest a linear
relation between log of modulus and void content of mix and proposed the
time-temperature superposition principle. This principle suggests that
there is an equivalency between time of loading and temperature.

23
Several researchers have investigated the stress dependence of
dynamic modulus. Cragg and Pell (67) reported stress dependence in
dynamic modulus but indicated that the change was small when compared
with temperature induced changes in modulus. Gonzalez et al. (64)
concluded that instantaneous modulus of elasticity decreased with
increasing temperatures and increasing number of applications but was
not affected by magnitude of applied stress. Kallas and Riley (68)
reported stress independent moduli for stresses between 17 and 70 psi
and temperatures from 4.4 C to 38 C (40 F to 100 F). Monismith et al.
(39) reported stress independent moduli over the range of 100 to 125 psi
using repeated flexure tests. Constant dynamic moduli were reported by
Pell and Taylor (33) for stresses below 125 psi at a temperature of 10 C
(50 F). They also indicated that low voids, low temperatures or high
loading rates, and adequate quantities of binder and filler were
conducive to linear response.
The effect of loading frequency on modulus has also been the
subject of several investigations. Yeager and Wood (69) described the
behavior of test specimens subjected to different frequencies. For
faster frequencies there is little time for flow and the mixture
behavior is more elastic. Slower load rates result in larger total
strains and lower calculated moduli. They reported a seven-fold
increase in modulus at 4.4 C (40 F) and a stress of 50 psi as the rate
of loading was increased from 1 to 12 cycles per second. Barksdale (70)
studied stress pulses applied by moving wheel loads and developed
relationships of stress pulse times as a function of vehicle velocity
and depth beneath the pavement surface. He developed charts that have

24
been subsequently recommended for determining loading times for dynamic
laboratory tests (71).
Investigations have also been conducted to evaluate different test
methods and differences in tensile and compressive properties. Kallas
(72) compared dynamic moduli of dense graded mixtures with normal air
voids at 4.4, 21.1, and 37.8 C (40, 70, and 100 F) and frequencies of 1,
4, and 16 cycles per second in tension, tension-compression, and
compression. He reported small differences in dynamic modulus in
tension, tension-compression, and compression for temperatures between
4.4 and 21.1 C (40 and 70 F) and frequencies between 1 and 16 cycles per
second. However, at 1 cycle per second and temperatures between 21.1
and 37.8 C (70 and 100 F), the dynamic tension or tension-compression
moduli averaged 1/2 to 2/3 of moduli in compression. The viscous
component of response was found to be considerably greater in tension
than in compression for frequencies of 1 to 16 cycles per second and
temperatures from 21.1 to 37.8 C (40 to 100 F). Wallace and Monismith
(73) compared moduli results from diametial indirect tension and
triaxial tests and analyzed the effect of anisotropy on both testing
procedures. They estimated that a more relevant assessment of modulus
could be obtained from diametral tests.
Although direct characterization is one approach to obtain mixture
parameters, the cost is prohibitive for most agencies (27), particularly
considering the variable properties of asphalts with temperature and
stress, and the fact that these properties can vary significantly for
different asphalts. Therefore, several investigators have established
relationships to predict mixture response parameters, based on
temperature, time of loading, and material characteristics from

25
conventional tests. Miller et al. (62) reviewed some of the methods
that have been presented for predicting modulus from physical and
mechanical properties of the mixture, which usually take the form of
nomographs or master equations. Two of the better known methods are:
1) the different versions of the Shell Nomographs, and 2) the Asphalt
Institute Bituminous Mix Modulus Predictive Equation. The Shell
Nomographs were developed from Van der Poel's original stiffness charts
for asphalt cements for a given load rate and temperature. Relation
ships were developed to translate bitumen stiffness to mixture stiffness
and these have been modified by several investigators. Modulus
prediction is based on asphalt stiffness as derived from softening point
and penetration, temperature, and frequency of loading. The asphalt
stiffness is then modified for mixture characteristics to obtain the
modulus of the mixture. The Asphalt Institute Equation is based on
tests performed on different mixtures at varying temperatures and fre
quencies. Modulus prediction is based on load frequency, void content,
filler content, asphalt viscosity at 21.1 C (70 F), temperature, and
asphalt content.
Note that these methods cannot account for the variable properties
of individual asphalts at low temperature. Asphalt properties at
different temperature are inferred from consistency measurements at high
temperatures, and in the case of the Shell Nomographs, these measure
ments are entirely empirical in nature. The temperature and shear
susceptibility of individual asphalts is ignored. As pointed out
earlier in this review, several researchers have shown that these
properties are a major factor in the response of the. mixture, and they
can vary significantly from asphalt to asphalt.

26
Ruth et al. (43) were the first to present modulus relationships
based on a direct evaluation of measured asphalt viscosities at
different temperatures. The dynamic modulus relationship was developed
for dense-graded mixtures and a loading time of 0.1 seconds. It
requires the asphalt viscosity at a specific temperature as determined
from measurements with the Schweyer rheometer. Viscosity measurements
are made at several temperatures and stress levels (comparable to those
induced in laboratory tests on mixtures), in order to develop asphalt
viscosity-temperature relations for input. Therefore, this prediction
method properly accounts for the effects of temperature and shear
susceptibility on mixture response, and thus shows the most promise for
accurate modulus determination. Furthermore, the tests required are
simple and inexpensive.
The evaluation of these relationships is one goal of the current
investigation. It has been recognized that there is much variability in
characterization parameters reported in the literature (63). Part of
the problem is that resilient modulus tests have not been standardized
and certain aspects of the test that have an important effect on results
have not been defined (66). One reason for the lack of standardization
is that little if any work has been done to identify test procedures and
data interpretation methods that yield modulus values which give
accurate prediction of measured response on full-scale asphalt concrete
pavements under varying temperature and loading conditions. In other
words, although the relationships between asphalt viscosities and
laboratory measured dynamic moduli seem adequate, it has not been
verified that laboratory generated moduli can be used to predict
response on full-scale pavements, particularly at low temperatures.

27
Ruth and Maxfield (42) also developed relationships between asphalt
viscosity and what he defined to be a static modulus. The static
modulus was calculated from results of constant stress creep tests from
which he also calculated a pseudo mix viscosity for creep strain
prediction. Ruth found good correlation between asphalt viscosity and
pseudo mix viscosity, which also correlated well with static modulus.
Thus he developed relationships for both static modulus and creep
strains that are based on the low temperature viscosity of the
asphalt. The creep strain prediction model is a function of the asphalt
viscosity (function of temperature), the applied stress and the failure
stress. The static modulus is representative of the material's long
term response, as for the case of thermally induced stresses.
It should be noted that creep prediction models presented by other
researchers were almost exclusively developed to predict rutting. Their
results will not be discussed because they have not considered low
temperature creep response.
With the emphasis on fatigue, only a limited amount of work has
been done to define the failure limits of asphalt concrete mixtures in
terms of stresses or strains. Even for low temperature transverse
cracking, failure parameters are usually presented in terms of a
limiting asphalt stiffness or a fracture temperature, determined from
correlations with observed cracking in the field.
Several researchers have reported that tensile strength and failure
strain of asphalt mixtures are dependent on rate of loading and
temperature, and all have determined that the tensile strength increases
as the temperature decreases and the rate of loading increases (9, 12,
36, 74, 75, 76). Ruth (14) reported that as the temperature decreases

28
the failure stress increases but remains constant below some transition
temperature which is dependent on asphalt properties. Heukelom (74)
presented evidence that the tensile strengths of mixtures are related to
asphalt properties. Various researchers have reported failure stresses
for conditions of low temperatures and fast loading rates. Finn (77)
reported fracture strengths of 290 to 580 psi for asphalts in bulk under
low temperature conditions and rapid loading. For asphalt mixtures
under the same conditions he reported strengths generally ranging from
400 to 700 psi. Ruth and Olson (78) reported failure stresses between
380 and 440 psi and chose a value of 400 psi as typical.
Ruth and Olson (78) reported that as the temperature decreases the
tensile strain at failure decreases. Tons and Krokosky (76) reported
that increasing asphalt content within practical limits had little
effect on strain at ultimate strength. They also stated that rate of
loading had little effect at low temperatures. Epps and Moni smith (36)
shov/ed that the strain at failure is related to the stiffness of the
mixture, the strain at failure decreasing with increasing mixture
stiffness. Pavlovich and Goetz (79) computed limiting strains from
axial deformations in direct tension tests and determined that
temperature is the most significant factor affecting limiting strains.
Strain rate has some effect but not as much as temperature. They found
the limiting strain at 60 C (140 F) was 300 to 500 times greater than
that at -27.5 C (-17.5 F). Sal am and Monismith (80) presented an
equation to determine the strain at failure for asphalt mixtures based
on the strain at failure of the asphalt, the asphalt stiffness, and the
mixture stiffness. The asphalt properties used are penetration and ring

29
and ball softening point, which are empirical in nature and performed at
high temperatures.
Ruth and Maxfield (42) tested different mixtures and determined
that measured failure strains are primarily a function of asphalt
viscosity. Ruth and Potts (81) found that the energy required to
fracture a specimen decreased with increasing viscosity. These findings
led to the development of relationships between asphalt viscosities and
strain and energy at failure. The relationships are unique, since they
can directly account for the properties of individual asphalts at
different temperatures.
Investigations on the thermal expansion and contraction charac
teristics of asphalt concrete mixtures have led to the following
conclusions (82, 83):
1. Different asphalts produce different amounts of expansion and
contraction.
2. The amount of shrinkage during cooling is more than the amount
of expansion during heating.
3. There are two different coefficients of expansion betv/een -10 F
and 140 F, and the transition temperature between the two was
found to vary between 70 F and 86 F. Values in the low and
hiqh range have been called the solid and fluid thermal
coefficients, respectively.
4. The greater the asphalt content, the greater the thermal
expansion and contraction.
5. In the fluid state, the amount of expansion (contraction)
depends on the degree of restraint, while in the solid state
the expansion is the same for free and friction conditions.

30
Jones
et al.
(83)
developed the following equation for
predicting the cubic coefficient of expansion in the solid
state:
V B V B
d ac ac + agg agg
Vmix
where,
Bmix
=
cubic thermal coefficient of expansion for
asphalt concrete
Bac
=
cubic thermal coefficient of expansion for
asphalt (glassy state)
Bagg
=
cubic thermal coefficient of expansion for
aggregate
vmix
=
volume of. asphalt plus aggregate
vac
=
volume of asphalt
vagg
=
volume of aggregate.
Assuming isotropic properties the linear coefficient of thermal
contraction (a) is:
Bmix
mix
2.4 Properties of Foundation Materials
The characterization of soils and granular base and subbase
materials, for both analysis and performance, is required data for all
structural pavement design methods, and is often treated with
considerable simplification. Geotechnical engineers often feel that
structural engineers have little interest in those parts of their work
below the ground level, and their feelings are certainly justified in
the case of pavements (84). Of the 185 papers presented to the past
five International Conferences on the Structural Design of Asphalt
Pavements only 15 have been concerned in any detail with the mechanical

31
properties of soils and granular materials, and much of this work has
concentrated on the prediction of rutting.
Soil behavior depends on many factors including water content, dry
density, stress level, stress states, stress path, structure, stress
history, and soil moisture tension (85). Although the relative effect
of these variables can be investigated in the laboratory, the cost of
even a limited number of tests may be prohibitive for most highway
projects.
Reference 27 describes tests that are typically used to
characterize soils and granular materials for pavement analysis,
including CBR, plate load tests, and triaxial testing. These tests are
usually performed for a limited set of conditions that in all proba
bility do not encompass the variable conditions encountered during the
life of the pavement. Furthermore, the parameters obtained from such
tests are dependent on sample preparation, testing procedures, and data
interpretation methods, all of which have not been standardized.
Several researchers have presented relationships of resilient
modulus as a function of dry density and moisture content for specified
soils, based on laboratory testing. However, even these types of rela
tionships are of limited value since they may ignore certain variables
that may have a significant effect on response. Also there is little
evidence that response parameters determined from conventional tests, or
any other test, provide for accurate prediction of response of full-
scale pavements.
A detailed review of work that has been done to define the relative
effects of different factors on soil response, or the relationships
developed to predict response parameters measured in the laboratory,

32
will not be presented here. This information is beyond the scope of the
work presented in the dissertation. However, some of the more recent
work done will be reviewed in an attempt to present a general view of
the state of knowledge in the area of characterizing soils for pavement
response prediction.
References cited later show that stresses and strains within the
asphalt concrete layer can be predicted fairly accurately by using an
effective modulus for the soil in an elastic layer analysis. They also
state that it is unlikely that stresses within the soil layers can be
predicted using a similar approach. Even for the simpler case of
predicting stresses in the asphalt concrete layer, the problem of deter
mining a suitable effective modulus remains.
It has become evident in recent years that it is difficult or even
impossible to predict the behavior of pavements solely from laboratory
test data (86). Therefore, there has been much more emphasis on full-
scale pavement testing. A great deal of work is being done to determine
response parameters from nondestructive testing devices such as
Dynaflect and Falling Weight Deflectometer, as evidenced by the many
papers presented on this subject in the latest International Conference
on the Bearing Capacity of Roads and Airfields (87), and the 1985
sessions of the Transportation Research Board. This approach involves
the calculation of effective layer moduli based on measured surface
deflections. The approach has advantages and disadvantages. One
advantage is that many tests can be performed quickly and easily over
several miles of roadway under in situ conditions. Therefore, modulus
relationships can be developed for widely varying conditions. However,
surface deflections alone may not yield unique solutions and the

33
peculiarities of the different testing devices may result in loading
conditions that are not directly comparable to moving wheel truck
loads. In addition, the method is only suitable for existing pavements
and the results obtained will represent only the behavior during the
particular time tested.
Maree et al (86) presented an approach to determine layer moduli
based on a device they developed to measure deflections at different
depths within the pavement structure. They suggested that effective
moduli for use in elastic layer theory can be determined from correction
factors or shift factors established from field measurements using their
device at different times of the year and under different conditions.
They determined effective moduli from tests performed at different load
levels and under different environmental conditions using their
multidepth deflectmeter. They found that moisture condition has a
significant effect on response and its effect can he as important as
stress state. They also demonstrated that the modulus of individual
materials depends on the modulus of the underlying layers. The stress
dependence of several pavement materials was also demonstrated from the
field data. They found that most granular materials behave in a stress-
stiffening way, while most subgrade materials (e.g. weathered shale)
behave in a stress-softening way. Comparison of field and 1aboratory
data showed that although the trends apparent in the field were also
apparent in the laboratory, the constant-confining pressure triaxial
tests overestimated the modulus of the base materials tested.
Accurate prediction of stresses and strains within the foundation
layers is necessary for the prediction of rutting and stability failures
within these layers. Luhr and McCullough (88) state that moduli of

34
unbound materials should vary both horizontally and vertically in the
pavement structure according to equations equating modulus with state of
stress (i.e. Mr = Ao^8 for fine-qrained materials and = k1k2 for
granular materials). They stated that although this variation can be
satisfactorily represented by finite element models (F.E.M.), these
models have the following disadvantages: 1) they usually require large
amounts of computer time; 2) the variability in pavement performance
data may make their precision superfluous; and 3) F.E.M.'s are usually
too complex and consume too much time to be used routinely in pavement
management systems. They concluded that elastic layer theory with an
equivalent effective modulus gives reasonable response prediction.
Brown and Pappin (89) seem to disagree. They developed a contour model
to predict non-linear behavior and lack of tensile strength in soils and
incorporated this model into the finite element program SENOL. Develop
ment and limited verification of their model involved both laboratory
tests and full-scale testing of pavements under controlled conditions.
Based on a parametric study and limited measurements they determined the
following:
- deflections and strains in the asphalt layer can be reasonably
predicted using an equivalent effective modulus in elastic layer
theory;
- it is unlikely that elastic layer theory can be used to predict
stresses and strains within the unbound layers;
- they suggest that the K-0 approach is less satisfactory than a
linear elastic solution; and
- the simplest approach for design calculations is the use of a
linear elastic system provided adequate equivalent moduli are used.

35
They suggest that detailed nonlinear analysis using the SEMOL model
would lead to adequate selection of these moduli.
It seems clear that more work is needed to determine suitable
parameters for soil response and to develop models for more accurate
prediction of stresses and strains within the soil layers. Finite
element models should lead to more satisfactory results but the stress-
strain relationships still have to be improved. However, accurate
measurements on full-scale pavements to develop such models are lacking.
2.5 Prediction of Thermal- and Load-Induced Stresses,
Strains, and Failure In Asphalt Concrete Pavements
Several investigators have developed models to predict thermal
stresses and strains in asphalt concrete. Hills and Brien (8) developed
a simple calculation procedure to predict thermally induced stresses.
Their solution was based on a restrained, infinitely long pseudoelastic
beam exposed to a uniform temperature drop. Lateral restraint was
neglected. Limited experimental work showed that the method gave
reasonable predictions. Haas and Topper (13) used the basic equation
presented by Hills and Brien to develop a procedure to calculate thermal
stresses that recognizes the temperature and stiffness gradients that
exist in actual pavements. Shahin (90) modified the Hills and Brien
equation to accommodate coefficients of contraction not constant with
temperature.
Christison et al. (7) used five different analyses for thermal
stress computation and compared predicted and observed values. He
concluded that the pseudoelastic beam yields reasonable results but
points out certain difficulties with this analysis: 1) the predicted
stress depends on the time interval used in the calculation; and 2) the

36
method does not allow for stress relaxation subsequent to the time
interval in which stresses are computed.
Monismith et al. (9) developed a thermal stress equation for an
infinite viscoelastic slab and complete restraint. Stress is calculated
as a function of depth, time, and temperature. A relaxation modulus is
required from uniaxial creep tests.
Ruth et al. (43) presented a stress equation that considers rate of
cooling, creep rate, and variation in modulus with temperature. The
model requires the asphalt viscosity-temperature relationship from which
all calculations are made. The method is unique in that it can account
for individual asphalt properties.
The prediction of thermal cracking in asphalt pavements is usually
based on comparing the accumulated thermal stress with the tensile
breaking strength of the asphalt concrete. Several investigators have
compared stresses as predicted by the various models with observed
cracking.
Burgess et al. (12) reported that the method of Hills and Brien
correlated well with cracking observed in the St. Anne Test Road.
Christison (7) compared results from pseudoelastic beam, viscoelastic
beam, and viscoelastic slab analyses to cracking observed at St. Anne.
He determined that thermal cracking could be predicted by using the
computed stresses from either analysis at 1/2-inch depth.
Haas and Topper (13) indicated that Moni smith's method appeared to
predict unusually high stresses, which may be due to his assumption of
infinite lateral extent.
Models to predict thermal stress cracking have also been
presented. The computer program COLO (91) was developed based on the

37
Hills and Brien equation. It uses a stress criteria of 200 psi. Shahin
and McCullough (92) developed a model to predict the amount of thermal
cracking. The model predicts temperatures, thermal stresses, low-
temperature cracking, and thermal fatigue cracking. They indicated that
comparisons of predicted cracking with measurements at the Ontario and
St. Anne Test Roads were reasonable. Lytton and Shanmugham (93)
developed a mechanistic model based on fracture mechanics to predict
thermal cracking of asphalt concrete pavements. The model assumes
cracking is initiated at the surface of the pavement and propagates
downward as temperature cycling occurs. The prediction of transverse
cracking and cracking temperature can also be attempted with the Haas
model (94) and the Asphalt Institute Procedure (95). Keyser and Ruth
(58) found absolutely no correlation between actual cracking in 6- to 9-
year old pavements in Quebec, and cracking predicted using the latter
two models.
Ruth et al. (43) developed a thermal cracking model based on their
stress equation. The model computes stresses, creep strains, and
applied creep energy, which are all used as failure parameters.
Predictions of cracking temperatures for a Pennsylvania D.Q.T. test road
were obtained which identified the two cracked sections in the test
road.
It should be noted that although indirect comparisons have been
made with observed field cracking, little if any measurements exist of
actual contraction and deformation of asphalt concrete pavements during
cooling.
Over the years, various solutions have been presented to predict
stresses, strains, and deflections within the pavement system due to

38
applied wheel loads. In 1885, Boussinesq presented the mathematical
solution for a concentrated load on a boundary of a semi-infinite
body. Love extended this solution to solve for a distributed load on a
circular area. Burmister (96) was the first to present a solution for a
multi-layered elastic system, and developed solutions for specific two-
and three-1ayered systems. Schiffmann (97) extended Burmister's
solutions to include shear stresses at the surface. Elastic solutions
for layered systems have been presented in the form of tables, graphs,
and equations that include a wide range of parameters (98). Also,
several computer programs have been developed for elastic layer
analysis.
With the advent of the finite element method, more sophisticated
models have been introduced, including viscoelastic analysis (99), Brown
and Pappin's contour model (89), and Yandell's mechano-1attice analysis
(100). Many of the finite element models have been developed for the
problem of rut prediction, and more recently to predict response within
the soil layers more accurately.
For the purpose of predicting stresses, strains and deflections
within the asphalt concrete layer, there seems to be general agreement
that elastic layer analysis is suitable. Although there is some
question of its ability to predict response in the underlying layers,
there has been considerable verification of its ability to predict
response in bound layers.
Avital (101) discusses a series of computer programs available for
the analysis of multi-layered systems. Barksdale and Hicks (32) present
a general description of multi-layered systems and finite element
approaches, along with extensive references on their detailed

39
development. They recommend the use of elastic layered systems for
pavement analysis since only two variables are needed (modulus and
Poisson's ratio).
The moderators for the last International Conference on the
Structural Design of Asphalt Pavements (84) indicated that use of linear
elastic theory for determining stresses, strains, and deflections is
reasonable as long as the time-dependent and nonlinear response of the
paving materials are recognized. They noted that the papers presented
at the conference confirm that multilayer elastic models generally yield
good results in layers containing binders. It is interesting to note
that of the 16 papers relating to analyzing pavements, 14 used
multilayer elastic theory and the two that used viscoelastic procedures
reduced their viscoelastic idealization to equivalent elastic layered
systems.
Ros et al. (102) measured strains and pressures at different levels
in a variety of trial sections subjected to standard loads at varying
speeds and temperatures. They found good correspondence between
measured values and values computed with an elastic layer program
(BISAR, Ref. 103), using properties determined in situ and in the
laboratory. Correlation was especially good at high asphalt
stiffness. Halim et al. (104) performed tests on reinforced and
unreinforced flexible pavements in a test pit and found that elastic
layer theory (BISAR) provides a reliable tool to predict flexible
pavement response. They suggested that use of a calibration factor for
stress-dependent materials is more efficient and less time consuming
than more sophisticated models. Waterhouse (105) measured axial
stresses vertically below the central axis of a circular load and

40
determined that elastic theory gave reasonable prediction of stress
distribution. However, as pointed out in section 2.4 by Brown and
Pappin (89), it is unlikely that elastic theory can predict stresses
within the soil layers, although they did find good correspondence
within the asphalt layer and at the surface of the subgrade using
elastic theory. Several other researchers have also found good
correspondence with measured results using elastic layer theory (e.g.
106, 107).
As mentioned earlier, the approach to pavement failure considered
in this dissertation is based on the idea that pavement cracking is
caused by brittle failure induced by short term repetitive loads and
thermal stresses that occur during cool weather when the asphalt
stiffness is high. Therefore, pavement design and analysis methods
should be based on the rheological properties of the asphalt which are
used to predict thermal and dynamic load stresses and strains at thermal
conditions typical of the lowest temperatures expected. The predicted
values can then be compared to the failure limits of the material for a
direct evaluation of failure. This method was proposed by Ruth et al.
(15).
Avital (101) implemented a computer program (CRACK) to handle this
interaction of load and temperature induced stresses. The program
combines an elastic layer computer program for response prediction under
dynamic load and a thermal program for thermal stress and creep strain
prediction. The program requires temperature conditions, traffic
volumes, rheological properties of the asphalt, and the pavement struc
ture characteristics (layer thickness, modulus, and poisson's ratio).

41
This approach is totally different from conventional fatigue
approaches. Therefore, the different distress prediction models that
have been developed will not be presented here. These models usually
attempt to predict pavement life by using different empirically based
sub-models that predict fatigue cracking, rutting, and thermal cracking
separately. The most sophisticated of these are the different versions
of the VESYS model presented by Kenis et al. (108).

CHAPTER III
EQUIPMENT AMD FACILITIES
3.1 Description of Test Pit Facility
A test pit facility located at the Office of Materials and Research
of the Florida Department of Transportation (FDOT) was used in this
research project. The facility included a 15 ft. long, 13'-4" wide, 6'-
2" deep concrete pit with a test area of 8 ft. by 12 ft. The test pit
made it possible to construct a layered system of materials to simulate
a flexible pavement system in the field. It had the following features:
- control of water level in the test pit;
- a hydraulic loading system that could apoly both static and
dynamic loads anywhere within the 8 ft. by 12 ft. test area;
- two linear displacement transducers for deflection measurements;
- a 20,000 lb. capacity load cell; and
- a four-channel continuous plotter used for both the transducers
and the load cell.
A detailed description of this facility is given in Research Report S-l-
63, Civil Engineering Department, Engineering and Industrial Experiment
Station, University of Florida. The facility was formerly used for the
evaluation of Florida base course materials using a variation of the
plate bearing test. However, the following modifications were necessary
to make it suitable for testing complete flexible pavement systems:
- a proper method for distributing hot asphalt concrete mix within
a reasonable time period to allow for proper compaction and to
avoid premature cooling of the mix;
42

43
- a cooling system with proper insulation to provide the capability
of testing at different temperatures; and
- a proper measurement and data acquisition system for deflection,
strain, and temperature measurements.
Therefore, a good deal of work was done to design, procure materials,
and construct equipment to overcome these deficiencies.
3.2 System for Hot Mix Asphalt Distribution
A triangular shaped hopper that spans the 8 ft. width of the test
pit and distributes material through an adjustable opening at its bottom
was designed to distribute the hot mix. A picture of the hopper is
shown as Figure 3.1. The hopper was made of steel angle and plate and
stands approximately 2'-6" high. It had a level capacity of 33 cf which
made it possible to place a four-inch lift in the test pit in only one
pass. The hopper was designed for completely manual operation and ran
on steel angle rails that were installed in the test pit. A concrete
pad was placed immediately north of the hopper to allow use of a dump
truck for asphalt concrete distribution.
3.3 Pavement Cooling System
Different alternatives were evaluated to select a pavement cooling
system for installation in the test pit area. An air cooling unit was
chosen as the most suitable system. A system of this type is clean,
essentially maintenance free, and provides for reasonably accurate
temperature control. A local mechanical engineer was hired to design
and prepare equipment specifications for a suitable air cooling system.

Figure 3.1: Hopper for Asphalt Hot Mix Distribution

45
The system was designed to cool six-inch thick pavements to a
temperature of -10 C (15 F) at a rate of 3.3 C (5 F)/hour, as measured
at a depth of 1/4-inch from the surface of the pavement. Temperature
control was achieved by manually controlling the cooling unit. A fully
automatic system with greater cooling capacity was originally
considered, but its cost was prohibitive. In any case, automatic
controls are of limited value, since temperature gradients would be
present in the pavement as long as the unit was running.
The cooling system consists of a direct expansion, low temperature
refrigeration system. The evaporator (Larkin ELT-300) was located
directly in the test pit and cooled the pavement by recirculating cold
air across the test pit surface. Temperature control was achieved by
lowering the pavement temperature below the required test temperature,
stopping the refrigeration system, and allowing the test pit temperature
to drift upward. The condensing unit (Larkin CS 0750L1), which housed a
7.5 HP compressor, sat outside the building's north wall. Refrigerant
hoses and electrical cables from the condensing unit to the evaporator
unit were connected through holes drilled in the north wall of the test
pit. Drainage was accomplished with a heated drain pipe, which was
passed through a hole drilled in the test pit wall. A layout of the
test pit cooling system is shown in Figure 3.2.
An insulated cover for the test pit area was designed and
constructed. The cover had a solid wood frame which enclosed the test
pit area. The top of the cover consisted of five removable wood panels
that spanned the 8-ft. width of the test area and were supported by a
ledger on the cover's frame. One panel had a one-ft. diameter hole to
allow for placement of the loading ram. The panel dimensions are such

Figure 3.2: Layout of Test Pit Cooling System

47
that the panel with the hole can be moved to three different positions,
thereby providing for three different loading positions during cooling.
All sections of the cover, including the frame and the panels, were
insulated with six inches of polystyrene, as recommended by the
contractor. Once the frame and panels were in place, all the joints
were sealed with clay and tape to reduce moisture migration and
infiltration into the test pit during operation. A nylon sheet was
placed over the entire cover to further reduce infiltration. A picture
of the cover, completely installed, is shown in Figure 3.3. Figure 3.4
shows the cover removed for access to the test pit.
3.4 Measurement System for Pavement Response
The test pit facility was formerly used exclusively to evaluate
Florida base course materials using a variation of the plate bearing
test. These tests required a loading system, capability for two
deflection measurements, and a recording device for the load and two
displacements. Considerably more extensive measurements were required
for complete evaluation of asphalt concrete pavements at different
temperatures. Static and dynamic deflection and strain measurements at
different points in the pavement were required to define the pavement
response during loading, and the contraction of the pavement during
cooling. Temperature measurements were also required. Therefore, a
measurement and data acquisition system was designed and installed for
this purpose.

48
Figure 3.3: Insulated Test Pit Cover Completely Installed
Figure 3.4: Insulated Test Pit Cover With Panels P^moved

49
3.4.1 Measuring Instruments
Linear variable differential transformers (LVDT's) were purchased
to obtain static and dynamic deflection measurements at different points
on the pavement surface. Schaevitz model DCD-200 LVDT's, with a range
of 0.20 in. (0.5 cm) and an output of 50 V/in. (19.7 V/cm) were
used. Two dual-output power supplies were purchased to operate these
units. All LVDT's were individually calibrated using a micrometer and a
voltmeter.
Two LVDT support systems were designed and constructed: one for
use with the plate loading system, and the other for use with a dual
wheel loading system designed for use in the test pit. Figure 3.5 shows
a plan view of the test pit area with the LVDT support system used for
plate testing. The system consisted of wood LVDT mounts supported by
1.5 in. diameter pipes that spanned the eight-foot width of the test
pit. The mounts could be positioned to obtain deflection measurements
at any specified distance from the load, and the entire system could be
moved for testing at different positions. The LVDT's were spring-loaded
and could be adjusted vertically by way of an adjustment screw on the
wood mount.
Figure 3.6 shows the LVDT support system used for loading with the
dual wheels. The longitudinal support was a laminated two-by-four-inch
beam which was located underneath the axle of the dual wheel system.
The entire length of the beam was grooved to accept the vertically
adjustable LVDT mounts shown in Section B-B (Figure 3.6). Thus the
mounts could be positioned at any specified distance from the load.
The same mounts used for plate testing (Figure 3.5) were used to obtain
transverse deflections with the dual wheel system. These mounts were

50
PLAN VIEW
Section A-A
Figure 3.5: LYDT Support System for Plate Loading

51
PLAN VIEW
SECTION B-B
Figure 3.6: LVDT Support System for Dual Wheel Loading

52
supported on 1 1/2-inch diameter pipes that rested on longitudinal beams
as shown in Figure 3.6. The supports could all be moved for testing at
different positions.
During initial cooling trials, the LVDT's malfunctioned at low
temperatures, even when covered with heavy insulation. The low
operating temperature of the cooling unit would eventually penetrate the
insulation and cause these units to malfunction. Therefore, a heating
system was installed to maintain the LVDT's at fairly constant temper
ature during cooling. Variable output heater wire (0 to 4 watts/ft. of
wire) was wrapped around the individual LVDT's, which were then covered
with a 1/2-inch layer of rubber foam insulation. A voltage regulator
was used to control the output of the heater wire, which was adjusted as
necessary to maintain the LVDT's at constant temperature of about 25 C
(77 F) during cooling. A picture illustrating how the LVDT's were
prepared for testing is shown in Figure 3.7. A picture of the entire
LVDT support system with the dual wheel loading system in place and
ready for testing is shown in Figure 3.8.
Strain measurements were obtained with two-inch bonded wire strain
gages (Micro-Measurements EA-G6-20CBW-120). Two methods were used to
position the gages at a given location. For surface strain measure
ments, the gages were mounted at specified points on the pavement after
it was placed and compacted. Several gages were also installed for
measurements at the bottom of the asphalt concrete. These gages were
first mounted on asphalt concrete cores (4-inch diameter and 2 1/2-
inches high) and then were positioned at specified locations on the
compacted base material. The cores were prepared in the laboratory,
using an asphalt concrete mixture similar to the one used for the rest
of the pavement.

en
co
Figure 3.7: LVOT Prepared for Tests at Low Temperatures

54
a) Frontal View
b) Diagonal View
Figure 3.8: Test Pit Pavement Completely Instrumented and Ready
for Testing With Dual Wheels

55
The strain gages were mounted with epoxy directly on the asphalt
concrete surface. The following procedure was used to prepare the
surface and mount the gages. The asphalt concrete surface was prepared
by sanding; first with a belt sander, and then with progressively finer
sandpaper until the surface was "glass-smooth". Clear tape was then
used to lift off all loose particles from the surface. Cleaning with
the tape was repeated until the tape was completely clean when lifted
off the surface. A thin layer of epoxy was then applied to the clean
surface and the gage was positioned, taking care to remove any air
bubbles trapped underneath the gage. A thin layer of epoxy was also
applied to the surface of the gage, for protection and to aid in
bonding. A sheet of cellophane was placed on top of the gage and clear
tape was used to hold the gage in position until the epoxy set. The
cellophane was used to prevent possible damage from the tape adhering
directly to the gage. Once the epoxy set, the tape was removed and the
strain gage wires were soldered to the gage. The completely installed
gage was covered with a piece of masking tape followed by a piece of
duct tape for protection. A picture of the two-inch strain gages
mounted on asphalt concrete cores is shown in Figure 3.9.
3.4.2 Data Acquisition System
A data acquisition system was designed and installed which was
capable of monitoring and recording ten dynamic deflection measurements,
ten dynamic strain measurements, 20 temperature measurements, and load
magnitude and time of loading. As mentioned earlier, only one recording
device, a Gould model 2400 strip chart recorder, was available in the
test pit facility, since this was all that was needed to evaluate base

Figure 3.9: Two-Inch Strain Gages Mounted on Asphalt Concrete

57
course materials. This high speed recorder had four channels, but only
three channels with amplifiers were available for use. Two channels
were used in conjunction with LVDT's to obtain deflection measurements,
and the third channel was used with the load cell to monitor and control
load magnitude and time of loading.
Five dual-beam digital oscilloscopes were purchased to record
additional* deflection and strain measurements. The oscilloscopes used
were Nicolet Explorer Series 2090 with model 201 plug-in units. These
instruments had the capability of monitoring and recording displacements
or strains continuously with time. All five oscilloscopes had a
¡
| temporary recording system for indefinite storage of measurements taken
within a specified time interval (i.e. one sweep of the oscilloscope).
In addition, three of the five oscilloscopes were equipped with a floppy
disc recording system for permanent data storage. Permanent storage of
i
I data obtained with the other two oscilloscopes was accomplished by using
an X-Y plotter (Hewlett-Packard Model 7046B). Once data were temporar-
¡
ily stored for a given series of loading cycles, they were immediately
output to a calibrated X-Y plotter for permanent recording.
I
Four oscilloscopes, two with floppy disc recording systems and two
without, were used in conjunction with eight LVDT's to obtain deflection
i
measurements. These eight measurements, plus the two on the strip chart
recorder provided for ten simultaneous deflection measurements at
i
I
1 different points on the pavement.
The following equipment was used in conjunction with the strain
i
gages to obtain strain measurements:
- a Vi shay/Ellis (V/E) 21 AK switch, balance, and calibration
module;

58
- a V/E 20 AJMLH strain gage indicator; and
- a digital oscilloscope with floppy disc recording system.
These units could handle 10 strain gages, in either a 1/4-, 1/2-, or
full-bridge arrangement. However, only one gage could be monitored at
any given time with the strain gage indicator. Output from the
indicator was sent to the digital oscilloscope for continuous recording
with time.
Deflection and strain measurements stored on floppy discs, were
later output to calibrated X-Y plotters. The plotters were calibrated
for an average LVDT output, since each LVDT had a slightly different
calibration. The measurements, as determined from the X-Y plotter
output, were then adjusted for the calibration factor of the particular
LVDT used. All cables going to the LVDT's and the strain gages were
passed through an access hole drilled through the side of the insulated
test pit cover.
A schematic diagram of the data acquisition system is shown in
Figure 3.10. A picture of the system is shown in Figure 3.11. Figure
3.12 is a picture of a typical recording of deflections on a digital
oscilloscope and Figure 3.13 shows a typical output recorded on an X-Y
plotter.
Temperature measurements were obtained with a Fluke Model 2240A
Datalogger. This unit used thermocouple wires to receive and record
temperature measurements for up to 20 different positions at one time.
It could record temperatures at specified, time intervals or could be
triggered to record at any given time. The unit automatically records
the date and time of the readings.

Record on
X-Y
Plotter
X-Y
Plotter
Floppy Discs
Record on
Floppy Discs
*Pi rV
High-Speed
Strip-Chart
Recorder
Switch
Power Supplies
t15V
15V
Switch
Box
1 2 3 4.
5 6 7 8

J
Input Voltage
to LVDT'S
Strain 1
Indicator Balance( &
GaQe Calibration
Module
*
To and From
To and From
Y Load Cell
To and From
LVDT'S 9&10
Remote Control Ten Strain Gages
for simultaneous
Operation of
Digital Oscilloscopes
Output Voltage from LVDT'S
to Digital Oscilloscopes
Figure 3.10: Schematic Diagram of Data Acquisition System

60
Figure 3.11: Data Acquisition System in Test Pit Facility
Figure 3.12: Typical Deflection Recording on Digital Oscilloscope

Figure 3.13: Typical Deflection Output on X-Y Plotter

62
3.5 Loading System: Rigid Plate Load vs. Flexible Dual Wheels
Two loading devices were used during the course of this research:
a rigid plate loading and flexible dual wheel loading. Rigid plate
loading was accomplished with a 12-inch diameter steel plate. A cage
was used in conjunction with the plate to evenly distribute the load
over the plate's area, and to increase the plate's rigidity. The plate
was always set on a thin layer of hydrocal (plaster) for levelling and
to evenly distribute the load. A diagram of the rigid plate loading
system is shown in Figure 3.14.
A set of small jet aircraft wheels (Piper Aircraft 31T) were used
for flexible dual wheel loading. The wheels were purchased from a
second-hand dealer for use in the test pit. An axle, which was
compatible with the existing loading system, was designed and machined
for the wheels. The wheels carried a pressure of 100 psi and were
designed to operate at 3,750 lbs. However, the wheels easily carried
5,000 lbs. each for a total of 10,000 lbs. on the dual wheel system.
The actual loaded area was determined from wheel imprints made in the
laboratory at different load levels. A picture of the dual wheel system
is shown in Figure 3.15. Note that it was necessary to attach cables to
the wheels to prevent them from rotating about their vertical axis.
A hydraulic loading system, which could apply static and dynamic
loads, vas used with both rigid plate and dual wheel loading. Loading
cycles could be preset for any combination of loading time and rest
period. The time required for the load to be fully applied could not be
controlled, and was dependent on the distance the loading ram had to
travel. Therefore, the load was applied faster with the rigid plate
than with the flexible wheels.

Figure 3.14: Rigid Plate Loading System

Figure 3.15: Flexible Dual Wheel Loading System

65
After extensive experience with the rigid plate and the dual wheel
devices, several advantages and disadvantages were observed for each.
These are as follows:
I. Rigid Plate Loading.
A. Advantages
- loading time could be controlled very accurately, since the
load came on almost instantaneously;
- the loading area was circular and constant; and
- the position of the load was always known because it was
difficult for the plate to move during loading; and
- there was less wear and tear on the loading system since very
little ram movement was required for loading.
B. Disadvantages
- very high shear stresses were induced at the edge of the
plate, causing it to sink and forcing a plane of failure;
- analysis of a rigid plate on a multi-layer system was a major
problem, since there was no computer program available that
accurately predicted stresses and strains under a rigid
load. An analytical procedure was developed using an elastic
layer analysis, but it proved to be extremely tedious.
II. Flexible Dual Wheel Loading.
A. Advantages
- this loading was close to flexible type loading (constant
pressure) and could be modeled more easily with existing
computer programs;
- dual wheel loading was more representative of actual truck
loads in the field; and

66
- the wheels were easier to position than the plate, since they
did not require setting with hydrocal.
B. Disadvantages
- the load did not come on instantaneously since the wheels had
to deform before a load was applied. Furthermore, the time
required for the load to come on, depended on load
magnitude. Therefore, it was difficult to set the loading
time and to evaluate creep strain accumulation for dynamic
loading conditions. For this reason, all creep tests were
performed using static loads.
- up to 2 1/2 inches of ram movement was required for loading,
which caused greater wear and tear in the loading system;
- the loading area varied with load and was not circular (more
difficult to model); and
- the wheels tended to roll during loading so that the exact
position of load was not known.
It v/as evident that neither of the loading devices was perfect, but
the disadvantages associated with the plate were overwhelming. The
loading was not representative of wheel loads and the analysis procedure
required tremendous amounts of time. Therefore, the dual wheel loading
system was used for the majority of tests performed.

CHAPTER IV
EFFECT OF ENCLOSED CONCRETE
TEST PIT ON PAVEMENT RESPONSE
4.1 Introduction
The test pit used in this research was made up of 8-inch concrete
walls and a 12-inch concrete slab, which enclosed a volume of 8 ft. by
12 ft. by 6 ft. deep. The layered pavement system was placed and tested
within this volume. Most analytical solutions and computer programs,
consider the layered system (or soil mass) to be infinite in lateral
extent and semi-infinite in depth. Three-dimensional finite element
programs could model the test pit, but as discussed later, these pro
grams were found to be either too expensive or inaccurate. Therefore, a
study was undertaken to evaluate the effects of the test pit floor and
walls on the response of the layered system to an applied load, and to
establish a methodology to account for these effects.
4.2 Preliminary Analysis
An initial attempt was made to predict the measured response of the
asphalt concrete pavement with an elastic layer computer program. The
pavement deflection and strain measurements used for this analysis were
obtained at a temperature of 18.3 C (65 F), using the plate loading
system at 10,000 lbs. and 0.1 sec. loading time (see Section 7.1).
Parameters for the sand subgrade and 1imerock base were determined from
plate load tests performed in the test pit. Conventional analytical
solutions, which consider the pavement layers to be infinite in lateral
67

68
extent and semi-infinite in depth, were used to obtain these
parameters. A sand subgrade modulus of 14,000 psi was calculated using
Boussinesq's theory for a rigid circular load on a semi-infinite mass.
Using Burmister's two-layer theory for similar conditions, a limerock
modulus of 75,000 psi resulted (see Table 5.11 and accompanying
discussion). This procedure is commonly used to obtain modulus values
from plate bearing tests performed in the field. The modulus values
obtained were essentially equivalent moduli, which represent the
behavior of all the materials below the tested surface. These values
are sometimes used to predict the response of the entire pavement system
using elastic layer analysis.
An asphalt concrete modulus of 145,000 psi was calculated for a
temperature of 18.3 C (65 F), using previously established correlations
with measured asphalt viscosity (see Appendix A). Therefore, the
following moduli, Poisson's ratios, and layer thicknesses were used in
an elastic layer computer program (BISAR) to predict pavement response:
Layer Poisson's
Material
Thickness (in.)
Modulus (psi)
Ratio
Asphalt Concrete
4 1/8
145,000
0.35
Limerock Base
6 3/4
75,000
0.40
Sand Subgrade
semi-infinite
14,000
0.40
Figure 4.1 shows the measured deflections and the deflections
predicted for the system above (Predicted 1). Although the shapes of
the deflection basins matched reasonably well outside the loaded area,
the measured deflections were grossly overpredicted. It appeared that
the effect of the floor was not properly accounted for by simply using
an effective layer modulus for the sand subgrade. A second computer run

o
6
30
T
36
T
DISTANCE FROM LOAD CENTER (INS.)
Cl

70
was made with a semi-infinite concrete embankment {E = 3,500,000 psi
and u = 0.20) underneath the sand layer. The sand layer was assigned a
finite depth of 48 in. and all other parameters were unchanged. The
predicted deflection basin for this case is also shown in Figure 4.1
(Predicted 2), and shows that the concrete foundation had a very
significant effect on the pavement's response. However, even with the
concrete embankment at a depth of 48 in., the measured response was
overpredict.ed, which made it clear that there were other factors
affecting the pavement response that were not being accounted for in the
analysis.
A series of program runs was made to determine the depth at which
the concrete floor had no effect on response. The depth of the sand
layer was varied from 48 in. to infinity, while maintaining all other
parameters constant. The results of this analysis are shown in Figure
4.2. Note that even at a depth of 120 in. (10 ft.) the concrete floor
had a significant effect on the predicted pavement response.
The following conclusions were drawn from this preliminary
analysis:
- the effect of the concrete floor must be directly accounted for
in the analysis procedure. It cannot be accounted for by simply
using an equivalent layer modulus determined from plate load
tests;
- the walls may have an effect on the response of the pavement.
The measured deflections were overpredicted, even when a concrete
embankment was introduced. Therefore, the wall effect needed to
be evaluated and accounted for in the analysis; and

DISTANCE FROM LOAD CENTER (INS.)

72
- the effect of the floor and walls must also be accounted for when
analyzing plate test data. Layer moduli determined from
analytical solutions for systems of semi-infinite depth, are not
suitable for pavement response prediction.
Therefore, a study was undertaken to evaluate the effect of the test pit
floor and walls on the response of the subgrade, the 1imerock base, and
the complete pavement system.
4.3 Effect of Test Pit Constraints
4.3.1 Analytical Model
Although the elastic layer theory computer program can model a
floor, it cannot model walls. The program considers all materials
infinite in the lateral direction. In addition, the program can only
handle flexible loads and the plate loading system is rigid. Therefore,
several available finite element computer programs were considered to
evaluate the effect of the test pit constraints on pavement perfor
mance. The AXSYM computer program was chosen for this purpose.
AXSYM is a three-dimensional finite element program written by
E. L. Wilson at the University of California at Berkeley. The program
is for solution of axisymmetric stress-deformation problems using
nonlinear stress-strain characteristics. It is specifically designed
for analysis of vertically loaded circular footings resting on or
beneath the surface of a soil mass. Only linear elastic stress-strain
characteristies were used in conjunction with the program. The basic
difference between the AXSYM model and the test pit is that AXSYM models
the walls as a circular tank, whereas the test pit is rectangular. This

73
was not considered a major problem and the effects could be defined and
approximated using this model.
Two other programs were also investigated; YBFE1 and SAPIV. YBFE1
is a two-dimensional (plane strain) finite element program developed for
soil-structure interaction problems. Preliminary program runs using
YBFE1 revealed that this program was unsuitable for predicting flexible
pavement response. The error was probably introduced by the plane-
strain nature of the model and the type of finite element used. SAPIV
is a three-dimensional finite element program developed for structural
dynamics but can be used to model homogeneous masses by means of a brick
or plate element. This program would have been most accurate in
modeling the test pit, but preliminary attempts at running the program
showed that an excessive amount of computer space was needed. This
space was not available on the current version of SAPIV at the
University of Florida. In addition, the cost of running the program was
prohibitive for the purposes of this project.
4.3.2 Effect of Constraints on Subgrade Response
Closed form solutions are available for the vertical displacement
of a rigid circle on both a semi-infinite mass and on a finite layer.
The following equations may be used to calculate these displacements:
Semi-infinite:
Ip Pavg(a)
pz = U-2) Pavgjal
Finite Layer:
where
' z F
p vertical displacement of rigid circle (ins.)
y Poisson's ratio
Pavg average pressure on the rigid circle (psi)

74
a radius of circle (ins.)
E Young's modulus (psi)
Ip influence coefficient: function of u and depth of
finite layer.
Subgrade moduli were calculated with these equations to show the effect
of assuming different finite layer depths and different Poisson's
ratios. Plate deflections measured in the test pit were used in the
calculations. The modulus values calculated are shown in Table 4.1.
As expected, the modulus increases with decreasing Poisson's ratio
and increases with increasing layer depth. However, the main purpose of
this comparison is to show that by assuming an infinite layer as opposed
to a finite layer, errors in excess of 20 percent may result in subgrade
modulus calculations. Similarly, errors in excess of 20 percent may
result in the modulus if a Poisson's ratio of 0.5 is assumed as opposed
to 0.3. Therefore, when calculating modulus for pavement response
prediction using plate load data, it is necessary to use the finite
layer solution. Poisson's ratio values are difficult to determine, but
values of 0.3 to 0.4 are usually considered reasonable for granular
materials. A Poisson's ratio of 0.3 was assumed for the sand subgrade
in the test pit, since laboratory tests by other researchers indicated
that this was a typical value for the Fairbanks Sand.
AXSYM was used to determine the effect of the test pit walls on the
response of the sand subgrade. The sand subgrade was assumed to be a
finite layer of 48-inch thickness. A subgrade modulus of 15,420 psi was
calculated for an assumed Poisson's ratio of 0.3 (see Table 4.1). The
following computer runs were made with these parameters to determine the
wall effect:

75
Table 4.1: Sand Subgrade Modulus for Different Layer Depths and
Poisson's Ratios
Modulus Values:
Sand Subgrade
Poisson's
Depth of Finite Layer
(in.)
Ratio
Semi-Infinite
24
36
48
0.2
17,890
14,590
16,020
16,610
0.3
16,960
13,410
14,530
15,420
0.4
15,650
12,100
13,290
13,940
0.5
13,980
10,260
11,510
12,160
Note:
(a) Calculated
using average 12-in.
plate deflection at 15
psi on 5th
loading cycle.

76
- wall at 15 ft., friction!ess;
- wall at 8 ft., frictionless;
- wall at 4 ft., frictionless; and
- wall at 4 ft., full friction.
It should be noted that rigid plate loading and finite element grids of
similar geometry were used in all runs.
The plate deflections as well as the deflection basins predicted by
the program, were identical for all cases, indicating that the wall had
absolutely no effect on the response. However, the deflections
predicted by the AXSYM program were about 14 percent less than predicted
by closed form solution (6.55 E-3 vs. 7.59 E-3 in.). It seemed like the
finite element grid used in the AXSYM runs was not fine enough.
Therefore, the number of elements was doubled and the program was
rerun. Although the program solution was closer, it underpredicted
deflections by about ten percent (6.83 E-3 vs. 7.59 E-3 in.). However,
one interesting point is that deflections remained unchanged away from
the loaded plate for the increased element grid.
Based on these results it seemed apparent that the type of finite
element used by AXSYM could not properly handle the high stress
concentrations at the edge of the rigid circle. Therefore, one cannot
put reliance on the rigid plate deflections predicted by AXSYM, except
on a relative basis. It also seems that the error introduced by the
high stress concentrations on these elements does not affect the
deflections away from the loaded area.
Several runs were made with the elastic layer computer program to
verify the accuracy of the AXSYM program away from the rigid loaded
area. A finite layer of 48 in. was used with a modulus of 17,000 psi

77
and a Poisson's ratio of 0.3. For the AXSYM program, a flexible load
was used and the walls were placed at 8 ft. and assumed frictionless.
The elastic layer program predicted deflections that were identical to
the deflections determined by closed form solution. The AXSYM solution
was identical to the elastic layer solution outside the loaded area and
was within 3 percent of the elastic layer solution underneath the load.
One additional AXSYM run was made to insure that the rigid and
flexible plate AXSYM solutions gave the same results away from the
loaded area, since the comparison of AXSYM and the elastic layer
solution was done for a flexible load. This comparison showed that the
AXSYM rigid and flexible plate solution predict identical deflections
beyond 2 in. of the loaded area.
The following conclusions were made after having verified the
deflections predicted by AXSYM:
- the walls have absolutely no effect on the response of the sand
subgrade to a load applied at its surface; and
- the floor has a definite effect on the sand subgrade response. A
finite layer closed form solution should be used to determine the
subgrade modulus from plate bearing test data in the test pit.
4.3.3 Effect of Constraints on Limerock Base Response
The effect of the concrete floor on the response of the limerock
base was investigated by a series of runs with the elastic layer
solution computer program. Two systems were analyzed: a 6.75 in.
limerock base over a semi-infinite subgrade; and a 6.75 in. limerock
base over a finite subgrade of 48 in. on a concrete embankment. Three
limerock base moduli were used for each system: 30,000, 60,000, and

78
100,000 psi. All other material properties were the same for all runs
and are given in Table 4.2.
The predicted deflection basins for each system are presented in
Table 4.2. The deflection differences for systems with and without a
concrete embankment (or floor) are also shown in Table 4.2. These
differences indicate that the effect of the concrete floor was to reduce
the deflections by an amount that is relatively independent of the
stiffness of the 1imerock base (approximately 2.5 E-3 in.).
A comparison of the deflection basins for the three cases studied
is presented in Figure 4.3. Clearly, the effect of the concrete floor
is considerable and must be accounted for in the analysis.
The following AXSYM runs were made to determine the effect of the
walls on the 1imerock base response:
- wall at 7 ft., no friction, rigid plate;
- wall at 7 ft., no friction, flexible plate;
- wall at 4 ft., no friction, rigid plate; and
- wall at 4 ft., full friction, rigid plate.
The following pavement system was used in the analysis:
Modulus (psi)
Poisson's Ratio
Thickness (in.)
Limerock:
90,000
0.40
6.75
Sand:
14,530
0.30
36
This system was underlain by a rigid base. A relatively high 1imerock
modulus was chosen for the analysis, since this stiffer system would be
affected to a greater degree by wall friction. A pressure of 50 psi on
a 12-inch diameter area was applied in all cases.

Table 4.2: Effect of Concrete Floor on Surface Deflections for Different Base Stiffnesses
Surface Deflections (in.)
# of Layers
Moduli
(psi)
Poisson's
Ratios
Thickness
(In.)
Distance From Center
of the Plate (in.)
0
4
6
9
12
18
24
30
36
48
2
30,000
0.4
6.75
2.43E-2
2.17E-2
1.70E-2
1.13E-2
8.74E-3
5.89E-3
4.39E-3
3.50E-3
2.91E-3
2.19E-3
15,420
0.3
SM-INF
3
30,000
0.4
6.75
15,420
0.3
48.0
2.18E-2
1.92E-2
1.45E-2
8.83E-3
6.27E-3
3.48E-3
2.06E-3
1.26E-3
7.79E-4
2.84E-4
3,500,000
0.2
SM-INF
Difference
2.5E-3
2.5E-3
2.5E-3
2.47E-3
2.47E-3
2.41E-3
2.33E-3
2.24E-3
2.24E-3
1.91E-3
2
60,000
0.4
6.75
1.89E-2
1.72E-2
1.44E-2
1.08E-2
8.68E-3
6.00E-3
4.46E-3
3.52E-3
2.90E-3
2.17E-3
15,420
0.3
SM-INF
3
60,000
0.4
6.75
15,420
0.3
48.0
1.64E-2
1.47E-2
1.20E-2
8.33E-3
6.27E-3
3.64E-3
2.17E-3
1.32E-3
8.10E-4
2.92E-4
3,500,000
0.2
SM-INF
Difference
2.5E-3
2.5E-3
2.4E-3
2.47E-3
2.41E-3
2.36E-3
2.29E-3
2.20E-3
2.09E-3
1.88E-3
2
100,000
0.4
6.75
1.59E-2
1.47E-2
1.28E-2
1.02E-2
8.49E-3
6.05E-3
4.53E-3
3.55E-3
2.92E-3
2.16E-3
15,420
0.3
SM-INF
3
100,000
0.4
6.75
15,420
0.3
48.0
1.35E-2
1.23E-2
1.04E-2
7.81E-3
6.12E-3
3.74E-3
2.28E-3
1.39E-3
8.52E-4
3.03E-4
3,500,000
0.2
SM-INF
2.4E-3 2.4E-3 2.4E-3 2.39E-3 2.37E-3 2.31E-3 2.25E-3 2.16E-3 2.07E-3 1.86E-3
Difference

DEFLECTION, IN.
DISTANCE FROM CENTER OF PLATE, IN.

81
Figure 4.4 shows a comparison between deflection basins for the
wall at 7 ft. with no friction and the wall at 4 ft. with full
friction. This comparison gives a direct indication of the effect of
having the wall at 4 ft. as opposed to having no wall. As shown in the
figure, the effect of the wall was to shift the deflection basin upward
by a small amount. The actual deflections for each case, given in Table
4.3, show that the deflection difference between the two basins is about
0.23 E-3 in., or about 2.2 percent of the plate deflection. This effect
is relatively insignificant, especially considering that the accuracy of
our measurements was somewhere in this range.
An elastic layer program run was made to evaluate the accuracy of
the AXSYM solution for this two-layer case. The elastic layer program
run gave a deflection basin identical to the AXSYM flexible plate run
with wall at 7 ft. The basin from the AXSYM rigid plate run with wall
at 7 ft. was also identical to these basins outside of the loaded
area. Figure 4.5 shows the deflection basins plotted for these three
runs. Mote that the rigid plate deflections seem low relative to the
flexible plate, again showing AXSYM's inability to handle the high
stress gradients induced at the edge of the plate. There is also some
discrepancy under the load between the AXSYM flexible plate solution and
the elastic layer solution, but this is small. In general, it seemed
that the AXSYM solution was accurate and could be used to evaluate the
wall effects on a relative basis.
Therefore, the following conclusions were made concerning the
effect of the test pit constraints on the 1 imerock base response:
- the floor effect is significant and must be accounted for, but
the effect is independent of 1imerock base modulus; and
- the wall effect is insignificant and can be ignored.

PREDICTED DEFLECTIONS (INS.)
DISTANCE FROM LOAD CENTER (INS.)
Figure 4.4: Effect of Test Pit Walls on Limerock Base Response

Table 4.3: Predicted Deflections Using AXSYM
Deflections (E-3 in.)
Position
Distance
From
Load Center
of Wall
0.0
6.0
7.5
9.0
12.0
16.0
20.0
24.0
30.0
36.0
42.0
48.0
4 ft. (NF)*
10.19
10.17
8.06
7.09
5.46
3.82
2.61
1.74
0.90
0.41
0.17
0.10
A-4 ft. (FF)*
10.16
10.14
8.03
7.06
5.43
3.78
2.56
1.68
0.82
0.32
0.08
0.0
B-7 ft. (NF)*
10.38
10.36
8.25
7.27
5.65
4.01
2.79
1.92
1.06
0.54
0.25
0.08
B A
0.22
0.22
0.22
0.23
0.22
0.23
0.23
0.24
0.24
0.22
0.18
0.08
* NF Ho Friction
FF Full Friction

PREDICTED DEFLECTIONS
5E-:
10E:
15E-:
DISTANCE FROM LOAD CENTER (INS.)
oo
-p>
Figure 4.5: Comparison of AXSYM and Elastic Layer Theory Solutions

85
4.3.4 Effect of Constraints on Three-Layer System
A series of runs was made with the elastic layer computer program
to evaluate the effect of the concrete floor on the pavement system
response. Three asphalt concrete moduli were used: 145,000 psi (room
temperature), 1,500,000 psi (very cold pavement), and 3,500,000 psi
(Portland cement concrete). Two runs were made for each of these
pavement systems: one with the floor at a depth of 48 in. and the other
with a semi-infinite subgrade layer. Figure 4.6 shows the layer
thicknesses and properties used for each system.
The predicted deflection basins for each of these pavement systems
are tabulated in Table 4.4. The deflection differences resulting from
the presence of the floor are also listed in the table. For all cases,
the effect of the floor is to shift the deflection basin upward (i.e.
the deflections decrease by a uniform amount). In addition, the
decrease in deflection is relatively independent of the stiffness of the
asphalt concrete layer. The magnitude of the decrease is approximately
3.5 E-3 in. for the pavement at room temperature and 3.2 E-3 in. for the
Portland cement concrete pavement. This decrease, of course, is for a
particular set of support conditions and will be different for another
set. The magnitude of the decrease would also change with magnitude of
load and load configuration (e.g. 16-inch plate vs. 12-inch plate or
dual vs. single load). The main point is that the effect of the floor
in the test pit is significant and must be accounted for, whether by
estimating the magnitude or preferably by modeling the floor with
appropriate boundary conditions.
The test pit wall effect on the three-layer system was determined
using the AXSYM computer program. Again three pavement modulus values

(b) With SemiInfinte Subgrade
(a) Underlain by Rigid Base
Figure 4.6: Three-Layer Systems as Modeled for Analysis

Table 4.4: Tabulated Deflection Basins to Show Effect of Test
Pit Floor on Pavements of Different Stiffness
Predicted
Deflections (E-
-3 in.)
System
Distance
From Load Center (in.
)
Description
0.0
0.4
6.0
9.0
12.0
18.0
24.0
30.0
36.0
48.0
o
Semi-
Infinite
21.1
19.5
17.1
13.8
11.3
9.03
7.13
5.78
4.80
3.54
LO
rH
Floor
0 48 in.
17.5
16.0
13.6
10.3
8.32
5.60
3.78
2.54
1.68
0.70
Difference
3.6
3.5
3.5
3.5
3.48
3.43
3.35
3.24
3.12
2.84
o
o
o
Semi-
Infinite
14.0
13.4
12.7
11.5
10.4
8.41
6.88
5.70
4.79
3.54
o
o
ir>
S
1^
FI oor
0 48 in.
10.7
10.1
9.43
8.29
7.14
5.25
3.79
2.70
1.89
0.87
UJ
Difference
3.3
3.3
3.27
3.26
3.26
3.16
3.09
3.0
2.90
2.67
o
o
o
Semi-
Infinite
11.9
11.5
11.1
10.3
9.49
8.01
6.74
5.68
4.83
3.60
CD
o
in
*
CO
FI oor
0 48 in.
8.7
8.33
7.90
7.13
6.36
4.93
3.72
2.75
1.99
0.86
UJ
Difference
3.2
3.17
3.20
3.17
3.13
3.08
3.02
2.93
2.84
2.74

88
were used: 145,000 psi, 1,500,000 psi, and 3,500,000 psi. For each
pavement system, the following AXSYM runs were made: frictionless wall
at 7 ft., frictionless wall at 4 ft., and full friction wall at 4 ft.
Figure 4.7 shows the thicknesses and properties used for each layer and
illustrates the systems as modeled. Mote that for all cases the sand
was modeled as a 48-inch layer underlain by a rigid base.
The predicted deflection basins for each of these runs are tabu
lated in Table 4.5. These results show that for all pavement stiff
nesses, the effect of the wall was to shift the deflection basin upward
by a small amount. Note that the wall effect was considered to be the
difference in deflection between the case of the frictionless wall at 7
ft. and the case of the full friction wall at 4 ft. The effect of the
wall increased with increasing pavement stiffness and was relatively
small for all cases. For the range of modulus values used in the test
pit (100,000 psi to 1,500,000 psi), it seems that the wall effect will
be essentially constant for a given stress level and support condi
tions. A shift factor of 1.0 E-3 in. could be used for a load of 10,000
lbs. and the support conditions used. It may be that this value is
adequate for most support conditions used in the test pit, but one
program run should be made to check this value once the subgrade and
limerock moduli are determined.
The following conclusions were drawn concerning the effect of the
test pit constraints on the three-layer system:
- the floor effect must be considered in the analysis of measured
deflection basins, but the effect is independent of the asphalt
concrete modulus for a given set of foundation conditions;

Figure 4.7: Pavement System Models to Determine Wall Effect

90
Table 4.5: Effect of Test Pit Malls on Surface Deflections for
Pavements of Different Stiffness
Predicted Deflections (E
-3 in.
)
System
Distance
From Load Center
(in.)
Description
Plate
7.5
9.0
12.0
16.0
20.0
24.0
30.0
36.0
42.0
48.0
A
Wall 0 7 ft.
Frictionless
13.19
10.91
9.76
7.93
6.11
4.72
3.65
2.47
1.63
1.03
0.60
o
o
o
B
Wall 0 4 ft.
12.90
10.62
9.47
7.67
5.88
4.53
3.51
2.43
1.74
1.38
1.26
in
Frictionless
R
Ui
C
Wall 0 4 ft.
Full Friction
12.37
10.09
8.93
7.10 .
5.26
3.85
2.76
1.54
0.72
0.23
0.05
A C
0.82
0.82
0.83
0.33
0.35
0.37
0.89
0.93
0.91
0.30
0.55
A
Wall 07 ft.
Frictionless
7.95
7.48
7.11
6.33
5.29
4.35
3.55
2.59
1.33
1.24
0.80
O
o
B
o
Wall 0 4 ft.
7.70
7.25
6.90
6.12
5.13
4.25
3.53
2.71
2.15
1.83
1.73
m
Frictionless
ii
LU
C
Wall 0 4 ft.
Full Friction
6.77
6.29
5.93
5.12
4.06
3.10
2.28
1.31
0.60
0.16
0.01
A C
1.18
1.19
1.18
1.21
1.23
1.25
1.27
1.28
1.23
1.08
0.79
A
Wall 07 ft.
Frictionless
6.33
6.08
5.S5
5.34
4.61
3.92
3.31
2.52
1.87
1.33
0.91
O
o
B

Wall 0 4 ft.
6.18
5.93
5.71
5.22
4.56
3.95
3.42
2.81
2.38
2.13
2.06
o
in
Frictionless
ro
ii
C
Wall 0 4 ft.
Full Friction
4.87
4.62
4.39
3.86
3.13
2.44
1.82
1.06
0.49
0.13
0.0
A C
1.46
1.46
1.46
1.48
1.48
1.48
1.49
1.46
1.38
1.20
0.91

91
- the wall effect was significant for the three-layer system and
must be accounted for to achieve correspondence between measured
and predicted deflection basins;
- the effect of the walls was to decrease deflections uniformly,
however, the effect was independent of asphalt concrete
modulus. Therefore, the wall effect can be accounted for by a
simple shift factor determined from computer runs using AXSYM.
4.4 Methodology to Account for the Effect of Test Pit
Constraints on Pavement Response Prediction
Based on the evaluation of the effects of the test pit constraints
(Section 4.1 to 4.3), a methodology was established to account for these
effects when interpreting data from load tests performed in the test
pit. Data interpretation methods were established for plate load tests
performed on the subgrade and base layers, and for plate load tests and
dual wheel load tests performed on the asphalt concrete layer.
4.4.1 Rigid Plate Loading on the Subgrade
The previous analysis showed that the test pit floor had a
significant effect on subgrade response which could not be properly
accounted for by simply using an equivalent modulus determined from the
solution for a semi-infinite system. In addition, the test pit walls
were found to have absolutely no effect on the response of the subgrade
to a load applied at its surface. Therefore, a finite-layer solution
for a rigid circular plate was determined to be suitable for the
evaluation of plate test data on the subgrade in the test pit.

92
4.4.1.1 Procedure to Determine Subgrade Modulus in the Test Pit
The following equation should be used to determine the subgrade
modulus from plate deflections (Ret. 98):
E Ip Pavg (a)
pz
where, pz vertical displacement of rigid circle (in.)
Pavg average pressure on the rigid circle (psi)
a radius of circle (in.)
Ip influence coefficient: function of Poisson's ratio
and depth of finite layer (Ref. 98)
E Young's modulus (psi).
4.4.2 Rigid Plate Loading on the Reinforcing Base Layer
The test pit floor was found to have a significant effect on the
load response of the base layer which could not be accounted for by
using an equivalent modulus. However, the effect was found to be
independent of base layer moduli within reasonable values. The test pit
walls were found to have an insignificant effect on the base response.
Therefore, the analysis method used may be infinite in lateral extent
but needs to account for the effect of the floor. Several approaches
were considered based on these findings.
The lack of a proper analytical tool (i.e. closed form solution or
computer program) is one problem associated with determining the rein
forcing layer modulus of a finite two-layer system, based on plate load
deflection measurements. Burmister's theory provides such a tool for a
two-layer semi-infinite system, where Poisson's ratio is assumed to be
0.5 for both layers. However, moduli determined using this solution did

93
not properly account for the effect of the test pit floor. Modulus
values determined from 12-inch plate tests were 25 percent less than
moduli from 16-inch plate tests.
The major source of error in this approach is the assumption of an
equivalent subgrade modulus based on the semi-infinite layer solution.
An equivalent modulus based on plate test deflections is only equivalent
for the same loading configuration and layer geometry that existed
during testing. That is, the equivalent modulus for the case of a 12-
inch plate on the sand subgrade will only give equivalent plate
deflections for similar conditions. Once the limerock base layer is
introduced, these equivalent moduli will give neither equivalent
deflection basins nor equivalent maximum deflections.
Two elastic layer theory computer runs were made to illustrate this
point. Figure 4.8 shows the layer thicknesses and material properties
used for the two runs. Equivalent subgrade moduli were calculated for
these runs based on 12-inch plate deflections measured in the test
pit. For the semi-infinite case, a modulus of 16,960 psi was obtained,
and for the finite layer case, a modulus of 14,530 psi was calculated
(See Table 4.1). The limerock modulus was chosen as 90,000 psi.
Figure 4.9 shows the predicted deflection basins for each of these
systems. As shown in the figure, neither the deflection basin nor the
maximum deflection are the same. This clearly shows that using an
equivalent subgrade modulus and Burmister's theory to determine the base
layer modulus is inadequate, since two totally different base moduli
would result for systems with and without a floor.
The computer program AXSYM models the floor and the walls of the
test pit, as well as rigid plate loading, but earlier analyses showed

SEMI-INFINITE
(a)
Figure 4.8: Equivalent Systems Based on Maximum Plate Deflection on Subgrade

DEFLECTION (INS.)
DISTANCE FROM LOAD CENTER (INS.)
Figure 4.9: Comparison of Response of Equivalent Systems Based on Maximum Plate Deflection on Subgrade

96
that this program could not accurately predict deflections under a rigid
load. Therefore, AXSYM could not he used to determine layer moduli from
measured plate deflections.
Considering these findings and the analytical tools available, the
following methods were established to determine the base modulus from
plate deflections measured in the test pit:
1. A method was developed to adjust the deflections measured in
the test pit to determine the deflection that would be obtained
for the same load on a two-layer semi-infinite system with no
walls (i.e. a Burmister system). The adjusted deflection was
then used in Burmister1s theory to obtain the base layer
modulus.
2. A method was developed to approximate rigid plate loading with
an elastic layer computer program. The modulus of the base
layer was determined by using this model and matching the
measured deflections by trial and error.
The first method listed is an alternative way of using Burmister's
theory to calculate base modulus. The idea is to adjust the measured
deflections to obtain an equivalent deflection for a two-layer semi
infinite system with a Poisson's ratio of 0.5. Figure 4.10 shows a
comparison of the actual test pit system and the Burmister system. This
comparison shows that two adjustments to the measured deflection are
required to obtain an equivalent deflection for use in Burmister's
theory: one to account for the effects of the floor and the walls, and
another to account for the change in response due to the difference in
Poisson's ratio.

A. System in Test Pit
Figure 4.10: Comparison of Test
LIMEROCK
V= 0.5
: 6 %"
SUBGRADE
E = 15,420
v- 0.5
SEMI-INFINITE
B. Burmister System
it System and Burmister System

98
Earlier analyses showed that the floor effect on the two-layer
system response was independent of base layer modulus, and the wall
effect was negligible. Also, the effect of the floor was found to be
uniform for the entire deflection basin. Therefore, for any given
subgrade modulus, an elastic layer program can be used to determine the
deflection change caused by the floor on the base response. This floor
effect is added to the measured deflection, since the effect of the
floor is to reduce deflections under load. Mote that this change would
be different for different subgrade moduli and different loading
configurations.
The effect of lower Poisson's ratio may also be determined using
the elastic layer program. Maintaining all other parameters constant, a
lower Poisson's ratio will increase deflections under load. Therefore,
the deflection change caused by the difference in Poisson's ratio is
subtracted from the measured deflection.
This approach was used to determine the modulus of the base layer
from measurements made in the test pit. The average measured deflection
on the limerock base for 50 psi loading on a 12-inch plate was
13.7 E-3 in. Two elastic layer theory computer runs were made to
determine the effect of the floor on the response: one for a two-layer
system with the floor at 48 in., and another for a two-layer semi
infinite system. The following parameters were used for both runs:
E = 15420 psi, y = 0.3 for the subgrade; and E = 50,000 psi and y = 0.4
for the base. The thickness of the limerock was 6.75 in. The effect of
the floor was 2.5 E-3 in., as determined from these runs.
The following series of runs were made to determine the effect of
Poisson's ratio on deflections for different base layer moduli:

99
Table 4.6: Computer Runs to Determine Poisson's Ratio Effect
System
Number Modulus (psi)
Poisson'
Ratio
s
Modulus (psi)
Poisson's
Ratio
I. Limerock
30,000
0.4
30,000
0.5
Sand
15,420
0.3
Vb
15,420
0.5
II. Limerock
60,000
0.4
60,000
0.5
Sand
15,420
0.3
VS.
15,420
0.5
III. Limerock
100,000
0.4
100,000
0.5
Sand
15,420
0.3
vs
15,420
0.5
The only variable
in these runs
was the
base modulus, since this value
was yet unknown.
The absolute change in
deflection as well as
the
percentage change
as determined
from these three runs, are given below:
System
Defl ection
Percent Maximum
Number
Change (in.)
Defl ection
I
1.8
E-3
7.4%
II
1.1
E-3
5.8%
III
0.8
E-3
5.0%
Unlike the floor effect, the effect of Poisson's ratio was not
independent of the base layer modulus. However, as shown in the table,
the percentage change of maximum deflection was relatively constant for
base moduli ranging from 30,000 psi to 100,000 psi. Therefore, an
average percentage change in maximum deflection was calculated for an
assumed base layer modulus. A value of six percent was used, corre
sponding to a base modulus of 60,000 psi (System number II).
The maximum measured deflection adjusted for floor effects was 16.2
E-3 in. (13.7 + 2.5). Therefore, the difference in deflection due to
Poisson's ratio was six percent of this, or about 1.0 E-3 in. This
adjustment was subtracted from the measured test pit deflection adjusted

101
DISTANCE FROM PLATE CENTER (INS.)
Figure 4.11: Stress Distribution Under Rigid Plate on Semi-
Infinite Mass: 50 psi Average Pressure

102
calculating the total load induced by each section and dividing by the
area over which this load acts. The stress rings were then used in the
elastic layer program to simulate the stress distribution under a rigid
plate. The deflection results of the different runs were combined to get
the predicted rigid plate deflections. The stresses within each ring
were adjusted until the resulting deflections were equal for all points
under the loaded area, indicating that a fl at (rigid) plate was being
modeled. The process was repeated for different moduli until the pre
dicted and measured deflections matched.
this procedure was used to calculate the modulus of the subgrade for
comparison to closed form solution for a finite layer of 48 in. and a
Poisson's ratio of 0.3 (E = 15,420 psi). The two values were identical
which confirmed the reliability of the procedure.
The same approach was used to determine the base layer modulus. The
subgrade was again modeled as a 48-inch finite layer (E = 15,420, u =
0.3), and the base layer was 6 3/4-in. thickness with a poisson's ratio
of 0.4. The base layer modulus was adjusted by trial and error until the
predicted plate deflections matched the measured. Several iterations
were required to obtain the proper stress distribution under the rigid
pi ate.
A base layer modulus of 53,000 psi was calculated. This value is
very close to the value of 52,550 psi predicted by the first method
discussed using Burmister's theory.
The rigid plate approximation procedure is probably most accurate,
since the floor and the Poisson's ratios can be modeled directly.
However, this method is extremely tedious and time consuming. For each
modulus value attempted, it was necessary to determine the stress

103
distribution under the plate by trial and error. Therefore, the method
was considered impractical.
4.4.2.1 Procedure to Determine the Base Layer Modulus in the Test Pit
Two procedures were established to determine the base layer modulus
from plate deflections measured in the test pit. The first method
involved adjusting the measured deflections so that Burmister's theory
could be used. The second method involved approximating the stresses
under a rigid plate using an elastic layer computer program and
determining the base modulus by trial and error until the measured
deflections were matched. Modulus values predicted by both methods were
almost identical. However, the second method was very tedious and was
considered impractical. Therefore, the first method is recommended to
determine the base layer modulus.
The following procedure should be followed:
1. Calculate the subgrade modulus as per Section 4.4.1.1.
2. Determine the effect of the floor on deflections as follows:
a. Assume a reasonable base layer modulus and make two runs
with the elastic layer computer program: one for a two-
layer semi-infinite system, and one for a two-layer system
underlain by a rigid base. The rigid base can modeled using
a concrete modulus (3,500,000 psi) or higher.
b. Calculate the difference between the maximum deflections
predicted by the program for these two systems. This
difference is the floor effect.
3. If the Poisson's ratio for the suhgrade or base layers are
assumed different than 0.5, their effect on deflections relative

104
to a value of 0.5, should be determined as follows (Note: Skip
step 3 if Poisson's ratios of 0.5 are assumed):
a. Assume a reasonable base layer modulus and make two runs
with the elastic layer program: one with the assumed
Poisson's ratios for base and subgrade, and one with a
Poisson's ratio of 0.5 for both layers. Both cases should
be modeled as two-layer semi-infinite systems.
b. Calculate the difference between the maximum deflections
predicted by the program for these two runs.
c. Calculate the Poisson's ratio effect as a percentage of the
maximum deflection of the pavement system with the assumed
Poisson's ratios.
4. Adjust the measured plate deflection for floor effect by adding
to it the value calculated in step 2.b above.
5. Calculate the effect of Poisson's ratio as the percent of the
adjusted deflection calculated in step 4, using the percentage
calculated in step 3.c.
6. Subtract the value determined in step 5 from the adjusted
deflection calculated in step 4. This is the final adjusted
deflection.
7. Calculate the base layer modulus using the final adjusted
deflection in Burmister's theory (Ref. 27).
4.4.3 Predicting Pavement System Response in the Test Pit
The test pit floor and walls were found to have a significant effect
on the load response of the three-layer pavement system, which as
explained earlier cannot be accounted for by using equivalent modulus

105
values for the base and/or subgrade layers. Therefore, the analysis
method used to evaluate oavement deflection and strain measurements in
the test pit needs to account for these effects directly.
The earlier discussion {see Section 4.4.2) relating to the lack of
proper analytical tools for the evaluation of finite two-layer systems is
also applicable for finite three-layer systems. The problem is even more
acute for the latter, especially considering that deflection and strain
predictions are needed at many points throughout the pavement and not
just on the loaded plate. Evaluation using an adjustment procedure
similar to the one developed for the finite two-layer system is clearly
out of the question in terms of practicality. Furthermore, even if it
were practical, to the author's knowledge, there are no solutions
available to predict the stress and strain distribution in a multi
layered system loaded by a rigid circle. The AXSYM computer program can
make these predictions, but was found to be inaccurate for our purposes.
Therefore, the rigid plate approximation procedure (established and
explained in Section 4.4.2), using the elastic layer computer program, is
the only available solution that is suitable. However, the elastic layer
program cannot model walls, so the wall effect has to be accounted for
separately. Fortunately, the effect of the walls was to decrease
deflections uniformly, and was independent of asphalt concrete modulus.
Therefore, the wall effect can be accounted for by a simple shift factor
determined from computer runs using AXSYM.
The rigid plate approximation procedure was used to predict the load
response of the pavement system in the test pit. Pavement deflection and
strain measurements used for this analysis were obtained at a temperature
of 18.3 C (65 F), using the plate loading system at 10,000 lbs. and

106
0.1 seconds loading time. Previously determined subgrade and limerock
base moduli were used. The pavement system was modeled as follows:
Layer Poisson's
Material Thickness (in.) Modulus Ratio
Asphalt Concrete
4 1/8
(a)
0.35
Limerock Base
6 3/4
53,000
0.40
Sand Subgrade
48
15,420
0.30
Concrete Embankment
semi-infinite
3,500,000
0.20
(a) Modulus of asphalt concrete to be determined by trial and error.
The modulus of the asphalt concrete layer was adjusted by trial and error
until the predicted plate deflection matched the measured plate
deflection, adjusted for wall effect. The wall effect for this system
was determined to be 1.0 E-3 in. in Section 4.3.4.
This procedure resulted in a predicted asphalt concrete modulus of
170,000 psi, which is within 12 percent of the modulus of 145,000 psi
determined from correlations with measured asphalt viscosity. This
prediction is very reasonable, and it should be emphasized that these
results were obtained from the rational evaluation of direct
measurements. Therefore, it seems that the analysis methods used to
evaluate the measurements are reasonable.
Deflections and strains were measured at several points throughout
the pavement and these were used to further evaluate the analysis
methods. Figure 4.12 shows a comparison of the measured and predicted
deflection basins. The plate deflections match exactly, since the
asphalt concrete modulus was adjusted to match the measured plate
deflection. In addition, there is good correspondence for the entire
deflection basin.

DEFLECTION (INS.)
5E-3
10E-3
15E-3
Figure 4.12: Measured vs. Predicted Deflection Basins at 18.3 C (65 F)

108
Table 4.6 shows the measured strains and the strains predicted using
the elastic layer program. Strains are shown for both 10,000 lb. and
7,000 lb. loadings. Two sets of predicted strains are given: one for an
asphalt concrete modulus of 170,000 psi which was determined by matching
the measured plate deflection, and one for an asphalt concrete modulus of
145,000 psi which was determined from asphalt viscosity correlations.
The latter set of strains was approximated by multiplying the former by a
factor of 170,000/145,000.
As shown in the table, the trend of the predicted strains agreed
very well with the measured strains. The strain magnitudes also agreed
well, although the strains predicted using the modulus obtained by
matching plate deflection (170,000 psi) were lower than the measured.
The strains approximated using the modulus determined from asphalt
viscosity correlations (145,000 psi) were almost identical to the
measured strains. Considering the possible errors involved in adjusting
the predicted deflections to match the measured, more reliance should be
placed on the strain measurements for evaluation, so the value of 145,000
psi seems most reasonable.
In any case, it seems like good predictions of measured deflections
and strains in the asphalt concrete can be obtained by using the analysis
methods presented. However, as explained earlier, the rigid plate
approximation procedure is extremely tedious and time consuming. The
procedure is impractical. Furthermore, no other rational method is known
to be available. Primarily for this reason, rigid plate loading was
abandoned, and a set of flexible dual wheels were designed and
constructed for use in the test pit.

109
Table 4.7: Measured and Predicted Surface Strains
Surface Strains (micro-strain)
Load
(lbs.)
Strain
Distance From Load
Center
(in.)
8 3/32
12 1/16
17
15/16
28 5/16
Measured Strains
5.4
34.9
34.4
10,000
Predicted^ Strains
8.9
30.0
30.0
19.4
Predicted^) Strains
10.4
35.2
35.2
22.7
Measured Strains
9.0
28.7
28.2
18.6
7,000
Predicted^ Strains
6.2
21.0
21.0
13.6
Predicted^) strains
7.3
24.6
24.6
16.0
(a) The following material properties used for predictions:
Modulus
(psi)
Poisson's
Ratio
Thickness
(in.)
Asphalt Concrete
170,000
0.35
4.125
Limerock
53,000
0.40
6.75
Sand*
15.420
0.30
48.0
* Underlain by rigid base.
(b) Strains predicted for system in (a) above were adjusted by a linear
factor of 170,000/145,000 to approximate strains for a similar
system, but with an asphalt concrete modulus of 145,000 psi.

110
The elastic layer computer program can be used directly to model the
flexible dual wheels. Therefore, rational predictions can be made by
using results from this program adjusted for the wall effect, which can
be determined using the AXSYM computer program. Although an example is
not given here, analyses of response measured with the dual wheel loading
system resulted in excellent correspondence between predicted and
measured response. These analyses are given in a later chapter. The
AXSYM program cannot be used directly to model the dual v/heel loading
system, since it can only model an axisymmetrical loading system.
4.4.3.1 Summary Procedure to Evaluate Deflection and Strain Measurements
in the Pavement System in the Test Pit
Rational procedures were established to predict deflections and
strains in a layered pavement system enclosed in a concrete test pit, and
loaded by either a rigid circular plate or by flexible dual wheels.
Prediction of rigid plate loading involves approximating the stresses
under the rigid plate by using the method described in Section 4.4.2
along with an elastic layer computer program. This method is
impractical, and therefore, rigid plate loading should not be used, if at
all possible. The flexible dual wheels can be modeled directly with the
elastic layer program. Aside from the difference in modeling the load,
the procedures for evaluating pavement response in the test pit are the
same for both loading devices.
The following steps should be followed:
1. Calculate the subgrade modulus as per Section 4.4.1.1.
2. Calculate the base layer modulus as per Section 4.4.2.1.
3. Determine the effect of the walls on deflections as follows:

Ill
a. Calculate an asphalt concrete modulus using previously
established correlations of modulus with measured asphalt
viscosity and temperature.
b. Make two AXSYM computer runs with the above parameters and
known layer thicknesses: one using a frictionless wall
(free vertical movement) at seven feet from the load, and
one with a full-friction wall (no nodal displacement) at
four feet from the load. A flexible, 12-inch diameter load
can be used with a load equal to the applied load.
c. Calculate the difference between the maximum deflections
predicted by the program for these two systems. This is the
wall effect.
4. Using the same parameters, make an elastic layer computer run,
modeling the rigid base with a concrete embankment
(E = 3,500,000 psi or higher) to determine deflections and
strains as required.
a. Use the rigid plate approximation procedure described in
Section 4.4.2 if the rigid plate loading system is used.
b. Use circular flexible loads to model the dual wheel loading
system.
5. Subtract the wall effect determined in step 3.c from the
deflections obtained in step 4. The strains were determined to
be unaffected by the walls. Therefore, these are the predicted
deflections and strains.
6. Compare the predicted deflections and strains with the measured
values. Deflection and strain distributions (basins) should be
plotted for comparison, rather than comparing results on a point

112
by point basis. More emphasis should be placed on the
comparison of strains and the shape of the deflection basin than
on absolute magnitude of deflections, but the measured and
predicted values should match as exactly as possible. If the
two do not match, proceed to steo 7. A certain amount of
judgement and experience is required to determine when the best
match is achieved.
7. Repeat steps 4 through 6 for different asphalt concrete moduli
until good agreement is obtained between measured and predicted
response. Base and subgrade moduli may also have to be adjusted
slightly, particularly if conditions have changed relative to
when plate load tests were performed on these materials.
This procedure should result in reasonable asphalt concrete moduli
for the structural analysis of asphalt concrete pavements using elastic
layer theory.

CHAPTER V
MATERIALS AND PLATE
TESTING PROCEDURES
5.1Introduction
A three-layer pavement system, consisting of a sand subgrade, a
crushed limerock base, and an asphalt concrete surface was placed and
compacted in the test pit. Standard laboratory tests were performed to
characterize each of these materials. Plate load tests were performed
along with conventional density and moisture measurements to determine
the properties of the subgrade and limerock materials in situ. In
addition, low-temperature rheology tests were performed on asphalt
samples recovered from the test pit pavement.
5.2Laboratory Tests
5.2.1 Fairbanks Sand Subgrade
Fairbanks sand was used for a subgrade material in the test pit.
This sand is used extensively in the test pit facility for evaluating
different base course materials and is known to have uniform proper
ties. Therefore, laboratory tests were not performed on the sand.
Typical laboratory test results for this material are given in Table
5.1.
5.2.2Crushed Limerock Base
A crushed limerock from the Ocala formation was used for the base
layer. Compaction tests, sieve analyses, and limerock bearing ratio
113

114
Table 5.1: Laboratory Test Results: Fairbanks Sand
Limerock Bearing Ratio (LBR) 31
Optimum Moisture Content 12.3
Dry Density of Optimum Moisture 108.9
Sieve Analysis
Sieve Size Percent Passing
# 10 100
# 40 90
# 60 59
#200 5

115
(LBR) tests were conducted to evaluate this material. These test
results are given in Table 5.2. A relatively high LBR value of 192 was
obtained, which indicates that this material is strong compared to other
limerocks used in Florida. Florida specifications require an LBR of
only 100 for use in base course materials.
Results of washed and unwashed sieve analyses are also shown in
Table 5.2 to illustrate the difference in gradation that can occur
between the two procedures. In particular, note difference in the
percentage passing the number 200 sieve. A washed sieve analysis should
always be used to characterize Florida limerocks.
5.2.3 Asphalt Cement and Asphalt Concrete
The asphalt concrete used in the test pit was obtained from
Whitehurst Construction Company in Gainesville. An S-I plant mix from a
local project was used. Table 5.3 lists the aggregate types and blends
used, as well as the job mix formula and the specification range for
this mix.
Samples of the asphalt concrete placed in the test pit were taken
for laboratory testing. The Abson recovery procedure was used to ex
tract the asphalt from the mixture and laboratory tests were performed
on the recovered asphalt and the aggregate. Results of sieve analyses,
Marshall analysis, and asphalt and asphalt mixture properties are given
in Table 5.4. The sieve analyses results show that the percent passing
was high (in some cases greater than the specification limits) for
almost all sieve sizes, indicating that the aggregate blend was somewhat
fine. In addition, the asphalt content of 7 percent.was higher than the
6 percent called for in the job mix formula. Despite these anomalies,

116
Table 5.2: Laboratory Test Results: Ocala Formation Limerock
Test
Test Number
1
2
3
Avg.
Limerock Bearing Ratio (LBR)
201
180
195
192
Optimum Moisture Content (%)
10.2
10.3
11.7
10.7
Dry Density at Optimum Moisture
123.5
123.3
121.3
122.7
Sieve Analysis
Sieve Size
Percent
Passing*
Avg. (Washed)
Avg.
(Unwashed)
3 1/2
100
100
2
98
84
1 1/2
96
83
1
88
77
3/4
84
74
1/2
80
69
3/8
77
64
4
73
53
10
63
39
40
43
24
60
33
17
200
22
4
* Average of three analyses.

117
Table 5.3: Source of Materials and Job Mix Formula for Asphalt Concrete
Material Material
Number
Type
Producer
Pit No.
1
#67 Stone
Florida Rock Industries
34-106
2
#89 Stone
Florida Rock Industries
34-106
3
#140 Screenings
Florida Rock Industries
34-106
4
Local Sand
Whitehurst Construction Co.
Alachua, FL
Percentage by Weight Total Aggregate Passing Sieves
B1 end
25%
30%
20%
25%
Job Mix
Spec
Number
1
2
3
4
Formula
Range
3/4
100
100
100
100
100
100
1/2
72
100
100
100
93
88-100
3/8
47
96
100
100
86
75-93
No. 4
5.9
50.7
92.2
100
60
47-75
No. 10
2.8
8.3
72.5
100
43
31-53
No. 40
2.5
2.2
36.9
87.7
31
19-35
No. 80
2.2
1.6
19.6
23.3
11
7-21
No. 200
1.5
1.4
10.8
2.8
3.7
2-7
Sg. Gr.
2.335
2.370
2.510
2.480
2.415

Table 5.4: Test Pit Asphalt Concrete Properties
GRADATION
Sieve Size
Percent Passing
Sample 1
Sampl e 2
Job Mix Formula
Specification
3/4"
100
100
100
100
1/2"
96.24
98.00
93.00
88-100
3/8"
92.40
96.40*
86.00
75- 93
#4
71.92
75.85*
60.00
47- 75
#10
53.26*
56.39*
43.00
31- 53
#40
35.68*
37.19*
31.00
19- 35
#80
13.73
14.17
11.00
7- 21
#200
5.23
4.92
3.70
2- 7
Bitumen
7.00*
7.45*
6.00
6+ 0.5
* Mote: exceeded specification limits

DENSITY
Sample No.
1 2
3 4
5
6
Density (pcf)
136.6 137.1
137.7 137.4
137.2
136.2
Average 137
.2 pcf
AIR VOID CONTENT
AV = 7.0%
VOLUME OF MINERAL AGGREGATE VOIDS
VMA
-
16.5%
MARSHALL TEST
Sampl e Mo. 2
4
6
Average
Stability (lbs.) 2371
2140
2136
2216
FIow 10
9
10
10
MECHANICAL PROPERTIES
Sample No.
1
3
Temperature (F)
41
77
Tensile Strength (psi)
324.8
105.4
Modulus (psi)
140,231
31,515
ABSON RECOVERY
Sample No.
1
2
Penetration at 77 F
84
70
Viscosity at 77 F (Poises)
2,330
3,412

119
the results of the Marshall analysis indicated that the mixture was
good. An average stability of 2,200 lbs. was obtained with an average
flow of 10.
Low-temperature rheology tests were performed at different
temperatures on asphalt cement samples recovered from the mixture used
in the test pit. These samples were taken when the mixture was
initially placed and compacted. Absolute and constant power viscosities
were computed from measurements with the Schweyer rheometer at 60, 25,
15, 5, -5, and -10 C (140, 77, 59, 41, 23, and 14 F). These results are
shown in Table 5.5. Linear regression analysis of constant power
viscosity (n 10q) with absolute temperature resulted in the following
equation:
109 n100 = 168-492 65.928 log T (Eqn. 5.1)
where R2 = 0.992
n100 constant power viscosity Pa.-sec.
T = temperature, K.
Low-temperature rheology tests were also performed on asphalt
samples recovered from the test pit mixture after all pavement response
tests were completed. Absolute and constant power viscosities, computed
from tests at 25, 15, 5, and -5 C (77, 59, 41, and 23 F), are given in
Table 5.6. Linear regression analysis of constant power viscosity with
absolute temperature resulted in the following equation:
log n10Q = 148.271 57.484 log T (Eqn. 5.2)
where R2 = 1.0
^100 constant power* viscosity, Pa.-s.
T temperature, K

120
Table 5.5: Rheology and Penetration of Asphalt Recovered From
Test Pit During Initial Placement: September, 1982
Temperature
C (F)
Absol ute
Viscosity
n j (Pa-s)
Com pi ex
Flow (C)
Constant Power ^
Vi scosity
n10Q (pa-s)
60 (140)
2.893 E2


25 (77)
9.247 E4
0.78
2.151 E5
15 (59)
1.006 E6
0.81
2.647 E6
5 (41)
1.850 E6
0.58
2.520 E7
-5 (23)
1.216 E7
0.64
1.589 E8
-10 (14)
2.417 E7
0.52
1.211 E9
(a) Penetration at 25 C (77 F) = 77
(b) n
100
100
n
c-1
c+1

Table 5.6 Rheology and Penetration of Asphalt Recovered From
Test Pit After All Testing: September, 1985
Temperature
C (F)
Absol ute
Viscosity
n^Pa-s)
Com pi ex
Flow (c)
Constant Power ^
Viscosity
n100 {Pas)
60 (140)
8.350 E2


25 (77)
2.176 E5
0.65
1.145 E6
15 (59)
7.216 E5
0.59
7.093 E6
5 (41)
3.044 E6
0.53
6.727 E7
-5 (23)
2.200 E7
0.60
4.766 E8
(a) Penetration at 25 C (77 F)
45

122
Temperature-viscosity relationships for asphalt recovered both
during initial placement, and after all testing was completed, are shown
in Figure 5.1. Clearly, the asphalt viscosities were higher after
testing, indicating that the asphalt hardened in the test pit.
Therefore, Equation 5.1 was used to compute constant power viscosities
for analyzing pavement response measurements made close to the time that
the pavement was originally placed. Equation 5.2 was used to calculate
constant power viscosities for analyzing oavement response measurements
taken later.
5.3 Material Placement and Compaction
A Fairbanks sand subgrade, crushed limerock base, and asphalt
concrete layer were placed and compacted in the test pit. The sand
subgrade was already in place, since this material was previously used
to evaluate different base course materials in the test pit. However,
the top 4 in. of the subgrade was replaced and recompacted.
The subgrade and base were compacted at optimum moisture content.
The optimum moisture content and dry density at optimum were determined
for the Fairbanks sand and for the crushed limerock by the modified
AASHO (T-180) procedure (see Tables 5.1 and 5.2). The moisture in the
stockpiled materials was adjusted by either drying or wetting until
optimum moisture was reached. The moisture content of the stockpiled
materials was determined from representative samples taken at the center
and edges of the stockpile. The speedy moisture device was used for
this purpose. If the material was too wet, it was spread with a front
end loader for drying and then remixed until the material was at
optimum. If too dry, the amount of water required to bring the moisture

Viscosity (Pa.Sec.)
123
Temperature
Figure 5.1: Viscosity Temperature Relationships for Asphalt
Recovered from the Test Pit

124
content up to optimum was determined, added to the material, and mixed
in a concrete mixer. The weight of material to obtain the required
layer thickness was calculated based on the density at optimum
moisture. The material was weighed on a heavy-duty scale and placed in
the test pit for compaction.
A vibratory compactor was used on all materials placed in the test
pit. The sand was compacted until 100% of dry density at optimum was
achieved. Density measurements were obtained with a Troxler nuclear
density gage and samples of compacted material were taken for moisture
determination. After all plate load tests were performed on the sand,
6 3/4 in. of crushed 1imerock base were placed and compacted in one
layer. Several chalk lines were drawn along the inside walls of the
test pit to monitor the thickness of the layer. The thickness of the
layer was also checked using density measurements obtained from nuclear
density tests. Also, SDeedy moisture tests were performed periodically
during compaction. The measured and calculated thicknesses agreed
closely. Ninety-five percent of laboratory density v/as achieved for the
1imerock layer.
A 4 1/8-in. layer of asphalt concrete was placed and compacted
after all plate tests were performed on the 1imerock base. Prior to
placement, asphalt concrete cores instrumented with strain gages (see
Section 3.4.1) were positioned on the surface of the 1 imerock base. In
addition, two wood pegs with thermocouples were hammered into the base,
and positioned at different distances above the surface of the base
layer. These thermocouples provided temperature measurements at
different points within the asphalt concrete layer. A layer of Celotex
was placed along the perimeter of the test pit wall to prevent any

125
restraining effect of the wall on the pavement. Lines were drawn on the
Celotex at different elevations to monitor the thickness of the layer.
The asphalt concrete hot mix was distributed using the hopper
system described in Section 3.2. The compacted thickness of the layer
was approximated by leveling the uncompacted material and stepping on it
to approximate the reduction in thickness that would occur with
compaction. The mixture was compacted using a vibratory compactor. To
prevent the compactor from sticking to the mixture, the bottom of the
compactor was sanded until smooth, and the contact area was kept cool by
using wet rags during compaction. Compaction was continued until it was
obvious that the pavement had cooled to the point where further compac
tion would have no effect.
The Troxler nuclear density device was used to determine the
density of the asphalt concrete as compacted in the test pit. The
majority of measurements were made with the backscatter method (non
destructive) while a few measurements were made by direct transmission,
which provides increased measurement accuracy. In addition, a four-inch
diameter core was taken from the test pit for density, asphalt content,
and air void content determinations. The density values determined by
the backscatter method were correlated to those determined by direct
transmission and from the core. The resulting average density was
133 pcf and was very consistent throughout the test pit area. The air
void content as determined from the core was seven percent. The density
achieved was 97 percent of the Marshall comoaction.

126
5.4 Material Properties In Situ
Static and dynamic plate load tests were performed in the test pit
to determine the modulus of the Fairbanks sand subgrade and the limerock
base. In addition, in-place density measurements were made with the
Troxler nuclear density gage and samples were taken for moisture content
determinations. Two series of plate tests were performed: one
immediately after the sand and the limerock were placed and compacted,
and another after all tests on the asphalt pavement were completed and
the pavement was removed. Plate tests procedures and results are
presented in this section.
5.4.1 Plate Load Test Procedures
The following basic procedure was used for all plate tests
performed on the sand and the limerock:
1) The plate was placed at a specified location. Molding plaster
(hydrocal) was used to insure uniform loading on a level
surface.
2) The loading mechanism was placed over the plate's center.
3) A seating load was applied and released and all instrumen
tation was zeroed. A seating load of 60 to 100 lbs. was
apDlied to the sand subgrade. This load was uncontrollable
because of the constraints in setting up the loading
mechanism. A seating load of approximately 1,200 lbs. was
applied to the limerock base.
4) The material was loaded incrementally up to a maximum stress
level, which was close to what the material would encounter in
a pavement system under a standard truck load. After removing

127
Table 5.7: Load Increments Used for Plate Load Tests:
Fairbanks Sand Subgrade
Fairbanks
Sand Subgrade
12"
PI ate
16"
Plate
30"
PI ate
Load
(lbs.)
Stress
(psi)
Load
(lbs.)
Stress
(psi)
Load
(lbs.)
Stress
(psi)
Static Load
0
0
0
0
0
0
800
7.1
800
4.0
800
1.1
900
8.0
900
4.5
1,200
1.7
1,000
8.8
1,000
5.0
2,000
2.8
1,100
9.7
1,200
6.0
3,000
4.2
1,300
11.5
1,400
7.0
5,000
7.1
1,500
13.3
1,800
9.0
7,000
9.9
1,700
15.0
2,200
11.0
9,000
12.7


2,600
13.0
10,600
15.0


3,000
15.0


Dynamic Load
1,250
11.0
2,325
11.6



Table 5.8:
Load Increments Used for Plate Load Tests:
Limerock Base
Limerock Base
12"
PI ate
16"
Plate
Load
Stress
Load
Stress
(lbs.)
(psi)
(lbs.)
(psi)
Static Load
0
0
0
0
800
7.1
800
4.0
1,200
10.6
1,200
6.0
1,600
14.1
1,800
9.0
2,000
17.7
2,400
12.0
2,500
22.1
3,000
15.0
3,000
26.5
4,000
19.9
3,500
30.9
5,000
24.9
4,500
39.8
6,000
29.8
5,650
50.0
8,000
39.8


10,050
50.0
Dynamic Load
5,200 46
8,800
43.8

129
the seating load, all instrumentation was zeroed prior to
testing. Each increment of static loading was maintained
until there was essentially no increase in deflection before
proceeding to the next load increment. This was determined
visually by continuously monitoring the deflection readings on
the chart recorder. After the maximum load was applied for a
particular cycle, it was released completely before initiating
the next loading cycle. Tables 5.7 and 5.8 give the
incremental loads used for each material and plate size. In
all cases the first increment of load was 800 lbs. This was
inevitable, since it was the minimum load that could be
applied once the hydraulic loading system was activated.
5) This incremental loading sequence was repeated for a total of
five cycles.
6) After the last static load cycle was applied, 100 dynamic load
repetitions were applied. The stress levels used for dynamic
loading on each material are also given in Tables 5.7 and 5.8.
7) Once all testing was complete a nuclear density test was
performed and a soil sample was taken for moisture content
determination.
5.4.2 Plate Tests Immediately After Placement
Three 12-inch and three 16-inch diameter plate load tests were
performed on both the sand subgrade and 1imerock base shortly after they
were placed and compacted. In addition, one 30-inch diameter plate was
performed on the sand. Figure 5.2 shows the positions of each plate
load test performed on the Fairbanks sand and Figure 5.3 shows the

130
Figure 5.2: Location of Plate Load Tests: Fairbanks Sand Subgrade
Figure 5.3: Location of Plate Load Tests: Limerock Base

131
positions of the tests performed on the limerock base. The moisture
content (w) and dry density (y^) measured at each test position is also
shown on these figures.
Stress-deflection curves were plotted for each plate load test
performed. Typical examples of these plots for tests performed on the
sand subgrade are shown in Figure 5.4 and 5.5 (12- and 30-inch plate,
respectively). Both the 12- and 16-inch plate tests on the sand
resulted in a linear stress-deflection relationship after the first
loading cycle, indicating that this material was stress independent
within the range of stresses applied. However, a curvilinear
relationship was observed for the 30-inch plate on the sand, even though
the same stress levels were used. The apparent reason for the nonlinear
response observed with the larger plate is discussed later in this
section.
Static and dynamic moduli for Fairbanks sand subgrade were
calculated based on resilient plate deflections, using the procedure
outlined in Section 4.4.1.1, with an assumed Poisson's ratio of 0.3.
The results of these calculations are given in Table 5.9. The
calculated moduli were very consistent for both the 12-and 16-inch plate
tests, except for test number one with the 16-inch plate, where a higher
water content and lower dry density were measured. Almost no difference
(0.5 percent) was observed between the average static and dynamic
modulus for the 12-inch plate tests, and only a small difference (4.6
percent) was observed for the 16-inch plate tests.
Since the stress-deflection relationship for the 30-inch plate test
was curvilinear, two modulus values were calculated: a resilient modulus
and a tangent modulus. Although the resilient modulus was unreasonably

6
\Q
o^'*
,ied Stress vs. _
rbanks ^and Subgrade
f\ec
Xo*
*.*
Applleu _
Fai "!/'i Sand


Table 5.9: Modulus Values Immediately After Placement: Fairbanks Sand Subgrade
12
" Plate
16
" Plate
30
" Plate
(1 Test)
£(a)
Water
Dry Unit
g(a)
Water
Dry Unit
g( a)
Water
Dry Unit
Test
Test
Content
Weight
Content
Weight
Content
Weight
Mo.
Type
(psi)
w-%
(pcf)
(psi)
w-%
(pcf)
(psi)
w-%
(pcf)
Static^)
14470
12900(d)
0
0
t-H
in
1
Dynamic(c)
15640
11.3
110.5
13620{d)
15.0
107.0
12.0
112.0
Static
15850
15170
6350(f)
2
Dynamic
14620
12.1
107.0
14150
11.4
111.0
Static
14840
14860
3
Dynamic
13000^)
12.5
109.0
14420
13.6
111.0
Static
15,050
15,000
15,100
12.0
112.0
Average
12.0
108.8
12.5
111.0
Dynamic
15,130
14,300
(a) Resilient modulus (c) Determined on the 100th repetition (e) Tangent modulus at 15 psi
(b) Determined on the fifth of load (f) Resilient modulus at
loading cycle (d) This value not included in average 15 psi

135
low, the tangent modulus is almost identical to the values determined
with the smaller plate sizes. This seems to indicate that the observed
nonlinearity was caused by the plate itself rather than being an actual
material characteristic. Because of its size, the 30-inch plate tends
to bridge more surface irregularities and it could not be set with
hydrocal, but was leveled with sand instead. Also, the 30-inch plate is
less rigid than the 12-inch plate loading system and much higher loads
are required to apply equivalent stress levels. These factors may be
the cause of the higher relative deflections measured at the lower
stress levels with the 30-inch plate. Apparently, once the plate
settles at higher stress levels, the measured relative response is the
same as measured with the smaller plates, which seems to be
representative of the actual response of the subgrade.
A typical stress-deflection curve for a 16-inch plate test
performed on the limerock base layer is shown in Figure 5.6. Both the
12- and 16-inch plate tests resulted in a curvilinear stress-deflection
relationship. It seems reasonable to assume that the nonlinearity is
directly attributable to the limerock, since the subqrade responded
linearly when tested independently.
Static and dynamic moduli for the limerock base were calculated
based on resilient plate deflections, using the procedure outlined in
Section 4.4.2.1, with an assumed Poisson's ratio of 0.4. A subgrade
modulus of 15,420 psi, based on typical 12-inch plate test results on
the subgrade, was used for all calculations. Since the stress-
deflection relationships were curvilinear, three modulus values were
calculated for the limerock: a resilient modulus for the entire stress
range; and initial tangent modulus for the lower stress levels; and a

Figure 5.6: Applied Stress vs. Deflection: 16-in. Plate on Limerock Base

137
final tangent modulus for the higher stress levels. The deflections
used to calculate these three modulus values are shown in Figure 5.7.
As shown on this figure, the stress-deflection relationships could
generally be divided into two straight line segments which intersected
at approximately 40 percent of the maximum stress applied.
The modulus values calculated for the limerock are given in Table
5.10. Considering the possible variation in thickness and subgrade
conditions for the different test positions, the moduli obtained were
very consistent for both the 12- and 16-inch plate tests. The average
static and dynamic resilient moduli were within seven percent of each
other for the 12-inch plate tests and within three percent for the 16-
inch plate tests. There was only a three percent difference between the
static resilient moduli calculated from the 12-inch plate tests versus
the 16-inch plate tests, and only a seven percent difference in the
dynamic resilient moduli. Hoy/ever, less variation was observed between
individual values for the 12-inch plate tests than for the 16-inch plate
tests. This was expected, since there is more room for error with the
larger plate. The 16-inch plate tends to bridge more surface irregular
ities, the plate is less stiff, and greater loads are required to apply
equivalent stress levels. Furthermore, a greater percentage of the
stress bulb falls Y/ithin the limerock layer for the 12-inch plate
loading system, which reduces the error introduced by differences in
subgrade conditions. Therefore, the 12-inch plate is recommended for
future testing.

Figure 5.7: Deflections Used to Calculate Limerock Moduli

Table 5.10 Modulus Values Immediately After Placement: Limerock Base
12" Plate
16" Plate
F(a)
r-res
F(b)
ttani
F(c)
Ltanf
Water
Dry Unit
E(a)
tres
p(b)
ttani
p(c)
Ltanf
Water
Dry Unit
Test
Test
Content
Weight
Content
Weight
No.
Type
(psi)
(psi)
(psi)
w-%
(pcf)
(psi)
(psi)
(psi)
v-%
(pcf)
Static^
48500
34850
63500
50250
24500
77100
1
Dynamic(
43800


9.5
117.4
63350


9.7
116.7
Static
55700
37450
89300
56800
27650
83250
2
Dynamic
51900


9.2
117.7
47100


10.5
117.2
Static
53450
40400
73250
56000
24500
77100
3
Dynamic
51000


10.4
116.7
47500


9.3
117.6
Static
52550
37550
75350
54350
25550
79150
Average
Dynamic
48900
9.7
117.3
52650
9.8
117.2
(a) Resilient Modulus
(b) Initial Tangent Modulus (Low Stresses)
(c) Final Tangent Modulus (High Stresses)
(d) Determined on fifth loading cycle
(e) Determined on 100th repetition of load

140
The initial tangent moduli for the limerock were higher for the 12-
inch plate tests than for the 16-inch plate tests, and correspondingly,
the final tangent moduli were lower. However, these moduli are
dependent on how the data are interpreted, where a relatively small
difference in the interpreted slope of the stress-deflection curve
results in a significant difference in the calculated tangent modulus.
In addition, the greater difference was observed for the initial tangent
modulus, indicating that plate seating or surface irregularities may
have been partially responsible. Evidence is presented in Chapter VII
that indicates that the initial tangent modulus from the 12-inch plate
tests results in better response prediction at lower stress levels. In
any case, the observed difference in moduli is small in terms of
response prediction, as evidenced by the consistency of the calculated
resilient moduli for the entire stress range. Furthermore, it will
later be shov/n that using an equivalent resilient modulus in an elastic
layer analysis results in good pavement response prediction for higher
stress levels (see Chapter VII). Therefore, it seems like an equivalent
resilient modulus for the limerock is suitable to obtain good pavement
response prediction.
Static and dynamic modulus values were calculated for the sand
subgrade and the limerock base using conventional procedures to
illustrate the errors that can result by obtaining equivalent moduli
assuming the subgrade layer is semi-infinite. Resilient subgrade moduli
were obtained from plate deflections using Boussinesq's theory for a
rigid circle on a semi-infinite mass. These equivalent subgrade moduli
were then used directly in Burmister's two-layer theory to obtain

141
limerock moduli from resilient plate deflections. The results of these
calculations are given in Tabel 5.11.
The limerock modulus values calculated by these methods are
anywhere from 50 to 100 percent greater than the values determined by
accounting for the effects of the test pit constraints. Furthermore,
totally different values were obtained for the 12- and 16-inch plates,
indicating that the analysis did not properly account for the difference
in response of the two plate sizes. Attempts to model pavement response
using these values indicated that they were unsuitable. Clearly, an
equivalent modulus cannot be used in a semi-infinite layer analysis to
obtain reinforcing layer moduli or to predict pavement response in the
test pit. This is equally true for deflection measurements taken in the
field where one pavement may be underlain by rock and another by muck.
This shows the need to define foundation conditions at depth for
pavements if suitable modulus values are to be determined for proper
structural analysis.
The procedures presented in Chapter IV resulted in reasonable
modulus values for pavement response prediction in the test pit. There
is no reason why the procedures could not be similarly applied in the
field. However, foundation conditions would have to be properly
defined.
5.4.3 Plate Tests After Pavement Removal
The asphalt concrete pavement was removed from the test pit after
all response tests at different temperatures were completed. The
pavement was removed one layer at a time and plate tests were performed
on the limerock base and the sand subgrade to define their properties.

142
Table 5.11: Modulus Values Without Accounting for Test Pit Constraints
12" !
PI ate
16" Plate
Test
Number
Test
Type
Equival ent
Subgrade
Modulus
(osi)
Resilient
Limerock
Modul us
(psi)
Equivalent
Subgrade
Modulus
(psi)
Resilient
Limerock
Modulus
(psi)
Static
13975
69875
12420
112910
1
Dynamic
14400
57830
13150
150285
o
Static
14750
81,040
14,450
96630
c
Dynamic
13560
74920
13520
97285
o
Static
14160
74530
14300
107250
o
Dynamic
12040
96320
13820
98715
Static
14295
75150
13720
105600
Average
Dynamic
13330
76360
13530
98000
(a) Excluded from average

143
Densities and moisture contents were also determined at this time. As
much care was taken as possible not to disturb the surface of the
limerock and the sand upon removal, but some disturbance was inevitable.
Three 12-inch plate tests were performed on the sand subgrade and
five 12-inch plate tests were Derformed on the limerock base. Only the
12-inch plate was used since it was found in earlier tests to give the
most consistent results (see Section 5.4.2). Figure 5.8 shows the
positions of the tests performed on the sand subgrade and Figure 5.9
shows the positions of the tests performed on the limerock base. The
measured moisture contents (w) and dry densities (Yd) are also shown on
these figures. Test positions one and two on the sand subgrade and one,
two, and three for the limerock base correspond to positions where
extensive load testing was done on the asphalt concrete pavement. The
other plate tests were performed at locations where load tests were not
performed.
Static Moduli for the sand subgrade were calculated based on
resilient plate deflections, using the procedure outlined in Section
4.4.1.1, with an assumed Poisson's ratio of 0.3. The results of these
calculations are given in Table 5.12. Note that the subgrade moduli
under the previously loaded areas (tests numbers one and two) were
higher than the value for the unloaded area (test number three).
Apparently, the sand densified locally under repeated loading. The
modulus value for the unloaded area was close to the values measured for
the original plate tests on the sand.
The average moisture content of the sand was about 10.4 percent,
which is lower than the average moisture of 12.5 percent measured when
the sand subgrade was first placed and compacted.

144

W= 10.2%
W =
10.9%
X
X
Test #3
Test #2
Test #1
>-N
Q
O
O
X
X
W= 11.2%
W = 9.5%
w =
10.0%
X
X
W= 10.5%
w =
10.4%
Figure 5.8: Location of Plate Load Tests: Fairbanks Sand Subgrade
Figure 5.9: Location of Plate Load Tests: Limerock Base

Table 5.12:
Modulus Values After Pavement Removal:
Fairbanks Sand Subgrade
Test
Number
E(a)
(psi)
Water
. Content
w-%
Dry Unit
Weight
(pcf)
x(b)
18400
10.0
Not Measured
2(b)
19050
10.0
Mot Measured
3(c)
16000
9.5
Not Measured
(a) Resilient Modulus
(b) Load tests performed on pavement at this location
(c) Ho load tests performed on pavement at this location

146
Moduli for the limerock base were calculated based on resilient
plate deflections, using the procedure outlined in Section 4.4.2.1, with
an assumed Poisson's ratio of 0.4. A subgrade modulus of 18,730 psi was
used for the plate tests performed at locations where pavement response
tests had been performed (test numbers one, two, and three). This value
was the average of the two moduli determined from plate tests on the
subgrade at locations where pavement response tests had been
performed. For plate tests at locations where no load tests were
performed on the pavement, a subgrade modulus of 15,420 psi was used,
which is the value calculated from the original plate tests on the sand
(Section 5.4.2).
The calculated limerock moduli are given in Table 5.13. These
values are siginficantly lower than the limerock moduli determined when
the limerock was initially placed and compacted (see Table 5.10). This
reduction seems to be directly related to the difference in water
content and dry density in the limerock. The original water content was
2.0% lower and the original dry density was 4.0 pcf higher than when the
pavement was removed. The reason for these changes is difficult to
determine with available data. However, it appears that loading had
little effect on the observed changes in limerock properties. The
modulus values were about equal for both loaded and unloaded areas, as
were the measured water contents and dry densities. Therefore, it seems
like some other phenomenon was responsible for these changes.
There are several possibilities for the observed increase in
moisture. Capillary rise may have occurred which transferred moisture
from the sand subgrade to the limerock base. This seems reasonable,
since there is a significant amount of fines in the crushed limerock

147
Table
5.13: Modulus
Values After Pavement
Removal:
Limerock Base
Test
Number
Subgrade
Modul us
(psi)
Limerock. .
Modulus'9'
(psi)
Water
Content
w-%
Dry Unit
Weight
(pcf)
! (b)
18730
30400
12.0
113.8
2(b)
18730
34850
11.6
114.2
3(b)
18730
34850
11.6
112.4
4(c)
15420
35000
11.7
113.2
5(c)
15420
35400
11.5
113.0
Average
34100
11.7
113.3
(a) Resillent Modulus
(b) Load tests performed on pavement at this location
(c) No load tests performed on pavement at this location

148
(> 20 percent passing the no. 200 sieve) and a decrease in subgrade
moisture was measured. However, the capillary rise would have occurred
during the first year the pavement was in the test pit, and based on
pavement response measurements at the surface, there appeared to be no
change in limerock modulus during this first year. Moisture migration
may have also occurred under successive cooling cycles, where a thermal
gradient was established in the pavement system and moisture migrated
from the warmer to the cooler region. Finally, the pavement was
accidentally flooded twice while in the test pit, and some moisture may
have gone into the limerock.
The decrease in density is also difficult to explain. When the
pavement was removed, the limerock had visibly consolidated under the
loaded areas. Even though part of this consolidation was known to have
occurred in the sand, it seems unlikely that the limerock expanded with
loading (i.e., dilation could not have occurred over the entire
layer). Errors in density measurements with the nuclear density device
may partially account for the observed reduction in density. It was
impossible not to disturb the surface of the limerock when the pavement
was removed, so that density measurements were generally made over areas
having surface irregularities. Other possibilities are that some of the
fines may have migrated out of the limerock with time and successive
cooling cycles or the limerock may have expanded slightly with increased
moisture or with freezing. Unfortunately, a final gradation was not
obtained for the limerock, since these findings were not anticipated.
Although there are various uncertainties including disturbance,
possible reduction in thickness of the limerock layer, and possible
drying of the limerock once the pavement was removed, two things are

149
clear from these latter plate test results: 1) the 1 imerock modulus
decreased; and 2) the subgrade modulus increased relative to when they
were initially placed and compacted. It will be shown in Chapter IX
that the same conclusions were reached based on pavement response
measurements taken close to the time the pavement was removed.

CHAPTER VI
PROCEDURES
6.1 Dynamic Plate Load Tests at Ambient Temperatures
These procedures correspond to the results presented in Chapter
VII. The tests were all performed at ambient temperatures using the
rigid plate loading system, and only dynamic load tests were
performed. Therefore, the procedures used for these tests were fairly
straightforward. These were as follows:
1. The 12-inch diameter plate was placed in position and leveled
by using a thin layer of hydrocal (see Figure 3.14).
2. All strain gages were connected and the LVDT support system was
set in place (see Figure 3.5). The LVDT's were then positioned
at specified distances from the plate.
3. The load-unload time was set by trial and error. Dynamic loads
were applied and recordings of deflection and load with time
were made with the digital oscilloscopes and the strip chart
recorder, respectively. The time of loading and rest period
between loads were determined from these recordings and were
adjusted as necessary to obtain the required load-unload
times. Since the plate loading system was very rigid, the load
was applied and released almost instantaneously, which resulted
in a square-wave loading pattern. Also, the load-unload times
did not have to be readjusted for different, load levels.
150

151
4. Fifty dynamic load repetitions at 10,000-lbs. were applied as a
seating load.
5. Dynamic load response measurements were made at load levels of
1,000, 4,000, 7,000, and 10,000 lbs., beginning with the
greatest load first.
6. For each load level, deflection and strain measurements were
recorded with the digital oscilloscopes for several successive
repetitions of dynamic load. Since the data acquisition system
could only monitor one strain gage at a time, dynamic loads
were applied until recordings were made for all ten strain
gages being monitored. Therefore, each strain gage was
monitored at a slightly different time.
6.2 Low-Temperature Pavement Response Tests
6.2.1 Introduction
These procedures correspond to the results presented in Chapter
VIII. The tests performed can be broken down as follows: thermal
response of the pavement during cooling; dynamic load tests at different
pavement temperatures; creep tests to determine the permanent
deflections and creep strains accumulated under static loads; and
dynamic load tests performed during creep tests to determine the effect
of permanent deformations on the dynamic load response of the
pavement. These tests were performed at three different positions in
the test pit pavement, and except for the temperature order of testing,
the procedures used were the same at all three tests positions.
Table 6.1 summarizes the temperature order in which these tests
were performed. As shown in this table, dynamic load tests were

152
Table 6.1: Summary of Order of Testing
Test
Series
Order of
Testing
Position
1
Position
2
Position
3
Dynamic
Load
1
0
C
(32 F)
0
C (32 F)
0
C (32 F)
2
6.7
C
(44 F)
6.7
C (44 F)
6.7
C (44 F)
Tests
3
13.3
c
(56 F)
13.3
C (56 F)
13.3
C (56 F)
4
0
c
(32 F)


Creep
1
13.3
c
(56 F)
6.7
C (44 F)
0
C (32 F)
Tests
2
6.7
c
(44 F)
0
C (32 F)
6.7
C (44 F)
3
0
c
(32 F)
13.3
C (56 F)
13.3
C (56 F)
Notes:
(a) Temperatures shown are nominal test temperatures. The measured
averaae pavement temperatures for each test series are given in
Table 8.2.
(b) All dynamic load tests and creep test were completed at one
position before going to the next. The positions were tested in
the following order: Position 3, Position 2, Position 1.

153
performed at all three test temperatures before any creep tests were
performed. This was done to eliminate any effect that creep might have
had on the dynamic response of the pavement. Dynamic load tests were
performed sequentially at 0.0 C (32 F), 6.7 C (44 F), and 13.3 C (56 F)
at all three test positions. There were two reasons for this: 1) the
thermal response characteristics of the pavement could be monitored for
a continuous cooling cycle; and 2) tests were performed more quickly.
By performing tests in this order, the pavement was cooled only once to
0.0 C (32 F) and tests at the higher temperatures were performed as the
pavement warmed up. Table 6.1 also shows that creep tests were
performed in a different sequence of temperature at each test
position. This was done to observe the effect of creep at one
temperature on the dynamic load response and creep response at a
different temperature.
6.2.2 Pavement Cooling and Initial Dynamic Load Tests
The following procedure was followed at all test positions:
1. The dual wheel loading system was placed in position (see
Figure 3.15).
2. All strain gages were connected in a half-bridge arrangement
with an external dummy located in an insulated box outside of
the test pit. This allowed the use of the half-bridge
arrangement, which is more stable than a quarter-bridge, while
still monitoring strains induced by cooling.
3. The LVDT's were prepared for 1ow-temperature tests and were
then positioned at specified locations on the pavement surface
(see Chapter hi).

154
4. The test pit cover was then sealed for cooling.
5. The strain gages were balanced and zeroed and an initial
temperature reading was taken prior to starting the cooling
unit.
6. After starting the cooling unit, temperature and strain
measurements were taken at specified time intervals. The time
intervals used are shown in the tables of pavement temperatures
during cooling presented in Appendix B.
7. The pavement was cooled until the temperature at the bottom of
the asphalt concrete layer was at 0.0 C (32 F). At this time,
the strains in the pavement were recorded and the cooling unit
was turned off.
8. Since the top of the pavement was colder than the bottom, the
pavement temperature was allowed to equalize so that the
properties of the asphalt concrete layer were uniform during
dynamic load tests. During this initial equalization period,
the temperature of the bottom of the pavement remained close to
0.0 C (32 F).
9. Once the top and bottom of the pavement were within about 1.5 C
(2.7 F), a final strain reading was made and dynamic load tests
were begun.
10.The strain gage dummy was changed from the gage located outside
the test pit to one located on the pavement surface. The dummy
gage was located at the north end of the pavement, perpendicu
lar to the axis of loading, so that it would not be affected by
applied loads. Each gage was then rebalanced with the new
dummy.

155
11. The load-unload time was set by trial and error for a load
level of 10,000 lbs., and 50 dynamic load repetitions were
applied as a seating load.
12. Dynamic response measurements were made at load levels of
1,000, 4,000, 7,000, and 10,000 lbs., beginning with the
greatest load first. The loading time had to be reset for each
load level since the time required to reach a given load level
increased with increasing load. This is because greater wheel
deformation was required to reach higher load levels. The
actual load-unload times used for each load level are presented
in Section 8.3.3 (Figure 8.28).
13. Once all dynamic load tests were performed at 0.0 C (32 F), the
pavement was allowed to warm up to the next test temperature
(6.7 C/44 F), and step 12 was repeated. Similarly, when
dynamic load tests were completed at 6.7 C (44 F), the pavement
was allowed to warm up to 13.3 C (56 F), where dynamic load
tests were again performed as described in step 12. Additional
cooling was unnecessary since the top and bottom of the
pavement remained at the same temperature as the pavement
warmed up.
6.2.3 Creep Test Procedures
As indicated earlier (see Table 7.1), the temperature order of
creep tests was different for tests performed at different test
positions. However, at any given temperature the creep test procedures
were the same. The same procedure described earlier was used to cool
the pavement down to a given test temperature. The cooling unit was run

156
until the bottom of the pavement was at or slightly below the test
temperature. Then the cooling unit was turned off and the temperature
of the top and bottom of the asphalt concrete layer were allowed to
stabilize. To warm up to a given test temperature from a cooler
temperature, the pavement temperature was simply allowed to increase
slowly with the insulated cover in place.
Once the pavement was at a specified test temperature, creep tests
were performed. Immediately prior to performing any creep tests,
dynamic load tests at 10,000 lbs. were performed, and deflections and
strains were recorded for several successive repetitions of dynamic
load. These initial measurements could then be compared to dynamic load
measurements after different durations of static load applications to
determine the effect of creep on the dynamic load response of the
pavement. In addition, these initial measurements provided a reference
from which permanent deflections and creep strains could be monitored.
Creep tests were then performed by apnl.ying different durations of
10,000-lb. static load. Load durations of 50, 50, 400, and 500 seconds
were applied sequentially with a rest period between loads equal to four
times the load duration. Dynamic load tests at 10,000 lbs. were
performed after each rest period, and deflections and strains were
recorded for several repetitions of dynamic load for each series of
tests. Permanent deflections and creep strains could then be determined
for any given duration of static load application by comparing the
deflections and strains recorded during dynamic load tests performed at
different times. The difference between the unloaded values of
deflections or strains measured at different times was the permanent
deflection or creep strain accumulated during that time.

CHAPTER VII
PAVEMENT RESPONSE AT
AMBIENT TEMPERATURES
Prior to installing the cooling unit, dynamic plate load tests were
performed to evaluate pavement response at ambient temperatures ranging
from 18.3 C (65 F) to 25.6 C (78 F). These tests also served to
evaluate the loading system and the measurement and data acquisition
system for the first time.
A cross-section of the pavement system in the test pit, including
the respective layer thicknesses is shown in Figure 7.1. This was the
pavement system used for all tests performed in this dissertation.
A summary of the tests performed, including the loading rates and
load levels used, is given in Table 7.1. This table also lists the test
position and date of testing for each series of tests. All tests were
performed with a 12-inch diameter rigid steel plate.
Four load levels were used for each series of tests to observe
changes in pavement system response with increasing load. A maximum
load of 10,000 lbs. was used, which is 1,000 lbs. greater than a
standard truck load. Therefore, pavement response was observed over the
entire range of loads encountered in an actual roadway.
Two loading rates were used during the initial tests performed in
March, 1983 in an attempt to evaluate the effect of different loading
rates on pavement response. For the fast loading rate, the load was
applied almost instantaneously, kept on for 0.1 sec. and kept off for
157

158
Table 7.1: Summary of Dynamic Plate Load Tests at Ambient Temperatures
Test
Series No.
Pavement
Temperature
Load Rate
(sec. on/sec. off)
Load Level
(kips)
Test
Position
Date
1
18.3 C/65 F
0.1/0.4
1,4,7 & 10
Center^
3/83
2
21.7 C/71 F
1.0/4.0
1,4,7 & 10
Center^3)
3/83
3
20.6 C/69 F
0.1/0.4
1,4,7 & 10
Center^
4/84
4
25.6 C/78 F
0.1/0.4
1,4,7 & 10
South^
4/84
(a) Plate was set at the center of the test pit.
(b) Plate was set three feet south of center.

159
0.4 sec. For the slow loading rate, the load was kept on for 1.0 sec.
with a 4.0 sec. rest period.
The results of all plate load tests performed at ambient
temperatures are presented and discussed in this chapter. They are
presented in numerical order according to the test series numbers given
in Table 7.1, which also corresponds to the chronological order in which
the tests were performed.
7.1 Initial Plate Load Tests at Fast Loading Rate
Test series number one, listed in Table 7.1, were the first dynamic
load tests performed on the pavement system in the test pit. These
tests were Derformed approximately six months after the asphalt, concrete
layer was placed and compacted. The position of the LVDT's during
testing are shown in Figure 7.1, where the LVDT support system was
omitted for clarity. The actual distances of the LVDT's from the center
of loading are given in Figure 7.2, which is a plan view of the test
area and shows the position of the strain gages installed on the
pavement's surface. Note that one strain gage was installed for each
LVDT position (except at the plate), and the instruments were positioned
such that duplicate readings were obtained. Two additional strain gages
were located along the longitudinal axis (12-ft. axis) of the test pit
for comparison. A total of ten deflection and ten strain measurements
were taken for all tests performed.
7.1.1 Dynamic Load Test Results
The measured deflections and strains for this series of tests are
given in Tables 7.2 and 7.3. The measurements shown are the average of

160
Figure 7.1: Test Pit Diagram:
Elevation

161
.
8'
>
AXIS OF SYMMETRY
>'
Figure 7.2: Test Pit Diagram: Plan

162
Table 7.2:
Surface Deflections:
Fast Loading Rate
Surface Deflections ^
(E-3 in.)
Load
Distance From Center
of Load (in.)
(lbs.)
0"
8 5/16"
12 1/6"
18 5/16"
28 1/4"
10,000
15.3
10.1
6.9
4.7
1.8
7,000
12.0
7.7
5.65
3.4
1.6
4,000
7.75
4.9
3.7
2.15
0.8
1,000
1.65

0.75

0.0
Notes:
(a) Readings shown are
average of
two LVDT's
(b) Average
pavement temperature:
18.3 C (65 F)
Table 7.3:
Surface Strains:
Fast Loading Rate
Surface Strains ^ (micro-strain)
Load
Distance From Center of Load (in.)
(lbs.)
8 3/32"
12 1/6"
17 15/16"
28 5/16"
10,000
5.4
34.9
34.4
7,000
9.0
28.7
28.2
18.6
4,000
5.8
20.4
19.7
12.9
1,000
2.3
6.4
7.0
4.3
Notes:
(a) Readings shown are average of two strain gages
(b) Average pavement temperature: 18.3 C (65 F)

163
two measurements taken on either side of the load. The readings were
repeatable and nearly identical for gages located at equal distances,
but on opposite sides of the center of loading.
The measured deflection basins for the different load levels are
shown in Figure 7.3. The plate deflection is shown as perfectly flat
and the deflection basin is defined by four additional measurements.
The measured surface strain distributions are shown in Figure 7.4.
Plots of load versus deflection and strain for the different measurement
positions (Figures 7.5 and 7.6, respectively) indicate that the pavement
response was slightly nonlinear in a stress-stiffening way. This
behavior is typical of pavements with granular bases and subgrades, and
is consistent with the nonlinear response observed on the plate tests
performed on the limerock base (see Chapter V).
7.1.2 Elastic Layer Simulation and Evaluation of Results
An elastic layer theory computer program (BISAR) was used to
predict the measured deflections and strains at the 10,000-lb. load
level. Attempts to predict this measured response are covered in detail
in Chapter IV. These attempts led to the establishment of procedures to
properly evaluate measurements made in the test pit. It was shown that
the floor and walls of the test pit have significant effects on load
response and a methodology was established to account for these effects
when analyzing test pit measurements (see Section 4.4). A procedure was
also established to model rigid plate circular loading using BISAR.
Good predictions of measured deflections and strains were obtained by
using these procedures. Therefore, it was shown that the response of
the asphalt concrete pavement system in the test pit could be accurately

DEFLECTION (x 103 INS.)
DISTANCE FROM CENTER OF LOAD (INS.)
Figure 7.3: Measured Deflection Basins: Fast Loading Rate
164

Figure 7.4: Measured Strain Distributions: Fast Loading Rate

166
Figure 7.5: Load-Deflection Relationships: Fast Loading Rate

167
Figure 7.6: Load-Strain Relationships: Fast Loading Rate

168
predicted using elastic layer theory, when effective layer moduli are
properly determined using the procedures presented in Section 4.4.
It was also shown that the modulus of the asphalt concrete as
determined from correlations with measured asphalt viscosity resulted in
excellent prediction of measured deflections and strains when used in
elastic layer theory. These correlations are given in Appendix A
(Equations A.l and A.2). Constant power viscosities for use in these
relationships were determined from measurements made with the Schweyer
rheometer on asphalt samples recovered from the mixture in the test
pit. Based on these measurements, a viscosity-temperature relationship
was developed (Equation 5.1, Section 5.2.3) from which constant power
viscosity can be calculated at any given temperature.
The results of pavement response prediction for these initial plate
tests were presented in Chapter IV, but will be summarized here for the
sake of continuity. The pavement system in the test pit was modeled as
shown in Figure 7.7 for elastic layer analysis. The BISAR elastic layer
theory computer program was used. An asphalt concrete modulus of
145,000 psi was calculated from Equations 5.1 and A.l for a temperature
of 18.3 C (65 F). A subgrade modulus of 15,420 psi was calculated based
on plate load test results using the procedure outlined in Section
4.4.1.1, with an assumed Poisson's ratio of 0.3. A limerock base
modulus of about 53,000 psi was calculated from plate load test results
using the procedure outlined in Section 4.4.2.1, with an assumed
Poisson's ratio of 0.4. These layer moduli were used in BISAR to
predict the pavement response using the procedure outlined in
Section 4.4.3.1. The rigid plate approximation procedure developed and

Figure 7.
rrrwrj'U./Crf
mw asphalt concrete
4V
u-
&c?' LIMEROCK
BASE
-'>0 0
o v*v.v
0 0 0-0. 9 tf6 *O o 4a 0 0 O o 4
i:.*0' *
V
?oo ei00 0 0
0_. > O A ^ A M
0 O 0 O 0 0
j O
a 0. 4 0
1 9
o 0'O t
0 9*0
o o o (
'. o o o
O ; 0 0 o O 0
5 O *
oe 9 o
) o
o
00
oc
c o o
9 O -y- *:i
O A AO O b O a> '
_ O O w0 w
**- a**
o<
4 o c
4
0
' -O 0
0 0
> 0 O e 0 0
o o 4
j d -
O
i 9 O A *
0O 4
. c 4 *
O 0 J
o
>;
t'
o
a
o a o o
SAND SUBGRADE
63/4"
48
en
eo
PORTLAND CEMENT
i
SEMI-INFINITE
: Test Pit Pavement System as Modeled for Elastic Layer Analysis

170
described in Section 4.4.2 was used. The response predicted by the
program was compared to the measured pavement response in the test pit.
A comparison of measured and predicted deflection basins for a load
of 10,000 lbs. is shown in Figure 4.12. Measured and predicted strains
are compared in Table 4.5. These comparisons show that both deflections
and strains were accurately predicted by the program with the parameters
determined above.
7.2 Initial Plate Load Tests at Slow Loading Rate
Test, series number 2, listed in Table 7.1, were performed a few
days after test series number 1. The tests were performed at the same
loading position. The instrumentation layout used for these tests was
the same as for the fast loading rate, which was shown in Figures 7.2.
7.2.1 Dynamic Load Test Results
The measured deflections and strains for this series of tests are
given in Tables 7.4 and 7.5. The measurements shown are the average of
two readings taken on either side of the load. As for the earlier
tests, the readings were repeatable and nearly identical for gages
located at equal distances, but on opposite sides of the load. Measured
deflection basins and surface strain distributions are shown in
Figure 7.8 and 7.9, respectively.
7.2.2 Evaluation of Results
The deflection basins shown in Figure 7.8 clearly show that
excessive plate indentation was occurring for the slow loading rate. As
the load was increased, the plate deflections seemed to increase

DEFLECTION (x 10~3INS.)
DISTANCE FROM CENTER OF LOAD (INS.)
Figure 7.8: Measured Deflection Basins: Slow Loading Rate

Figure 7.9: Measured Strain Distributions: Slow Loading Rate

173
Table 7.4: Surface Deflections: Slow Loading Rate
Surface Deflection (x 10"3 in.)
Load
Distance
From Center of Load (in.)
(lbs.)
0 in.
8 5/16 in.
12 1/16 in. 18 5/16 in.
28 1/4 in.
10,000
17.6
6.8
4.2 4.35
1.7
7,000
14.0
5.3
2.9

4,000
9.0
5.0
3.4
0.6
1,000
5.4
2.6
1.7 1.0
0.0
Notes:
(a) Readings shown are average of two LVDT's
(b) Average pavement temperature: 21.7 C (71 F)
Table 7.5: Surface Strains: Slow Loading Rate
Surface Strains
(micro-strain)
Load
Distance From
Center of Load (in.)
(lbs.)
8 3/32
in. 12 1/16
in. 17 15/16 in.
28 5/16 in.
10,000
47.6
87.9
67.6
41.0
7,000
39.6
76.0
54.2
32.0
4,000
37.8
51.4
36.8
21.9
1,000
25.9
33.4
21.5
14.0
Notes:
(a) Readings shown are average of two strain gages
(b) Average pavement temperature: 21.7 C (71 F)

174
proportionately, while the deflections next to the plate remained
essentially constant. This effect is more clearly illustrated by the
load-deflection relationships for the different measurement positions.
These are shown in Figure 7.10. The load-deflection relationships for
the positions next to the plate (8- and 12-inches from load center) were
almost vertical for loads greater than 4,000 lbs., while the
relationship for the plate had a much more gradual slope. Apparently,
the high shear stresses at the edge of the rigid plate, which are
theoretically infinite, combined with the increased loading time, caused
the plate to sink into the asphalt concrete layer.
This phenomenon appeared to have two effects on the pavement system
response: 1) it prevented the asphalt concrete layer from acting as a
structural unit in flexure, since yielding was taking place on a
vertical plane at the edge of the plate; and 2) it caused a volume of
asphalt concrete to be transferred from underneath the plate to the
surrounding areas. Load-strain relationships for the different
measurement positions are shown in Figure 7.11. This figure shows that
the strain response at a point two in. from the edge of the plate (eight
in. from load center), was totally different from the strain response
further from the plate. As the load increased, the rate of increase in
strain at this position decreased dramatically as compared to the other
positions, indicating that strains, and thus stresses, were not
increasing in proportion to the applied load. This suggests that the
behavior of the asphalt concrete layer in the vicinity of the plate was
discontinuous in flexure.
Figures 7.12 through 7.15 are comparisons of deflection basins for
the fast and slow loading rates for the different load levels used. The

175
Figure 7.10: Load-Deflection Relationships: Slow Loading Rate

176
Figure 7.11: Load-Strain Relationships: Slow Loading Rate

DISTANCE FROM CENTER OF LOAD (INS.)
Figure 7.12: Deflection Basin Comparison for Fast and Slow Loading Rates: 10,000 lbs.

DEFLECTION (x 10~3INS.)
DISTANCE FROM CENTER OF LOAD (INS.)
0 6 12 18 24 30 36
Figure 7.13: Deflection Basin Comparison for Fast and Slow Loading Rates: 7,000 lbs.

DEFLECTION (x 10~31NS.)
DISTANCE FROM CENTER OF LOAD (INS.)
Figure 7.14: Deflection Basin Comparison for Fast and Slow Loading Rates: 4,000 lbs.

DISTANCE FROM CENTER OF LOAD (INS.)
Figure 7.15: Deflection Basin Comparison for Fast and Slow Loading Rates: 1,000 lbs

181
volume transfer of asphalt concrete referred to earlier is evident in
these comparisons, particularly at the higher load levels. Figures 7.12
and 7.13 show that there is a bulging of material next to the plate for
the slow loading rate. This was probably caused by creep and an
increase in compressive stresses under the plate caused by yielding and
subsequent stress redistribution.
It is interesting to note that the pavement system response was
almost identical for the two loading rates away from the loaded area.
This indicates that the response of the foundation materials was
independent of loading rate and that the observed difference in response
was mainly related to the behavior of the asphalt concrete as influenced
by the rigid plate. Furthermore, plate sinking at the slow loading rate
was even observed at the 1,000-lb. load level, as shown in Figure 7.15
and the highly nonlinear deflection and strain response observed during
the first 1,000 lbs. of load (see Figures 7.10 and 7.11). Therefore,
pavement response for this loading rate cannot be evaluated with the
rigid plate, even at the lower load levels.
In summary, the high shear stresses at the edge of the rigid plate,
combined with the increased loading time for the slow loading rate
resulted in plate indentation and volume transfer from underneath the
plate to the surrounding area. This behavior was very complex and
obviously could not be predicted with an elastic layer model or any
other available model known to the author. In any case, it would be of
little value to predict this response since it was totally
uncharacteristic of the response of an actual wheel load. Therefore,
further analysis of these data viere deemed unnecessary. The major

182
outcome of these tests was to show that the rigid plate loading system
is inadequate for evaluating pavement response at slower loading rates.
7.3 Additional Plate Load Tests at Fast Loading Rate
Additional plate load tests were performed at two loading positions
to further evaluate pavement response at ambient temperatures and the
analytical procedures developed on the basis of the initial plate test
results. These tests were performed at 20.6 C (69 F) and 25.6 C (78 F)
and are listed in Table 7.1 as test series three and four, respec
tively. Tests were not performed to evaluate pavement response at
slower loading rates because results of earlier tests indicated that
rigid plate loading was inadequate for this purpose (see Section 7.2).
Figure 7.16 shows the loading positions and instrumentation layout
used for each test series. Test series three (Table 7.1) was performed
at the center of the test pit, which is the same position where the
initial plate tests were performed. Tests series four was performed
three feet directly south of the test pit center, which was previously
unloaded. The position of the LVDT's and strain gages during testing
are also shown in Figure 7.16. A total of 10 deflections and 10 strain
measurements were taken at each position for all tests performed.
7.3.1 Dynamic Load Test Results
Tables 7.6 through 7.9 give the measured deflections and strains
for test series three and four (Table 7.1). The measurements shown are
the average of two readings taken on either side of the load. As in
earlier tests, the measurements were repeatable and nearly identical for
gages located at equal distances but on opposite sides of the load.

Figure 7.16: Location of Plate Loading Positions and Strain and Deflection Measurements

184
Table 7.6: Measured Surface Deflections at 20.6 C (69 F):
Center Plate Loading Position
Surface Deflection ^ (x 10"
3 in.)
Load
Distance
From Load Center (in.)
(lbs.)
0 in.
8 5/16 in.
12 1/16 in.
18 5/16 in.
28 1/4 in.
10,000
14.6
8.5
6.0
4.0
1.7
7,000
11.7
6.3
4.7
3.0
1.2
4,000
7.4
4.1
2.9
2.0
0.9
1,000
1.9
1.0
0.5

0
Notes:
(a) Readings shown are average of two LVDT's
(b) Loading rate: 0.1 seconds load time/0.4 seconds unload time
Table 7.7: Measured Surface Strains at 20.6 C (69 F):
Center Plate Loading Position
Surface
Strains^
(micro-strain)
Load
Distance From Load Center (in.)
(lbs.)
8 3/32 in.
12 1/16 in.
17 15/16 in.
28 5/16 in.
10,000
0
70
67
42
7,000
0
58
53
32
4,000
0
37
33
20
1,000
0
11
9
6
Notes:
(a) Readings shown are average of two strain gages
(b) Loading Rate: 0.1 seconds loading time/0.4 seconds unload time

185
Table 7.8: Measured Surface Deflection at 25.6 C (78 F):
South Plate Loading Position
Surface Deflection (x
10"3 in.)
Load
Distance
From Load
Center (in.)
(lbs.)
0
8
12
18
28
10,000
15.2
7.0
4.7
2.9
0.8
7,000
11.5
5.5
3.5
2.0
0.6
4,000
7.6
3.4
2.3
1.4
0.5
1,000
2.4
1.0
0.7
0.3
0
Notes:
(a) Readings shown are average of two LVDT's
(b) Loading Rate: 0.1 seconds load time/0.4 seconds unload time
Table 7.9: Measured Strains at 25.6 C (78 F):
South Plate Loading Position
Strains^3^) (micro-strain)
Load Distance From Load Center (in.)
(lbs.)
10,000
60.0
75.0
45.0
51.0
17.5
17.5
7,000
46.5
65.0
35.5
40.0
13.0

4,000
32.5
41.0
29.0
30.0
11.0
11.0
1,000
10.0
15.0
10.0
8.5
3.0
3.0
Motes:
(a) Readings shown are average of two strain gages, except at zero in.
from load center.
(b) Loading Rate: 0.1 seconds load time/0.4 seconds unload time.
(c) Readings shown are from two strain gages located at the pavement/
limerock interface directly underneath the center of the load. The
gages were not located at exactly the same location, so the larger
readings correspond to the gage that is closest to the center of
load.

186
The measured deflection basins and surface strain distributions for
load tests oerformed at 20.6 C (69 F) are shown in Figures 7.17 and
7.18, respectively. Similar relationships are shown in Figures 7.19 and
7.20 for the tests performed at 25.6 C (78 F). In all cases, plate
deflections are shown as perfectly flat and the deflection basin is
defined by four additional measurements.
7.3.2 Elastic Layer Simulation and Evaluation of Results
While the deflection basins at 20.6 C (69 F, Figure 7.17) seemed
reasonable and compatible with deflections measured earlier at 18.3 C
(65 F, Figure 7.3), the basins at 25.6 C (78 F, Figure 7.18) indicate
that excessive plate indentation occurred at this higher temperature.
Similarly, the strain distributions at 25.6 C (78 F, Figure 7.20) were
incompatible with the strains measured at lower temperatures (Figures
7.4 and 7.19). Whereas the strains should be higher for the pavement at
higher temperature (since the asphalt stiffness is lower), the measured
strains were significantly lower. Apparently, plate sinking resulted in
discontinuous behavior of the asphalt concrete layer, where stresses
were not effectively transmitted away from the load.
A better evaluation of the results can be made by comparing the
deflection basins and strain distributions for these plate load tests at
20.6 and 25.6 C (69 and 78 F), and the initial plate load tests at
18.3 C (65 F), were all performed at the fast loading rate. The
deflection basin comparisons for these tests, at different load levels,
are shown in Figures 7.21 through 7.24. Similar comparisons of the
strain distributions in Figures 7.25 through 7.28.

DEFLECTION ( x 103 ins)
DISTANCE FROM LOAD CENTER (ins)
0 6 12 18 24 30 36
00
Figure 7.17: Measured Deflection Basins at 20.6 C (69 F)

DISTANCE FROM LOAD CENTER (ins)
Figure 7.18: Measured Deflection Basins at 25.6 C (78 F)

Figure 7.19: Measured Strain Distributions at 20.6 C (69 F)

Figure 7.20: Measured Strain Distributions at 25.6 C (69 F)

DISTANCE FROM LOAD CENTER (ins)
0
Figure 7.21: Deflection Basin Comparison at 10,000 lbs.

DEFLECTION (E-3 ins)
DISTANCE FROM LOAD CENTER (ins)
Figure 7.22: Deflection Basin Comparison at 7,000 lbs.

DEFLECTION (E-3 ins)
DISTANCE FROM LOAD CENTER (ins)
Figure 7.23: Deflection Basin Comparison at 4,000 lbs.

DISTANCE FROM LOAD CENTER (ins)
0 6 12 18 24 30 36
Figure 7.24: Deflection Basin Comparison at 1,000 lbs.

SURFACE STRAINS (jjie) TENSION
Figure 7.25: Strain Distribution Comparison at 10,000 lbs.

Figure 7.26: Strain Distribution Comparison at 7,000 lbs.

Figure 7.27: Strain Distribution Comparison at 4,000 lbs.

DISTANCE FROM LOAD CENTER (ins)
Figure 7.28: Strain Distribution Comparison at 1,000 lbs.

199
The deflection basin comparisons show that at all load levels, the
deflection basin at 20.6 C (69 F) was very similar in shape to the one
at 18.3 C (65 F), but was shifted slightly upward (i.e., slightly lower
deflections were measured at the higher temperature). Comparisons of
the strain distributions at the same two temperatures show that the
measured strains were significantly higher at 20.6 C (69 F) than at
18.3 C (65 F) for all load levels. Assuming that foundation conditions
were the same for both tests, the observed difference in strains was as
expected, while the observed deflection difference was opposite of what
was expected. However, a period of one year elapsed between the two
tests and the limerock modulus may have changed with time or moisture
changes. Furthermore, this position was loaded extensively at fast and
slow loading rates prior to testing at 20.6 C (69 F) which may have
densified one or all three of the pavement layers.
BISAR was used to predict the measured deflections and strains at
20.6 C (69 F) and thereby deduce changes in the individual pavement
layer properties which would explain the observed differences in
response between the initial tests at 18.3 C (65 F) and the tests
performed one year later at 20.6 C (69 F). This analysis would also
serve to evaluate the asphalt concrete dynamic modulus prediction
equations (Appendix A) at a higher temperature.
The procedure outlined in Section A.4.3.1 was used along with the
rigid plate approximation procedure developed and described in Section
4.4.2 to predict pavement response. The pavement system was again
modeled as shown in Figure 7.7. An asphalt concrete modulus of
124,400 psi was calculated for use in the program from Equations 5.1 and
A.l for a temperture of 20.6 C (69 F). Different values of limerock

200
base and sand subgrade moduli were used in the program to obtain the
values that resulted in the best correspondence between measured and
predicted response at 10,000 lbs. A sand subgrade modulus of 15,420 psi
(y = 0.3) and a limerock base modulus of 53,000 psi (y = 0.4), which
were determined from the original plate load tests performed on these
materials, were used in the initial analysis. As it turned out, these
values resulted in the best response prediction of the different
combination of values attempted.
Comparisons of measured and predicted deflections and strains using
these values are shown in Figure 7.29 and 7.30. Although the maximum
deflection was overpredicted by approximately 11 percent, the measured
strains were predicted almost exactly. When higher moduli were used for
the limerock and the sand subgrade to get a better deflection
prediction, the predicted strains were grossly in error. This is
illustrated in Figures 7.31 and 7.32, which are comparisons of measured
and predicted deflections and strains using a limerock modulus
(75,000 psi) that was tuned so that the plate deflections would match
exactly. Although better correspondence was obtained between measured
and predicted deflections (Figure 7.31), the predicted strains are not
even close to the measured (Figure 7.32).
These results seem to indicate that the properties of the sand and
the limerock were unchanged and that the observed change in response was
mainly related to the behavior of the asphalt concrete. However, the
asphalt concrete modulus as predicted from asphalt viscosities measured
from samples recovered when the pavement was first placed (Equation
5.1), resulted in excellent strain predictions. This was the same
equation used to predict the asphalt concrete modulus which resulted in

DEFLECTION (E-3 ins)
DISTANCE FROM LOAD CENTER (ins)
Figure 7.29: Measured vs. Predicted Deflections at 10,000 lbs.: E2 = 53,000 psi, 20.6 C (69 F)

100
80
60
40
20
x Measured
o Predicted: EAC= 124,400 psi v=0.35
Elr= 53,000 psi v = 0.40
Ess= 15,420 psi v = 0.30
rs>
o
ro
30
: Measured vs. Predicted Strains at 10,000 lbs.: E2 = 53,000 psi, 20.6 C (69 F)

DISTANCE FROM LOAD CENTER (ins)
Figure 7.31: Measured vs. Predicted Deflections at 10,000 lbs.: E¡> = 75,000 psi, 20.6 C (69 F)

SURFACE STRAINS (p.e) TENSION
Figure 7.32: Measured vs. Predicted Strains at 10,000 lbs.: E2 = 75,000 psi, 20.6 C (69 F)

205
good deflection and strain prediction for the earlier tests performed at
18.3 C (65 F). Therefore, it seems like the flexural properties of the
asphalt concrete have not changed.
One logical explanation for the discrepancy between measured and
predicted deflections at 20.6 C (69 F) is that the asphalt concrete may
have developed anistropic properties as a result of previous loading at
this position. The initial tests performed at this position,
particularly at the slow loading rate, may have compressed the asphalt
concrete layer and decreased its vertical compressibility under the
plate. A 12-inch diameter circular depression was clearly visible after
these initial plate tests were performed. Therefore, the asphalt
concrete responded more stiffly vertically than radially. Since BISAR
cannot account for anisotropic properties, it overpredicted the measured
deflections, even though it predicted the measured radial strains almost
exactly.
These results indicate that more reliance should be placed on
measured strains than on measured deflections for the evaluation of
pavement response. Errors in interpreting deflection measurements can
result from uncertain foundation conditions, anisotropy, or the volume
transfer phenomenon observed earlier for the slow loading rate.
Although these may have some effect on measured strains, the effect is
less. In addition, strains are a direct measure of the response of the
material, making their interpretation much less dependent on the model
used to predict response. Thus a more direct evaluation of response can
be made. It is interesting to note that most researchers rely mainly on
deflection data alone to evaluate pavement response, since deflections
are much simpler to obtain.

206
However, the most important finding from this analysis was that the
difference in measured strains between tests at 18.3 C (65 F) and tests
at 20.6 C (69 F), was predicted almost exactly by changing only the
asphalt concrete modulus as determined from asphalt viscosity
measurements. This provides additional verification of the relation
ships for dynamic modulus of asphalt concrete developed Ruth et al. (43)
and presented in Appendix A.
As mentioned earlier, the measured pavement response at 25.6 C
(78 F) indicated that excessive plate indentation occurred, which
resulted in discontinuous behavior of the asphalt concrete layer. The
deflection basin and strain distribution comparisons shown in Figure
7.21 through 7.24 and 7.25 through 7.28, respectively, show that the
response measured at the higher temperature was totally incompatible
with the response measured at 18.3 and 20.6 C (65 and 69 F). The shape
of the deflection basin at 25.6 C (78 F) is very different from the
shapes of the deflection basins at lower temperatures, and it is obvious
that this difference cannot be accounted for by the change in asphalt
concrete modulus caused by the higher temperature. The very sharp
change in the slope of the 25.6 C (78 F) deflection basins immediately
next to the plate indicates that the plate was sinking into the asphalt
concrete layer (see Figures 7.21 through 7.24). Apparently, this plate
sinking caused the asphalt concrete layer to behave discontinuously in
flexure, as evidenced by the relatively low strains measured at this
temperature. As shown in Figures 7.25 through 7.28, the strains
measured at 25.6 C (78 F) were significantly lower than those measured
at 20.6 C (69 F) at all load levels, which was contrary to what was

207
expected. This behavior indicates that stresses were not effectively
transmitted away from the plate by the asphalt concrete layer.
A BISAR comDUter run was made with an asphalt concrete modulus of
90,000 psi, determined from Equation 5.1 and A.l for a temperature of
25.6 C (78 F) and the sand and limerock moduli found to give the best
prediction of measured response in previous analyses (15,420 psi and
53,000 psi, respectively). As expected, the measured strains were
grossly overpredicted by the program, as was the measured deflection
basin. The shapes of the predicted and measured deflection basins were
not even close. It was obvious that the measured response could not be
predicted using elastic layer theory, so additional attempts were not
made.
A major finding from these tests was to set an upper temperature
limit for pavement response evaluation with the rigid plate.
Apparently, for temperatures greater than about 21 C (70 F), excessive
plate identation can occur as a result of the high shear stresses at the
edge of the rigid plate, even at the fast loading rate. Therefore, the
plate should not be used at temperatures higher than 21 C (70 F).
Furthermore, rigid plate loading is not recommended for pavement
evaluation at any temperature because the analytical procedures required
to model rigid plate loading are extremely tedious and time consuming.
It should be noted that strain measurements were obtained at the
bottom of the asphalt concrete for the test performed at 25.6 C
(78 F). Unfortunately, the measurements obtained from these gages could
not be evaluated because of the uncharacteristic response that was
observed. However, the measurements seemed very reasonable and the
gages seemed to function well.

?.08
Load-deflection and load-strain relationships were plotted for the
pavement response measurements taken at 20.6 C (69 F). These plots are
shown for the different measurement positions, in Figures 7.33 and 7.34,
respectively. Both the deflection and strain relationships indicate
that the pavement response was nonlinear in a stress-stiffening way.
This is consistent with the response observed for the initial tests
performed at 18.3 C (65 F) and with the nonlinear response observed on
the plate tests performed on the limerock base (see Chapter V).
BISAR was used to predict the pavement response at 4,000 lbs. to
see if the observed nonlinearity could be accounted for by changing only
the limerock base modulus. The procedure outlined in Section 4.4.3.1
was used along with the rigid plate approximation procedure to predict
response using different limerock moduli until the best correspondence
was achieved between measured and predicted response at 4,000 lbs. The
best prediction was obtained with the following parameters.
Asphalt Concrete:
E =
124,400
psi
u = 0.35
Limerock Base:
E =
40,000
psi
v = 0.40
Sand Subgrade:
E =
15,420
psi
y = 0.30
Note that the modulus of the sand and the asphalt concrete are the same
as those used to predict response at the 10,000-lb. load level. Only
the limerock modulus was changed.
Comparisons of measured and predicted deflections and strains are
shov/n in Figures 7.35 and 7.36, respectively. As shown in these
figures, these parameters resulted in excellent predictions of both
deflections and strains, indicating that the observed nonlinearity was
mainly caused by the limerock base. Plate load tests performed on the
limerock also indicated that this material responded non! inearly in a

CQ
C
-s
fO
co
co
n>
o>
CO
c
-s
ro
Cl
O
O
a.
i
a
a>
o
o
c+
o
3
O
a
(/>
rr
*o
V)
O
rf
ro
o
oi
D
m
n
r~
m
O
H
O
z
m
i
w
5
c/)
to
o
28" From Load
Center
18" From Load
Center
12" From Load
Center
8" From Load
Center
Plate

210
Figure 7.34: Measured Load-Strain Relationships at 20.6 C (69 F)

DEFLECTION (E-3 ins)
DISTANCE FROM LOAD CENTER (ins)
6 12 18 24 30 36
Figure 7.35: Measured vs. Predicted Deflections at 4,000 lbs.: E2 = 40,000 psi, 20.6 C (69 F)

SURFACE STRAINS (jjie) TENSION
Figure 7.36: Measured vs. Predicted Strains at 4,000 lbs.: E2 = 40,000 psi, 20.6 C (69 F)

213
stress-stiffening way (see Section 5.4.2). Furthermore, an initial
secant modulus of 37,550 was calculated from the 12-inch diameter plate
tests results, and a value of 25,550 psi was determined with the 16-inch
plate (see Table 5.10). The value obtained from the 12-inch plate
results is within about six percent of the value determined by tuning to
the measured pavement response (37,550 psi vs. 40,000 psi).
Therefore, it seems like good prediction of pavement response can
be obtained at lower stress levels by using initial tangent modulus
values calculated from 12-inch plate test results, using the procedures
and data interpretation methods presented in Chapters IV and V (see
Sections 4.4.2.1 and 5.4.2). In addition, the asphalt concrete modulus
predicted from asphalt viscosity correlations gives accurate response
prediction over the entire range of stresses applied in the test pit.
7.4 Summary
The results presented in this chapter indicated that load-induced
deflections and strains can be accurately predicted using elastic layer
theory when effective layer moduli are properly determined. Good
predictions of load response were obtained using asphalt concrete moduli
determined from previously established correlations with measured
asphalt viscosity, and effective moduli for the sand and limerock
determined from plate load tests. It was also shown that the effects of
the test pit floor must be accounted for when determining the sand and
limerock moduli, and the effect of both the test pit floor and walls
must be accounted for when predicting the load response of the asphalt
concrete pavement. The analytical procedures developed and described in
Chapter IV served this purpose well.

214
A thorough evaluation of the measurements indicated that measured
strains are much better than deflection for evaluating the load response
of the pavement. Strains provide a direct measurement of the behavior
of the asphalt concrete, while deflections give an indication of the
response of the entire pavement system. Therefore, proper evaluation of
deflection data is much more dependent on the analytical models used.
Furthermore surface deflections are affected more by foundation
conditions at depth.
The nonlinearity observed for plate load tests performed on the
1imerock was confirmed by the measured load response of the pavement.
However, it was shown that accurate response prediction could be
obtained using linear analysis with an effective 1 imerock modulus
(secant modulus) determined from plate deflections at higher stress
levels. It was also shown that pavement response at lower load levels
could be accurately predicted by using a 1imerock secant modulus
determined at lower stress levels.
The plate test results also indicated that rigid plate loading is
not a good tool for the evaluation of asphalt concrete pavements. At
temperatures greater than about 25 C (77 F) the high shear stresses at
the edge of the plate resulted in excessive plate indentation even at
the fast loading rate. For a loading time of one second, plate
indentation and volume transfer were observed at 21.7 C (71 F).
Although there did not appear to be plate indentation at the fast
loading rate at temperatures below 21.1 C (70 F), the analysis
procedures required to model the plate proved to be extremely tedious
and time consuming. There are no known models that give accurate
prediction of a rigid circular area on a layered system. A method to do

215
this was developed using an elastic layer analysis program, but was
found to be very tedious. Therefore, if at all possible, rigid plate
loading systems should not be used to evaluate asphalt concrete
pavements. After these plate tests were completed, a flexible dual
wheel loading system was designed and constructed for use in the test
pit.

CHAPTER VIII
LOW-TEMPERATURE
PAVEMENT RESPONSE
8.1 Preliminary Tests With the Rigid Plate
Once the cooling system was installed and operational, the test pit
pavement was subjected to several cooling cycles to evaluate the
performance of the cooling unit and the instrumentation system at low
temperatures. In addition, dynamic load tests were performed in an
attempt to evaluate pavement system response at low temperatures. The
rigid plate loading system was used because the dual wheels were not yet
available. These initial tests led to several findings concerning the
cooling unit capability, problems and deficiencies with the
instrumentation and loading systems at cold temperatures, and a low-
temperature limit for pavement evaluation.
The cooling system capability was determined by allowing the
cooling unit to run until the pavement temperature would not go any
lower. A temperature of -6.7 C (20.0 F) was achieved at the bottom of
the 4 1/8-inch asphalt concrete layer with a pavement surface temper
ature of -17.8 C (0.0 F). Measured cooling rates are presented in
conjunction with pavement response measurements in a later section.
During the first cooling cycle, an attempt was made to measure the
vertical movement at different points on the pavement surface. The
LVDT's were individually insulated with a loosely-fitting, cylindrically
shaped piece of 1/2-inch rubber foam insulation and the space between
the LVDT and the rubber foam was filled with expanding foam insulation.
216

217
The LVDT's were positioned and the test pit cover was sealed for
cooling. Deflections were monitored during cooling, but the LVDT's
began to drift and eventually went out of range. It was unclear whether
this drift was caused by the contraction of the LVDT support system,
movement of the pavement, or a reaction of the LVDT's to temperature
changes. It was probably a combination of all three.
An attempt to perform load tests at -6.7 C (20 F) indicated that
several LVDT's and the load cell malfunctioned as a result of the cold
temperatures. When the temperature of the pavement stabilized, one end
of the tests pit cover was removed and the LVDT's were adjusted manually
until each was brought back into range. However, several LVDT's were
not functioning and the load cell was not functioning properly. The
load cell was indicating a load of about 1000 lbs. when no load was
applied, and measurements of applied loads did not correspond to
approximate measurements made with a hydraulic gage.
Apparently, insulation alone was not enough to prevent these
instruments from malfunctioning. Therefore, the LVDT's and the load
cell were stripped of all insulation and a heating system was installed
to maintain these at a constant temperature of about 25 C (77 F, see
Section 3.4.1). Thus, the effect of temperature on their performance
was eliminated.
With the heating system installed and the instruments in testing
position, the test pit cover was again sealed for cooling. Deflections
were monitored during cooling, but excessive drift (> 1/4-inch) was
again observed and the LVDT's eventually went out of range, even though
these were kept at constant temperature. Although part of the movement
could be attributed to the pavement, it seemed apparent that these large

218
deflections were mainly a result of the contraction of the LVOT support
system. In any case, there was no way to tell what part of the measured
deflection was caused by the movement of the pavement. Therefore, it
was clear that the movement of the pavement could not be monitored with
an LVDT system located inside of the test pit cover. However, the
heating system served its purpose well. When the LVDT's were adjusted
manually, all were completely operational.
Dynamic load tests were performed during this cooling cycle at
temperatures of -6.7, 0.0, 3.3, 6.7, 11.7, and 15.6 C (20, 32, 38, 44,
53, and 60 F). At each temperature, tests were performed at 1, 4, 7,
and 10 kips, and deflections and strains were measured at different
points on the pavement surface. Data from these tests were reduced and
plotted, and limited analyses were performed to evaluate pavement
response. However, because of noise in the strain measurements and
problems encountered with the loading system, these data were of limited
value and are not presented.
Excessive noise was observed in the measured dynamic strains,
making it difficult to evaluate these measurements. The noise was
traced to improperly grounded cables and the fact that a 1/4-bridge
strain gage arrangement was used for dynamic load tests. Although
shielded cable was used in all cases, the shield was improperly grounded
at a point where a connection was made to extend the cable length, it
was also necessary to use a 1/2-bridge arrangement to help eliminate the
noise, since with this arrangement, noise from a common source cancels
itself when introduced to the two active arms of the bridge. A 1/4-
bridge arrangement was used initially because it resulted in higher

219
resolution and it was not necessary to compensate for temperature
effects during dynamic loading.
The second problem encountered concerned the transverse movement of
the loading ram on the reaction beam. Apparently, repeated applications
of 10,000-lb. loads, caused the loading ram to move along the reaction
beam and jammed the swivel head against the LVDT's. The swivel head
partially jammed several LVDT's and may have affected all the deflection
readings. Unfortunately, this problem was not discovered until all
tests were complete, since the readings appeared normal during
testing. However, when the data were reduced, it was obvious which
LVDT's were jammed. This problem was solved by installing four heavy-
duty C-clamps which were designed to prevent the loading ram from moving
along the reaction beam.
Although these problems rendered the measurements unsuitable for
detailed analysis, several observations were made concerning the
pavement's response. Deflections and strains measured at -6.7 C (20 F)
were extremely low compared to measurements made above freezing. The
measurements were also low compared to predictions made with elastic
layer analysis using reasonable modulus values for the pavement
layers. It seemed clear that the moisture in the limerock base and
possibly part of the sand subgrade froze, and significantly increased
the stiffness of pavement system. These results indicated that future
tests to evaluate pavement response should be performed at temperatures
above freezing, since a partially frozen foundation would introduce
another unknown into the analysis.
Deflection and strain measurements at temperatures from 0.0 C
(32 F) to 15.6 C (60 F) were incomplete and noise in the strain

220
measurements made these readings questionable. However, the readings
that were obtained appeared to be reasonable when compared to
predictions made using elastic layer analysis with reasonable modulus
values for the pavement layers. Additional analyses did not seem
productive, especially considering that little reliance could be placed
on their outcome. Furthermore, the limited analyses that were
performed, once again emphasized that the analysis of rigid plate
loading was extremely tedious and time consuming. For this reason
alone, additional tests were not performed with the rigid plate.
Even though data obtained from these preliminary tests did not
directly serve to evaluate asphalt concrete pavement response at low
temneratures, they did help in evaluating the capabilities, limitations,
and problems with the cooling, loading, and instrumentation systems.
Findings from the tests led to modifications to these systems that were
necessary for future testing. It was also found that movement of the
pavement during cooling could not be monitored with a system of LVDT's
located inside of the test pit cover. A system of LVDT's should be
located in the subgrade for future test pit installations if these
movements are to be monitored. Finally, it was determined that load
tests for pavement evaluation should not be performed with the rigid
plate nor at temperatures below 0.0 C (32 F).
After these preliminary tests were completed, the test pit pavement
was prepared for tests with the flexible dual wheels. The test pit
cover was dismantled and all instrumentation was disconnected.
Preparation of the pavement for tests with the dual wheels, as well as
test results are presented in subsequent sections of this chapter.

221
8.2 Reinstrumentation for Tests With the Dual Wheels
The test pit pavement was completely reinstrumented for low-
temperature pavement response tests with the dual wheel loading
system. An LVDT support system was designed and constructed to measure
ten deflections along the longitudinal and transverse axes of the dual
wheels (see Section 3.4.1). The support system could be moved to
different test locations, and the LVDT's could be positioned to measure
deflections at any specified distance from the load.
Twenty-six strain gages were installed as shown in Figure 8.1, to
measure thermally- and load-induced strains in the pavement. The gages
were positioned such that the strain response of the pavement was
adequately defined at three test positions located at three, five, and
seven feet from the south wall of the test pit. Since only ten strain
gages could be monitored at one time, cables were permanently attached
to all strain gages so that any particular set of gages could be
connected for measurement by simply switching the cables on a set of
terminal posts. As shown in Figure 8.1, the strain gages were divided
into pairs, and one four-conductor cable was soldered to each pair and
given a cable number. Thus, each strain gage was identified by a cable
number and the color coded wires in the cable.
8.3 Results of Tests With the Dual Wheel Loading System
8.3.1 Introduction
Dynamic load tests and creep tests were performed to evaluate the
elastic and inelastic response characteristics of the pavement at
temperatures of 0.0, 6.7, and 13.3 C (32, 44, and 56 F). In addition,
thermally-induced strains were monitored during cooling at different

STRAIN GAGE NUMBER
CABLE NUMBER
VIII
16x
xx xx
U U.
I II
VII
2 Ox
x24
15x
X
19x
x23
XII
X X
X
X
XXX
X
26
X
5J5
_ 8^
9 10 11 12
.25 x .
III
Tv
V
VI
XlT
13x
17x
x21
14x
IX
XI
18x
x22
Scale:
12"
NOTE: DIMENSIONS OMMITED FOR CLARITY.
A. TWO GAGES LOCATED AT PAVEMENT/LIME ROCK INTERFACE:
ONE TRANSVERSE AND ONE LONGITUDINAL
Figure 8.1: Layout of Strain Gages and Cables in the Test Pit

223
points in the pavement to evaluate its thermal response characteris
tics. Detailed procedures for these tests are presented in Chapter
VI. These tests were performed at each of the three tests positions
shown in Figure 8.2. Load tests could not be performed north of test
position 3 because the cooling unit evaporator was located at the north
end of the test pit.
8.3.2 Pavement Response During Cooling
At each of the three test positions, temperatures and strains were
measured at different points in the pavement as it was cooled from room
temperature to 0.0 C (32 F). Therefore, the thermal strain response of
the pavement was monitored during three cooling cycles. Since only ten
strain gages could be monitored at one time, a different set of strain
gages was monitored during each cooling cycle. Figures 8.3, 8.4, and
8.5 show the positions of the ten strain gages that were monitored
during cooling cycles at test positions 1, 2, and 3, respectively.
These figures also show the positions of the 16 thermocouples used to
monitor temperatures.
The temperatures and strains measured during cooling at each of the
three test positions are given in Appendices B and C, respectively. The
time intervals necessary to adequately define the thermal response of
the pavement were determined during initial cooling trials which were
performed to evaluate the cooling unit. The time intervals between
measurements were increased as the rate of temperature drop decreased.
Measured cooling curves (temperature vs. time) at test positions 1,
2, and 3 are shown in Figures 8.6, 8.7, and 8.8, resnectively. In each
figure, curves are shown for the temperature of the top and bottom of

2'0" 2l0"
( ) ( ) ( 1
C ) CZ3 CUD
POSITION I POSITION 2 POSITION 3
5'.0"
Figure 8.2: Location of Test Positions in the Test Pit

09
x Surface Thermocouples
O- Strain Gages
X
3
X
7
13-16(a)
? 6
5 4
2C$bi O
6
x x
5 8
07
X
08 2 &
9-12^
N
X
6
12
I 1
(a) FOUR THERMOCOUPLES LOCATED AT DEPTHS OF 1 3/8,2 1/8, 3 1/8, & 4 1/8 INS.
(b) LOCATED AT BOTTOM OF PAVEMENT: LONGITUDINAL DIRECTION
(c) LOCATED AT BOTTOM OF PAVEMENT: TRANSVERSE DIRECTION
Figure 8.3: Thermocouple and Strain Gage Location During Cooling: Test Position 1

y
x Surface Thermocouples
O- Strain Gages
x 08
X
3
06
7&
13-16(a)
4
2 0 1
3 5
x O
0x0 O
Oxo x
1
4
5 8
07
N
X
X
2 &
6
g.12(a)
09
12
1 1
(a) FOUR THERMOCOUPLES LOCATED AT DEPTHS OF 1 3/8, 2 1/8, 3 1/8 & 4 1/8 INS.
Figure 8.4: Thermocouple and Strain Gage Location During Cooling: Test Position 2

x Surface Thermocouples
O* Strain Gages
*
5
x O
8
1
x
4
O
3
x O
4
1
O
6
O
0
o
2
O
x
2&
9-12(a>
7
O
x
6
9
O
- N
12
i 1
(a) FOUR THERMOCOUPLES LOCATED AT DEPTHS OF 1 3/8, 2 1/8, 3 1/8 & 4 1/8 INS.
ro
ro
-'~l
Figure 8.5: Thermocouple and Strain Gage Location During Cooling: Test Position 3

30
25
20
15
10
5
0
-5
10
X Average Surface Temperature
O Average Temperature at Bottom
of Asphalt Concrete Layer
Average Temperature of Asphalt
Concrete Layer
1 i I I i i I I i I I I l I w
2 6 10 14 18 22 26
TIME (HOURS)
ro
ro
CO
Figure 8.6: Measured Cooling Curves: Test Position 1

TEMPERATURE (C)
Figure 8.7: Measured Cooling Curves: Test Position 2

Figure 8.8: Measured Cooling Curves: Test Position 3

231
the asphalt concrete layer, and for the average temperature of the
layer. These figures show that the rate of temperature drop was
greatest during the first two hours of cooling and decreased sharply
thereafter. The thermal gradient between the top and bottom of the
asphalt concrete layer rose sharply within the first two hours of
cooling, then remained fairly constant. This is better illustrated by
the plots of temperature gradient as a function of time shown in Figure
8.9. For each cooling cycle, the temperature gradient reached a maximum
after about two hours of cooling, then gradually decreased with time.
The temperature gradient dropped sharply when the cooling unit was shut
off. At that time, the top of the pavement quickly warmed up and the
bottom remained at constant temperature (see Figures 8.6, 8.7, and 8.8).
Plots of measured longitudinal thermal strains versus temperature
during cooling are shown in Figures 8.10, 8.11, and 8.12 for cooling
cycles at test positions 1, 2, and 3, respectively. Similar plots for
transverse strains measured during cooling at test positions 1, 2, and 3
are shown in Figures 8.13, 8.14, and 8.15, respectively. In all cases,
the strains were plotted as a function of the temperature measured at or
near the respective strain gage. These plots illustrate several
characteristics of the thermal response of the pavement during cooling.
Higher contraction strains were measured for those gages located
near the edge of the pavement in both the longitudinal and transverse
directions (see gage 5, Figure 8.10; gage 5, Figure 8.12; gage 9, Figure
8.13; gage 9, Figure 8.14; and gage 9, Figure 8.15). This was expected,
since the frictional resistance provided by the underlying base
decreases as one gets closer to the edge of the pavement.

Figure 8.9: Change in Temperature Gradient During Cooling

MEASURED STRAINS DURING COOLING (E-6 IN./IN.)
233
>k
300- -
200
100
z
o
tn
Z
UJ
t-
TEMPERATURE (C)
X M
O 2
D 3
0 4
O 5
O 6
(1) See Figure 8.3 for
Gage Location
-800
Measured Longitudinal Strains vs. Temperature:
Test Position 1
Figure 8.10:

234
Figure 8.11: Measured Longitudinal Strains vs. Temperature:
Test Position 2

MEASURED STRAINS DURING COOLING (E-6 IN./IN.)
235
Figure 8.12: Measured Longitudinal Strains vs. Temperature:
Test Position 3

.MEASURED STRAINS DURING COOLING ,(E-6 IN./IN.)
236
Figure 3.13: Measured Transverse Strains vs. Temperature:
Test Position 1

MEASURED STRAINS DURING COOLING (E-6 IN./IN.)
I t i i i i '
237
Figure 8.14: Measured Transverse Strains vs. Temperature:
Test Position 2

MEASURED STRAINS DURING COOLING (E-6 IN./IN.)
t t i i l
238
Figure 8.15: Measured Transverse Strains vs. Temperature:
Test Position 3

239
A close examination of the strain-temperature plots show that there
were three distinct changes in the thermal response characteristics of
the pavement during cooling. Initially, the rate of contraction with
respect to temperature was very low for all surface gages. The reason
for this was that the thermal stresses developed in the asphalt concrete
were not yet high enough to overcome the frictional resistance between
the asphalt concrete layer and the underlying base. This was mainly
because the asphalt viscosity near room temperature is relatively low,
so that the asphalt concrete has a high capacity to relax stresses
induced thermally or otherwise.
An increase in contraction rate occurred for all surface gages
after about two hours of cooling. This increase in rate appears as a
distinct change in slope in the strain-temperature plots (Figures 8.10
to 8.15). This break was observed in both the longitudinal and
transverse directions. It is interesting to note that the break
occurred at the same time, but at different temperatures, for gages
located at different positions on the pavement. Therefore, it seems
like the change in thermal response was related more to the overall
behavior of the pavement than to the response characteristics of the
asphalt concrete itself. Apparently, after about two hours of cooling,
the thermal stresses induced in the pavement were enough to overcome
frictional resistance and thus initiate a significant amount of movement
or contraction in the pavement. Measured cooling curves (Figures 8.6 to
8.8) and plots of thermal gradient with time (Figure 8.9) indicated that
this change occurred when the average pavement temperature was about
eight to ten degrees C (46.4 to 50.0 F) and when the temperature
gradient reached a maximum.

240
The second change in the thermal response of the pavement occurred
after several hours of continuous cooling. This change appears as the
second break in the strain-temperature plots, and is more evident for
some gages than for others. As for the first change in response, the
break occurred at the same time, but at different temperatures, for
gages located at different positions on the pavement. Also, the change
was different for different gages. Some gages stopped contracting (e.g.
see gage 4, Figure 8.10; gage 0, Figure 8.12; and most transverse gages
shown in Figures 8.13 to 8.15), some gages showed a sharp increase in
compression (e.g. see gage 6, Figure 8.10; gage 0, Figure 8.11; and gage
1, Figure 8.12), and some gages showed a sharp decrease in compression
(e.g. see gages 3 and 5, Figure 8.11; gage 2, Figure 8.12; and gage 6,
Figure 8.15). It should be emphasized that these changes were occurring
while the pavement continued to cool.
The reason for this change in response is not totally clear, but
the fact that the response of all gages changed at the same time
indicates that it was related to some overall change in the response of
the pavement. Also, the fact that some gages decreased in compression,
or stopped contracting, even though the pavement temperature continued
to drop, indicates that there was probably some bending occurring in the
asphalt concrete layer. The measured cooling curves (Figures 8.6 to
8.8) indicate that at the time this second change was observed (11 hours
at test position 1, 10 hours at test position 2, and 12 hours at test
position 3) the average pavement temperature was approximately -5 C
(23 F) for all three cooling cycles. Figure 8.9 shows that the
temperature gradient was lowering at these times and was approximately
15 C (28 F). The observed effect was also possibly related to the

241
change in thermal gradient, since reductions in pavement temperature
alone should simply increase contraction. However, at this point it was
difficult to speculate as to the reason for the observed thermal
response.
An examination of measured thermal strain distributions at
different times during cooling provide a better picture of how the
pavement actually contracted. Figures 8.16, 8.17, and 8.18 show
longitudinal strain distributions measured during cooling at test
positions 1, 2, and 3, respectively, while the corresponding transverse
strain distributions are shown in Figures 8.19, 8.20, and 8.21,
respectively. Note that in each one of these figures, there are four
distributions plotted, which correspond to the times where distinct
changes in thermal response were observed in the pavement. Therefore,
by examining these distributions in sequence, one can see how thermal
strains were accumulated with time in the pavement. It should be noted
that the changes in response were observed at different times during
different cooling cycles, because the temperature at the start of each
cooling cycle was different (see Figures 8.6, 8.7, and 8.8). Therefore,
the thermal strain distributions were also plotted at different times at
each test position.
The thermal strain distributions presented in Figures 8.16 to 8.21
indicate, that in both the longitudinal and transverse directions, the
measured contraction at the edge of the pavement was always greater than
towards the center of the pavement. As explained earlier, this was
expected, because there was less pavement restraint near the edge of the
pavement.

MEASURED STRAINS DURING COOLING (E-6 IN./IN.)
242
Figure 8.16: Longitudinal Strain Distributions During Cooling:
Test Position 1

MEASURED STRAINS DURING COOLING (E-6 IN./IN.)
t t t
243
Figure 8.17: Longitudinal Strain Distributions During Cooling:
Test Position 2

MEASURED STRAINS DURING COOLING (E-6 IN./IN.)
III! II* '
244
Figure 8.18: Longitudinal Strain Distributions During Cooling:
Test Position 3

MEASURED STRAINS DURING COOLING (E-6 IN./IN.)
i i i t *
245
Figure 8.19: Transverse Strain Distributions During Cooling:
Test Position 1

MEASURED STRAINS DURING COOLING (E-6 IN./IN.)
i l 1
246
Figure 8.20
Transverse Strain Distributions During Cooling:
Test Position 2

MEASURED STRAINS DURING COOLING (E-6 IN./IN.)
l i l *
247
Figure 8.21: Transverse Strain Distributions During Cooling:
Test Position 3

248
However, aside from this general observation, the measured thermal
strain distributions were different from the distributions expected for
a pavement under normal conditions (i.e. level, homogeneous, and with
uniform frictional resistance between the layers). Under such
conditions, the resistance to thermal contraction would increase
uniformly from zero at the edge of the pavement to some maximum value
either at the center of the pavment or at some specified distance from
the edge of the pavement. Therefore, the measured thermal contraction
strains would decrease uniformly from a maximum value at the edge of the
pavement to a minimum value at or near the center of the pavement.
Although the theoretical distribution described above was generally
observed at all three test positions in the transverse direction (see
Figures 8.19, 8.20, and 8.21), it was clearly not observed in the
longitudinal direction. Figures 8.16, 8.17, and 8.18 show that the
longitudinal thermal strain distributions measured during cooling were
not uniform. Apparently, either differential friction between the
asphalt concrete and the base, or some other phenomenon probably related
to the thermal gradient in the asphalt concrete layer (or a combination
of both) were resulting in the differential contraction measured in the
longitudinal direction.
The accumulation of thermal strains with time gave some clues as to
the mechanism that led to the observed distributions. Note that for all
three cooling cycles (Figures 8.16, 8.17, 8.18), the shape of the first
and second longitudinal strain distributions (in order of time) was
about the same. This indicated that the differential strain
distribution was established during the first two hours of cooling, then
remained essentially the same until the second change in thermal

249
response occurred. This indicates that the mechanism may be related to
the thermal gradient in the asphalt concrete layer, since the thermal
gradient reached a maximum during the first two hours of cooling (Figure
8.9). The thermal gradient decreased comparatively slowly after two
hours, and no change in the shape of the thermal strain distribution
occurred for these conditions under continued cooling.
The next change in thermal response occurred anywhere from 10 to 12
hrs. of cooling, depending on the cooling cycle. This change is evident
by comparing the second and third longitudinal strain distributions for
each cooling cycle. The most significant changes are shown in Figures
8.17 and 8.18, which show the greatest changes in the shape of the
longitudinal strain distribution. The section to the left of the center
of the test pit (six feet from south wall) went into compression, while
the area to the right of center decreased in compression, and at one
point, the surface of the pavement actually went into tension.
The most interesting thing about the observed change in thermal
response was that there were very small changes in pavement temperature
during the time the strain changes were observed. The respective
cooling curves shown in Figures 8.6 to 8.8, show that the cooling rate
after about 11 hrs. was almost negligible (about 0.3 C/0.6 F per hr.).
Yet even with these small temperature reductions, relatively large and
unusual changes in strains were observed in the pavement.
The repeatability of the measurements indicated that the
measurements were good. Several gages were monitored during different
cooling cycles and their response during all cycles was almost
identical. Figures 8.22 and 8.23 show that the strain-temperature
relationship for two gages that were monitored during different cooling

250
Figure 8.22: Comparison of Measured Thermal Strains for Different
Cooling Cycles: Six Feet from South Wall

MEASURED STRAINS DURING COOLING (E*6 IN./IN.)
I l l I I I' 9
251
Figure 8.23: Comparison of Measured Thermal Strains for Different
Cooling Cycles: 8.33 Feet from South Wall

252
cycles were almost identical (Note that two gages with very different
thermal response characteristics were chosen for these comparisons). As
shown in these figures, the magnitude of strain was very close during
all cooling cycles. In addition, the observed changes in thermal
response occurred at the same time during all cooling cycles. This
latter observation indicated that the longitudinal strain distributions
measured during different cooling cycles could be superimposed to obtain
the longitudinal strain distributions at different times for the entire
12-ft. length of pavement.
The superimposed distributions are shown in Figure 8.24. As shown
in this figure, the correspondence of strains measured during different
cooling cycles and at different points in the pavement was excellent.
Therefore, this figure provided an excellent opportunity to observe the
sequence of strain accumulation for the entire pavement.
As observed earlier for the individual distributions, Figure 8.24
shows that the differential pattern of strains established after 2 hrs.
of cooling, remained essentially the same until the second change in
thermal response was observed. This is evident by noting the very
similar shapes of the first and second strain distributions with respect
to time, which are defined in the figure by x's and circles,
respectively.
The effect of the next change in thermal response is evident by
comparing the second and third strain distributions, which are defined
in Figure 8.24 by the circles and squares, respectively. Clearly, the
pattern of strain accumulation changed during this time. A relatively
sharp increase in compressive strains occurred in the section from about
four to six feet from the south wall of the test pit. At the same time,

MEASURED STRAINS DURING COOLING (E-6 IN./IN.)
I I i '
253
Figure 8-24: Longitudinal Strain Distributions During Cooling

254
an even sharper decrease in compressive strains occurred to the north of
this section where the surface of the pavement actually went into
tension at about nine feet from the south wall of the test pit.
As mentioned earlier, the most interesting thing about these
changes in response, was that there was almost no change in pavement
temperature during the time that they occurred. The cooling rate was
negligible so that the thermal stresses accumulated during this time
were probably also negligible. The small temperature reduction that did
occur should have resulted in either no change in strain or a small
uniform increase in compressive strains. Therefore, the change in
thermal response observed was truly unusual. But so was the
differential strain distribution observed before these changes
occurred. It is possible that the two were related.
There are two ways, aside from changes in moisture, for a change in
strain to occur in any structure: with a change in temperature or with
an application of load. As mentioned before, the change in temperature
was very small and could not explain the observed response. This
implies that the strain changes occurred as a result of an applied
load. The only load on the pavement during cooling was the weight of
the asphalt concrete itself, and the only way that the weight of the
asphalt concrete could have caused the observed changes in strain, was
if the pavement was uplifted from the base prior to the time the changes
were observed. Therefore, it seems like the effect of cooling, prior to
the time this change in strain response was observed, was to uplift the
asphalt concrete layer from the base.
It will be shown in Section 8.3.3 that highly uncommon load
response was observed in the pavement, which seemed to be related to the

255
unusual thermal response just described. Extremely high deflections
under load were observed at 0.0 C (32 F) at test positions 1 and 3 (see
Figure 8.24). In section 8.3.4, it will be shown that the high
deflections could only be attributed to an uplifted pavement, which
substantiates the observations made above. It is interesting to note
that relatively higher compressive strains were observed during cooling
at both test positions 1 and 3, until the time the strain pattern
changed. These observations did not help to better define the mechanism
that led to the observed thermal response, but it did reinforce the
concept that there was some correspondence between the thermal response
and the load response of the pavement.
8.3.3 Dynamic Load and Creep Response at Different Temperatures
For all dynamic load tests and creep tests performed, deflection
and strain measurements were taken at ten different points in the
pavement during load tests. Figures 8.25, 8.26, and 8.27 show the
positions of the LVDT's and strain gages monitored at test positions 1,
2, and 3, respectively. These instruments were positioned so that the
pavement deflection and strain distributions were adequately defined in
both the longitudinal and transverse directions.
Table 6.1 summarized the order in which dynamic load tests and
creep tests were performed. For the sake of clarity, the order of
testing was given in terms of nominal test temperatures in this table.
A summary of the measured pavement temperatures during testing is given
in Table 8.1. As shown in Table 6.1, dynamic load tests were performed
at all three test temperatures before any creep tests were performed.
This was done to eliminate any effect that creep might have had on the

36"
24" O
18"
O
24" 18"
24"
PAVEMENT: ONE
LONGITUDINAL AMD
ONE TRANSVERSE
3'. O"
36" O

X LVDT AND STRAIN GAGE
O LVDT ONLY
STRAIN GAGE ONLY
Jfahi
SCALE:
NOTE: DIMENSION SHOWN
INDICATES DISTANCE
FROM CENTER CF
TEST POSITION
Figure 8.25: LVDT and Strain Gage Location During Load Tests: Test Position 1

X LVDT AND STRAIN GAGE
O LVDT ONLY
STRAIN GAGE ONLY
X 24"
X
36"
X
16"
CENTER OF
TEST POSITION 2
X
24"
XI8"
££ \
X 36"
SCALE:I
NOTE: DIMENSION SHOWN
INDICATES DISTANCE
FROM CENTER OF
TEST POSITION
Figure 8.26: LVDT and Strain Gage Location During Load Tests: Test Position 2

X LVDT AND STRAIN GAGE
O LVDT ONLY
STRAIN GAGE ONLY
7'.0"
X 24"
a13"
12"
CENTER OF
TEST POSITION 3
X X XXX
24" 12" 0" 8" 16"
NOTE: DIMENSION SHOWN
INDICATES DISTANCE
FROM CENTER OF
TEST POSITION
X 18"
SCALE: ||
X 36"
Figure 8.27: LVDT and Strain Gage Location During Load Tests: Test Position 3

259
Table 8.1: Summary of Average Pavement Temperatures During Testing
Average Pavement
Temperatures During Testing (C/F)
Nominal
Test Position Number
Temper
ature
1 2
3
Response
Creep Response Creep
Response
Creep
0.0/32.0
0.6/33.1
1.2/34.2 -0.4/31.3 0.8/33.4
-0.5/31.1
-0.2/31.6
0.0/32.0
(Repeat)
0.5/32.9



6.7/44.0
7.5/45.5
7.5/45.5 6.9/44.5 7.2/44.8
5.8/42.5
5.9/42.6
13.3/56.0
13.3/56.0
14.0/56.0 12.9/55.3 13.4/56.1
12.6/54.7
12.2/53.9

260
dynamic response of the pavement. Dynamic load tests were performed
sequentially at 0.0 C (32 F), 6.7 C (44 F), and 13.3 C (56 F) at all
three test positions, mainly as a matter of convenience. By performing
tests in this order, the pavement was cooled only once to 0.0 C (32 F)
and tests at the higher temperatures were performed as the pavement
warmed up. Creep tests were performed in a different sequence at each
test position in order to observe the effect of creep at one temperature
on the dynamic load response and creep response at another temperature.
Deflections and strains measured during dynamic load tests at each
of the three test positions are given in Appendix D. At all test
temperatures, load tests were performed at levels of 1000, 4000, 7000,
and 10,000 lbs. The actual load-unload times used for dynamic testing
are shown in Figure 8.28. For all load levels, the maximum load was
left on for 0.1 sec., hut the time to reach the maximum load increased
as the maximum load increased. This was because a greater amount of
wheel deformation was required to reach higher loads. The rest period
was set at four times the total load time at the 10,000-lb. load level,
since Ruth and Maxfield (42) found that in laboratory samples all
delayed elastic response was recovered after a rest period equal to four
times the loading time.
Data collected during creep tests at each test position are
presented in Appendix E. For all test temperatures, permanent
deflections and creep strains were measured after different durations of
10,000-lb. static load applications. Load durations of 50, 50, 400, and
500 seconds were applied sequentially with a rest period between loads,
equal to four times the load duration. The permanent deflections and
creep strains measured were the residual deflections and strains after

LOAD
ro

Figure 8.28: Load-Unload Times for Dynamic Loading with Dual Wheels

262
the rest period. In addition, dynamic load tests at 10,000 lbs. were
performed before creep tests and after each rest period to observe the
effect of creep on the load response of the pavement. Since at any
given load level there is a wide range of stresses throughout the
pavement, the creep rate at different stress levels could be observed by
simply monitoring creep strains at several locations. Therefore, creep
tests were performed only at the 10,000-lb. load level. Dynamic load
tests were also only performed at 10,000 lbs. since only relative
changes in dynamic load response were being monitored. Also, the number
of tests performed had to be limited because the pavement temperature
was changing with time. Therefore, the time to perform tests at a given
temperature was limited.
For each test position, measured longitudinal and transverse
deflection basins and strain distributions were plotted for the four
load levels used during dynamic load tests at each of the three test
temperatures. To illustrate more clearly the data that were obtained
and how these were analyzed, a complete series of plots for tests
performed at test position 3 is shown in Figures 8.29 through 8.40.
These figures are the final interpretation of the dynamic response
measurements, so an explanation of how these figures were arrived at is
necessary before proceeding with the presentation of results. This
explanation is presented in the following paragraphs, and applies
equally for all test positions. Plots for the other two test positions
are presented later in this section as necessary to illustrate specific
effects that were observed.
It should first be noted that although only six longitudinal and
four transverse deflections and strains were measured at each position

Distance From Center of Test Position (ins )
South-
Figure 8.29: Measured Longitudinal Deflections at 0.0 C (32 F): Test Position 3
283

Figure 8.30: Measured Longitudinal Strains at 0.0 C (32 F): Test Position 3

Distance From Center of Test Position (ins )
Figure 8.31: Measured Transverse Deflections at 0.0 C (32 F): Test Position 3

West
Figure 8.32: Measured Transverse Strains at n.O C (32 F): Test Position 3

Distance From Center of Test Position (ins )
Figure 8.33: Measured Longitudinal Deflections at 6.7 C (44 F): Test Position 3

South
Figure 8.34: Measured Longitudinal Strains at 6.7 C (44 F): Test Position 3
268

Distance From Center of Test Position (ins )
Figure 8.35: Measured Transverse Deflections at 6.7 C (44 F): Test Position 3

West
f>0
O
Figure 8.36: Measured Transverse Strains at 6.7 C (44 F): Test Position 3

Distance From Center of Test Position (ins )
Figure 8.37: Measured Longitudinal Deflections at 13.3 C (56 F): Test Position 3

South
FO
ro
Figure 8.38: Measured Longitudinal Strains at 13.3 C (56 F): Test Position 3

West
Distance From Center of Test Position (ins )
Figure 8.39: Measured Transverse Deflections at 13.3 C (56 F): Test Position 3

Figure 8.40: Measured Transverse Strains at 13.3 C (56 F): Test Position 3

275
(see Figures 8.25, 8.26, and 8.27), each deflection basin and strain
distribution was defined by twice that number of data points (see
Figures 8.29 to 8.40). This was because the deflections and strains
were assumed to be symmetrical about the center of loading, and
measurements taken at different distances on either side of the load
were used to define the measured distribution on the opposite side.
Initial analyses of dynamic load response data indicated that the
wheels moved during dynamic load tests. Therefore, the exact position
of the wheels during loading was not known and had to be interpreted for
each set of deflection and strain measurements. The figures presented
for test position 3 (Figures 8.29 through 8.40) show that for almost all
tests performed, the measured strains and deflections were not
symmetrical about the center of the test position, where the wheels were
originally placed. However, the figures also show, that as expected,
the measurements were in all cases symmetrical. The point of symmetry,
which was determined by trial and error for each set of measurements
(i.e. for each load level and temperature), was interpreted as the
center of loading. It is interesting to note that the center of loading
was determined from deflection and strain measurements independently,
and in all cases the tv/o centers corresponded exactly. This indicated
three things: that the method used to establish the center of loading
was valid; that the precision of the measurements was good; and that
there was good correspondence between measured deflections and measured
strains.
It seemed clear from these results that there were deficiencies in
the loading system that caused the wheels to move longitudinally with
repeated loading, and in some cases, caused the wheels to rotate as the

276
load was increased. Apparently, the clamping system could not prevent
the reaction beam from moving longitudinally under repeated 10,000-lb.
loads. Also, the reaction beam was not stiff enough in torsion to
prevent wheel rotation. Figure 8.29 shows the center of loading for
each load level and how this center migrated southward as the load was
increased. This figure clearly illustrates that the wheels were rolling
as the reaction beam rotated in torsion with increasing load. Figure
8.31 shows that for this case, the wheels also moved slightly to the
east as the load was increased.
These problems with the loading system introduced additional
uncertainty in the data, since the exact center of loading was not
known. This uncertainty resulted in very tedious data interpretation
procedures. Therefore, if at all possible, the loading system should be
improved by either stiffening the existing system or by installing a new
system with a dual reaction beam. Also, the clamping system should be
improved so that the loading ram can be positioned more easily.
Because the wheels moved during dynamic load tests, the center of
loading was different for different sets of deflection and strain
measurements (see Figures 8.29 to 8.40). This made it difficult to
compare measurements taken at different temperatures, positions, etc.
Therefore, after the center of loadings were determined for each set of
data, the deflection basins and strain distributions were shifted so
that the center of loading coincided with the center of the test
position. These shifted drawings could then be easily compared to each
other. All deflection basins and strain distributions presented after
Figure 8.40 were shifted in this way.

277
There were two outstanding and unusual features that characterized
the measured dynamic load response of the oavement: 11 for all test
positions, the deflections at 0.0 C (32 F) were higher (about twice as
high at test positions 1 and 3) than those measured at the higher
temperatures; and 2) the response of the pavement was very different for
the three test positions, even though the pavement section was
uniform. These features are illustrated in Figures 8.41 to 8.52, which
show comparison of measured longitudinal and transverse deflections and
strains at different temperatures for each of the three test
positions. These comparisons are presented for the 10,000-lb. load
level only, since the same effect was observed at all load levels.
Separate comparisons of response at the different test positions were
not presented because the difference in response for the different
positions is quite obvious from the figures already introduced (e.g.
compare Figures 8.41, 8.45, and 8.49).
Assuming that the only effect of temperatures was on the stiffness
of the asphalt concrete, the observed response was totally
uncharacteristic, and was in fact, contrary to the expected behavior of
asphalt concrete pavements. A comparison of the measured response with
the response predicted by elastic layer theory will demonstrate this
point. Figures 8.53 and 8.54 show comparisons of longitudinal
deflections and strains at different temperatures, as predicted by the
BISAR elastic layer theory computer program for the loading conditions
and pavement configuration in the test pit. The only variable used for
these computer runs was the modulus of the asphalt concrete, which was
determined for each temperature level from correlations with measured
asphalt viscosity. Further details on how the predictions were made are

Distance From Load Center (ins)
Figure 8.41: Comparison of Measured Longitudinal Deflections at Different Temperatures:
Test Position 1, 10,000 lbs.

Tensile Strain (E-6ln/ln)
South
Figure 8.42: Comparison of Measured Longitudinal Strains at Different Temperatures:
Test Position 1, 10,000 lbs.
279

Distance From Load Center (ins)
Figure 8.43:
Comparison of Measured Transverse Deflecti
Test Position 1, 10,000 lbs.
ons at Different Temperatures:

Tensile Strain (E-6in/in)
West
Figure 8.44: Comparison of Measured Transverse Strains at Different Temperatures:
Test Position 1, 10,000 lbs.

Distance From Load Center (ins)
<
Figure 8.45: Comparison of Measured Longitudinal Deflections at Different Temperature
Test Position 2, 10,000 lbs.

South
ro
cx>
OJ
Figure 8.46: Comparison of Measured Longitudinal Strains at Different Temperatures:
Test Position 2, 10,000 lbs.

Distance From Load Center (ins)
Figure 8.47: Comparison of Measured Transverse Deflections at Different Temperatures:
Test Position 2, 10,000 lbs.
284

West
42
ro
co
cn
Figure 8.48: Comparison of Measured Transverse Strains at Different Temperatures:
Test Position 2, 10,000 lbs.

Distance From Load Center (ins)
Figure 8.49: Comparison of Measured Longitudinal Deflections at Different Temperatures:
Test Position 3, 10,000 lbs.

Tensile Strain (E-6in/in)
-North
PO
00
Figure 8.50: Comparison of Measured Longitudinal Strains at Different Temperatures:
Test Position 3, 10,000 lbs.

Distance From Load Center (ins)
West
Figure 8.51: Comparison of Measured Transverse Deflections at Different Temperatures:
Test Position 3, 10,000 lbs.
288

Tensile Strain ( E-G in/in)
Figure 8.52: Comparison of Measured Transverse Strains at Different Temperatures:
Test Position 3, 10,000 lbs.

Distance From Load Center (ins)
South
Figure 8.53: Comparison of Predicted Longitudinal Deflections at Different Temperatures: 10,000 lbs.

Tensile Strain (E-6In/In)
100
South
x 0.0 C(32F), E, = 750,000psi
O 6.7 C (44F), E, = 490,000psi
I3.3C(56F), E, = 320,000psl
NOTE! For all temperatures,
E2= 25,000 psi
E3= 25, OOOpsI
280
Compressive Strain (E-6ln/ln)
Figure 8.54: Comparison of Predicted Longitudinal Strains at Different Temperatures: 10,000 lbs.
291

292
omitted, since the comparisons made from these predictions are only
shown to give a general indication of the expected change in response of
the pavement at different temperatures. As shown in Figure 8.53, the
effect of lowering the pavement temperature was to decrease predicted
deflections by a small amount. Figure 8.54 shows that as the
temperature was lowered, the predicted strains decreased. This reponse,
of course, is logical, since the modulus of the asphalt concrete
increases as the temperature is lowered.
The measured dynamic load response shown in Figures 8.41 to 8.52
was so dramatically different from the expected, and varied so much from
position to position, that it was first thought that the measurements
were erroneous. However, a more detailed examination of the results
indicated/that the measurements were consistent and repeatable. It was
evident from the deflection basins and corresponding strain
distributions already presented, that the measured deflections and
strains were compatible. For cases where unusually high deflections
were recorded, either unusually high or inconsistent strains were also
recorded (e.g. see Figures 8.49 and 8.50). Thus, the anomalous response
was recorded by both sets of instruments. Also, the longitudinal
measurements were totally consistent with the transverse (e.g. see
Figures 8.49 and 8.51). Finally, a series of tests were repeated for a
case where unusual response was observed, and the results indicated that
the observed measurements were repeatable. At test position 1, the
pavement was recooled to 0.0 C (32 F) and dynamic load tests were
repeated to double-check the original measurements. The results of the
two series of tests were almost identical. It seemed clear from these

293
observations that the measurements obtained were a true representation
of the dynamic load response of the pavement.
The uncommon response observed at different temperatures and
different positions in the test pit, indicated that changes in
temperature were having a significant effect on the pavement's behavior
that could not be explained by the expected changes in the stiffness of
the asphalt concrete layer. In other words, aside from increasing the
pavement stiffness and possibly inducing thermal stresses in the
pavement, the temperature reduction apparently had another effect on the
pavement in the test pit. This other effect resulted in deflections and
strains, and thus stresses, at lower temperatures, that were more than
double those expected and normally used for design. This is evident by
comparing the response measured at 0.0 C (32 F) at test position 3
(Figures 8.49 and 8.50) with the response predicted by elastic layer
theory for this temperature (Figures 8.53 and 8.54).
The high stresses resulting from this temperature effect combined
with the fact that the asphalt concrete is more brittle at lower
temperatures, may create a very critical condition in the pavement. If
this phenomenoni occurs in the field, it could explain the cause of
certain early pavement failures as well as the source of longitudinal
wheel-path cracking. Therefore, this observation was very significant,
and it is important to determine the mechanism that led to the observed
response. However, before specul ating on the cause of this phenomenon,
the creep test results will be presented, since these results provided
additional insight into the observed behavior.
Longitudinal and transverse permanent deflection and creep strain
distributions were plotted for all creep tests performed at each

294
temperature and test position. To better illustrate the data that were
obtained, a series of plots for creep tests performed at 0.0 C (32 F) at
test position 3, is presented in Figures 8.55 through 8.58. The
distributions shown in these figures are the residual deflections and
strains in the pavement after specified durations of 10,000-lb. static
load, followed by rest periods equal to four times the load duration.
The permanent longitudinal and transverse deflections after load
durations of 50, 100, 500, and 1000 seconds are shown in Figures 8.55
and 8.57, respectively, while the corresponding creep strains are shown
in Figures 8.56 and 8.58, respectively. (Hote that the actual load
durations applied were 50, 50, 400, and 500 seconds. However,
cumulative distributions were plotted, since these are more
illustrative.) Similar distributions were plotted for each temperature
and test position.
At each temperature and test position, plots were also made of
measured longitudinal and transverse deflection basins and strain
distributions for 10,000-lb. dynamic load tests performed after each
duration of static load. This was done to observe the change in dynamic
load response caused by the permanent deflections and creep strains
induced in the pavement. Obviously, a large number of plots and
comparisons were made and it would take too much space, and would be of
questionable value, to present them all. Therefore, only those plots
that demonstrate specific effects will be presented.
As was the case for the measured dynamic load response, the
measured creep response of the test pit pavement was totally different,
and in many cases, contrary to the response expected in pavements under
normal conditions. As temperature decreases, less permanent deflections

South
NOTE:
Distance From Center
-Lio
Of Test Position (ins.)
---5
' \x
' \
\ N
' \ x
TIMES SHOWN INDICATE THE
DURATION OF 10,000-lb.
STATIC LOAD. THE PERMANENT
DEFLECTIONS SHOWN ARE THE
RESIDUAL DEFLECTIONS AFTER
A REST PERIOD EQUAL TO FOUR
TIMES THE LOAD DURATION.
\b
V
\
20 / a 500 sec.
-/ /
y* /
P 01000 sec.
35 / r
. /
47 /
-/5 /
50
- r
55 I
/
/
6(V
/
North
Deflections (E-3ins.)
ro
cn
Figure 8.55: Permanent Longitudinal Deflections at 0.0 C (32 F): Test Position 3

South
Tensile Strain (E-6in/in)
400
^1000 sec.
N>'
600
NOTE: TIMES SHOWN INDICATE THE
DURATION OF 10,000-lb.
STATIC LOAD. THE CREEP
STRAINS SHOWN ARE THE
RESIDUAL STRAINS AFTER
A REST PERIOD EQUAL TO FOUR
TIMES THE LOAD DURATION.
ro
UD
cn
IOOO
! Compressive Strain (E-6In/In)
Figure 8.56: Longitudinal Creep Strains at 0.0 C (32 F): Test Position 3

West
NOTE:
Distance From Center
Of Test Position (ins)
-5
6 12 /Td X24
IOO sec.
East
TIMES SHOWN INDICATE THE
DURATION OF 10,000-lb.
STATIC LOAD. THE PERMANENT
DEFLECTIONS SHOWN ARE THE
RESIDUAL DEFLECTIONS AFTER
A REST PERIOD EQUAL TO FOUR
TIMES THE LOAD DURATION.
500 sec.
1000 sec.
Figure 8.57: Permanent Transverse Deflections at 0.0 C (32 F): Test Position 3

Tensile Strain (E-6in/in)
400
200 Distance from center
of test position (ins)
400
NOTE: TIMES SHOWN INDICATE THE
DURATION OF 10,000-lb.
STATIC LOAD. THE CREEP
STRAINS SHOWN ARE THE
RESIDUAL STRAINS AFTER
A REST PERIOD EQUAL TO FOUR
TIMES THE LOAD DURATION.
600
800
1000
Compressive Strain (E-6In/in)
Figure 8.58: Transverse Creep Strains at 0.0 C (32 F): Test Position 3
298

299
and creep strains are expected, because the asphalt concrete becomes
stiffer and Its behavior Is more elastic. This was not the case for the
pavement In the test pit. Figures 8.59 through 8.64 show comparisons of
permanent deflections and creep strains measured at different
temperatures for each test position. Note that only the cumulative
permanent deflections and creep strains after all load durations (which
totaled 1,000 sec.) were plotted for these comparisons. These figures
show that for any given temperature, the magnitude of the creep response
was totally different for the three test positions. The figures also
illustrate that the relationship between temperature and amount of
permanent deflection and creep strains was different for each test
position. At test position 3 (Figures 8.63 and 8.64), the creep
response increased with increasing temperature, which is totally
contrary to expected response. There was no clear pattern for the
measurements at test position 2 (Figures 8.61 and 8.62). Probably the
most unusual result was that the magnitudes of the permanent deflections
and creep strains observed at 0.0 C (32 F) at test position 3, were much
greater than those measured at any other temperature or position.
At first glance, the measurements seemed questionable, but a closer
examination of the observed creep response indicated that the
measurements were reasonable and consistent. In all cases, the
permanent deflection measurements were compatible with corresponding
creep strain measurements. Where unusually high permanent deflections
were measured, correspondingly high creep strains were also measured.
For example, not only were the highest permanent deflections measured at
0.0 C (32 F) at test position 3, but the measured creep strains were
also higher than for any other conditions. The longitudinal and

South-*
NOTE
GJ
O
o
Figure 8.59: Comparison of Permanent Longitudinal Deflections at Different Temperatures: Test Position 1

South
NOTE:
Tensile Strain (E- 6in/in)
400
200
s
1 1 42 36 30 24 ,/|8 12
zf
Tensile strains at
bottom of pavement
~i - North
42
*1 A
200
400
Distance from center
of test position (ins)
CREEP STRAINS SHOWN ARE
RESIDUAL STRAINS AFTER A
1000-sec. DURATION OF 10,000-lb.
STATIC LOAD FOLLOWED BY A
4000-sec. REST PERIOD.
600
eoo
x 0.0 C (32 F)
O 6.7 C (44F)
13.3 C(32F)
IOOO
f Compressive Strain (E-6 in/ln)
Figure 8.60: Comparison of Longitudinal Creep Strains at Different Temperatures: Test Position 1

South<-
NOTE:
Distance From Center
'i-IO
Of Test Position (ins)
-5
-- 40
PERMANENT DEFLECTIONS SHOWN
ARE RESIDUAL DEFLECTIONS AFTER
A 1000-sec. DURATION OF 10,000-lb.
STATIC LOAD FOLLOWED BY A 4000-sec.,
REST PERIOD.
45
50
55
60
X 0.0C(32F)
o 6.7 C(44F)
13.3 C(56Ft
--65
f Deflections (E -3 ins)
42
-H>-North
CO
o
PO
Figure 8.61: Comparison of Permanent Longitudinal Deflections at Different Temperatures: Test Position 2

Tensile Strain (E-6in/in)
400
200
400
co
o
CO
NOTE: CREEP STRAINS SHOWN ARE
RESIDUAL STRAINS AFTER A
1000-sec. DURATION OF 10,000-lb
STATIC LOAD FOLLOWED BY A
4000-sec. REST PERIOD.
600
800
x 0.0 C (32 F)
O 6.7 C (44F)
13.3 C(32F)
1000
Compressive Strain (E-6In/In)
Figure 8.62: Comparison of Longitudinal Creep Strains at Different Temperatures: Test Position 2

Figure 8.63: Comparison of Permanent Longitudinal Deflections at Different Temperatures: Test Position 3

South
NOTE:
Figure 8.64: Comparison of Longitudinal Creep Strains at Different Temperatures: Test Position 3

306
transverse measurements were also compatible in all cases. This is
evident in Figures 8.55 through 8.58, which show a case where the most
unusual creep response was observed.
Therefore, it appears that the creep test results substantiate the
observation made from the dynamic load test results: that aside from
changing the pavement stiffness and inducing thermal stresses in the
pavement, temperature changes were having another, possibly more
significant effect, on the pavement's behavior. In fact, further
analyses indicated that the dynamic load response and the creep response
of the pavement were closely related. When the two sets of data were
analyzed together, the measurements made sense and they orovided some
insight into the phenomenon that led to the observed behavior.
Comparisons of the dynamic load response with the creep response of
the pavement for each temperature and test position indicated that the
two sets of measurement were closely related. It was evident from these
comparisons that the magnitudes of permanent deflections and creep
strains were directly related to the dynamic load response of the
pavement immediately prior to the time creep tests were performed. More
specifically, the magnitude of creep response was related to the
relative change in stiffness of the pavement system from immediately
before to immediately after creep tests, as determined by the dynamic
load response of the pavement. The greater this change in load
response, the greater the amount of permanent deflections and creep
strains. These observations will be demonstrated by using the
measurements obtained at 0.0 C (32 F), since the most unusual results
were observed at this temperature. However, simil ar effects were also
observed at the higher test temperatures. Furthermore, only comparisons

307
of longitudinal deflections are presented, since as mentioned earlier,
deflection and strain measurements were found to be compatible, as were
longitudinal and transverse measurements. Therefore, it would be
superfluous to present comparisons of strains or of measurements in the
transverse direction.
A comparison of longitudinal deflection basins measured at 0.0 C
(32 F) for each test position is shown in Figure 8.65. These
measurements were taken immediately before creep tests were performed.
As shown in this figure, the apparent stiffness of the pavement was much
lower at test position 3, where much higher deflections were recorded.
The measured deflections were very similar for positions 1 and 2.
Figure 8.66 shows a comparison of the total amount of permanent
deflection recorded at each position at the same test temperature.
Clearly, much higher permanent deflections were recorded at test
position 3, which seemed to indicate that there was a direct
relationship between the initial dynamic response or stiffness of the
pavement system, and the magnitude of permanent deflections. This was
not so obvious for test positions 1 and 2, which had very similar
initial dynamic response, yet higher permanent deflections were recorded
at test position 2. The difference in permanent deflections for these
two positions can be explained by comparing the change in response from
before to after creep tests were performed.
Figures 8.67, 8.68, and 8.69 show comparisons of the dynamic load
response of the pavement at different times for tests performed at 0.0 C
(32 F) at tests positions 1, 2, and 3, respectively. Mote that
deflection basins are shown for initial dynamic load tests as well as
for dynamic load tests performed immediately before and immediately

Distance From Center of Test Position (ins,)
Figure 8.65: Comparison of Dynamic Load Response Immediately Prior to Creep Tests
for Different Test Positions: 0.0 C (32 F)

South
Distance From Center
42,-, 36 30 24
i (¡^
-Â¥=Â¥=
-10
Of Test Position (ins)
-5
6 12 18 30 36 42
\
\ '
\
\
\
\
O
\
\
\
/
\
/
O'
\
\
\
\
k
\
\
\
o-g
- 10
15 NOTE: PERMANENT DEFLECTIONS SHOWN
ARE RESIDUAL DEFLECTIONS AFTER
A 1000-sec. DURATION OF 10,000-lb.
STATIC LOAD FOLLOWED BY A 4000-sec.
REST PERIOD.
-O'
-North
20
25
30
35
40
45
50
55
i

/
/
c/
X TEST POSITION I
O TEST POSITION 2
TEST POSITION 3
/
/
\ #5
f Deflections (E-3 ins)
CO
o
LO
Figure 8.66: Comparison of Permanent Longitudinal Deflections for Different Test Positions: 0.0 C (32 F)

South
Distance From Center of Test Position (ins.)
CO
h-*
O
Figure 8.67: Comparison of Dynamic Load Response at Different Times: 0.0 C (32 F), Test Position 1

Distance From Center of Test Position (ins.)
South
Figure 8.68: Comparison of Dynamic Load Response at Different Times: 0.0 C (32 F), Test Position 2

Distance From Center of Test Position (Ins.)
Figure 8.69: Comparison of Dynamic Load Response at Different Times: 0.0 C (32 F), Test Position 3

313
after creep tests. The initial dynamic load test results will be
temporarily ignored. These figures show that for positions 2 and 3,
where permanent deflections were higher than for position 1, there was
also a greater change in dynamic load response, from before to after
creep tests, than at position 1. It seemed that as permanent
deflections and creep strains were accumulated, one or more of the
pavement layers settled and caused the pavement to respond more
stiffly. Conversely, the magnitude of permanent deflections and creep
strains were dependent on the capacity of one or more of the pavement
layers to settle.
These observations were not unusual, and were in fact, logical.
One would expect higher permanent deflections and creep strains for a
pavement system that responds less stiffly. Less stiff response implies
that higher stresses are being induced in the asphalt concrete for any
given load, and higher stresses imply higher creep rates. It also makes
sense that the pavement responded more stiffly as more permanent
deflections and creep strains were accumulated, since this implies that
either settlement, consolidation, or both were taking place, and these
are normally associated with an increase in stiffness.
These observations resulted in increased confidence in all the
measurements obtained. Therefore, it appeared that the creep test
results substantiated the observation made from initial dynamic load
test results: that aside from changing the stiffness of the asphalt
concrete and possibly inducing thermal stresses in the pavement,
temperature changes were having another, possibly more significant
effect, on the pavement's behavior.

314
However, the fact remained that this effect resulted in very
unusual pavement response. The measured response varied considerably
from position to position, and the highest dynamic deflections and creep
response were observed at the lowest test temperature. Further analysis
of the data provided additional clues to explain the phenomenon that
resulted in the observed pavement behavior.
Creep test results pointed to other factors that had an effect on
the dynamic load response and creep response of the pavement. It was
found that the order, with respect to temperature, in which creep tests
were performed, had a significant effect on the response of the
pavement. As mentioned earlier although initial dynamic load tests were
performed in the same order of temperature at all test positions, creep
tests were performed in a different order at each test position. As
shown in Table 8.1, the order of creep tests at test position 1 was the
reverse order of the creep tests performed at test position 3.
Therefore, a comparison of the measured resDonse for these two positions
will help demonstrate the effects of the order of temperature on the
measured response of the pavement.
The initial dynamic response at 0.0 C (32 F) at test positions 1
and 3 was very similar and unusually high (see Figures 8.67 and 8.69).
However, the dynamic response immediately prior to creep tests was
totally different at the same two positions at the same temperature
(again see Figures 8.67 and 8.69). Figure 8.66 shows that the permanent
deflections recorded at these two positions at 0.0 C (32 F) were also
totally different, where much higher permanent deflections were recorded
at test position 3. A similar, but opposite effect was observed for
these two positions at 13.3 C (56 F). As shown in Figure 8.59, a

315
maximum permanent deflection of 9.0 E-3 in. was recorded at 13.3 C
(56 F) at test position 1, while Figure 8.63 shows that a maximum
permanent deflection of only 1.0 E-3 was recorded at test position 3 for
the same temperature. Apparently, the order of creep testing had a
significant effect on the response of the pavement.
A brief review of the test procedures used at each test position
will help expose the factors that affected the response. At both test
positions, the pavement was cooled to 0.0 C (32 F) first and initial
dynamic load tests were performed. Initial dynamic load tests were then
performed sequentially at 6.7 C (44 F) and 13.3 C (56 F) as the pavement
warmed up. At this point, creep tests were begun, and the procedure was
different for the two test positions. At test position 1, the following
procedure was followed:
- creep tests were performed at 13.3 C (56 F) first, immediately
following the initial dynamic load tests at this temperature;
- the pavement was then cooled to 6.7 C (44 F) and creep tests were
performed; and
- finally, the pavement was cooled to 0.0 C (32 F) and creep tests
were performed.
At test position 3, the pavement was cooled directly to 0.0 C (32 F)
immediately after initial dynamic load tests were performed at 13.3 C
(56 F), and creep tests were performed at the lower temperature. Creep
tests were then performed sequentially at 6.7 C (44 F) and 13.3 C (56 F)
as the temperature warmed up.
These procedures show that there were two major factors, resulting
from the difference in the temperature order of testing, that may have
caused the observed differences in response between the two positions:

316
1) the method of achieving a given test temperature in the pavement; and
2) the difference in amount of permanent deflection and creep strains
that were induced in the pavement prior to performing additional creep
tests at any given temperature. Apparently, not only was the absolute
temperature a factor in the response of the pavement, but how the test
temperature was arrived at was also a factor. Permanent deflections and
creep strains, and possibly settlement, induced at one temperature may
have also affected the observed response at another test temperature.
In addition, these two factors may have been interrelated.
Although it was clear from the measurements obtained that these two
factors influenced the response of the pavement, it was impossible to
separate or even accurately define their effects with the data
available. Thermal strains during cooling were only measured during
initial cooling cycles at each of three positions, since the observed
effects were totally unexpected. Also, as mentioned earlier, changes in
pavement elevation could not be measured during cooling because of the
effect of temperature changes on the LVDT's and the LVDT support
system. Therefore, an absolute reference of strains and deflections was
not maintained throughout testing, and the individual effects of these
factors on the pavement's response could not be determined.
Another observation was made about the pavement behavior based on
the creep test results and the dynamic load tests performed during creep
tests. As more permanent deflections and creep strains were induced,
the response of the pavement became very similar at all three tests
positions. This is illustrated in Figures 8.70, 8.71, and 8.72, which
show comparisons of measured longitudinal defl ection basins for the
different test positions at 13.3 C (56 F), 6.7 C (44 F), and 0.0 C

Distance From Center of Test Position (ins.)
Figure 8.70: Comparison of Dynamic Load Response After Creep Tests for
Different Test Positions: 13.3 C (56 F)

Distance From Center of Test Position (ins.)
Figure 8.71: Comparison of Dynamic Load Response After Creep Tests for
Different Test Positions: 6.7 C (44 F)

Distance From Center of Test Position (ins.)
Figure 8.72: Comparison of Dynamic Load Response After Creep Tests for
Different Test Positions: 0.0 C (32 F)

320
(32 F), respectively. The deflection basins shown were measured after
all creep tests were performed at the respective temperatures. Although
the initial response of the pavement was very different for different
positions, Figures 8.70 and 8.71 show that the response after creep was
almost identical at all test positions for the two higher tests
temperatures. Figure 8.72 shows there was some discrepancy at 0.0 C
(32 F) but the response was much closer than it was for initial dynamic
load tests at this temperature. The reason for the discrepancy at this
lower temperature is that permanent deflections and creep strains
induced at 0.0 C (32 F) were not enough to seat the pavement completely
at test position 3, where creep tests were performed at 0.0 C (32 F)
first.
It is also interesting to note that the deflection basin approached
at all test positions, had the expected pattern with respect to
temperature. That is, the measured deflections decreased slightly as
the temperature decreased. The same patterns were also observed for the
measured strain distributions.
8.3.4 Combined Effect of Thermal and Load Response
The measured thermal response, dynamic load response, and creep
response of the pavement described earlier in this chapter appeared to
be related. All three sets of measurements showed unusual response
characteristics. The thermal response of the pavement, as determined by
the strains measured during initial cooling cycles, was much different
from the response expected for a pavement under normal conditions.
Observed changes in strain distributions with time indicated that
continued cooling of the pavement resulted in some combination of

321
contraction and bending which may have caused the asphalt concrete layer
to separate and uplift from the base at one or more points.
The initial dynamic load response of the pavement at different
temperatures was totally contrary to the response expected for pavements
under normal conditions. Deflections and strains measured under load at
0.0 C (32 F), were in some cases more than double those measured at
higher temperatures. Furthermore, the dynamic load response varied
significantly for different test positions, even though the pavement
section was uniform.
The creep test results substantiated the results observed during
initial dynamic load tests. Comparitively high permanent deflections
and creep strains were measured for cases where high deflections and
strains were measured under dynamic loads. Dynamic load tests performed
during creep tests showed that there was a settling or stiffening effect
with sustained applications of static load. In addition, the dynamic
load response of the pavement after all creep tests were performed, was
found to be very similar for all test positions. This latter response,
which was approached with increased settlement at all test positions,
showed the expected changes in response with respect to temperature.
Dynamic load tests performed during creep tests showed that the
effect of temperature on load response depended on the method of cooling
used to arrive at a given temperature. These tests also showed that
response at a given temperature may have been affected by the amount of
permanent deflection and creep strain accumulated at a different
temperature. It is possible that these two effects were interrelated.
These observations left little doubt that aside from the expected
changes in asphalt concrete stiffness and possibly inducing thermal

322
stresses in the pavement, temperature changes were having another,
possibly more significant effect on the pavement. Furthermore, the
effect was detrimental, since it resulted in a weaker pavement system,
as evidenced by the high deflections and strains observed under load.
However, it was still unclear what changes occurred in the pavement
system to cause the observed response. Therefore, the measured thermal
response and load response of the pavement were used to formulate a
hypothesis to explain this phenomenon.
There are two possible ways in which a reduction in temperature
resulted in a weaker pavement system: 1) by weakening the base or
subgrade; or 2) by causing the asphalt concrete layer to deform in such
a way that it separated from the base. Of course, it is also possible
that both these effects occurred at the same time. However, based on
the observed response it seems like the second possibility occurred in
the test pi t.
Although it is possible that a temperature reduction at the surface
of the pavement could weaken or reduce the modulus of the foundation
materials, the measured response indicated that this was not the case.
There are two ways in which a reduction in the modulus can occur in the
foundation materials: 1) an increase in moisture in the limerock base;
or 2) a decrease in capillary tension in either the sand subqrade or the
limerock base. Either of these effects could occur as a result of
upward moisture migration due to the thermal gradient created in the
pavement system during cooling. However, even if these changes did
occur, their effect would probably not be enough to explain the
magnitude of the deflections and strains measured at lower
temperatures. These changes would have occurred only in the top two

323
feet or so of the foundation materials, so that drastic reductions in
moduli would be necessary to effect the observed changes in load
response.
Several other observations indicated that the properties of the
foundation materials did not change during cooling. First, the changes
in response were not observed at all test positions, and it seems
unlikely that weakening of the foundation materials occurred in
localized areas. The moisture content and density of the sand subgrade
and limerock base were measured before the asphalt concrete layer was
placed and after it was removed. In both cases, these properties were
uniform throughout the pavement area (see Chapter V). Second, very
significant changes in pavement response were observed within a period
of a few hours, and for relatively small increases in temperature. As
shown in Figure 8.49, there was a drastic reduction (60%) in measured
deflections under load, when the pavement warmed up from 0.0 C (32 F) to
6.7 C (44 F). This change in temperature occurred within a few hours.
A reduction in base layer moisture could not have occurred during this
time, and no other effect related to changes in foundation material
properties could explain the observed increase in pavement stiffness.
Third, the fact that the load response after creep tests was very
similar for all three test positions, indicated that the same foundation
conditions existed under each test position. Finally, creep tests were
performed using the same level of static load for different time
durations, and permanent deflections continued to accumulate with
increased time of loading. It is difficult to believe that these time-
dependent permanent deflections were a result of densification of the
granular subgrade and base materials under a static load. There were

324
additional observations that indicated that the properties of the
foundation materials were not affected by cooling. However, this point
has already been illustrated by the evidence presented above. There
fore, by a process of deduction, it appears like the unusual response
observed was caused by an uplifted asphalt concrete layer.
The load-deflection curves for initial dynamic load tests performed
at each test position, provided additional evidence that the response
observed was caused by an uplifted asphalt concrete layer. Figures
8.73, 8.74, and 8.75 show comparisons of load deflection curves for the
three test positions at 0.0 C (32 F), 6.7 C (44 F), and 13.3 C (56 F),
respectively. These figures show that the load-deflection response was
approximately the same at all test temperatures at test position 2,
where lower deflections were observed. However, at test positions 1 and
3, the load-deflection response changed significantly as the pavement
temperature increased. As shown in Figure 8.73, the load-deflection
response is linear for these two positions at 0.0 C (32 F), where very
high deflections were recorded. As the temperature increased, the load-
deflection response at these two positions approached the response
observed at test position 2. For test position 1, the response was very
similar to test position 2 once the temperature was raised to 6.7 C
(44 F, Figure 8.74) and remained about the same at 13.3 C (56 F, Figure
8.75). The change was more gradual for test position 3. As the
temperature increased the load-deflection response remained about the
same at lower load levels, but the pavement responded more stiffly at
higher load levels. Furthermore, the response at the higher load levels
was very close for all three test positions. This is evident by noting
that the slopes of the load-deflection curves at 6.7 and 13.3 C (44 and

325
Figure 8.73: Comparison of Load-Deflection Relationships for
Different Test Positions: 0.0 C (32 F)

326
Figure 8.74: Comparison of Load-Deflection Relationships for
Different Test Positions: 6.7 C (44 F)

327
Figure 8.75: Comparison of Load-Deflection Relationships for
Different Test Positions: 13.3 C (56 F)

328
56 F), are almost the same at the higher load levels for the three test
positions. Also note that the break in the load-deflection curve at
test position 3 occurred at 7,000 lbs. at 6.7 C (44 F) and at 4,000 lbs.
at 13.3 C (56 F).
The fact that the relative response approached the same values at
all three test positions indicated that the foundation conditions were
uniform. The changes in load-deflection response observed at test
positions 1 and 3 indicated that the pavement was uplifted from the base
and was then settling as the temperature was increased. This was
particularly wel1-ill ustrated at test position 3. The linear response
observed at 0.0 C (32 F) at this position (see Figure 8.73) indicated
that even after 10,000 lbs. were applied, the asphalt concrete layer did
not make contact with the base. The nonlinear response observed at
6.7 C (44 F) at test position 3, indicated that the asphalt concrete
layer made contact after 7,000 lbs. were applied. The relative response
of the pavement system from 7,000 lbs. to 10,000 lbs. was the same for
all positions, indicating that full contact was being made. A similar
argument can be made at 13.3 C (56 F) where the asphalt concrete layer
apparently made contact after only 4,000 lbs. of load were applied.
It seems clear from these observations that the unusual response
observed was caused by the contraction and bending characteristics of
the asphalt concrete layer under a thermal gradient and continued
cooling. These characteristics apparently caused the asphalt concrete
layer to separate and lift off from the base at two or more points.
However, the actual mechanism that led to this behavior was unclear.
There are several ways in which the pavement may have been uplifted
from the base. One possibility is thermal curling of the asphalt

329
concrete resulting from the thermal gradient induced in the surface
layer. Curling is a well recognized phenomenon in Portland cement
concrete slabs, but is normally not considered in the analysis of
asphalt concrete pavements. However, it is possible that at lower
temperatures the stiffness of the asphalt concrete is high enough for
curling or a similar mechanism to occur. This mechanism may be
different from that observed in Portland cement concrete pavements
because asphalt concrete is viscoelastic and its properties change with
temperature. One obvious difference is that once curling occurs and
stresses are induced in the asphalt concrete by the weight of the
uplifted slab, the asphalt concrete will begin to creep and the uplifted
section will begin to settle. It is possible that this type of behavior
was partially responsible for the thermal response observed in the test
pit. Other possibilities exist that may be related to differential
friction between the asphalt concrete and the base, the initial contour
of the surface of the base, and nonhomogeneity in the asphalt concrete
layer. In any case, no matter what hypothesis is established at this
point, it would be little more than speculation, since it could not be
verified with the available data.
It is difficult, if not impossible, to determine what the strains
measured during cooling imply in terms of actual contraction and bending
that may have occurred in the asphalt concrete layer. For complex
movements, such as those that probably occurred in the pavement during
cooling, strain measurements alone mean very little without also knowing
the vertical and horizontal displacements at different points in the
pavement. Unfortunately, deflections could not be monitored during
cooling because of the effect of temperature on the LVDT's and the LVDT

330
support system. Therefore, these observations showed the importance of
installing a system to monitor deflections durinn cooling for future
pavement installations.
These results also emphasized the importance of considering the
combined effects of temperature and load in the failure analysis of
asphalt concrete pavements. The high load-induced deflections and
strains, and thus stresses in the pavement, combined with the fact that
asphalt concrete is more brittle at lower temperatures, may result in a
very critical condition in the pavement. Therefore, if the effects
observed in the test pit occur in the field, it could lead to early
pavement failures. The effect may also explain the cause of
longitudinal wheel-path cracking, which cannot be explained using
conventional analysis procedures.

CHAPTER IX
RESPONSE PREDICTION
AT LOW TEMPERATURES
9.1 Dynamic Load Response
It was shown in Chapter VIII, that pavement cooling caused the
asphalt concrete layer to uplift from the base. This phenomenon
resulted in highly unusual dynamic load response that in some cases was
totally contrary to the response expected for pavements under normal
conditions. It was obvious that this load response which was observed
during initial dynamic load tests, could not be modeled using conven
tional layered system analysis. Furthermore, the support conditions
existing during these tests could not be defined from the available
data. Therefore, the initial dynamic load response could not be
model ed.
However, further analyses indicated that the measured deflections
and strains approached the same values at all test positions with
continued application of static load. In addition, the response
approached at each test temperature showed the expected changes in load
response with temperature. These observations indicated that for each
test position the asphalt concrete layer would eventually reach a point
where it would be in full contact with the base. This latter condition
is, of course, suitable for modeling with conventional layered system
analysis.
Therefore, the dynamic load test results were examined carefully to
determine which measurements were made when the asphalt concrete layer
331

332
was in a fully settled condition. Two criteria were used for this
purpose: 1) cases were chosen where the changes in load response
observed during creep tests were small; and 2) cases were chosen where
the deflection and strain measurements were symmetrical about the center
of loading. Results of dynamic load tests performed immediately prior
to creep tests at test position 1 met both of these criteria for all
three test temperatures. Apparently, the temperature order in which
creep tests were performed at this position did not cause the asphalt
concrete layer to separate from the base (see Section 8.3.3).
Therefore, these measurements were used to evaluate the capability of
elastic layer theory to predict the load response of the pavement using
laboratory generated material parameters. It should be noted that
dynamic load tests were only performed at the 10,000-lb. load level
during creep tests, so that response at lower load levels could not be
eval uated.
An elastic layer theory computer program (BISAR) was used to model
the pavement system in the test pit. A cross-section of the test pit
pavement was shown in Figure 7.1 and the pavement was modeled as shown
in Figure 7.7 for elastic layer analysis. Initial estimates of asphalt
concrete moduli at different temperatures were determined from
previously established correlations with measured asphalt viscosity.
These correlations are presented in Appendix A (Equations A.l and
A.2). Viscosities for use in these relationships were determined from
measurements made with the Schweyer rheometer on asphalt samples
recovered from the mixture in the test pit after the pavement was
removed. Based on these measurements a viscosity-temperature
relationship was developed (Equation 5.2, Section 5.2.3) from which

333
constant power viscosity can be calculated at any given temperature.
The following asphalt concrete moduli were calculated for the three test
temperatures, using equation 5.1 and A.1:
Test Temperature Asphalt Concrete Modulus
0.0 C (32 F) 750,000 psi
6.7 C (44 F) 490,000 psi
13.3 C (56 F) 320,000 psi
Initial estimates for the sand subgrade and limerock base moduli were
obtained from the plate load tests performed when these materials were
first placed and compacted (E=53,000 psi for the limerock, and E=15,420
psi for the sand subgrade).
These initial estimates of modulus were used in BISAR to predict
the deflections and strains in the pavement for a 10,000-lb. dual wheel
load, using the procedures developed and described in Section 4.4.3.1.
There was considerable discrepancy between predicted and measured
response at all three temperatures, for this initial prediction.
Therefore, the modulus values for the three layers were adjusted
systematically until the good correspondence was achieved between
predicted and measured deflections and strains at each temperature.
The best correspondence between predicted and measured response was
obtained when the following modulus values were used:
Asphalt Concrete Limerock Base Sand Subqrade
Temperature
Modulus (psi)
Modulus (psi)
Modulus (psi)
0.0 C (32 F)
750,000
25,000
25,000
6.7 C (44 F)
490,000
25,000
25,000
13.3 C (56 F)
320,000
25,000
25,000
Poisson's ratios of 0.35, 0.40, and 0.30 were assumed for the asphalt

334
concrete, limerock base, and sand subgrade, respectively. Comparisons
of measured and predicted deflections and strains using these parameters
are shown in Figures 9.1 through 9.12. Four comparisons are presented
for each test temperature: longitudinal deflections, longitudinal
strains, transverse deflections and transverse strains. These figures
show that for all test temperatures, very good predictions of measured
deflections and strains were obtained in both the longitudinal and
transverse directions.
Note that these predictions were made using asphalt concrete moduli
determined from correlations with measured asphalt viscosities at
different temperatures. Other modulus values were attempted, but none
resulted in as good a prediction of measured response.
The moduli used for the sand subgrade and the 1imerock base
indicated that the properties of these materials were different than
when they were originally placed. The sand subgrade had apparently
densified with time, while the limerock base was less stiff. Plate load
tests performed on the limerock and the sand substantiated this finding
(see Section 5.4). The sand subgrade had apparently densified under
repeated dynamic loading. The decrease in limerock modulus appeared to
be directly related to a higher moisture content than when it was
originally placed.
It was interesting to note that the best prediction of measured
response was achieved with the same modulus values for the sand subgrade
and limerock base at all test temperatures. This indicated that the
properties of the foundation materials were not changing with tempera
ture. The same conclusion was reached by analyzing the results of all
dynamic load tests and creep tests performed (see Section 8.3.4), but
these analyses are a bit more definitive.

Dislance From Load Center (ins)
Figure 9.1: Comparison of Measured and Predicted Longitudinal Deflection Basins at 0.0 C (32 F)

Tensile Strain (E 6In/In)
South
Figure 9.2: Comparison of Measured and Predicted Longitudinal Strain Distributions at 0.0 C (32 F)

Dislance From Load Center (ins)
West
Figure 9.3: Comparison of Measured and Predicted Transverse Deflection Basins at 0.0 C (32 F)

Tensile Strain (E-6ln/ln)
Wesi
loo
Distance from
load center (ins.)
too
x Measured
O Predicted
E, = 750,000 psl
E2 25,000 psl
E3= 25,000psl
200
Compressive Strain (E-6in/in)
East
Figure 9.4: Comparison of Measured and Predicted Transverse Strain Distributions at 0.0 C (32 F)

South
Distance From Load Center (ins)
Figure 9.5: Comparison of Measured and Predicted Longitudinal Deflection Basins at 6.7 C (44 F)

South
Tensile Sirain (E-6ln/ln)
Figure 9.6: Comparison of Measured and Predicted Longitudinal Strain Distributions at 6.7 C (44 F)

Distance From Load Center (ins)
West
Figure 9.7: Comparison of Measured and Predicted Transverse Deflection Basins at 6.7 C (44 F)

Tensile Strain (E-6ln/ln)
West
too
Bol tom of
pavement
100
Distance from
load center (Ins.)
x Measured
|B0 O Predicted
E, =490,000psi
E2= 25,000psi
E, = 25,000psl
200
280
Compressive Strain (E-6ln/in)
Figure 9.8: Comparison of Measured and Predicted Transverse Strain Distributions at 6.7 C (44 F)

Dislance From Load Center (ins)
Figure 9.9: Comparison of Measured and Predicted Longitudinal Deflection Basins at 13.3 C (56 F)

Tensile Strain (E-65n/Jn)
South
North
Figure 9.10: Comparison of Measured and Predicted Longitudinal Strain Distributions at 13.3 C (56 F)

West
Distance From Load Center (ins)
Figure 9.11: Comparison of Measured and Predicted Transverse Deflection Basins at 13.3 C (56 F)

Tensile Strain (E-6ln/ln)
i oo
Distance from
load center (Ins.)
Bottom of pavement
x Measured
ISO
O Predicted
'J
E, =320,000psl
E2= 25, OOOpsI
zoo
E3 = 25, OOOpsI
2B0
Compressive
Strain (E-6ln/ln)
Fiqure 9.12: Comparison of Measured and Predicted Transverse Strain Distributions at 13.3 C (56 F)

347
9.2 Thermal Response
It was shown in Section 8.3.2 that the thermal response of the test
pit pavement during cooling was unusual and different from the response
expected for pavements under normal conditions. A differential pattern
of thermal strains was observed during cooling which apparently caused
the asphalt concrete layer to contract and bend in such a way that it
separated and uplifted form the base at two or more points. This uplift
effect was verified by the results of dynamic load tests and creep tests
performed on the pavement.
It was obvious that the complex movements that probably occurred in
the pavement could not be modeled using conventional analyses. These
analyses consider either contraction or thermal stress development for a
uniform temperature drop, or curling induced by thermal gradients.
Uniform frictional resistance between the layers, homogeneous materials,
and horizontal interfaces are also normally assumed in conventional
analyses. The behavior observed in the test pit was probably some
combination of contraction and curling that occurred under nonuniform
frictional resistance between the layers. This behavior may also have
been affected by other factors including the variation in asphalt
concrete properties with temperature, and the profile of the limerock
base. Therefore, a model would have to be developed that includes those
effects that led to the observed thermal response.
However, as mentioned in Chapter VIII (Section 8.3.4) the mechanism
that led to the observed thermal response could not be determined from
the available data. For the complex movements that probably occurred in
the pavement during cooling, strain measurements alone mean very little
without also knowing the vertical and horizontal displacements at

348
different points in the pavement. Unfortunately, these displacements
could not be obtained. Therefore, it would be inefficient to develoD a
model at this time without knowing those effects that truly influenced
the thermal response. Furthermore, the model could not be verified with
the available data. Thus, the measured thermal response was not
predicted.
9.3 Creep Response
Because of the uplift effect induced by pavement cooling, the
measured creep strains and permanent deflections could not be directly
evaluated. The stress distribution in the pavement must be predicted if
creep strains are to be predicted. Since the support conditions could
not be properly defined with the data available, the stresses in the
pavement could not be predicted using conventional models.
It was shown in Section 9.1 that measured deflections and strains
under load were accurately predicted for dynamic load tests performed
immediately prior to creep tests at test position 1. This implied that
the stress distribution predicted by the model was also reasonable at
this position. However, it was obvious by comparing the predicted
stresses to the measured creep strains that the two were not directly
related. This was because, as mentioned earlier, the creep strains
measured were actually the residual creep strains after the load was
removed and a rest period equal to four times the load duration had
elapsed. Therefore, these strains included both the creep strains
induced under load and the creep strains recovered during the rest
period after the load was removed. The stress distribution in the
asphalt concrete layer during loading was not necessarily proportional

349
to the stress distribution once the load was removed. Therefore, at any
given point the creep strains accumulated during loading were probably
not proportional to the strains recovered during unloading. This was
evident from the measured creep strains, where compressive creep strains
were measured at points where tensile stresses were predicted during
loading (see Figures 8.60 and 8.62). This implies that the compressive
stresses upon unloading were greater than the tensile stresses during
loading. Unfortunately, the data were insufficient to define the stress
conditions in the pavement at different times, so these measurements
could not be evaluated.

CHAPTER X
CONCLUSIONS AND RECOMMENDATIONS
10.1 Conclusions
10.1.1 Pavement Testing and Evaluation Methods
The pavement response tests in this investigation were performed in
a test pit facility that was enclosed by a concrete floor and concrete
walls. Deflection and strain measurements were measured for load tests
performed in the test pit at different temperatures, using both a rigid
plate and a flexible dual wheel loading system. The evaluation of these
measurements led to several conclusions concerning the testing and
evaluation of asphalt concrete pavements in the test pit facility which
may be useful to other researchers who perform tests in similar
facilities. These conclusions are as follows:
1. Rigid plate loading is not a good tool for evaluating asphalt
concrete pavement response. The high shear stresses occurring
at the edge of the plate caused excessive plate indentation
under load at temperatures greater than 21.1 C (70 F) for the
pavement section tested. In addition, a model could not be
found that gives accurate predictions of deflections and
strains for a rigid circular area on a layered system.
Although a suitable procedure was developed to do this using
flexible loads in an elastic layer theory computer program, it
was found to be extremely tedious and time-consuming for
routine analysis (see Chapter IV).
350

351
2. A flexible loading system, such as the dual wheels used in this
study, is suitable for asphalt concrete pavement evaluation.
Advantages and disadvantages to using the rigid plate load and
the flexible dual wheel load were outlined in Section 3.5. It
was concluded that neither loading system was perfect but the
disadvantages associated with the rigid plate were
overwhelming.
3. The test pit floor had a significant effect on the load
response of the sand subgrade, the 1imerock base, and the
three-layer pavement system. Furthermore, this effect could
not be accounted for by using an equivalent subgrade modulus
determined from the solution for a semi-infinite mass. It was
shown that using an equivalent modulus resulted in serious
errors when evaluating the load response of the 1imerock base
or the asphalt concrete pavement.
4. The test pit walls had a significant effect on the response of
the pavement system. This effect should be directly accounted
for when evaluating the load response of the pavement using
elastic layer models.
5. The procedures developed and described in Chapter IV, to
account for the effects listed in conclusions three and four,
resulted in accurate prediction of the load response of the
pavement under both the rigid plate and flexible dual wheel
loading system.
6. Two-inch strain gages epoxied directly to the surface of the
asphalt concrete mixture performed well. Two-inch strain gages
mounted on Marshall-sized cores and placed at the bottom of the

352
asphalt concrete layer also performed well. Apparently, the
2.5-inch height of the core provided adequate protection for
the gage against heat and compaction when the four-inch asphalt
concrete layer v/as placed.
7. Measured strains are much better than measured deflections for
evaluating the load response of the pavement. Strains provide
a direct measurement of the behavior of the asphalt concrete,
while deflections give an indication of the response of the
entire pavement system. Therefore, proper evaluation of
deflection data is much more dependent on the analytical models
used. Measured deflections are affected more by boundary
effects and foundation conditions at depth than are measured
strains.
10.1.2 Thermal and Load Response of Asphalt Concrete Pavement
1. Measurements of the thermal and load response of a four-inch
asphalt concrete pavement in the test pit indicated that
temperature differentials produced by rapid pavement cooling
caused the asphalt concrete layer to contract and bend in such
a way that it separated and uplifted from the base at a minimum
of two locations. However, the actual mechanism that led to
this uplift phenomenon could not be determined from the
measurements obtained. The phenomenon appeared to be some
combination of contraction and curling which was at least
partially related to the thermal gradient induced in the
pavement during cooling.

353
2. The uplift effect resulted in load-induced deflections,
strains, and stresses at 0.0 C (32 F) that were in some cases
over twice as high as those expected for a pavement exhibiting
elastic behavior. These high load-induced stresses and
strains, combined with the fact that asphalt concrete is more
brittle at lower temperatures, may result in a very critical
condition in the pavement.
3. If the uplift effect observed in the test pit occurs in typical
pavements, it may explain the cause of some pavement failures
and the cause of longitudinal wheel path cracking. This
emphasizes the importance of considering the variable
properties of asphalt with temperature and the combined effects
of temperature changes and load in the failure analysis of
asphalt concrete pavements.
4. The pavement curling effect suggests that the cracking
potential of asphalt concrete pavements would be reduced by
using thinner asphalt concrete lifts with stiffer foundation
materials. This assumes that the uplift effect is
predominately caused by curling of the asphalt concrete layer
under a thermal gradient. The thermal gradient would be less
in a thinner section so that curling would be reduced or
eliminated.
5. It was obvious that conventional models could not predict the
measured thermal response of the pavement in the test pit.
Conventional models that predict thermal stresses and
contraction of asphalt concrete pavements normally consider
contraction with no thermal gradient assuming elastic or

354
viscoelastic beams or slabs of uniform properties. Models
exist to predict curling of Portland cement concrete slabs, but
curling alone could not explain the observed behavior.
6. Additional instrumentation to monitor deflection and strains is
needed to adequately define the thermal response of the
pavement. Since the observed thermal effect may lead to early
pavement failures, the mechanmism that leads to this phenomenon
and the factors involved need to be defined, so that a suitable
model can be developed to predict the observed behavior.
7. Elastic layer theory resulted in accurate prediction of load-
induced deflections and strains measured in a pavement system
that is typical for Florida. Accurate predictions were
obtained for both rigid plate and flexible dual wheel loading
systems at temperatures ranging from 0.0 C (32 F) to 21.1 C
(70 F), when suitable effective layer moduli were used for
input.
8. Asphalt concrete moduli determined from the correlations with
asphalt viscosity presented in Appendix A resulted in accurate
prediction of measured deflections and strains at all
temperatures and load levels tested. Furthermore, viscosity
measurements were made when the pavement was first placed and
after it was removed, which indicated that the asphalt cement
hardened in the test pit. For load tests performed closer to
the time the pavement was first placed, modulus values
predicted from the earlier viscosity measurements gave accurate
predictions. For load tests performed close to the time the

355
pavement was removed, modulus values predicted from the later
viscosity measurements gave accurate predictions.
9.Reasonable modulus values for the sand subgrade and limerock
base layers were determined from plate load tests performed on
the respective material s. The use of proper analytical tools
to evaluate the plate load test data was critical in the
determination of these moduli. Procedures for the proper
evaluation of plate test data were developed and described in
Chapter IV. These procedures emphasize that foundation
conditions at depth must be accounted for directly for proper
evaluation. An equivalent subgrade modulus cannot be used,
since it may lead to serious error when evaluating pavement
response and in determining moduli for the structural layers.
10. Although the response of the asphalt concrete pavement was
slightly nonlinear in a stress-stiffening way, the use of a
nonlinear model to predict response was unnecessary. Accurate
predictions were obtained with a linear elastic layer model
when suitable effective layer moduli were used for input.
11. A direct evaluation of the creep strain measurements could not
be made. Because of the uplift effect observed during cooling,
the stresses in the asphalt concrete layer could not be
predicted for most creep tests performed. Therefore, the creep
strains could not be predicted.
12. Failure stresses and strains could not be evaluated because it
was not possible to induce failure in the pavement in the test
pit under a well defined set of conditions.

356
10.2 Recommendations
The 1ow-temperature pavement response tests performed in the test
pit emphasized the need to consider the variable properties of asphalt
with temperature and the combined effect of temperature changes and load
in the failure analysis of asphalt concrete pavements. Therefore,
additional pavement sections should be installed in the test pit to
further define the thermal response, load response, and failure
parameters of asphalt concrete pavements. The following recommendations
are presented for tests to be performed on future test pit
installations:
1. The next pavement section to be installed in the test pit
should be chosen such that failure can be induced by exceeding
the stress limit of the asphalt concrete layer under an applied
wheel load. By inducing failure in this way, the failure
stresses necessary to cause failure can be identified for
comparison to failure stresses measured in laboratory
specimens. The section should be chosen such that the maximum
tensile stress in the asphalt concrete layer is about 300 psi
as predicted by elastic layer analysis for a pavement
temperature of 0.0 C (32 F). The modulus values determined for
the materials in this dissertation can be used for these
analyses.
2. The thermal response characteristics of the pavement should be
thoroughly defined to identify the mechanism that led to the
uplift effect of the asphalt concrete layer. However, more
definitive measurements will be required for this purpose. As
before, strains should be monitored at different points on the

357
surface of the pavement during cooling. However, additional
strain gages should be installed at the bottom of the asphalt
concrete layer for better definition of the contraction
characteristics of this layer. In addition, vertical and
horizontal deflections need to be monitored during cooling at
different points on the pavement for proper evaluation of the
measured thermal strains. A system to obtain these deflections
needs to be designed and installed.
3. The thermal response of the pavement should be completely
defined prior to performing any load tests. Therefore, the
pavement should be exposed to several cooling and warm-up
cycles where deflections and strains are monitored at different
points in the pavement during each cycle. Deflections and
strains should be monitored during both cooling and warm-up
cycles.
4. After the thermal response of the pavement is well defined,
dynamic load tests and creep tests should be performed on the
pavement at different temperatures. These tests should be
performed under conditions that can be fully defined for
modeling with conventional layer system analysis. Therefore,
extreme care should be taken to cool the pavement slowly so
that the asphalt concrete layer remains in contact with the
foundation materials. Cooling procedures to achieve this
should be determined from the results of the thermal response
tests performed earlier.
5. Dynamic load tests should be performed at loads well under the
design failure load to evaluate the dynamic load response of
the pavement at different temperatures.

358
6. Creep tests at different temperatures should be performed at
loads well below the design failure load. Static load
durations can be used with rest periods similar to those
applied for creep tests performed in this investigation.
However, the creep strain accumulation should be monitored when
the load is on, and the creep strain recovery should be
monitored during unloading.
7. Once all pavement response tests are completed, the pavement
should be carefully cooled to the design failure temperature
and the design failure load should be applied dynamically in an
attempt to induce failure. The load should be cycled
continuously and the dynamic load response of the pavement
should be recorded periodically to observe any changes in
response that may give an indication that the pavement has
failed. Permanent deflections and creep strains should also be
monitored during these tests, since failure may be induced by
cumulative creep under cyclic load rather than by excessive
stress.

APPENDIX A
RELATIONSHIPS BETWEEN ASPHALT CONCRETE PROPERTIES
AND ASPHALT CEMENT PROPERTIES
A.l Dynamic Modulus
Ruth et al. (43) presented relationships between measured asphalt
viscosity and the dynamic modulus of the asphalt concrete mixture at
different temperatures. These relationships are as follows:
For < 9.19 E8 Pa*s
log Eq j = 7.18659 + 0.30677 log n^g (Equation A.2)
For n^gg > 9.19 E8 Pa*s
log Eg j = 9.51354 + 0.04716 log n10g (Equation A.l)
where, Eg ^ dynamc modulus 0f the asphalt concrete
n100 aPParent viscosity of the asphalt at constant
power of 100 watts/m3
The relationships were developed for dense-graded mixtures tested
in indirect tension for a loading time of 0.1 seconds. They require the
constant power viscosity of the asphalt at a specific temperature as
determined from measurements with the Schweyer rheometer. Viscosity-
temperature relationships for the asphalt can be developed by obtaining
viscosity measurements at several temperatures and stress levels.
A.2 Creep Strain Rate
Ruth et al. (43) also presented relationships between the creep
strain rate of the mixture at a given stress level and the
359

360
pseudo-viscosity of the mixture as predicted from measured asphalt
viscosity. These relationships are as follows:
a
cr n0.01mix
where, ecr creep strain rate for a given stress (a)
n0 Olmix Pseu(l0 mix viscosity at a constant power
of 0.01 watts/m3
and, log ^0>01mix = 9.25549 + 0.36647 log n1Q0
Note that n^g is determined as described in Section A. 1. The creep
strain for a given duration of load is determined as follows:
e - lt = x *
cr cr n0.Olmix
where, ecr creep strain accumulated during time t.
- other variables were defined earlier.

APPENDIX B
PAVEMENT TEMPERATURES DURING COOLING

362
Table B.l: Pavement Temperatures During Cooling: Test Position 1
Time
(Hours/Mlnutes)
Thermocouple
0
1
2
3
4
Number
0
35
0
30
0
30
0
30
0
1
C
19.7
14.9
9.1
6.1
4.2
2.9
2.8
2.1
1.2
F
67.5
58.8
48.4
43.0
39.6
37.2
37.0
35.8
34.2
2
C
19.5
12.2
4.9
1.9
-0.3
-1.7
-2.4
-3.3
-4.2
F
67.1
54.0
40.8
35.4
31.5
28.9
27.7
26.1
24.4
3
C
19.4
15.5
10.2
7.5
5.4
4.0
3.1
1.8
0.5
F
66.9
59.9
50.4
45.5
41.7
39.2
37.6
35.2
32.9
C
19.0
13.0
7.6
5.1
3.2
1.9
1.2
0.3
-0.5
F
66.2
55.4
45.7
41.2
37.8
35.4
31.2
32.5
31.1
C
19.2
10.2
2.6
-0.2
-2.0
-3.1
-3.4
-4.7
-5.8
F
66.6
50.4
36.7
31.6
28.4
26.4
25.9
23.5
21.6
6
C
19.5
13.7
7.2
4.2
2.0
0.4
-0.5
-1.6
-2.6
F
67.1
56.7
45.0
39.6
35.6
32.7
31.1
-29.1
27.3
7
C
19.7
9.4
0.4
-2.8
-5.0
-6.1
-6.3
-7.2
-8.2
F
67.5
48.9
32.7
27.0
23.0
21.0
20.7
19.0
17.2
C
20.6
10.7
2.3
-1.1
-3.4
-4.8
-5.4
-6.C
-7.4
F
69.1
51.3
36.1
30.0
25.9
23.4
22.3
20.1
18.7
C
19.2
18.3
13.3
9.8
7.2
5.4
4.1
3.1
2.1
F
66.6
64.9
55.9
49.6
45.0
41.7
39.4
37.6
35.8
in
C
19.0
18.8
15.8
12.7
10.2
8.3
5.8
5.8
4.7
F
66.2
65.8
60.4
54.9
50.4
46.9
44.2
42.4
40.5
u
C
18.9
18.8
17.6
15.4
13.4
11.7
10.2
9.0
8.0
F
66.0
65.8
63.7
59.7
56.1
53.1
50.4
48.2
46.4
12
C
18.7
18.7
18.3
17.0
15.5
14.0
12.7
11.6
10.6
F
65.7
65.7
64.9
62.6
59.9
57.2
51.9
52.9
51.1
13
C
19.3
18.7
13.9
10.1
7.4
5.4
4.1
3.0
1.8
F
66.7
65.7
57.0
50.2
45.3
41.7
39.4
37.4
35.2
14
C
19.3
19.0
15.9
12.7
10.2
8.2
6.5
5.2
4.0
F
66.7
66.2
60.6
54.9
50.4
46.8
43.7
41.4
39.2
15
C
19.1
19.0
17.9
15.7
13.5
11.6
10.1
8.9
7.7
F
66.4
66.2
64.2
60.3
56.3
52.9
50.2
48.0
45.9
16
C
18.9
18.9
18.5
17.2
15.5
14.0
12.0
11.4
10.4
F
66.0
66.0
65.3
63.0
59.9
57.2
54.7
52.5
50.7
Note: See Figure 8.3 for thermocouple location.

363
Table B.lextended
5
6
8
11
17
20
0
0
30
0
0
0
-0.5
-1.9
-4.6
-7.0
-8.8
1.2
31.1
28.6
23.7
19.4
16.2
34.2
-5.9
-7.2
-8.8
-10.7
-12.0
-.1
21.4
19.0
16.2
12.7
10.4
31.8
-1.6
-3.2
-4.1
-6.6
-8.6
0.6
29.1
26.2
24.6
20.1
16.5
33.1
-2.1
-3.3
-5.1
-7.3
-9.5
0.2
28.2
26.1
22.8
18.9
14.9
32.4
-7.8
-9.3
-10.4
-11.8
-13.1
-1.1
18.0
15.3
13.3
10.8
8.4
30.0
-4.5
-6.6
-7.7
-9.6
-11.3
-0.9
23.9
21.2
18.1
14.7
11.7
30.4
-10,1
-11.3
-12.5
-14.0
-14.6
-1.1
13.8
11.7
9.5
6.8
5.7
30.0
-9.4
-10.9
-11.6
-13.6
-14.8
-1.2
15.1
12.4
11.1
7.5
5.4
29.8
0.2
-1.4
-2.0
-5.6
-7.5
-0.3
32.4
29.5
28.4
21.9
18.5
31.5
2.8
1.2
+0.3
-3.2
-5.5
-0.2
37.0
34.2
32.5
26.2
22.1
31.6
6.1
4.5
2.3
0.2
-2.5
0.2
43.0
40.1
36.1
32.4
27.5
32.4
8.8
7.2
4.6
2.4
0.0
0.6
47.8
45.0
40.3
36.3
32.0
33.1
0.5
-0.4
-3.9
-6.3
-7.9
-0.8
31.1
31.3
25.0
20.7
17.8
30.6
1.8
0.9
-1.8
-4.4
-6.2
-0.6
35.2
33.6
28.8
24.1
20.8
30.9
5.6
3.9
1.5
-0.7
-2.9
-0.1
42.1
39.0
34.7
30.7
26.8
31.8
8.4
6.6
3.9
1.8
-0.3
0.5
47.1
43.9
39.0
35.2
31.5
32.9

364
Table B.2: Pavement Temperatures During Cooling: Test Position 2
Time
(Hours/Mlnutes)
Thermocouple
0
1
2
3
4
Number
0
35
0
30
0
30
0
30
0
1
C
17.4
12.6
7.7
5.3
3.6
2.3
1.0
-0.12
-1.2
F
63.3
54.7
45.9
41.5
38.5
36.1
33.8
31.6
29.8
2
C
17.0
10.1
3.5
0.7
-1.2
-2.5
-3.6
-*.5
-5.0
F
62.6
50.2
38.3
33.3
29.8
27.5
25.5
23.9
23.0
3
C
16.9
12.8
7.5
4.6
2.5
1.1
-0.3
-1.3
-2.2
F
62.4
55.0
45.5
40.3
30.5
34.0
31.5
29.7
28.0
4
C
16.4
12.2
7.4
4.8
3.1
1.7
rt o
-0.7
-1.7
F
61.5
54.0
45.3
40.6
37.6
35.1
32.5
30.7
28.9
5
C
16.2
7.7
0.7
-2.0
-3.7
-4.9
-6.0
-6.7
-6.7
F
61.2
45.9
33.3
28.4
25.3
23.2
21.2
19.P
19.9
5
C
16.7
11.0
5.0
+2.2
0.2
-1.2
-2.4
-3.
.-3.8
F
62.1
51.8
41.0
36.0
32.4
29.8
27.7
25.9
25.2
7
C
16.6
6.5
-1.4
-4.3
-6.1
-7.2
-8.1
-8.7
-9.1
F
61.9
43.7
29.5
24.3
21.0
19.0
17.4
16.3
15.6
8
C
17.1
7.1
0.0
-3.0
-5.0
-6.3
-7.5
-8.3
-8.9
F
62.8
44.8
32.0
26.6
23.0
20.7
18.5
17.1
16.0
Q
C
16.6
15.8
11.6
8.5
6.3
4.6
3.1
1.8
0.9
F
61.9
60.4
52.9
47.3
43.3
40.3
37.6
35.2
33.6
10
C
16.5
16.2
13.7
11.0
8.8
7.1
5.6
4.4
3.3
F
61.9
61.2
56.7
51.8
47.8
49.8
42.1
39.9
37.9
ll
C
16.4
16.3
15.3
13.4
11.6
10.1
8.8
7.6
6.5
F
61.5
61.3
59.5
56.1
52.9
50.2
47.8
45.7
43.7
12
C
16.3
16.3
15.9
14.8
13.4
12.2
11.0
9.9
8.9
F
61.3
61.3
60.6
58.6
56.1
54.0
51.8
49.8
8.9
13
C
16.3
15.3
11.1
7.8
5.4
3.5
2.0
0.8
-0.2
F
61.3
59.5
51.8
46.0
41.7
38.3
35.6
33.4
31.6
14
C
16.2
15.7
13.0
10.0
7.7
5.8
4.3
3.0
1.9
F
61.2
60.3
55.4
50.0
45.9
42.4
39.7
37.4
35.4
15
C
16.0
15.9
14.8
12.8
11.0
9.3
7.9
6.6
5.5
F
60.8
60.6
58.6
55.0
51.8
48.7
45.2
43.9
41.9
16
C
15.0
15.9
15.5
14.3
12.8
11.5
10.2
0.1
8.0
F
60.0
60.6
59.9
57.7
55.0
52.7
50.4
48.4
46.4
Notes: (a) See Figure 8.4 for thermocouple location.
(b) Defrost cycle came on just before 6 hours. Therefore, temperatures are higher than normal.

365
Table B.2extended
5
6
8
10
12
15
18
0
0
30
0
0
0
0
-2.7
-1.3
-4.0
-6.1
-7.1
-7.5
1.6
27.1
29.7
24.8
21.0
19.2
18.5
34.9
-5.3
-3.8
-7.8
-9.7
-10.9
-11.6
0.0
21.6
25.2
18.0
14.5
12.4
11.1
32.0
-3.2
-1.8
-4.8
-7.1
-8.7
-9.0
0.5
26.2
28.8
23.4
19.2
16.3
15.8
32.9
-2.7
-1.9
-4.0
-6.4
-7.7
-7.9
0.4
27.1
28.6
24.8
20.5 .
18.4
17.8
32.7
-6.2
-4.0
-9.6
-11.6
-12.4
-13.2
-1.0
20.8
24.8
14.7
11.1
9.7
8.2
30.2
-4.8
-3.5
-7.1
-9.6
-11.0
-11.3
-0.9
23.4
25.7
19.2
14.7
12.2
11.7
30;4
-9.2
-6.6
-11.6
-13.7
-14.7
-14.9
-1.1
15.4
20.1
11.1
7.3
5.5
5.2
30.0
-9.4
-5.8
-10.9.
-13.1
-14.4
-14.5
-1.2
15.1
21.6
12.4
7.3
6.1
5.9
29.8
-0.6
-1.7
-2.7
-4.4
-6.1
-6.7
-0.3
30.9
28.9
27.1
29.1
21.0
19.9
31.5
1.7
0.4
-0.6
-2.1
-4.0
-4.8
-0.2
35.1
32.7
30.9
28.2
24.8
23.4
31.6
4.8
3.4
2.1
0.8
-0.8
-2.0
0.2
40.6
38.1
35.8
33.4
30.6
28.4
32.4
7.2
5.8
4.3
2.7
1.4
0.4
0.7
45.0
42.4
39.7
36.9
34.5
32.7
33.3
-1.3
-1.7
-2.8
-5.8
-7.4
-7.6
-0.9
29.7
28.9
27.1
21.6
. 18.7
18.3
30.4
0.6
0.0
-0.5
-3.9
-5.5
-5.9
-0.6
33.1
32.0
31.1
25.0
22.1
21.4
30.9
3.9
2.9
1.7
0.1
-1.9
-2.6
-0.1
39.0
37.2
35.1
32.2
28.6
27.3
31.8
6.3
5.2
3.7
1.9
0.8
-0.2
0.5
43.3
41.4
38.7
35.4
33.4
31.6
32.9

366
Table B.3: Pavement Temperatures During Cooling: Test Position 3
Time
(Hours/Minutes)
Thermocouple
0
1
2
3
4
Number
0
30
0
30
0
30
0
30
0
1
C
22.8
18.5
12.3
9.4
CO
r-
6.9
6.2
4.7
3.3
F
73.0
65.3
54.1
48.9
46.0
44.4
13.2
40.5
37.9
2
C
22.7
16.3
8.3
5.0
3.2
2.8
2.6
1.1
1
o
F
72.9
61.3
46.9
41.0
37.8
37.0
36.7
34.0
30.9
3
C
22.6
19.2
13.3
10.1
8.2
7.5
7.0
5.8
4.3
F
72.7
66.6
55.9
50.2
46.8
45.5
44.6
42.4
39.7
4
C
22.5
18.0
11.9
9.2
7.8
7.1
6.4
4.9
3.3
F
72.5
64.4
53.4
48.6
18.0
44.8
43.5
40.8
37.9
5
C
22.2
13.3
3.4
0.2
-1.2
0.8
-0.6
-1.5
-2.8
F
72.0
55.9
38.1
32.4
29.8
30.6
30.9
29.3
27.0
C
22.4
17.3
9.8
6.6
5.0
4.4
4.0
2.7
1.2
F
72.3
63.1
49.6
43.9
41.0
39.9
39.2
36.9
34.0
C
22.4
13.0
2.4
-1.3
-2.6
-1.8
-1.4
CO
CM
1
-4.5
F
72.3
55.4
36.3
29.7
27.3
28.8
29.5
27.0
23.9
8
C
22.6
14.4
5.0
1.3
1
o
a*
-0.8
-0.5
-1.9
-3.5
F
72.7
57.9
41.0
34.3
30.9
30.6
31.1
28.6
25.7
C
22.5
22.0
16.9
13.4
10.7
9.1
8.9
7.3
5.9
F
72.5
71.6
62.4
56.1
51.3
48.4
46.9
45.1
42.6
10
C
22.5
22.4
19.5
16.3
13.8
12.0
10.9
9.9
8.7
F
72.5
72.3
67.1
61.3
56.8
53.6
51.6
49.8
47.7
11
C
22.4
22.5
21.4
19.1
17.1
15.4
14.1
13.1
12.1
F
72.3
72.5
70.5
66.4
62.8
59.7
57.4
55.6
53.8
12
C
22.4
22.4
22.0
20.7
19.2
17.8
16.5
15.5
14.6
F
72.3
72.3
71.6
69.3
66.6
64.0
61.7
59.9
58.3
13
C
22.3
21.9
16.7
12.5
9.8
8.3
7.4
6.6
5.5
F
72.1
71.4
62.1
54.5
49.6
46.9
15.3
43.9
41.9
14
C
22.2
22.1
19.0
15.3
12.5
10.7
9.6
8.7
7.7
F
72.0
71.8
66.2
59.5
54.5
51.3
19.3
47.7
45.9
15
C
22.2
22.2
21.1
18.7
16.4
14.5
13.2
12.2
11.3
F
72.0
72.0
70.0
65.7
61.5
58.1
55.8
54.0
52.3
16
C
22.1
22.2
21.8
20.4
18.7
17.0
15.7
14.7
13.8
F
71.8
72.0
71.2
68.7
65.7
62.6
60.3
58.5
56.8
Notes: (a) See Figure 8.5 for thermocouple location.
(b) Defrost cycle came on just before 6 hours. Therefore, temperatures are higher than normal.

367
Table B.3extended
5
6
8
10
12
15
18
21
24
payJ^-^2
0
0
0
0
0
0
0
0
0
0
1.1
0.1
-1.0
-2.4
-3.6
-5.0
-5.8
-6.0
-6.7
1.0
34.0
32.2
30.2
27.7
25.5
23.0
21.6
21.2
19.9
33.8
-3.1
-4.6
-6.1
-7.9
-9.1
-9.7
-10.9
-10.5
-10.7
-0.7
26.4
23.7
21.0
17.8
15.6
11.5
12.4
13.1
12.7
30.7
1.4
-0.6
-1.8
-4.3
-6.1
-6.4
-8.1
-7.7
-8.3
-0.1
34.5
30.9
28.8
24.3
21.0
20.5
17.4
18.1
17.1
31.8
1.0
-0.3
-1.2
-3.3
-4.8
-5.3
-6.7
-6.5
-7.5
-0.1
33.8
31.5
29.8
26.1
23.4
22.5
19.9
20.3
18.5
32.2
-4.8
-6.3
-9.6
-11.2
-12.0
-12.8
-13.9
-13.7
-13.8
-2.9
23.4
20.7
14.7
11.8
10.4
9.0
7.0
7.3
7.2
26.8
-1.4
-3.1
-5.3
-7.0
-8.5
-9.8
-10.8
-10.7
-11.0
-2.4
28.5
26.4
22.5
19.1
16.7
14.4
12.6
12.7
12.2
27.7
-6.9
-8.4
-11.2
-12.7
-13.8
-14.3
-15.0
-14.4
-2.9
19.6
16.9
11.8
9.1
7.2
6.3
5.0
6.1
5.0
26.8
-5.8
-7.2
-9.5
-11.5
-12.6
-13.1
-14.6
-13.7
-14.8
-3.0
21.6
19.0
14.9
11.3
9.3
8.4
5.7
7.3
5.4
26.6
3.4
1.6
-0.3
-2.3
-3.0
-4.7
-6.0
-6.3
-6.5
-0.9
38.1
34.9
31.5
27.9
26.6
23.5
21.2
20.7
20.3
30.4
6.3
4.4
2.3
0.1
-0.5
-25
-4.0
-4.4
-4.5
-0.7
43.3
39.9
36.1
32.2
31.5
27.5
24.8
24.1
23.7
30.7
9.9
8.0
5.7
3.1
2.1
0.7
-1.0
-1.6
-2.0
0.0
49.8
46.4
42.3
38.1
35.8
33.3
30.2
29.1
28.4
32.0
12.6
10.8
8.4
6.1
4.6
2.9
1.4
0.7
0.2
0.7
54.7
51.4
47.1
43.0
40.3
37.2
34.5
33.3
32.4
33.3
3.5
1.8
-1.3
-3.3
-5.6
-6.1
-7.4
-7.4
-8.2
-2.4
38.3
35.2
29.7
26.1
21.9
21.0
18.7
18.7
17.2
27.7
5.7
3.9
0.9
-0.8
-3.5
-4.4
-5.6
-5.7
-6.5
-2.0
42.3
39.0
33.6
30.6
25.7
24.1
21.9
21.7
70.3
28.4
9.4
7.6
4.7
2.4
0.5
-1.0
-2.2
-2.5
-3.2
-0.9
48.9
45.7
40.5
36.3
32.9
30.2
28.0
27.5
26.2
30.4
. 12.0
10.4
7.6
5.1
3.0
1.5
0.4
-0.1
-0.6
0.1
53.6
50.7
45.7
41.2
37.4
34.7
32.7
31.8
30.9
32.2

APPENDIX C
MEASURED THERMAL STRAINS DURING COOLING

369
Table
C.l: Measured Thermal
Strains During
Cooling:
Test 1
Position 1
Gage Humber
Time From Start
of Cooling (Hours/Minutes)
Thermocoupl e
Number
0
1
2
3
4
0
30
0
30
0
30
0
30
0
Readino
0
-100
-140
-165
-188
-205
-215
-230
-249
Temperature (F)
65.7
65.7
64.9
62.6
59.9
57.2
54.9
52.9
51.1
12
Adjusted Strain
0
-100
-140
-164
-186
-202
-211
-225
-242
Reading
0
-100
-140
-174
-207
-233
-248
-267
-289
Temperature (F)
65.7
65.7
64.9
62.6
59.9
57.2
54.9
52.9
51.1
12
Adjusted Strain
0
-100
-140
-173
-205
-230
-244
-282
Reading
0
-50
-75
-103
-135
-166
-183
-212
-237
Temperature (F)
66.9
57.1
47.1
42.1
38.7
36.3
35.6
34.2
32.7
1.4
Adjusted Strain
0
-47
-65
-89
-118
-146
-167
-190
-213
Reading
0
-27
-43
-57
-71
-85
-93
-103
-116
Temperature (F)
66.2
55.4
45.7
41.2
37.8
35.4
31.2
32.5
31.1
4
Adjusted Strain
0
-23
-32
-42
-53
-64
-71
-78
-90
4
Reading
0
-22
-43
-68
-91
-112
-123
-130
-143
Temperature (F)
67.5
58.89
48.4
43.0
39.6
37.2
37.0
35.8
34.2
i
Adjusted Strain
0
-20
-34
-55
-75
-93
-101
-110
-121
Reading
0
-86
-170
-240
-290
-320
-3 33
-350
-371
3
Temperature (F)
67.5
58.8
48.4
43.0
39.6
37.2
37.0
35.8
34.2
l
Adjusted Strain
0
-84
-161
-227
-274
-301
-314
-329
-349
Reading
0
-37
-50
-58
-70
-83
-91
-103
-115
Temperature (F)
66.4
52.7
41.2
36.4
33.1
30.9
30.1
28.0
26.4
4.5
Adjusted Strain
0
-31
-35
-38
-46
-56
-63
-72
-82
Reading
0
-45
-74
-113
-152
-186
-207
-228
-250
Temperature (F)
67.3
56.4
44.6
39.2
35.6
33.1
32.4
31.0
29.5
1,2
Adjusted Strain
0
-41
-62
-96
-131
-162
-132
-202
-221
Reading
0
-59
-105
-155
-200
-236
-268
-282
-307
Temperature (F)
67.1
54.0
40.8
35.4
31.5
28.9
27.7
26.1
24.4
2
Adjusted Strain
0
-54
-90
-134
-174
-207
-227
-248
-271
Q
Reading
0
-36
-84
-141
-193
-234
-258
-284
-313
Temperature (F)
66.9
59.9
50.4
45.5
41.7
39.2
37.6
35.2
32.9
3
Adjusted Strain
0
-34
-77
-130
-179
-217
-24)
-263
-289
Notes:
(a) Strains in micro-strain. Compression is negative.
(b) See Figure B.3 for gage and thermocouple location.
(c) Strains adjusted using temperature-induced apparent strain relationship furnished by strain gage
manufacturer.

370
Table
C.l-
extended
5
6
8
li
17
20
0
0
30
0
0
0
-279
47.8
-306
45.0
-350
40.3
-380
36.3
-383
32.0
-204
33.1
-270
-295
-334
-360
-358
-180
-225
47.8
-354
45.0
-397
40.3
-428
36.3
-461
32.0
-289
33.1
-216
-343
-381
-408
-436
-265
-281
29.7
-319
27.4
-373
23.3
-420
19.1
-449
15.5
-297
33.3
-253
-287
-335
-375
-397
-273
-141
28.2
-166
26.1
-201
22.8
-227
18.9
-289
14.9
-177
32.4
-111
-132
-162
-181
-236
-152
-171
31.1
-197
28.6
-253
23.7
-283
19.4
-281
16.2
-162
34.2
-145
-167
-216
-238
-230
-140
413
31.1
-446
28.6
-500
23.7
-536
19.4
-544
16.2
-284
34.2
-387
-416
-463
491
493
-242
-146
23.1
-175
20.7
-219
18.1
-257
15.0
-403
11.7
-353
31.2
-103
-133
-172
-204
-343
-327
-292
26.3
-328
23.8
-384
20.0
-422
16.1
-396
13.3
-178
33.0
-259
-291
-340
-371
-340
-154
-352
21.4
-389
19.0
-443
16.2
-478
12.7
-449
10.4
-141
31.8
-311
-344
-392
-420
-387
-116
-365
29.1
-407
26.2
-460
24.6
-500
20.1
-503
14.9
-272
33.1
-336
-374
-424
-457
-450
-248

371
Table C.
2: Measured Thermal
Strains During
Cooling:
Test
Position
2
Gage Humber
Time From Start
of Cooling (Hours/Minutes)
Thermocouple
0
1
2
1
4
Humber
0
30
0
30
0 30
0
30
0
0
Reading
Temperature (F)
0
61.5
-30
54.0
-40
45.3
-47
40.6
-61 -74
37.6 35.1
-8
32.5
-mo
30.7
-115
28.9
Adjusted Strain
0
-25
-29
-32
-43 -53
-63
-/3
-86
1
Reading
Temperature (F)
0
61.3
-36
48.1
-58
39.3
-75
34.5
-96 -115
31.5 29.1
-132
26.9
-145
25.3
-156
24.4
Adjusted Strain
0
-27
-41
-54
-70 -86
-100
-no
-120
2
Reading
Temperature (F)
0
61.5
-33
54.0
-65
45.3
-83
40.6
-103 -118
37.6 35.1
-131
32.5
-143
30.7
-153
28.9
Adjusted Strain
0
-28
-54
-68
-85 -97
-106
-116
-124
3
Reading
Temperature (F)
0
61.2
-48
45.9
-105
33.3
-134
28.4
-166 -190
25.3 23.2
-213
21.2
-230
19.9
-245
19.9
Adjusted Strain
0
-38
-81
-104
-131 -152
-172
-186
-201
4
Reading
Temperature (F)
0
63.3
-24
54.7
-51
45.9
-75
41.5
-101 -120
38.5 36.1
-133
31.8
-152
31.6
-166
.29.8
Adjusted Strain
0
-20
-41
-61
-84 -100
-115
-126
-138
5
Reading
Temperature (F)
0
61.2
-12
45.9
-34
33.3
-61
28.4
-87 -108
25.3 23.2
-12'
21.2
19.9
-153
19.9
Adjusted Strain
0
-2
-10
-31
-52 -70
-85
-96
-109
6
Reading
Temperature (F)
0
62.0
029
54.5
-51
45.4
-71
40.5
-95 -114
37.0 34.5
-132
29.6
-146
30.2
-159
28.5
Adjusted Strain
0
-24
-40
-56
-76 -92
-104
-118
-129
7
Reading
Temperature (F)
0
62.1
-50
52.1
-105
41.8
-146
37.0
-184 -214
33.7 31.3
-239
29.0
-261
27.3
-277
26.0
Adjusted Strain
0
-44
-91
-127
-161 -188
-210
-229
-243
8
Reading
Temperature (F)
0
62.4
-34
55.0
-88
45.5
-126
40.3
-162 -189
36.5 34.0
-213
31.5
-233
29.7
-250
28.0
Adjusted Strain
0
-30
-77
-110
-142 -166
-187
-205
-219
9
Reading
Temperature (F)
0
62.6
-71
50.2
-221
38.3
-320
33.3
-396 -448
29.8 27.5
-490
25.5
-523
23.9
-547
23.0
Adjusted Strain
0
-64
-203
-296
-368 -417
-456
-486
-509
Motes:
(a) Strains 1n micro-strain. Compression Is negative.
(b) See Figure 8.4 for gage and thermocouple location.
(c) Strain adjusted using temperature-Induced apparent strain relationship furnished by strain gage
manufacturer.
(d) Defrost cycle came on just before 6 hours, therefore, these readings are not typical.

372
Table C.2extended
5
6
8
10
12
15
18
0
0
0
0
0
0
0
-135
27.1
-152
28.6
-152
24.8
-171
20.5
-263
18.1
-310
17.8
-298
32.7
-103
-122
-116
-128
-216
-262
-274
-171
24.0
-186
26.7
-211
19.8
-244
15.8
-340
13.9
-406
13.0
-382
31.5
-134
-153
-167
-193
-285
-349
-356
-160
27.1
-168
28.6
-183
24.8
-207
20.5
-230
18.1
-233
17.8
-133
32.7
-128
-138
-147
-164
-183
-185
-89
-260
20.8
-264
24.8
-297
14.7
-332
11.1
-314
9.7
-268
8.2
-118
30.2
-218
-228
-243
-271
-250
-201
-90
-183
27.1
-202
29.7
-215
24.8
-242
21.0
-257
19.2
-268
18.5
-132
34.9 .
-151
-174
-179
-200
-212
-222
-110
-159
20.8
-164
24.8
-189
14.7
-207
11.1
-128
9.7
-59
8.2
+199
30.2
-125
-128
-135
-146
-64
8
+226
-169
25.8
-178
28.7
-203
24.1
-233
19.9
-245
17.2
-274
16.8
-143
32.8
-135
-148
-166
-189
-196
-225
-119
294
24.4
-306
25.1
-341
21.4
-374
17.5
-388
15.3
-383
14.5
-173
32.4
258
-271
-291
-326
-336
-329
-148
-269
26.2
-282
28.8
-309
23.4
-346
19.2
-356
16.3
-356
15.8
-144
32.9
-236
-252
-271
-301
-306
-305
-120
-573
21.6
-589
25.2
-635
18.0
-690
14.5
-715
12.4
-680
11.1
-268
32.0
-532
-554
-588
-636
-657
-619
-243

373
Table C.
3: Measured Thermal
Strains During Cooling:
Test Position 3
Cage Number
Time From Start
of Cooling (Hours/'llmites)
Thermocouple
Number
0
1
2
3
4
0
30
n
30
0
30
0
30
0
Reading
0
-56
-119
-161
-185
-198
-209
-223
-238
Temperature (F)
72
55.9
38.1
32.4
29.8
30.6
30.9
29.3
27.0
Adjusted Strain
0
-52
-101
-136
-157
-171
-132
-194
-206
Reading
0
-37
-45
-48
-51
-54
-51
-80
-102
Temperature (F)
72.2
60.2
45.8
40.5
37.9
37.7
37.2
35.1
32.5
Adjusted Strain
0
-35
-34
-33
-33
-36
-12
-59
-77
Reading
0
-15
-26
-54
-72
-78
-33
-oa
-117
Temperature (F)
72.0
55.9
38.1
32.4
29.8
30.6
30.9
29.3
27.0
Adjusted Strain
0
-11
-8
-29
-44
-51
-56
-59
-85
Reading
0
-27
-39
-41
-43
-43
-42
-52
-64
Temperature (F)
72.5
64.4
53.4
48.6
46.0
44.8
43.5
40.8
37.9
Adjusted Strain
0
-27
-34
-33
-33
-32
-29
-37
-46
Reading
0
-29
-46
-56
-63
-68
-72
-82
-97
Temperature (F)
72.5
64.4
53.4
48.6
46.0
44.8
43.5
40.8
37.9
Adjusted Strain
0
-29
-41
-48
-53
-57
-59
-67
-79
Reading
0
-96
-254
-353
-403
-415
-425
-457
-497
D
Temperature (F)
72.7
57.9
41.0
34.3
30.9
30.6
31.1
28.6
25.7
Adjusted Strain
0
-93
-239
-331
-376
-388
-390
-427
-463
Reading
0
-7
-3
-24
-40
-48
-5A
-72
-92
Temperature (F)
72.1
55.7
37.2
31.0
28.6
29.7
30.2
28.1
25.4
Adjusted Strain
0
-3
+16
+2
-10
-20
-26
-41
-57
Reading
0
-51
-103 -
147
-176
-188
-193
-225
-257
Temperature (F)
72.1
59.5
43.9
39.0
35.4
35.2
35.1
33.1
30.6
Adjusted Strain
0
-49
-91
-130
-155
-141
-177
-201
-230
Reading
0
-36
-88
-140
-172
-183
-191
-212
-238
Temperature (F)
72.3
55.4
36.3
29.7
27.3
28.8
29.5
27.0
23.9
Adjusted Strain
0
-32
-68
-112
-140
-153
-162
-180
-201
Reading
0
-56
-151
-239
-258
-324
-341
-377
-423
Temperature (F)
72.3
63.1
49.6
43.9
41.0
39.9
39.2
36.9
34.2
Adjusted Strain
0
-55
-143
-227
-283
-308
-324
-358
-401
Note:
(a) Strains In micro-strain. Compression 1s negative.
(bl See Flqure 8.5 for gage and thermocouple location.
(c) Strain adjusted using temperature-Induced apparent strain relationship furnished by strain gage
manufacturer.

374
Table C.3extended
5
6
8
10
12
15
18
21
24
26
0
0
0
0
0
0
0
0
C
0
-265
23.4
-293
20.7
-344
14.7
-384
11.7
-382
10.4
-389
9.0
-385
7.0
-364
7.3
-351
7.2
-229
26.8
-227
-251
-290
-324
-320
-323
-315
-295
-281
-197
-140
28.6
-173
26.1
-195
22.3
-247
18.6
-255
16.9
-268
15.7
-360
13.7
-412
13.3
-410
12.8
-357
28.5
-110
-139
-155
-201
-206
-216
-304
-357
-356
-321
-148
23.4
-175
20.7
-214
14.7
-246
11.7
-249
10.4
-235
9.0
-134
7.0
-64
7.3
-73
7.2
+127
26.8
-110
-133
-160
-186
-187
-169
-64
+5
-3
+160
-91
33.8
-115
31.5
-122
29.8
-148
25.5
-153
23.4
-146
22.5
-220
19.9
-283
20.3
-293
18.5
-258
32.2
-68
-89
-94
-114
-115
-107
-176
-240
-247
-233
-124
33.8
-151
31.5
-171
29.8
-206
25.5
-212
23.4
-209
22.5
-223
19.9
-245
20.3
-262
18.5
-208
32.2
-101
-125
-143
-172
-174
-170
-179
-202
-216
-183
-561
21.6
-615
19.0
-680
14.9
-740
10.9
-762
9.3
-767
8.4
-777
5.7
-782
7.3
-814
5.4
-518
26.6
-520
-570
-627
-679
-697
-700
-704
-713
-740
-485
-124
21.5
-156
18.8
-170
13.3
-216
10.1
-223
8.8
-201
7.6
-154
6.0
-123
6.7
-155
6.1
-38
26.8
-83
-110
-114
-153
-157
-132
-82
-52
-03
-6
-315
26.4
-361
23.5
-408
18.6
-467
15.2
-488
13.6
-492
11.7
-528
9.8
-503
10.0
-520
9.7
-347
27.4
-282
-323
-362
-414
-432
-432
-464
-445
-4(6
-316
-288
19.6
-331
16.9
-382
11.8
-435
8.6
-449
7.2
-438
6.3
-418
6.0
-399
6.1
-429
5.0
-216
26.8
-244
-282
-323
-369
-379
-366
-343
-327
-354
-184
-503
29.5
-564
26.4
-618
22.5
-686
18.7
-705
16.7
-703
14.4
-726
12.6
-703
12.7
-714
17.2
-429
27.7
-474
-531
-579
-640
-655
-649
-668
-645
-65F
-398

APPENDIX D
DYNAMIC LOAD RESPONSE MEASUREMENTS

Table D.l: Measured Deflections, Test Position 1, 0 C (32 F)
Load
Longitudinal Deflections
(E-3 in.)(a)
Transverse Deflections (E-3
in.)
(kips)
Distance
from Load
Center
(in.): N-S^k)
Distance
from Load
Center (in
.): E-Wic)
0 8(N)
12 (S)
18 (S)
24(N) 36(N)
13(W)
18(E)
24 (W)
36(E)
10
23.99 15.15
15.43
9.17
1.93 -.1
15.29
9.67
5.67
.2
7
15.42 9.47
9.26
6.27
.97 0
10.99
6.35
3.94
.2
4
10.38 6.63
6.51
3.86
.58 0
6.69
3.47
2.46
0
1
2.15 1.33
1.06
.87
0 0
1.34
0
.49
0
(a) Positive is downward deflection.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.

Table 0.2: Measured Strains, Test Position 1, 0 C (32 F)
Load
Longitudinal
Strains (micro-strain)*
[a)
Transverse Strains
(micro-
strain)
(kips)
Distance from Load
Center
(in.): N-S(b>
Distance
from Load Center (in.): E-W*c^
0BT(d)
OBL^
8(N)
12(S)
24(S)
36 (N)
12(E)
18 (W)
24(E)
36 (W)
10
30
258
-22
-28
48
23
-6
80
54
24
7
24
194
-18
-9
36
16
-4
55
39
17
4
16
117
-15
-6
24
10
-3
32
24
12
1
4.8
27
-5.5
-1
6
3
0
6
7
3
(a) Tension is positive.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.
(d) This gage located at the center of load, at the bottom of pavement, in the transverse direction.
(e) This gage located at the center of load, at the bottom of pavement in the longitudinal direction.

Table D.3: Measured Deflections, Test Position 1, 6.7 C (44 F)
Load
Longitudinal Deflections
(E-3 in
.) (a)
Transverse Deflections (E-3
in.)
(kips)
Distance
from Load
Center
(in.):
N-S(b)
Distance
from Load
Center (in
.): E-W(c)
0 8(N)
12(S)
18 (S)
24 (N)
36 (N)
13 (W)
18(E)
24(W)
36(E)
10
17.63 7.81
8.73
4.11
0.87
-.1
10.51
6.45
2.96
-.4
7
13.10 7.10
6.51
3.02
0.63
-.1
7.64
4.09
2.22
-.3
4
7.83 4.14
3.86
1.81
0.39
0
4.42
1.16
1.48
.1
1
1.59 .83
.87
0.48
0
0
.76
0
0
0
(a) Positive is downward deflection.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.

Table D.4: Measured Strains, Test Position 1, 6.7 C (44 F)
Load
Longitudinal
Strains (micro-strain)^
Transverse Strains
(micro-strain)
(kips)
Distance from Load Center
(in.):
N-S(b)
Distance
from Load Center (in.
): E-W(c)
0BT{d) 0BL{e)
8{N) 12(S)
24 (S)
36 (N)
12(E)
18 (W)
24(E)
36 (W)
10
-52 246
31 66
38
8
18
96
45
17
7
-32 191
22 47
29
6
16
66
32
13
4
-13 112
10 28
20
4
9
40
20
8
1
-2 30
2 7
6
1
2
10
5
2
(a) Tension is positive.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.
(d) This gage located at the center of load, at the bottom of pavement, in the transverse direction.
(e) This gage located at the center of load, at the bottom of pavement in the longitudinal direction.

Table D.5: Measured Deflections, Test Position 1, 13.3 C (56 F)
Load
Longitudinal Deflections
(E-3 in.)
(a)
Transverse Deflections (E-3
in.)
(kips)
Distance
from Load
Center
(in.): N-
S Distance
from Load
Center (in.
): E-W(c)
0 8(N)
12 (S)
18(S)
24(N)
36 (N)
13 (W)
18(E)
24(W)
36(E)
10
14.20 7.48
6.03
2.90
0.68
0
9.08
4.27
2.71
7
9.89 5.44
4.58
2.17
0.58
0
6.93
2.98
1.97

4
6.37 3.41
2.80
1.21
0.24
0
4.06
1.49
1.08
0
1
1.47 .71
.72
0.19
0.0
0
.76
0
0
0
(a) Positive is downward deflection.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.

Table
i D.6:
Measured
Strains, Test
Position 1, 13
.3 C (56 F)
Load
(kips)
Longitudinal
Strains (micro-
strain)
(a)
Transverse Strains
(micro-strain)
Distance from Load
Center (in.): N-
S(b)
Distance
from Load Center (in.)
: E-W{c>
0BT{di
OBL^
8 (N)
12 (S)
24(S)
36 (N)
12(E)
18(W)
24(E)
36 (VI)
10
-144
242
52
90
39
-6
35
103
40
13
7
-104
181
36
63
28
-3
25
71
29
11
4
-54
131
20
38
18
-1
15
44
19
7
1
-12
30
4
10
4
1
4
12
6
2
(a) Tension is positive.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.
(d) This gage located at the center of load, at the bottom of pavement, in the transverse direction.
(e) This gage located at the center of load, at the bottom of pavement in the longitudinal direction.

Table D.7: Measured Deflections, Test Position 1, 0 C (32 F), Repeat Test
Load
Longitudinal Deflections
(E-3 in
.)(a)
Transverse Deflections (E-3 in.)
(kips)
Distance
from Load
Center
(in.):
N-S{b)
Distance
from Load
Center (in.):
: E-W(c)
0
8( M)
12 (S)
18 (S)
24 (N)
36 (N)
13 (W)
18(E)
24(W)
36(E)
10
22.0
14.6
14.0
9.0
3.1
-.1
17.1
5.5
6.2
0
7
16.6
11.5
10.8
6.6
2.5
0
13.2
4.0
4.9
0
4
10.4
7.4
6.8
4.2
1.5
0
8.4
2.2
3.2
0
1
2.3
1.4
1.3
.8
.1
0
1.5
.7
0
0
(a) Positive is downward deflection.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.

Table D.8: Measured Strains, Test Position 1, 0 C (32 F), Repeat Test
Load
Longitudinal
Strains (micro-strain)
(a)
Transverse Strains
(micro-
strain)
(kips)
Distance from Load
Center
(in.): M-
S Distance
from Load Center (in.): E-VI^C^
OBT^ OBL^
8(N)
12 (S)
24(S)
36 (N)
12(E)
18(W)
24(E)
36(W)
10
52 206
10
20
38
22
87
58
39
16.5
7
49 154
-1
6
34
18
68
34
30
15
4
34 96
-13
-3
23
12
42
16
21
12
1
9 24
-5
-2
6
3
8
2
6
3
(a) Tension is positive.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.
(d) This gage located at the center of load, at the bottom of pavement, in the transverse direction.
(e) This gage located at the center of load, at the bottom of pavement in the longitudinal direction.

Table D.9: Measured Deflections, Test Position 2, 0 C (32 F)
Load
Longitudinal Deflections
(E-3 in
.) (a)
Transverse Deflections (E-3
in.)
(kips)
Distance
from Load
Center
(in.):
N-S{b)
Distance
from Load
Center (in
.): E-WC)
0
8(S)
12 (N)
16 (S)
24(N)
36 (S)
13 (W)
18(E)
24 (W)
36(E)
10
11.26
6.63
7.95
2.66
3.26
0
9.32
6.45
3.45
0.75
7
7.83
4.26
5.3
1.45
2.17
0
6.21
4.71
2.71
.5
4
4.28
2.13
2.89
.48
.97
0
3.34
1.49
.99
0
1
.7
.2
.4
0
0
0
.5
0
0
0
(a) Positive is downward deflection.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.

Table D.10: Measured Strains, Test Position 2, 0 C (32 F)
Load
(kips)
Longitudinal
Strains (micro-strain)
(a)
Transverse Strains
(micro-
strain)
Distance from Load
Center i
(in.): N-
Sib)
Distance
from Load Center (in.): E-W^0^
0
8(S)
12(N)
16 (S)
24 (N)
36 (S)
12(W)
18(E)
24 (W)
36(E)
10
-185
-25
5
48
20
14
-3
44
34
11
7
-138
-21
8
35
15
10
-3
35
24
9
4
-84
-14
5
20
9
7
-2
11
15
5
1
-20
-4
.5
4
2
2
-1
5
4
2
(a) Tension is positive.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.

Table D.ll: Measured Deflections, Test Position 2, 6.7 C (44 F)
Load
Longitudinal Deflections
(E-3 in
.) (a)
Transverse Deflections (E-3 in.)
(kips)
Distance
from Load
Center
(in.):
N-S(b)
Distance
from Load
Center (in.):
: E-W(c)
0 8(S)
12(N)
16 (S)
24(N)
36 (S)
13 (VI)
18(E)
24(W)
36(E)
10
10.53 4.26
5.78
-0.24
3.02
0.0
8.56
4.30
2.48
.3
7
7.94 3.08
5.31
-0.25
2.17
0.0
5.97
3.22
1.97
0.0
4
4.53 1.66
3.01
-0.24
1.1
0.0
3.34
1.49
0.2
0.0
1
.73 .12
.24
0.0
0.0
0.0
.48
0.0
0.0
0.0
(a) Positive is downward deflection.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.

Table D.12: Measured Strains, Test Position 2, 6.7 C (44 F)
Load
Longitudinal
Strains (micro-strain)
(a)
Transverse Strains
(micro-
strain)
(kips)
Distance from Load
Center
(in.): N-
s(b)
Distance
from Load Center (in.): E-W^
0
8(S)
12 (N)
16 (S)
24 (N)
36 (S)
12(W)
18(E)
24(W)
36(E)
10
-231
-22
51
55
31
14
16
65
43
15
7
-182
-18
38
40
25
12
14
50
34
11
4
-110
-14
23
24
14
6
7
28
20
6
1
-26
-4
5
5
4
1
.5
7
5
2
(a) Tension is positive.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.

Table D.13: Measured Deflections, Test Position 2, 13.3 C (56 F)
Load
Longitudinal Deflections
(E-3 in
.) (a)
Transverse Deflections (E-3
in.)
(kips)
Distance
from Load
Center
(in.):
N-S(b)
Distance
from Load
Center (in
.): E-W(c)
0 8(S)
12 (M)
16 {S)
24 (N)
36 (S)
13 (W)
18(E)
24 (W)
36(E)
10
9.55 3.08
5.78
-.24
1.21
0
8.12
3.97
2.22
.25
CO
7
7.10 2.13
4.10
-.20
.72
0
5.97
2.48
1.48
0
00
4
4.16 1.18
2.41
-.20
.24
0
3.34
1.24
.49
0
1
.49 0
.24
0
0
0
.72
-.12
0
0
(a) Positive is downward deflection.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.

Table D.14: Measured Strains, Test Position 2, 13.3 C (56 F)
Load
Longitudinal
Strains (micro-strain)(a)
Transverse Strains
(micro-strain)
(kips)
Distance from Load
Center
(in.):
N-S(b>
Distance
from Load Center (in.
): E-W(c)
0
8(S)
12 (N)
16 (S)
24 (N!)
36 (S)
12 (W)
18(E)
24 (W)
36(E)
10
-274
-6
56
67
32
10
30
86
49
15
7
-216
-5
43
49
24
8
24
62
37
12
4
-136
-5
26
30
15
5
14
37
23
7
1
-36
-2
7
8
4
1
2
9
6
2
(a) Tension is positive.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.
389

Table D.15: Measured Deflections, Test Position 3, 0 C (32 F)
Load
Longitudinal Deflections
(E-3 in.)
(a)
Transverse Deflections (E-3 in.)
(kips)
Distance
from Load
Center
(in.): N-
S(b)
Distance
from Load
Center (in.):
: E-W(c)
0 8(N)
12 (S)
16 (N)
24 (S)
40 (S)
13 (W)
18(E)
24(W)
36(E)
10
21.4 11.1
18.6
4.9
6.8
-0.6
13.5
12.2
5.7
3.2
7
15.52 9.16
11.41
3.62
3.29
-0.3
9.32
7.05
3.99
1.9
4
9.01 5.45
6.53
2.08
1.93
-0.1
4.87
2.98
2.07
0.9
1
1.4 0.72
1.04
0.0
0.0
0.0
0.0
0.0
0.0
0.0
(a) Positive is downward deflection.
(b) M is for north, S is for south of load center.
(c) E is for east, W is for west of load center.

Table D.16: Measured Strains, Test Position 3, 0 C (32 F)
Load
Longitudinal
Strains (micro-strain)^
Transverse Strains (micro-strain)
(kips)
Distance from Load
Center
(in.):
N-S(b^
Distance
from Load
Center (in.):
: E-Vlt
0
8( M)
12 (S)
16(N)
24 (S)
40 (S)
12 (W)
18(E)
24 (W)
36(E)
10
-272
44
-73
91
86
34
16
28
48
26
7
-220
14
-46
67
68
22
10
25
36
19
4
-132
-6
-28
38
40
13
2
14
20
10
1
-31
-4
-8
8
8
3
-2
3
5
2
(a) Tension is positive.
(b) M is for north, S is for south of load center.
(c) E is for east, W is for west of load center.

Table D.17: Measured Deflections, Test Position 3, 6.7 C (44 F)
Load
Longitudinal Deflections
(E-3 in
.) (a)
Transverse Deflections (E-3 in.)
(kips)
Distance
from Load
Center
(in.):
M-S^
Distance
from Load
Center (in.): E-W^c^
0
8(N)
12 (S)
16 (N)
24 (S)
40 (S)
13(W)
18(E)
24(W) 36(E)
10
9.7
9.3
4.4
5.3
1.1
0.0
7.0
3.4
3.0 0.3
7
7.3
6.7
3.3
3.9
0.7
0.0
4.9
2.0
1.9 0.2
4
4.3
3.9
1.7
2.0
0.3
0.0
2.6
0.7
0.8 0.1
1
0.8
0.7
0.3
0.2
0.0
0.0
0.2
0.0
0.0 0.0
(a) Positive is downward deflection.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.

Table D.18: Measured Strains, Test Position 3, 6.7 C (44 F)
Load
Longitudinal
Strains (micro-strain)^
Transverse Strains
(micro-strain)
(kips)
Distance from Load
Center
(in.):
M-S(b)
Distance
from Load Center (in.);
: E-W(c>
0
8(N)
12 (S)
16(N)
24 (S)
40 (S)
12 (W)
18(E)
24(W)
36(E)
10
-164
-54
48
18
30
9
8
44
30
14
7
-124
-41
34
12
22
8
6
33
22
11
4
-80
-27
22
8
14
4
4
19
14
8
1
-20
-7
5
2
4
1
1
4
4
2
(a) Tension is positive.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.

Table
D.19:
Measured Deflections, Test
Position 3,
13.3 C (56 F)
Load
Longitudinal Deflections
(E-3 in.)
(a)
Transverse Deflections (E-3
in.)
(kips)
Distance from Load
Center
(in.): N-
s(b)
Distance
from Load
Center (in.
): E-W{ci
0
8( N)
12(S)
16 (N)
24 (S)
40 (S)
13 (W)
18(E)
24 {W)
36(E)
10
10.28
8.29
4.83
3.96
1.11
0.1
6.74
2.98
2.32
0.30
7
7.54
6.07
3.41
2.90
0.82
0.0
4.78
2.08
1.58
0.2
4
4.11
3.33
0.71
0.63
0.24
0.0
2.63
0.74
0.74
0.0
1
0.78
0.63
0.24
0.07
0.0
0.0
0.19
0.0
0.0
0.0
(a) Positive is downward deflection.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.

Table D.20: Measured Strains, Test Position 3, 13.3 C (56 F)
Load
Longitudinal
Strains (micro-strain)
(a)
Transverse Strains
(micro-strain)
(kips)
Distance from Load
Center
(in.): N-
s(b)
Distance
from Load Center (in
.): E-W(c)
0
8(N)
12 (S)
16 (N)
24 (S)
40 (S)
12(W)
18(E)
24 (W)
36(E)
10
-194
-40
51
40
34
10
18
59
40
18
7
-154
-31
39
30
24
8
16
43
31
14
4
-98
-21
24
18
15
5
9
27
21
9
i
-24
-5
5
5
3
0
2
5
4
2
(a) Tension is positive.
(b) N is for north, S is for south of load center.
(c) E is for east, W is for west of load center.

APPENDIX E
CREEP TEST DATA

Table E.l: Measured Permanent Deflections For Different Times of 10,000-lb. Static Load Application,
Position Number 1, 13.3 C (56 F)
Load^
Longitudinal Deflections
(E-3 in.)(b>
Transverse Deflections (E-3 in.)
Duration
Distance
from Load
Center
(in.): N-S^
Distance
from Load
Center (in.): E-W^)
(Seconds)
0
8(M)
12 (S)
18 (S)
24(N) 36(M)
13(H)
18(E)
24(W) 36(E)
50
3.67
0.71
0.1
-0.24
-0.24 0.0
3.58
1.79
0.86 0.1
50
2.11
0.38
0.0
-0.39
-0.24 0.0
1.91
0.25
0.07 0.1
400
1.66
-0.28
0.25
0.14
0.0 0.0
1.96
0.30
0.30 0.1
500
1.76
0.0
0.43
-0.48
-0.58 0.0
1.53
0.Q4
0.25 0.2
(a) Loads were applied sequentially with a rest period of four times the load duration between loads.
(b) Measured permanent deflections are the residual deflections after the rest period. Positive is
downward.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.

Table E.2: Measured Creep Strains For Different Times of 10,000-lb. Static Load Application,
Position Number 1, 13.3 C (56 F)
Load^
Longitudinal
Strains
(micro-
strain)^)
Transverse Strains
(micro-strain)
Duration
Distance from
Load Center (in
.): N-S(c)
Distance
from Load Center (in.
): E-W(d)
(Seconds)
0BT(e)
0BL(f)
8 (N)
12 {S)
24(S) 36(N)
12(E)
18 (W)
24(E)
36 (W)
50
-212
148
48
60
-6 -5
42
47
-2
-2
50
-141
119
32
35
-6 -4
38
28
-4
-2
400
-342
254
-2
-11
-55 -41
39
-18
-61
-44
500
-104
122
9
1
-14 -3
46
-6
-23
-7
(a) Loads were applied sequentially with a rest period of four times the load duration between loads.
(b) Measured creep strains are the residual strains after the rest period. Tension is positive.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.
(e) This gage located at the center of load, at the bottom of pavement, in the transverse direction.
(f) This gage located at the center of load, at the bottom of pavement in the longitudinal direction.

Table E.3: Measured Dynamic Deflections at 10,000 lbs. After Different Times of Static Load Application,
Position Number 1, 13.3 C (56 F)
Load^
Duration
Longitudinal Deflections 1
[E-3 in.)(b>
Transverse Deflections (E-3
in.)
Distance
from Load
Center (in.): N-
s(c)
Distance
from Load
Center (in.
): E-W(d)
(Seconds)
0
8(N)
12 (S)
18 (S)
24 (N)
36 (N)
13 (W)
18(E)
24 (W)
36(E)
0
12.58
7.10
6.75
3.04
0.68
0.0
9.46
4.37
3.20
0.3
50
12.09
7.43
5.88
2.94
0.77
0.0
8.65
3.97
2.96
0.1
100
12.19
7.05
6.03
2.94
0.72
0.0
9.08
3.87
2.86
0.1
500
11.90
6.77
5.83
2.94
0.82
0.0
8.69
3.72
2.86
0.1
1000
11.60
6.67
5.64
2.90
0.92
0.0
8.50
3.47
2.71
0.2
(a) A rest period equal to four times the load duration was allowed before dynamic testing.
(b) Positive is downward.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.
399

Table E.4: Measured Dynamic Strains at 10,000 lbs. After Different Times of Static Load Application,
Position Number 1, 13.3 C (56 F)
Load^
Longitudinal
Strains
(micro-
strain)^)
Transverse Strains
(micro-strain)
Duration
Distance from
Load Center (in
.): N-S(ci
Distance
from Load Center (in.
): E-W(di
(Seconds)
0BT
OBL^
8(N)
12 (S)
24(S) 36(N)
12(E)
18 (W)
24(E)
36 (W)
0
-156
204
26
59
36 -3
11
78
37
14
50
-159
197
26
60
35 -5
21
82
36
13
100
-142
196
28
58
35 -5
21
77
34
12
500
-146
176
25
55
35 -4
28
78
33
13
1000
-156
175
24
54
33 -5
28
78
32
12
(a) A rest period equal to four times the load duration was allowed before dynamic testing.
(b) Tension is positive.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.
(e) This gage located at the center of load, at the bottom of pavement, in the transverse direction.
(f) This gage located at the center of load, at the bottom of pavement in the longitudinal direction.

Table E.5: Measured Permanent Deflections For Different Times of 10,000-lb. Static Load Application,
Position Humber 1, 6.7 C (44 F)
Load^
Longitudinal Deflections
(E-3 in.)
(b)
Transverse Deflections (E-3
in.)
Duration
Distance
from Load
Center
(in.): N-
Sic)
Distance
from Load
Center (in.
): E-V/i
(Seconds)
0
8(N)
12 (S)
18 (S)
24(N)
36 (N)
13(W)
18(E)
24 (W)
36(E)
50
0.49
0.0
-0.05
-0.05
-0.48
0.0
0.48
0.45
0.0
-0.10
50
0.20
-0.17
0.0
-0.07
-0.12
0.0
0.48
0.30
0.25
0.0
400
0.24
0.12
0.0
0.05
0.10
0.0
0.67
0.40
0.39
0.40
500
0.07
0.0
-0.07
-0.10
-0.10
0.0
0.12
0.35
0.20
0.30
(a) Loads were applied sequentially with a rest period of four times the load duration between loads.
(b) Measured permanent deflections are the residual deflections after the rest period. Positive is
downward.
(c) N is for north, S is for south of load center.
(d) E is for east, VI is for west of load center.

Table E.6: Measured Creep Strains For Different Times of 10,000-lb. Static Load Application,
Position Number 1, 6.7 C (44 F)
Load^
Longitudinal
Strains
(micro-
strain)
(b)
Transverse Strains
(micro-strain)
Duration
Distance from
Load Center (in
.): N-S(c)
Distance
from Load Center (in.
): E-W(d)
(Seconds)
0BT(e)
0BL(f>
8 (M)
12 (S)
24 (S)
36(N)
12(E)
18(H)
24(E)
36(W)
50
-50
26
12
14
1
1
11
11
2
4
50
-25
26
10
8
1
1
6
7
-2
0
400
-48
28
6
0
-3
-8
14
4
-2
5
500
-46
36
4
2
-2
-4
9
0
-1
4
(a) Loads were applied sequentially with a rest period of four times the load duration between loads.
(b) Measured creep strains are the residual strains after the rest period. Tension is positive.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.
(e) This qage located at the center of load, at the bottom of pavement, in the transverse direction.
(f) This gage located at the center of load, at the bottom of pavement in the longitudinal direction.

Table E.7: Measured Dynamic Deflections at 10,000 lbs. After Different Times of Static Load Application,
Position Number 1, 6.7 C (44 F)
Load^
Longitudinal Deflections
(E-3 in.)
(b)
Transverse Deflections (E-3
in.)
Duration
Distance
from Load
Center
(in.): N-
s(c)
Distance
from Load
Center (in.
): E-W
(Seconds)
0
8(N)
12 (S)
18 (S)
24(N)
36 (N)
13 (W)
18(E)
24(W)
36(E)
0
10.87
6.63
5.78
3.33
0.97
0
7.64
3.97
2.87
.2
50
10.18
6.58
5.78
3.28
1.11
0
7.52
3.97
2.96
.3
100
10.67
6.48
5.74
3.33
1.06
0
7.64
3.72
2.96
.3
500
10.67
6.53
5.74
3.26
1.11
0
7.55
3.57
2.96
.3
1000
10.87
6.67
5.74
3.33
1.01
0
7.74
3.87
2.86
.3
(a) A rest period equal to four times the load duration was allowed before dynamic testing.
(b) Positive is downward.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.

Table E.8: Measured Dynamic Strains at 10,000 lbs. After Different Times of Static Load Application,
Position Number 1, 6.7 C (44 F)
Load^
Longitudinal
Strains
(micro-
strain)^)
Transverse Strains
(micro-strain)
Duration
Distance from
Load Center (in
.): N-S{c)
Distance
from Load Center (in.
): E-W(d)
(Seconds)
0BT(e) 0BL(f)
8(N)
12 (S)
24(S) 36(N)
12(E)
18(H)
24(E)
36 (W)
0
-82 144
11
30
25
8
14
54
28
12
50
-80 142
11
30
26
8
15
53
28
13
100
-82 134
10
28
26
9
16
50
28
13
500
-84 138
12
29
26
8
16
53
28
12
1000
-90
13
31
27
9
19
54
27
14
(a) A rest
period equal to four
times
the load
duration
was
allowed before dynamic
testing.
(b) Tension
is positive.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.
(e) This gage located at the center of load, at the bottom of pavement, in the transverse direction.
(f) This gage located at the center of load, at the bottom of pavement in the longitudinal direction.

Table E.9: Measured Permanent Deflections For Different Times of 10,000-lb. Static Load Application,
Position Number 1, 0 C (32 F)
Load^
Longitudinal Deflections
(E-3 in.)(b)
Transverse Deflections (E-3
in.)
Duration
Distance from Load
Center
(in.): N-S^
Distance
from Load
Center (in.
): E-W(di
(Seconds)
0
8(N) 12(S)
18 (S)
24(N) 36 (N)
13 (W)
18(E)
24 (W)
36(E)
50
......
_
0.91
2.28
0.10
1.7
50
0.24
-0.05 -0.10
-0.24
-0.24 0.4
0.24
1.24
0.0
0.7
400
-0.15
0.05 -0.43
-0.53
-0.48 1.1
-0.19
0.84
-0.39
0.2
500
-0.20
-0.71 -0.63
-0.82
-1.11 -.5
0.24
0.60
-0.25
0.0
(a) Loads were applied sequentially with a rest period of four times the load duration between loads.
(b) Measured permanent deflections are the residual deflections after the rest period. Positive is
downward.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.

Table E.10: Measured Creep Strains For Different Times of 10,000-lb. Static Load Application,
Position Number 1, 0 C (32 F)
Load^
Longitudinal
Strains
(micro-
strain)^
Transverse Strains
(micro-strain)
Duration
Distance from
Load Center (in
.): N-S{c)
Distance
from Load Center (in.): E-W^
(Seconds)
OBT^
OBL^
8(N)
12 (S)
24 (S) 36 (N)
12(E)
18 (W)
24(E) 36(W)
50
24
34
14
14
9
7
5
14.5
-2 4
50
11
18
7
11
9
10
7
11.5
-2 5
400
13
40
18
18
16.5
41
18
17
6 15
500
27
41
13.5
18
16
66
15
20
10 14
(a) Loads were applied sequentially with a rest period of four times the load duration between loads.
(b) Measured creep strains are the residual strains after the rest period. Tension is positive.
(c) M is for north, S is for south of load center.
(d) E is for east, W is for west of load center.
(e) This gage located at the center of load, at the bottom of pavement, in the transverse direction.
(f) This gage located at the center of load, at the bottom of pavement in the longitudinal direction.

Table E.ll: Measured Dynamic Deflections at 10,000 lbs. After Different Times of Static Load Application
Position Number 1, 0 C (32 F)
Load^
Duration
Longitudinal Deflections
(E-3 in.)(b)
Transverse Deflections (E-3 in.)
Distance
from Load
Center
(in.): M-S^
Distance
from Load
Center (in.):
: E-W{d)
(Seconds)
0
8(N)
12 (S)
18 (S)
24(H) 36(N)
13 (W)
18(E)
24 (W)
36(E)
0
10.77
6.63
5.78
3.86
1.11 0
7.64
5.21
3.45
0.7
50
10.40
6.67
5.93
3.76
1.21 O
7.55
4.86
3.35
0.9
100
10.33
6.63
5.93
3.62
1.21 .1
7.52
4.56
3.35
0.5
500
10.48
6.82
5.78
3.72
1.35 0
7.64
4.17
3.30
0.5
1000
10.58
6.72
5.78
3.72
1.16 0
7.55
4.07
3.20
0.6
(a) A rest period equal to four times the load duration was allowed before dynamic testing.
(b) Positive is downward.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.

Table E.12: Measured Dynamic Strains at 10,000 lbs. After Different Times of Static Load Application,
Position Number 1, 0 C (32 F)
Load^
Longitudinal
Strains
(micro-
strain)^
Transverse Strains
(micro-strain)
Duration
Distance from
Load Center (in
.): N-S(c)
Distance
from Load Center (in
.): E-Vl(d)
(Seconds)
0BT(e)
0BL(f)
8(N)
12 (S)
24(S) 36(N)
12(E)
18 (W)
24(E)
36 (W)
0
-55
137
6
19
22
11
2
42
24
12
50
-51
129
5
17
21
10
4
39
23
12
100
-50
126
6
17
21
11
4
40
24
12
500
-49
125
5
16
20
10
5
40
23
12
1000
-51
124
7
17
21
10
9
40
22
12
(a) A rest period equal to four times the load duration was allowed before dynamic testing.
(b) Tension is positive.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.
(e) This gage located at the center of load, at the bottom of pavement, in the transverse direction.
(f) This gage located at the center of load, at the bottom of pavement in the longitudinal direction.
408

Table E.13: Measured Permanent Deflections For Different Times of 10,000-lb. Static Load Application,
Position Number 2, 6.7 C (44 F)
Load^a)
Longitudinal Deflections
(E-3 in
.)(b)
Transverse Deflections (E-3
in.)
Duration
Distance
from Load
Center
(in.):
N-S(c)
Distance
from Load
Center (in
.): E-W{di
(Seconds)
0
8(S)
12(N)
16 (S)
24(N)
36 (S)
13 (W)
18(E)
24 (W)
36(E)
50
0.98
-0.24
0.72
-0.19
-0.24
0.2
0.53
0.25
-0.25
-0.1
50
1.27
-0.24
0.63
-0.24
0.12
0.0
0.96
0.37
0.0
0.0
400
1.10
0.90
0.60
-0.12
-0.48
-0.1
0.81
0.35
-0.39
-0.1
500
0.61
-0.19
0.36
0.0
-.05
-0.1
0.76
0.0
0.12
-0.3
(a) Loads were applied sequentially with a rest period of four times the load duration between loads.
(b) Measured permanent deflections are the residual deflections after the rest period. Positive is
downward.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.

Table E.14: Measured Creep Strains For Different Times of 10,000-lb. Static Load Application,
Position Number 2, 6.7 C (44 F)
Load^
Longitudinal
Strains
(micro-
strain) ^)
Transverse Strains
(micro-strain)
Duration
Distance from
Load Center (in
.): N-S(ci
Distance
from Load Center (in.
): E-W(d>
(Seconds)
0
8 (S)
12(N)
16 (S)
24(N) 36(S)
12 (W)
18(E)
24 (W)
36(E)
50
-68
10
22
39
-5 -2
24
27
7
7
50
-25
2
3
6
4 -1
12
12
7
1
400
-44
13
6
12
-10 -5
40
44
1
12
500
-14
13
-6
5
-13 -3
34
0
7
10
(a) Loads were applied sequentially with a rest period of four times the load duration between loads.
(b) Measured creep strains are the residual strains after the rest period. Tension is positive.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.

Table E.15: Measured Dynamic Deflections at 10,000 lbs. After Different Times of Static Load Application,
Position Dumber 2, 6.7 C (44 F)
Load(a)
Longitudinal Deflections
(E-3 in.
) (b)
Transverse Deflections (E-3 in.)
Duration
Distance
from Load
Center
(in.): D
-S(c)
Distance
from Load
Center (in.):
: E-W{d)
(Seconds)
0 8 (S)
12(D)
16 (S)
24(D)
36 (S)
13 (VI)
18(E)
24(W)
36(E)
0
8.08 3.08
5.54
0.24
1.45
-0.2
7.17
3.97
1.97
0.7
50
8.42 3.08
2.53
0.24
1.45
-0.1
7.40
3.72
2.22
0.4
100
8.20 2.98
5.30
0.19
1.21
0.0
7.17
3.72
1.97
0.3
500
8.45 3.08
5.59
-0.05
1.21
0.0
7.26
3.67
2.17
0.0
1000
8.81 3.22
5.54
-0.22
1.59
0.0
7.21
3.77
1.97
0.3
(a) A rest period equal to four times the load duration was allowed before dynamic testing.
(b) Positive is downward.
(c) D is for north, S is for south of load center.
(d) E is for east, W is for west of load center.

Table E.16: Measured Dynamic Strains at 10,000 lbs. After Different Times of Static Load Application,
Position Number 2, 6.7 C (44 F)
Load^
Longitudinal
Strains
(micro-
strain)^
Transverse Strains
(micro-strain)
Duration
Distance from
Load Center (in
.): N-S(c)
Distance
from Load Center (in.
): E-W(d)
(Seconds)
0
8 {S)
12 (N)
16 (S}
24(M) 36(S)
12(W)
18(E)
24 (W)
36(E)
0
-206
-17
38
46
29
12
15
56
37
13
50
-206
-16
39
47
30
11
17
57
36
12
100
-208
-15
38
48
28
12
17
57
37
13
500
-312
-13
37
49
29
12
19
59
38
14
1000
-232
-11
40
50
30
11
21
62
39
13
(a) A rest
period equal to four
times
the load
duration
was
allowed before dynamic
testing.
(b) Tension is positive.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.

Table E.17: Measured Permanent Deflections For Different Times of 10,000-lb. Static Load Application,
Position Number 2, 0 C (32 F)
Load^
Longitudinal Deflections
(E-3 in.)
(b)
Transverse Deflections (E-3 in.)
Duration
Distance
from Load
Center
(in.): N-
S(c)
Distance
from Load
Center (in.):
: E-W{di
(Seconds)
0
8(S)
12(N)
16 (S)
24(N)
36 (S)
13(W)
18(E)
24 (W)
36(E)
50
3.55
4.26
2.17
3.38
1.21
0.0
2.39
3.22
1.48
8
50
1.35
1.42
0.36
1.69
0.0
0.0
.48
2.98
0.49
6.4
400
0.49
2.84
-0.12
5.80
-0.60
0.0
1.07
7.19
0.99
13.6
500
0.24
2.72
-0.12
8.46
0.12
-0.2
1.31
1.98
1.49
2.5
(a) Loads were applied sequentially with a rest period of four times the load duration between loads.
(b) Measured permanent deflections are the residual deflections after the rest period. Positive is
downward.
(c) N is for north, S is for south of load center.
(d) E is for east, W is for west of load center.

Table E.18:
Measured Creep Strains
Position Humber 2, 0 C
For Different Times of 10
(32 F)
,000-lb. Static
Load Application,
Load^3^
Longitudinal
Strains (micro-strain)^
Transverse
Strains
(micro-strain)
Duration
Distance from
Load
Center (in.): H-S^
Distance from
Load Center (in.):
E-W(c0
(Seconds)
0
8 (S)
12(H)
16(S) 24(N) 36(S)
12(W) 18(E)
24(W)
36(E)
50
-18
-22
-11
23 -0.4 12
23.5
4
14
9.5
50
8
-19
2
14 -3 8
14.5
6
10.5
9
400
35
-41
30