EVALUATION OF LAYER MODULI IN FLEXIBLE PAVEMENT SYSTEMS
USING NONDESTRUCTIVE AND PENETRATION TESTING METHODS
By
KWASI BADUTWENEBOAH
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
UNIVERSITY OF FLORIDA
1987
DEDICATED TO MY FAMILY, ESPECIALLY MY SISTER, ABENA KYEM
AND GRANDFATHER, ATTA KWAME, FOR THEIR CONTINUAL PRAYERS,
ENCOURAGEMENT AND SUPPORT DURING THE COURSE OF MY EDUCATION.
"A MIND IS A TERRIBLE THING TO WASTE."
ACKNOWLEDGMENTS
I would like to express my gratitude to Dr. Byron E. Ruth, chairman
of my supervisory committee, for his guidance, encouragement and con
structive criticisms in undertaking this research work. I am also
grateful to Drs. F. C. Townsend, J. L. Davidson, M. Tia, J. L. Eades,
and D. P. Spangler for serving on my graduate supervisory committee. I
consider myself honored to have had these distinguished men on my com
mittee.
I also owe sincere thanks to Dr. J. H. Schaub, chairman of the
civil engineering department, for the many times he gave help during the
course of my studies here, especially in my obtaining the grant award to
participate in the 1986 APWA Congress in New Orleans, Louisiana.
I would like to express my appreciation to the Florida Department
of Transportation (FDOT) for providing the financial support, testing
facilities, materials, and personnel that made this research possible.
I would like to thank the many individuals at the Pavement Evaluation
and Bituminous Materials Research sections of the Bureau of Materials
and Research at FDOT who contributed significantly to the completion of
this work. In particular, I am indebted to Messrs. W. G. Miley, Ron
McNamara, Ed Leitner, Don Bagwell, and John Purcell for giving so
generously of their time.
A very special word of thanks goes to Dr. David Bloomquist for his
significant contributions in conducting the in situ penetration tests
and for his helpful suggestions, advice, and friendship. The assistance
of Mr. Ed Dobson in the field work is also appreciated.
I would also like to thank Dr. F. Balduzzi of the Institute of
Foundation Engineering and Soil Mechanics of the Federal Institute of
Technology, Zurich, Switzerland, for inviting me to the institute, and
consequently helping me develop the interest to pursue active research
and further studies.
Last, but far from being the least, I would like to thank
Ms. Candace Leggett for her expertise and diligent skill in typing this
dissertation.
TABLE OF CONTENTS
Page
ACKNOWLEDGMENTS................................... ................. ii
LIST OF TABLES................. ................... ...... ..... ...... ix
LIST OF FIGURES.......................................... E..... xiii
ABSTRACT............o ......... .... o ....... ...................... xx
CHAPTER
1 INTRODUCTION ..... .......... ........................ .. 1
1.1 Background........ ... .... ........ .................... 1
1.2 Study Objectives ... .... .............. ... ... ...... .... ... 3
1.3 Scope of Study................................. ... .. 4
2 LITERATURE REVIEW .. ....................... ........... ... ... 6
2.1 Introduction......................... .............. 6
2.2 ElasticLayer Theory................... ................ 7
2.2.1 General............ ....... ...... ........... 7
2.2.2 OneLayer System................................ 8
2.2.3 TwoLayer System............................. 9
2.2.4 ThreeLayer System........................... 10
2.2.5 Multilayered or NLayered Systems.............. 11
2.3 Material Characterization Methods...................... 13
2.3.1 General......................................... 13
2.3.2 StateoftheArt Nondestructive Testing......... 15
2.3.2.1 General............................... 15
2.3.2.2 Static Deflection Procedures........... 16
2.3.2.3 SteadyState Dynamic Force
Deflection................................ 21
2.3.2.4 Dynamic Impact Load Response........... 24
2.3.2.5 Wave Propagation Technique............. 29
2.3.3 NDT DataInterpretation Methods................ 32
2.3.3.1 General ...... ...................... ....... 32
2.3.3.2 Direct Solutions ...................... 36
2.3.3.3 BackCalculation Methods............... 41
2.3.4 Other In Situ Methods...............t......... 44
Page
2.4 Factors Affecting Modulus of PavementSubgrade
Materials ................................................... 47
2.4.1 Introduction..................................... 47
2.4.2 Temperature.................................. .. 48
2.4.3 Stress Dependency............................. 49
3 EQUIPMENT AND FACILITIES.................................... 55
3.1 Description of Dynaflect Test System.................... 56
3.1.1 Description of Equipment....................... 56
3.1.2 Calibration.................... ............... 59
3.1.3 Testing Procedure .............................. 59
3.1.4 Limitations...................................... 60
3.2 Description of the Falling Weight Deflectometer
Testing System.......................................... 60
3.2.1 The 8002 FWD..................................... 61
3.2.2 The 8600 System Processor........................ 61
3.2.3 The HP85 Computer.............................. 63
3.2.4 Testing Procedure............................... 63
3.2.5 Advantages...... ... .............................. 64
3.3 BISAR Computer Program ................................. 64
3.4 Description of Cone Penetration Test Equipment.......... 65
3.5 Marchetti Dilatometer Test Equipment.................... 67
3.6 Plate Bearing Test...................................... 70
4 SIMULATION AND ANALYSES OF NDT DEFLECTION DATA.............. 71
4.1 BISAR Simulation Study................................ 71
4.1.1 General .......................................... 71
4.1.2 Dynaflect Sensor Spacing........................ 74
4.1.3 FWD Sensor Spacing............................... 74
4.2 Sensitivity Analysis of Theretical NDT
Deflection Basins....................................... 76
4.2.1 Parametric Study............................... 76
4.2.2 Summary of Sensitivity Analysis.................. 88
4.3 Development of Layer Moduli Prediction Equations........ 91
4.3.1 General .......................................... 91
4.3.2 Development of Dynaflect Prediction Equations.... 95
4.3.2.1 Prediction Equations for E .............95
4.3.2.2 Prediction Equation for E2 for
Thin Pavements............... .......... 103
4.3.2.3 Prediction Equations for E3............. 105
4.3.2.4 Prediction Equations for E4 ............ 108
4.3.3 Development of FWD Prediction Equations.......... 110
4.3.3.1 Prediction Equations for E ............. 110
4.3.3.2 Prediction Equations for E ........... 111
4.3.3.3 Prediction Equations for E ............. 113
4.3.3.4 Prediction Equations for E ............. 114
Page
4.4 Accuracy and Reliability of NDT Prediction Equations.... 119
4.4.1 Prediction Accuracy of Dynaflect Equations....... 119
4.4.1.1 Asphalt Concrete Modulus, E ............ 119
4.4.1.2 Base Course Modulus, E2,
for Thin Pavements...................... 123
4.4.1.3 Stabilized Subgrade Modulus, E ......... 123
4.4.1.4 Subgrade Modulus, E .................... 125
4.4.2 Prediction Accuracy of FWD Equations............. 127
4.4.2.1 Asphalt Concrete Modulus, E ............ 127
4.4.2.2 Base Course Modulus, E2.............. 129
4.4.2.3 Stabilized Subgrade Modulus, E ......... 132
4.4.2.4 Subgrade Modulus, E ................... 134
5 TESTING PROGRAM............. ................. .. ......... 136
5.1 Introduction..... o ..................................... 136
5.2 Location and Characteristics of Test Pavements.......... 137
5.3 Description of Testingn Procedures..................... 140
5.3.1 General .......... ...................... ....... 140
5.3.2 Dynaflect Tests........................................... ....... 142
5.3.3 Falling Weight Deflectometer Tests............... 142
5.3.4 Cone Penetration Tests.......................... 144
5.3.5 Dilatometer Tests..... .......................... 145
5.3.6 Plate Loading Tests................................. 145
5.3.7 Asphalt Rheology Tests.......................... 147
5.3.8 Temperature Measurements........................ 151
6 ANALYSES OF FIELD MEASURED NDT DATA......................... 153
6.1 General ... ... .................. .. .. ............... ... 153
6.2 Linearity of LoadDeflection Response.................. 153
6.3 Prediction of Layer Moduli.............................. 169
6.3.1 General.......................................... 169
6.3.2 Dynaflect Layer Moduli Predictions............... 171
6.3.3 FWD Prediction of Layer Moduli..................... 174
6.4 Estimation of El from Asphalt Rheology Data............. 178
6.5 Modeling of Test Pavements.............................. 181
6.5.1 General .................... ............. ..... ... 181
6.5.2 Tuning of Dynaflect Deflection Basins........... 182
6.5.3 Tuning of FWD Deflection Basins.................. 205
6.5.4 Nonuniqueness of NDT Backcalculation
of Layer Moduli ................................. 231
6.5.5 Effect of Stress Dependency.................... 233
6.6 Comparison of NDT Devices.............................. 236
6.6.1 Comparison of Deflection Basins.................. 238
6.6.2 Comparison of Layer Moduli...................... 253
6.7 Analyses of Tuned NDT Data.............................. 265
6.7.1 General ...................o .... ...... ....... 265
6.7.2 Analysis of Dynaflect Tuned Data................. 266
Page
6.7.2.1 Comparison of Measured and
Predicted Deflections................... 266
6.7.2.2 Development of Simplified Layer
Moduli Equations........................ 271
6.7.3 Analysis of FWD Tuned Data....................... 279
6.7.3.1 Comparison of Measured and
Predicted Deflections................... 279
6.7.3.2 Development of Prediction Equations..... 282
7 INTERPRETATION OF IN SITU PENETRATION TESTS.................. 288
7.1 General ................. ............................... 288
7.2 Soil Profiling and Identification....................... 289
7.3 Correlation Between ED and qc........................... 292
7.4 Evaluation of Resilient Moduli for Pavement Layers...... 302
7.4.1 General .......................................... 302
7.4.2 Correlation of Resilient Moduli with
Cone Resistance.................................. 304
7.4.3 Correlation of Resilient Moduli with
Dilatometer Modulus.............................. 308
7.5 Variation of Subgrade Stiffness with Depth.............. 312
8 PAVEMENT STRESS ANALYSES..................................... 316
8.1 General ................................................. 316
8.2 ShortTerm Load Induced Stress Analysis................. 318
8.2.1 Design Parameters................................ 318
8.2.2 Comparison of Pavement Response and
Material Properties.............................. 319
8.2.3 Summary ....................................... 330
9 CONCLUSIONS AND RECOMMENDATIONS.............................. 331
9.1 Conclusions................ ..... ...................... 331
9.2 Recommendations ........................ ......... 334
APPENDICES
A FIELD DYNAFLECT TEST RESULTS................................. 338
B FIELD FWD TEST RESULTS ...................................... 354
C COMPUTER PRINTOUT OF CPT RESULTS............................ 379
D COMPUTER PRINTOUT OF DMT RESULTS............................ 406
E RECOVERED ASPHALT RHEOLOGY TEST RESULTS...................... 432
F RECOMMENDED TESTING AND ANALYSIS PROCEDURES FOR THE
MODIFIED DYNAFLECT TESTING SYSTEM............................ 456
G PARTIAL LISTING OF DELMAPS1 COMPUTER PROGRAM................. 463
viii
Page
REFERENCES.......................................................... 478
BIOGRAPHICAL SKETCH................................................. 491
LIST OF TABLES
Table Page
2.1 Summary of Deflection Basin Parameters...................... 35
2.2 Summary of Computer Programs for Evaluation of
Flexible Pavement Moduli from NDT Devices.................... 42
4.1 Range of Pavement Layer Properties........................... 73
4.2 Sensitivity Analysis of FWD Deflections for ti = 3.0 in. .... 86
4.3 Sensitivity Analysis of FWD Deflections for t t ,
and t3....................................................... 87
4.4 Sensitivity Analysis of FWD Deflections for E =
600 ksi and ti = 3.0 in. ....... ...... ......... .......... 89
4.5 Pavements with Dynaflect E1 Predictions Having More
Than 10 Percent Error................ ....... ............. 121
4.6 Pavements with Dynaflect E2 Predictions Having More
Than 10 Percent Error.......................... ............ 124
4.7 Comparison of Actual and Predicted E3 Values for
Varying t ................................................... 126
4.8 Prediction Accuracy of Equation 4.18Error
Distribution as a Function of t ............................. 128
4.9 Prediction Accuracy of Equation 4.19Error
Distribution as a Function of t ............................ 130
4.10 Pavements with E1 Predictions Having 15 Percent
or More Error......................... .............. ....... 131
4.11 Pavements with E2 Predictions Having 20 Percent
or More Errors............................................... 133
5.1 Characteristics of Test Pavements.......................... 138
5.2 Summary of Tests Performed on Test Pavements................. 143
5.3 Plate Loading Test Results................................... 148
5.4 ViscosityTemperature Relationships of Recovered
Asphalt from Test Pavements.................................. 149
5.5 Temperature Measurements of Test Pavement Sections........... 152
6.1 Typical Dynaflect Deflection Data from Test Sections......... 170
6.2 Typical FWD Data from Test Sections......................... 172
6.3 Layer Moduli Using Dynaflect Prediction Equations............ 173
6.4 Layer Moduli Using FWD Prediction Equations.................. 176
6.5 Comparison Between NDT and Rheology Predictions
of Asphalt Concrete Modulus................................. 180
6.6 Comparison of Field Measured and BISAR Predicted
Dynaflect Deflections ....................................... 183
6.7 Dynaflect Tuned Layer Moduli for Test Sections............... 203
6.8 Predicted Deflections from Tuned Layer Moduli................ 204
6.9 Comparison of Field Measured and BISAR Predicted
FWD Deflections ..................................... ...... 206
6.10 FWD Tuned Layer Moduli for Test Sections..................... 227
6.11 Predicted FWD Deflections from Tuned Layer Moduli............ 228
6.12 Comparison Between ReCalculated and Tuned FWD
Layer Moduli............................................... 230
6.13 Illustration of Nonuniqueness of Backcalculation of
Layer Moduli from NDT Deflection Basin...................... 232
6.14 Comparison of Deflections Measured at Different Load
Levels .................................. ... ........ ........ 235
6.15 Comparison Between Tuned Layer Moduli and Applied
FWD Load............. ...................................... 237
6.16 Comparison of the Asphalt Concrete Modulus for
the Test Sections.............................................. 254
6.17 Comparison of the Base Course Modulus for the
Test Sections................................................ 258
6.18 Comparison of the Subbase Modulus for the
Test Sections................................................ 259
6.19 Comparison of the Subgrade Modulus for the
Test Sections ................................................ 260
6.20 Ratios of Dynaflect Moduli to FWD Moduli for Test Sections... 262
6.21 Correlation Between Measured and Predicted FWD
(9kip Load) Deflections ......... ..... ........... ....... 281
7.1 Relationship Between ED and qc for Selected Test
Sections in Florida....................................... 298
7.2 Correlation of NOT Tuned Base Course Modulus (E2)
to Cone Resistance .......................................... 305
7.3 Correlation of NDT Tuned Subbase Modulus (E3) to
Cone Resistance......................................... ..... 306
7.4 Correlation of NDT Tuned Subgrade Modulus (E4) to
Cone Resistance............................... ..... ............ 307
7.5 Relationship Between Resilient Modulus, ER and
Cone Resistance, qc......... ...... ... ..... ................... 308
7.6 Correlation of NDT Tuned Subbase Modulus to
Dilatometer Modulus.......................................... 309
7.7 Correlation of NDT Tuned Subgrade Modulus to
Dilatometer Modulus ................... ................. ..... 310
7.8 Relationship Between Resilient Modulus, ER and
Dilatometer Modulus, ED...................................... 311
7.9 Effect of Varying Subgrade Stiffness on Dynaflect
Deflections on SR 26A...................................... 314
8.1 Material Properties and Results of Stress Analysis
for SR 26B (Gilchrist County); a) Input Parameters
for BISAR; b) Pavement Stress Analysis....................... 320
8.2 Material Properties and Results of Stress Analysis
for SR 24 (Alachua County); a) Input Parameters
for BISAR; b) Pavement Stress Analysis....................... 321
8.3 Material Properties and Results of Stress Analysis
for US 441 (Columbia County); a) Input Parameters
for BISAR; b) Pavement Stress Analysis....................... 322
8.4 Material Properties and Results of Stress Analysis
for SR 15C (Martin County); a) Input Parameters
for BISAR; b) Pavement Stress Analysis....................... 323
8.5 Material Properties and Results of Stress Analysis
for SR 80 (Palm Beach County); a) Input Parameters
for BISAR; b) Pavement Stress Analysis....................... 324
8.6 Summary of Pavement Stress Analysis at Low Temperatures...... 326
8.7 Effect of Increased Base Course Modulus on Pavement
Response on SR 80; a) Input Parameters for BISAR;
b) Pavement Stress Analysis.................................. 329
xiii
LIST OF FIGURES
Figure Page
2.1 WellDesigned Pavement Deflection History Curve.............. 18
2.2 Typical Annual Deflection History for a Flexible Pavement.... 20
2.3 Typical Output of a Dynamic Force Generator.................. .. 22
2.4 Schematic Diagram of Impulse LoadResponse Equipment......... 25
2.5 Characteristic Shape of Load Impulse......................... 26
2.6 Comparison of Pavement Response from FWD and
MovingWheel Loads. a) Surface Deflections;
b) Vertical Subgrade Strains................................. 28
2.7 Empirical Interpretation of Dynaflect Deflection
Basin. a) Basin Parameters; b) Criteria...................... 34
2.8 Dynaflect Fifth Sensor DeflectionSubgrade
Modulus Relationship.................. ........ ......... 40
2.9 Temperature Prediction Graphs. a) Pavements More
Than 2 in. Thick; b) Pavements Equal to or Less
Than 2 in. Thick............................................ 50
3.1 Typical Dynamic Force Output Signal of Dynaflect............ 57
3.2 Configuration of Dynaflect Load Wheels and Geophones
in Operating Position........................................ 58
3.3 Schematic of FWD LoadGeophone Configuration and
Deflection Basin ............................................ 62
3.4 Schematic of Marchetti Dilatometer Test Equipment............ 69
4.1 FourLayer Flexible Pavement System Model.................... 72
4.2 Dynaflect Modified Geophone Positions........................ 75
4.3 Typical FourLayer System Used for the Sensitivity
Analysis ..................................................... 77
4.4 Effect of Change of El on Theoretical FWD (9kip Load)
Deflection Basin................................... ........ 79
4.5 Effect of Change of E2 on Theoretical FWD (9kip Load)
Deflection Basin ............................................. 80
4.6 Effect of Change of E3 on Theoretical FWD (9kip Load)
Deflection Basin ........................................... 81
4.7 Effect of Change of E4 on Theoretical FWD (9kip Load)
Deflection Basin............................................. 82
4.8 Effect of Change of tI on Theoretical FWD (9kip Load)
Deflection Basin............................................. 83
4.9 Effect of Change of t2 on Theoretical FWD (9kip Load)
Deflection Basin............................................. 84
4.10 Effect of Change of t3 on Theoretical FWD (9kip Load)
Deflection Basin............................................. 85
4.11 Effect of Varying Subgrade Thickness on Theoretical FWD
(9kip Load) Deflection Basin................................ 90
4.12 Variation in Dynaflect Deflection Basin with Varying
E2 and E3 Values with t, = 3.0 in. ......................... 93
4.13 Variation in Dynaflect Deflection Basin with Varying
E and E4 Values with t, = 3.0 in. .......................... 94
4.14 Relationship Between El and D1 D4 for t, = 3.0 in ....... 96
4.15 Relationship Between E1 and D1 D4 for t, = 6.0 in. ........ 97
4.16 Relationship Between El and Di D4 for tI = 8.0 in. ........ 98
4.17 Variation of Ki with ti for Different E2 Values.............. 100
4.18 Variation of K2 with t for Different E2 Values.............. 101
4.19 Relationship Between E2 and Di D4 for t, = 1.0 in. ........ 104
4.20 Comparison of E4 Prediction Equations Using Modified
Sensor 10 Deflections .................................... 109
4.21 Relationship Between E4 and FWD Deflections for Fixed
E E and E Values with t1 = 3.0 in. .................... 116
4.22 Relationship Between E4 and FWD Deflections for Fixed
El, E2, and E3 Values with t1 = 6.0 in. ..................... 117
5.1 Location of Test Pavements in the State of Florida........... 139
5.2 Layout of Field Tests Conducted on Test Pavements........... 141
6.1 Surface Deflection as
6.2
6.3
6.4
6.5
6.6
6.7
6.8
6.9
6.10
6.11
6.12
6.13
6.14
Surface
Surface
Surface
Surface
Surface
Surface
Surface
Surface
Surface
Surface
Surface
Surface
Surface
Load on SR 26A........... 155
Deflection
Deflection
Deflection
Deflection
Deflection
Deflection
Deflection
Deflection
Deflection
Deflection
Deflection
Deflection
Deflection
as a Function of Load on
Function
Function
Function
Function
Function
Function
Function
Function
Function
Function
Function
Function
Function
6.15 Comparison of Measured and Predicted Dynaflect
Deflections for SR 26AM.P. 11.912.........................
6.16 Comparison of Measured and Predicted Dynaflect
Deflections for SR 26BM.P. 11.205..........................
6.17 Comparison of Measured and Predicted Dynaflect
Deflections for SR 26CM.P. 10.168..........................
6.18 Comparison of Measured and Predicted Dynaflect
Deflections for SR 24M.P. 11.112..........................
6.19 Comparison of Measured and Predicted Dynaflect
Deflections for US 301M.P. 11.112.........................
6.20 Comparison of Measured and Predicted Dynaflect
Deflections for I10AM.P. 14.062...........................
6.21 Comparison of Measured and Predicted Dynaflect
Deflections for I10BM.P. 2.703............................
6.22 Comparison of Measured and Predicted Dynaflect
Deflections for I10CM.P. 32.071...........................
Load
Load
Load
Load
Load
Load
Load
Load
Load
Load
Load
Load
SR 26C...........
SR 24...........
US 301...........
US 441............
I10A............
I10B ............
I10C.............
SR 715............
SR 12............
SR 15C ...........
SR 26B............
SR 15A............
SR 15B...........
156
157
158
159
160
161
162
163
164
165
166
167
168
6.23 Comparison of Measured and Predicted Dynaflect
Deflections for SR15AM.P. 6.549..........................
6.24 Comparison of Measured and Predicted Dynaflect
Deflections for SR 15BM.P. 4.811...........................
6.25 Comparison of Measured and Predicted Dynaflect
Deflections for SR 715M.P. 4.722............................
6.26 Comparison of Measured and Predicted Dynaflect
Deflections for SR 715M.P. 4.720............................
6.27 Comparison of Measured and Predicted Dynaflect
Deflections for SR 12M.P. 1.485.............................
6.28 Comparison of Measured and Predicted Dynaflect
Deflections for SR 80Section 1............................
6.29 Comparison of Measured and Predicted Dynaflect
Deflections for SR 80Section 2........... .................
6.30 Comparison of Measured and Predicted Dynaflect
Deflections for SR 15CM.P. 0.055...........................
6.31 Comparison of Measured and Predicted Dynaflect
Deflections for SR 15CM.P. 0.065...........................
6.32 Comparison of Measured and
(Normalized to 1kip Load)
6.33 Comparison of Measured and
(Normalized to 1kip Load)
6.34 Comparison of Measured and
(Normalized to 1kip Load)
6.35 Comparison of Measured and
(Normalized to 1kip Load)
6.36 Comparison of Measured and
(Normalized to 1kip Load)
6.37 Comparison of Measured and
(Normalized to 1kip Load)
6.38 Comparison of Measured and
(Normalized to 1kip Load)
6.39 Comparison of Measured and
(Normalized to 1kip Load)
6.40 Comparison of Measured and
(Normalized to 1kip Load)
Predicted FWD Deflections
for SR 26AM.P. 11.912........... 208
Predicted FWD Deflections
for SR 26BM.P. 11.205........... 209
Predicted FWD Deflections
for SR 26CM.P. 10.168........... 210
Predicted FWD Deflections
for SR 26CM.P. 10.166........... 211
Predicted FWD Deflections
for SR 24M.P. 11.112............ 212
Predicted FWD Deflections
for US 301M.P. 21.585........... 213
Predicted FWD Deflections
for US 441M.P. 1.236........... 214
Predicted FWD Deflections
for I1OAM.P. 14.062............ 215
Predicted FWD Deflections
for I1OBM.P. 2.703............. 216
xvii
194
195
196
197
198
199
200
201
202
6.41 Comparison of Measured and
(Normalized to 1kip Load)
6.42 Comparison of Measured and
(Normalized to 1kip Load)
6.43 Comparison of Measured and
(Normalized to 1kip Load)
6.44 Comparison of Measured and
(Normalized to 1kip Load)
6.45 Comparison of Measured and
(Normalized to 1kip Load)
6.46 Comparison of Measured and
(Normalized to 1kip Load)
6.47 Comparison of Measured and
(Normalized to 1kip Load)
6.48 Comparison of Measured and
(Normalized to 1kip Load)
6.49 Comparison of Measured and
(Normalized to 1kip Load)
Predicted FWD Deflections
for I10CM.P. 32.071............ 217
Predicted FWD Deflections
for SR 15AM.P. 6.546............ 218
Predicted FWD Deflections
for SR 15AM.P. 6.549............ 219
Predicted FWD Deflections
for SR 15BM.P. 4.811............ 220
Predicted FWD Deflections
for SR 715M.P. 4.722............ 221
Predicted FWD Deflections
for SR 715M.P. 4.720........... 222
Predicted FWD Deflections
for SR 12M.P. 1.485............. 223
Predicted FWD Deflections
for SR 15CM.P. 0.055............ 224
Predicted FWD Deflections
for SR 15CM.P. 0.065........... 225
6.50 Comparison of Measured NDT Deflection Basins on SR 26C
M.P. 10.166.................................................. 239
6.51 Comparison of Measured NDT Deflection Basins on US 301
M.P. 21.585.................................................. 240
6.52 Comparison of Measured NDT Deflection Basins on US 441
M.P. 1.237................................................... 241
6.53 Comparison of Measured NDT Deflection Basins on SR 12
M.P. 1.485 ..................... ............ ........ ....... 242
6.54 Comparison of Measured NDT Deflection Basins on SR 26B
M.P. 11.205.................................................. 243
6.55 Comparison of Measured NDT Deflection Basins on SR 15A
M.P. 6.549................................................... 244
6.56 Comparison of Measured NDT Deflection Basins on SR 715
M.P. 4.722................................................... 245
6.57 Comparison of Measured NDT Deflection Basins on SR 26A
M.P. 11.912...................... ......... .................... 246
6.58 Comparison of Measured NDT Deflection Basins on SR 24
M.P. 11.112....................................................... 247
xviii
6.59 Comparison of Measured NDT Deflection Basins on I10A
M.P. 14.062.................................................. 248
6.60 Comparison of Measured NDT Deflection Basins on I10B
M.P. 2.703................................................... 249
6.61 Comparison of Measured NDT Deflection Basins on I1OC
M.P. 32.071................................................. 250
6.62 Comparison of Measured NDT Deflection Basins on SR 15B
M.P. 4.811.................................................... 251
6.63 Comparison of Measured NDT Deflection Basins on SR 15C
M.P. 0.055....... ......... ............................... 252
6.64 Relationship Between Asphalt Concrete Modulus, E and
Mean Pavement Temperature.................................... 256
6.65 Comparison of Dynaflect and FWD Tuned Layer Moduli........... 261
6.66 Relationship Between Predicted and Measured Dynaflect
Modified Sensor 1 Deflections.............................. 267
6.67 Relationship Between Predicted and Measured Dynaflect
Modified Sensor 4 Deflections.............................. 268
6.68 Relationship Between Predicted and Measured Dynaflect
Modified Sensor 7 Deflections............................... 269
6.69 Relationship Between Predicted and Measured Dynaflect
Modified Sensor 10 Deflections .............................. 270
6.70 Relationship Between E12 (Using Equation 6.8) and D D .... 274
6.71
6.72
6.73
6.74
6.75
7.1
7.2
7.3
7.4
7.5
Relationship Between E12 (Using Equation 6.9) and D1 D ....
Relationship Between E3 and D4 D ..........................
Relationship Between E4 and D o..............................
Simplified Flow Chart of DELMAPS1 Program ...................
Relationship Between E4 and FWD D6 and D7....................
Variation of qc and FR with Depth on SR 12...................
Variation of ED and KD with Depth on SR 12...................
Variation of qc and ED with Depth on SR 26A..................
Variation of qc and ED with Depth on SR 26C..................
Variation of qc and ED with Depth on US 301...................
275
276
277
280
287
290
291
293
294
295
xix
7.6 Variation of qc and ED with Depth on US 441.................. 296
7.7 Variation of qc and ED with Depth on SR 12................... 297
7.8 Correlation of ED with qc.................................. 300
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
EVALUATION OF LAYER MODULI IN FLEXIBLE PAVEMENT SYSTEMS
USING NONDESTRUCTIVE AND PENETRATION TESTING METHODS
By
Kwasi BaduTweneboah
December 1987
Chairman: Byron E. Ruth
Major Department: Civil Engineering
A research study was conducted to develop procedures for the eval
uation of layer moduli in flexible pavement systems using in situ non
destructive (NDT) and penetration tests. The Bitumen Structure Analysis
in Roads (BISAR) elastic layer computer program was used to simulate
Dynaflect and Falling Weight Deflectometer (FWD) loaddeflection
response for typical flexible pavements in the state of Florida. A
field testing program consisting of Dynaflect, FWD, cone penetration,
Marchetti Dilatometer, and plate bearing tests was conducted on fifteen
pavement sections in the state of Florida. Cores of the asphalt con
crete pavement were collected for laboratory lowtemperature rheology
tests. This provided a reliable and effective method of predicting the
asphalt concrete modulus.
A modified Dynaflect geophone configuration and simplified layer
moduli prediction equations which allow a layerbylayer analysis of
Dynaflect deflection measurements were developed. Multiple linear
regression equations with relatively good prediction accuracy were
obtained from analyses of FWD deflection data. Different layer moduli
values were obtained from the Dynaflect and FWD deflection basins for
the various test sections.
The penetration tests provided means for identifying the soils and
also assessing the variability in stratigraphy of the test sites. Good
correlations between cone resistance, qc, and dilatometer modulus, ED,
for sandy soils and soils above the water table were obtained. Pavement
layer moduli determined from NDT data were regressed to qc and ED for
the various layers in the pavement. The correlations were better with
qc than with ED, and also for the base and subbase layers than the
variable subgrade layer. The penetration tests can be used to supple
ment NDT evaluation of pavements especially in locating zones of weak
ness in the pavement or underlying subgrade soils.
The effects of moisture, temperature, and the properties of the
asphalt binder on the performance and response characteristics of
flexible pavements were demonstrated using shortterm loadinduced
stress analyses on five of the test pavements.
xxii
CHAPTER 1
INTRODUCTION
1.1 Background
In recent years, the use of layered elastic theory to evaluate and
design highway and airfield pavements has become increasingly popular
visavis existing empirical methods. The elastic layer approach, also
called mechanistic analysis, has obvious advantages over empirical
methods which are based on the correlation between the maximum deflec
tion under a load and pavement performance. It allows a rational eval
uation of the mechanical properties of the materials in the pavement
structure.
An essential part of the mechanistic process is determining real
istic elastic modulus values for the various layers in the pavement
structure. Current methods to determine the modulus of pavement
materials include various laboratory testing procedures, destructive
field tests, and in situ nondestructive tests (NDT). The problems
associated with the simulation of in situ conditions such as moisture
content, density, loading history and rate of loading of the pavement in
the laboratory are well known and recognized. Destructive field tests,
such as the California bearing ratio (CBR) and plate tests are expen
sive, timeconsuming, and generally involve trenching the pavement,
which has to be subsequently repaired.
Nondestructive testing generally involves applying some type of
dynamic load or shock waves to the surface of the pavement and measuring
2
the response of the pavement. Among such methods are various seismic
techniques and surface dynamic loading tests. The basic concept behind
seismic or wave propagation techniques is the use of vibratory loads and
the resulting identification and measurement of the waves that propagate
through the media. These methods have not gained wide acceptance,
partly because of the relative sophistication required in field opera
tion and in the interpretation of test data.
Surface loading tests generally involve the use of measured surface
deflections to backcalculate the moduli of the pavement layers. Among
the numerous types of devices used are the Dynaflect, Road Rater, and
Falling Weight Deflectometer (FWD). Such techniques have gained wide
spread popularity partly because they are simple, timeefficient, and
relatively inexpensive, and partly because of their ability to model
real traffic load intensities and durations. However, there are no
direct theoretical solutions available at present to evaluate the
various layer moduli of the pavement from the measured surface deflec
tions which generally represent the overall combined stiffness of the
layers. Instead, computerized iterative solutions, graphical solutions,
and nomographs are currently used to backcalculate pavement layer
moduli. All these techniques basically consist of using linearelastic
programs in which calculated versus measured deflections are matched by
adjustment of pavement layer moduli Evalues.
Those methods which are based on iterative procedures may need a
large amount of computer time to arrive at the correct moduli for the
pavement materials. In some cases, the required computer may not be
accessible (e.g., for direct field evaluation) or the expertise required
may not be available. Also, due to the inherent problems associated
with iteration methods, unique solutions cannot be guaranteed and dif
ferent sets of elastic moduli can produce results that are within the
specified (deflection or layer moduli) tolerance. In addition, elastic
layer programs generally assume an average (composite) modulus for the
subgrade layer without regard to the variation of the underlying soil
properties with depth. For sites with highly variable subgrade stiff
nesses, it becomes very difficult to analytically match measured deflec
tion basins using a composite modulus for the subgrade layer. There
fore, there is a need to find a more viable way to determine the E
values of pavement materials for a rational mechanistic analysis.
Recent advances in in situ testing in geotechnical engineering have
led to improvements in the determination of important soil parameters
such as strength and deformation moduli. Unfortunately, the application
of the improved techniques to evaluate or design pavements has been very
limited. The Marchetti Dilatometer test (DMT) offers significant pro
mise for providing a reliable and economical method for obtaining in
situ moduli of pavement layers, especially of the subgrade. There is
also the potential of determining in situ moduli from the cone penetra
tion test (CPT) since several correlations between different deformation
moduli and cone resistance have been reported in the geotechnical liter
ature. The CPT and DMT provide detail information on site stratifica
tion, identification, and classification of soil types which makes them
attractive tests for the evaluation and design of pavements.
1.2 Study Objectives
The primary objective of this study is to develop procedures for
the evaluation of material properties in layered pavement systems using
NDT deflection measurements. This includes the development of layer
moduli prediction equations from NOT deflections.
The secondary objective is to evaluate the feasibility of deter
mining the modulus of pavement layers and underlying subgrade soils
using in situ penetration tests and to evaluate the possible effects of
stratigraphy, water table and underlying subgrade soil properties on
surface deflections obtained from NDT.
1.3 Scope of Study
This investigation is primarily concerned with predicting pavement
layer moduli from nondestructive and penetration tests. It is hoped
that this will lead to improvements in the determination of layer moduli
for mechanistic evaluation and design of flexible pavement systems. The
initial part of the study consisted of developing layer moduli predic
tion equations from computersimulated Dynaflect and FWD deflection
data. A modified Dynaflect loadsensor configuration was utilized in
the theoretical analysis.
Field tests were conducted on fifteen pavement sections in the
state of Florida. Tests conducted consisted of Dynaflect, FWD, elec
tronic CPT, DMT, and plate bearing tests. Also, cores of asphalt con
crete pavement were collected for laboratory lowtemperature rheology
tests. These were used to establish viscositytemperature relationships
of the recovered asphalts which were then used to predict the moduli of
the asphalt concrete layers.
The field measured NDT data were analyzed to establish layer moduli
values for the test pavement sections. The layer moduli derived from
the Dynaflect and FWD nondestructive tests were compared with each other
and correlated to the results of the penetration tests. Simplified
layer moduli prediction equations were developed for the modified
Dynaflect testing system.
Five of the test pavements were selected for shortterm load
induced stress analysis using actual wheel loadings and low temperature
conditions. The effects of agehardened asphalt, soil type, moisture
content, weak base course and subgrade characteristics on layer stiff
nesses were evaluated to assess the stressstrain response of the
different pavements.
CHAPTER 2
LITERATURE REVIEW
2.1 Introduction
A mechanistic pavement design procedure consists of analyzing the
pavement on the basis of the predicted structural response (stresses,
strains, and deflections) of the system to moving vehicle loads. Pave
mentlayer thicknesses (surface, base, and subbase) are selected such
that the predicted structural response would be acceptable for some
desired number of load repetitions and under existing environmental
conditions. This approach is consistent with the conditions established
by Yoder and Witczak that
for any pavement design procedure to be completely
rational in nature total consideration must be
given to three elements. These elements are
(1) the theory used to predict failure or distress
parameter, (2) the evaluation of the pertinent
material properties necessary for the theory
selected, and (3) the determination of the rela
tionship between the magnitude of the parameter in
question to the failure or performance level
desired.
(133, p. 2425)
In the last several years, a concentrated effort has been made to
develop a more rational analysis and design procedures for pavements.
For flexible or asphalt concrete pavements, layered (7,19,20) and finite
element (33) theories have been used with some success to analyze pave
ment response. The use of either theory requires that the materials
that make up the pavement be suitably characterized. Layered and finite
element theories use Young's modulus and Poisson's ratio to characterize
the stressstrain behavior of pavement materials. While some success
has been made in developing design theories, their limitations must also
be understood. Most pavement material responses differ from the assump
tions of the theory used, and the "true" values of stress, strain or
deflection may differ from the predicted levels. However, a great deal
of engineering reliance is being placed upon the use of multilayered
linear elastic theory (133) in which the elastic modulus is an important
input parameter.
The thesis presented here is aimed at determining realistic modulus
values for the structural characterization of flexible pavement systems
using layered elastic theory. This chapter reviews previous work on
layersystem solutions, the methods of determining the elastic modulus
of pavementlayer materials, and some important factors influencing the
modulus of elasticity.
2.2 ElasticLayer Theory
2.2.1 General
The type of theory used in the analysis of a pavementlayered
system is generally distinguished by reference to three properties of
material behavior response (133). They are the relationship between
stress and strain (linear or nonlinear), the time dependency of strain
under constant stress level (viscous or nonviscous), and the degree to
which the material can rebound or recover strain after stress removal
(plastic or elastic). These concepts have been clearly elucidated by
Yoder and Witczak (133). Considerable effort has been expended to
analyze pavement response using the above concepts. For example, the
finite element method (33), elasticlayer analysis based on Burmister's
theory (18,19,20) and the viscoelastic layer analysis (7) are all based
on these three properties of material behavior. As previously noted,
the type of theory most widely used at the present time is the
multilayered linear elastic theory. The development of multilayered
elastic solutions is presented below.
2.2.2 OneLayer System
The mathematical solution of the elastic problem for a concentrated
load on a boundary of a semiinfinite body was given by Boussinesq in
1885 (13). His solution was based on the assumption that the material
is elastic, homogeneous, and isotropic. Boussinesq's equation (133;
p. 28) indicates that the vertical stress is dependent on the depth and
radial distance and is independent of the properties of the transmitting
medium. There are several limitations of this solution when applied to
pavements. For example, the type of surface loading usually encountered
in flexible pavements is not a point load but a load which is distri
buted over an elliptical area (133).
Further work with the Boussinesq equation expanded the solutions
for a uniformly distributed circular load by integration. Newmark (85)
developed influence charts for determination of stresses in elastic soil
masses. The charts are widely used in foundation work. Love (60) used
the principle of superposition to extend Boussinesq's solution to solve
for a distributed load on a circular area. Foster and Ahlvin (36)
presented charts for computing vertical stress, horizontal stress, and
vertical elastic strains due to circular loaded plates, for a Poisson's
ratio of 0.5. This work was subsequently refined by Ahlvin and Ulery
(4) to allow for an extensive solution of the complete pattern of
stress, strain, and deflection at any point in the homogeneous mass for
any value of Poisson's ratio.
Although most asphalt pavement structures cannot be regarded as
being homogeneous, the use of these solutions are generally applicable
for subgrade stress, strain and deflection studies when the modular
ratio of the pavement and subgrade is close to unity. This condition is
probably most exemplified by conventional flexible granular base/subbase
pavement structures having a thin asphalt concrete surface course (133).
Normally, in deflection studies for this type of pavement, it is assumed
that the pavement portion (above the subgrade) does not contribute any
partial deflection to the total surface deflection.
2.2.3 TwoLayer System
Since Boussinesq's solution was limited to a onelayer system, a
need for a generalized multiplelayered system was recognized.
Moreover, typical flexible pavements are composed of layers such that
the moduli of elasticity decrease with depth (133). The effect is to
reduce stresses and deflections in the subgrade from those obtained for
the ideal homogeneous case.
Burmister (18,19,20) established much of the ground work for the
solution of elastic layers on a semiinfinite elastic layer. Assuming a
continuous interface, he first developed solutions for two layers, and
he conceptually established the solution for threelayer systems. The
basic assumption made was full continuity between the layers, which
implies that there is no slippage between the layers. Thus, Burmister
assumed that the strain in the bottom of one layer is equal to the
strain at the top of the next layer, but the stress levels in the two
layers will differ as a function of the modulus of elasticity of each
layer.
2.2.4 ThreeLayer System
Although Burmister's work provided analytical expressions for
stresses and displacements in two and threelayer elastic systems, Fox
(38) and Acum and Fox (2) produced the first extensive tabular summary
of normal and radial stresses in threelayer systems at the intersection
of the plate axis with the layer interfaces. Jones (52) and Peattie
(89) subsequently expanded these solutions to a much wider range of
solution parameters. Tables and charts for the various solutions can be
found in Yoder and Witzcak (133) and Poulos and Davis (92). It should
be noted that the figures and tables for stresses and displacements have
been developed, respectively, for Poisson's ratios of 0.5 and 0.35, for
all layers, and on the assumption of perfect friction at all interfaces.
Hank and Scrivner (42) presented solutions for full continuity and
zero continuity between layers. Their solutions indicate that the
stresses in the top layer for the frictionless case (zero continuity)
are larger than the stresses presented for the case of full continu
ity. In an actual pavement, the layers are very likely to develop full
continuity; hence, full continuity between layers should probably be
assumed.
Schiffman (100) extended Burmister's solution to include shear
stress at the surface for a threelayer system. Mehta and Veletsos (73)
developed a more general elastic solution to a system with any number of
loads. They extended the solution presented by Burmister to include
tangential forces as well as normal forces.
2.2.5 Multilayered or NLayered Systems
A general analysis of a multilayered system under general condi
tions of surface loading or displacement, or both was developed indepen
\
dently by Schiffman (99) and Verstraeten (125). Schiffman (99) con
sidered the general solutions for stresses and displacements due to non
uniform surface loads, tangential surface loads, and slightly inclined
loads, but no numerical evaluations were presented. Verstraeten (125)
presented a limited analysis of the fourlayered elastic problem. He
first derived expressions for the stresses and displacements for the
general case and performed numerical calculations for the particular
case of fourlayered systems with continuous interfaces. The analysis
by Verstraeten included not only a uniform normal surface stress, but
also two types of surface shear stresses: (1) uniform onedirectional
shear stress and (2) a uniform centripetal shear stress.
Recently, the Chevron Research Corporation (74) and the Shell Oil
Company (32) have developed computer programs for multilayered solutions
of the complete state of stress and strain at any point in a pavement
structure. Notable programs of interest are the BISTRO and BISAR pro
grams by Shell (32), and the various forms of CHEVRON program by the
Chevron Research Corporation. These computer solutions are essentially
an extension of Burmister's work that permit the analysis of a structure
consisting of any number of layers supported by a semiinfinite sub
grade, and under various loading conditions. In reality, it is only the
CHEVRON NLAYER program (74) which is suitable for any number of layers.
All the others are restricted to a maximum number of layers. BISAR
(32), for example, can handle nine pavement sublayers of known thick
nesses plus the semiinfinite subgrade or bottom layer.
Several investigators have verified the validity of Burmister's
theory with the actual mechanical response of flexible pavements.
Foster and Fergus (37) have compared the results of extensive test
measurements on a clayey silt subgrade to theoretical stresses and
deflections based on Burmister's theory and reported satisfactory
agreement. The discrepancy between actual and theoretical stresses and
displacements can be mainly attributed to the assumption of a homoge
neous and isotropic material, the ratedependent behavior of some
materials such as asphalt, and a circular loaded area representing the
wheel load. Nielsen (86) has made a detailed study in this area. His
review of the magnitude and distribution of stresses within a layered
system revealed regions where vertical and shearing stresses were criti
cal. His studies concluded that the layeredelastic theory is in every
respect consistent and that it is possible to establish fundamental
patterns of pavement performance based upon this theory. This suggests
that the elasticity theory could be used more extensively.
The moderators of the Fifth International Conference on the Struc
tural Design of Asphalt Pavements (76) concluded that the use of linear
elastic theory for determining stresses, strains, and deflections is
reasonable as long as the timedependent and nonlinear response of the
paving materials are recognized. They noted that the papers presented
at the conference confirmed that multilayer elastic models generally
yield good results for asphalt concrete pavements.
Barksdale and Hicks (10) compared the multilayered elastic approach
with the finite element method and recommended the use of the former for
pavement analysis since only two variables are needed (modulus and
Poisson's ratio). Pichumani (91) used the BISAR computer program for
the numerical evaluation of stresses, strains, and displacements in a
linear elastic system. He demonstrated that predicted vertical, radial,
and shear stress distribution were noticeably affected by slight changes
in the assumed material moduli. Pichumani's work demonstrated the need
for proper and extensive material characterization.
2.3 Material Characterization Methods
2.3.1 General
The use of multilayered elastic theory has provided the engineer
with a rational and powerful basis for the structural design of pave
ments, for pavement evaluation, and for overlay design. In this theory,
the complete stress, strain, and displacement pattern for a material
needs only two material properties for characterization, namely the
elastic modulus (E), and the Poisson's ratio (u). Generally, the effect
of Poisson's ratio is not as significant as the effect of the modulus
(133, pp. 280282; 88; 59, p. 160). Thus, E is an important input
parameter for pavement analysis using the layer theory.
Many tests have been devised for measuring the elastic modulus of
paving materials. Some of the tests are arbitrary in the sense that
their usefulness lies in the correlation of their results with field
performance. To obtain reproducible results, the procedures must be
followed at all times. The various possible methods for determining the
elastic modulus of pavement materials include laboratory tests, destruc
tive field tests, and in situ nondestructive tests.
Laboratory methods consist of conducting laboratory testing on
either laboratorycompacted specimens or undisturbed samples taken from
the pavement. Yoder and Witzcak (133) describe various laboratory
testing methods with the diametral resilient modulus test (8), indirect
tension test (9), and the triaxial resilient modulus test (1) being the
most popular. The latter is useful for unbound materials such as base
course and subgrade soils, while the other two are for bound materials
like asphalt concrete and stabilized materials. Monismith et al. (78)
studied the various factors that affect laboratory determination of the
moduli of pavement systems. They concluded that
. it is extremely difficult to obtain the same
conditions that exist in the road materials (mois
ture content, density, etc.) and the same loadings
(including loading history) in the laboratory as
will be encountered in situ. . Thus the best
method of analysis would appear to be to determine
an equivalent modulus which when substituted into
expressions derived from the theory of elasticity,
will give a reasonable estimate of the probable
deformation.
(78, p. 112)
Destructive field tests include, among others, several different
plate load tests (8) and the California Bearing Ratio (CBR) test (8).
These tests require trench construction and subsequent repair of the
pavement, and like the laboratory test methods, usually call for an
elaborate and costly testing program. The delays associated with such
programs are prohibitive especially for routine pavement analysis
studies.
The third method involves the extraction of pavementlayer proper
ties from in situ nondestructive testing (NDT). NDT methods have gained
wide popularity in the last few decades because of their ability to
collect data at many locations on a highway or airfield in a short
time. Therefore, a great deal of research effort has been concentrated
on this area. A review of the various types of NOT equipment available
and the associated interpretation tools is presented below.
2.3.2 StateoftheArt of Nondestructive Testing
2.3.2.1 General. Nondestructive testing (NDT) consists of making
nondestructive measurements on a pavement's surface and inferring from
the responses the in situ characteristics related to the structural ade
quacy or loading behavior (79). Among such methods are various seismic
techniques (associated with time measurements) and surface loading tests
(associated with deflection measurements). The latter is more popular
because surface deflection is the most easily measured structural
response of a pavement. The idea of using deflection measurements to
evaluate the structural integrity of pavements dates back to 1938 when
the California Division of Highways used electrical gages implanted in
roadways to measure displacements induced by truck loads (134).
There are currently several NDT procedures being used for pavement
investigations. Each of the procedures can be placed into one of the
following four general classes:
1. Static forcedeflection,
2. Steadystate vibratoryy) dynamic forcedeflection,
3. Dynamic impulse forcedeflection, and
4. Wave propagation.
As their names imply, the first three categories are associated with
deflection measurements due to application of force or load. The fourth
categorywave propagationmeasures the length and velocity of force
induced waves traveling through the pavement system. A detailed
description and evaluation of many of these NDT devices and procedures
has been presented by Bush (21), Moore et al. (79), and Smith and Lytton
(109). In the following pages, a brief description of the principles
involved and equipment available for each class will be presented.
2.3.2.2 Static Deflection Procedures. Measurement systems that
determine the pavement response to slowly applied loads are generally
termed static deflection systems. In these systems, the loading.methods
may consist of slowly driving to or from a measurement point with a
loading vehicle, or by reacting against a stationary loading frame. The
maximum resilient or recoverable deflection at the surface of the pave
ment is measured.
The most commonly used equipment in this class is the various forms
of the Benkelman beam devices. Other equipment that had been used
include the plate bearing test (8), Dehlen Curvature Meter, Traveling
Deflectometer, LacroixLCPC Deflectograph, and the French Curviameter.
The last three devices are essentially automation of the Benkelman beam
principle. The French Curviameter, for example, measures both the
deflection and curvature of the pavement, under an 18kips rear axle
load, with tire pressure maintained at 100 psi (24). Most of the
automated devices have been used widely in Europe and other parts of the
world, except for the Traveling Deflectometer which was built for the
California Department of Transportation and has been in use by that
agency for several years (109).
The major advantages of the static deflection procedures are the
simplicity of the equipment and the large amount of data that has been
accumulated with these devices. The most serious problem with this type
of measurement technique is the difficulty in obtaining an immovable
reference point for making the deflection measurements. This makes the
absolute accuracy of this type of procedure questionable. In addition,
since most of the devices generally measure a single (maximum)
deflection only, it is impossible or difficult to determine the shape
and size of the deflection basin.
In spite of their shortcomings, the large amount of data developed
using static deflection techniques makes such procedures an important
part of structural pavement evaluation. For this reason, several inves
tigators have attempted deflection comparison and correlations from the
static devices with those measured by the dynamic devices. The
following is a list of concepts developed from the deflection response
of a pavement using static NDT (79):
1. For adequately designed pavements, the deflections during the same
season of the year remain approximately constant for the life of the
pavement.
2. There is a tolerable level of deflection that is a function of
traffic type, volume and the structural capacity of the pavement as
determined by the pavement's structural section. This tolerable
level of deflection can be established through the use of fatigue
characteristics of the pavement structure.
3. Overlaying of a pavement will reduce its deflection. The thickness
required to reduce the deflection to a tolerable level can be esta
blished.
4. The deflection history of a welldesigned pavement undergoes three
phases in its behavior (71). A typical curve representing these
phases is shown in Figure 2.1.
a. In the initial phase, immediately after construction, the pave
ment structure consolidates and the deflection shows a slight
decrease.
TRAFFIC
Figure 2.1
WellDesigned Pavement Deflection History Curve (71)
b. During the functional or service phase, the pavement carries the
anticipated traffic without undue deformation and the deflection
remains fairly constant or shows a slight decrease.
c. The failure phase occurs as a result of both traffic and envi
ronmental factors. In this phase the deflection increases
rapidly and there is a rapid deterioration resulting in failure
of the pavement structure.
5. The deflection history of a pavement system varies throughout the
year due to the effects of frost, temperature, and moisture. A
typical annual deflection history of a pavement subjected to frost
action, as shown in Figure 2.2, can be divided into the following
four periods (103):
a. The period of deep frost when the pavement is the strongest.
b. The period during which the frost is beginning to disappear from
the pavement structure. During this period, the deflection
rises rapidly.
c. The period during which the water from the melting frost leaves
the pavement structure and the deflection begins to drop.
d. The period during which the deflection levels off with a general
downward trend as the pavement structure continues to slowly dry
out.
6. For a given flexible pavement structure it is generally known that
the magnitude of the deflection increases with an increase in the
temperature of the bituminous surfacing material. This is due to a
decrease in the stiffness of the bituminous surfacing. The effect
of temperature varies with the stiffness of the underlying layers.
As the stiffness of the underlying layers increases, the effect of
Cn 32
Period of : Period of Slow
Deep Frost ) Strength Recovery
C)
0
,.J
LL
 I I 1 I I I I
DEC JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV
MONTH
Figure 2.2 Typical Annual Deflection History for a Flexible Pavement (103)
an increase in temperature on deflection of the total structure
decreases.
2.3.2.3 SteadyState Dynamic ForceDeflection. Essentially, all
steadystate dynamic deflection measurement systems induce a steady
state sinusoidal vibration in the pavement with a dynamic force genera
tor. The dynamic force is superimposed on the weight of the force gen
erator, resulting in a variation of force with time as shown in Figure
2.3. The magnitude of the peaktopeak dynamic force is less than twice
the static force to insure continuous contact of the vibrator with the
pavement (79). This means there must always be some amount of dead
weight or static force applied. As the dynamic peaktopeak loading is
increased, this preload must also be increased (109).
Deflections are usually measured with inertial motion sensors. For
pure sinusoidal motions at any fixed frequency, the output of such sen
sors is directly proportional to deflection. Thus, to measure deflec
tion it is only necessary to determine the calibration factor (output
per unit of deflection) for the measurement frequency. In general,
either an accelerometer or a velocity sensor may be used to measure
deflections. The latter type is commonly called a geophone and is the
type normally employed in dynamic deflection measurements.
There are several different types of steadystate dynamic deflec
tion equipment that are currently being used for nondestructive struc
tural evaluation of pavements. Only two of them have been used exten
sively and are available commerciallythe Dynaflect and the Road
Rater. The others have been designed and constructed by agencies
involved in pavement research, namely the U.S. Army Waterways Experiment
Station (WES), the Federal Highway Administration (FHWA), the Illinois
N)
o
0
U.
TIME +
Figure 2.3 Typical Output of a Dynamic Force Generator (79)
Department of Transportation, and the Koninklijke/Shell Laboratorium,
Amsterdam, Holland. Detailed descriptions of the various vibratory
equipment can be found in References 21, 79, 109. The Dynaflect was
used in this study and a description of the device will be given later
in this report.
When one considers the difficulty in obtaining a reference point
for deflection measurements, the real advantage of a steadystate dyna
mic deflection measurement system becomes apparent. An inertial refer
ence can be employed to measure dynamic deflections. That is, the mag
nitude of the deflection change (the peaktopeak value) can be compared
directly to the magnitude of the dynamic force change (peaktopeak
value). For a given value of dynamic force, the lower the deflection,
the stiffer the pavement is (79).
Although the dynamic response of a pavement system approaches its
static (or elastic) response at low frequencies, exactly what value of
driving frequency is low enough to determine the elastic characteristics
of a pavement is somewhat questionable. As the driving frequency be
comes low it becomes difficult to generate dynamic forces and the output
of inertial motion sensors becomes very small. These factors combine to
make it difficult to obtain accurate low frequency dynamic deflection
measurements (79). Other technical limitations of vibratory equipment
include the need for a heavy static preload for the heavier devices and
the nonuniform loading configurations (109).
The deflection measurements that result represent the stiffness of
the entire pavement section. Although some significant accomplishments
have been made in separating the effects of major parts of the pavement
structure, the separation of the effects of all of the various
components of the structure with deflection basin measurements has not
yet been accomplished (63). The study presented herein is aimed at
developing an approach that would allow a layerbylayer analysis of the
Dynaflect vibratory deflection basin.
2.3.2.4 Dynamic Impact Load Response. Essentially, all impact
load testing methods deliver some type of transient force impulse to the
pavement surface and measure its transient response. The equipment uses
a weight that is lifted to a given height on a guide system and is then
dropped. Figure 2.4 illustrates this schematically. By varying the
mass of the falling weight or the drop height or both, the impulse force
can be varied.
The width or duration of the loading pulse (loading time) is
controlled by the buffer characteristics, Figure 2.4, and it closely
approximates a halfsine wave (Figure 2.5). The duration of the force
is nominally 2530 msec, Figure 2.5, which approximates the load
duration of a vehicle traveling 40 to 50 mph (123). The peak magnitude
of the force can be determined approximately by equating the initial
potential energy of the system to the stored strain energy of the
springs (buffer system) when the mass is momentarily brought to rest
(11,105). Thus
F = (2Mghk)1/2 Eqn. 2.1
where
M = mass of the falling weight,
h = drop height,
k = spring constant, and
g = acceleration due to gravity.
M
*
0oo o00 o oo
OD~ I I^ 0 0
II,'Il
11111.
Figure 2.4 Schematic Diagram of Impulse LoadResponse Equipment (105)
*1\jj\
TIME
Figure 2.5
Characteristic Shape of Load Impulse (105)
The response of the pavement to the impulse loading is normally
measured with a set of geophones placed at varying radial distances from
the center of the plate. These deflection measurements can, in princi
ple, be used to characterize the structural properties of the pavement
layers.
Three manufacturers currently market impulse testing equipment in
the United States. These are the Dynatest, KUAB and Phoenix falling
weight deflectometers. The Dynatest fallingweight deflectometer (FWD)
is the most widely used impulse loading device in North America and
Europe (109). Its newest versionthe Dynatest 8000 FWD testing
systemwas used in this study and will be described later in this
report. Other experimental impulse testing devices have been evaluated
by Washington State University and Cornell Aeronautical Laboratory (79).
The impulse testing machines have several advantages over other NDT
instruments. The magnitude of the force can be quickly and easily
changed to evaluate the stress sensitivity of the pavement materials
being tested. Perhaps the greatest advantage is the ability to simulate
vehicular loading conditions. Several investigators (11,35,46) have
compared pavement response in terms of stresses, strains, and deflec
tions from an FWDimposed load to the response of a moving wheel load.
All these comparisons have shown that the response to an FWD test is
quite close to the response of a moving wheel load with the same load
magnitude. Figure 2.6 shows such an example of pavement response com
parison.
However, the deflection basin produced by an impulse loading device
is symmetrical about the load and Lytton et al. (63) have argued that
the deflection basin under a moving wheel is not symmetrical about the
750
E 500
00
OSL
O L.
M 250
U.0
LL O
0 
0
600
/
x< 0
400
/c
2 400
I LLA
na
n I
250 500
DEFLECTION (gm)
FROM MOVING WHEEL LOAD
a) Surface Deflections
200
400
VERTICAL STRAIN (x 106)
FROM MOVING WHEEL LOAD
b) Vertical Subgrade Strains
Figure 2.6
Comparison of Pavement Response from FWD and
MovingWheel Loads (35)
750
600
load in any pavement structure. Thus, the impulse load of a FWD is not
an exact representation of a moving wheel load. Moreover, the response
from the impact testing technique is similar to other types of dynamic
deflection testing in the sense that it represents characterization of
the entire structure. The technique does not provide information that
readily separates the effects of its various layers. Finally, the
parameters that cause plastic deformation in the structure are not
readily determinable from impact testing (79).
2.3.2.5 Wave Propagation Technique. Wave propagation provides
methods for the determination of the elastic properties of individual
pavement layers and subgrades. Unlike the three previous methods of
NDT, these methods are not concerned with the deflection response of the
pavement. Rather, they are concerned with the measurement of the velo
city and length of the surface waves propagating away from the load
surface (127).
There are two basic techniques for propagating waves through pave
ment structures: (1) steadystate vibration tests and (2) seismic
(impulse) tests. Generally, three types of waves are transmitted when a
pavement surface is subjected to vibration. These are
1. Compression or primary (P) waves,
2. Shear (S) waves, and
3. Rayleigh (R) waves.
The P and S waves are body waves while the R wave is a surface wave.
Raleigh waves are the dominant waves found in the dissipation of energy
input from a vibrator on a semiinfinite halfspace (75). Also, because
P and S waves attenuate rapidly with radial distance from the vibration
source, R waves are the typical waves measured in the wave propagation
technique.
Wave propagation theory is based upon the fact that in a homoge
neous isotropic half space subjected to an external disturbance, waves
travel at velocities that may be expressed as (59, p. 153; 79; 127)
Vs= (G) 2 E /2 Eqn. 2.2a
p 2(1 + u)p
V = E(1 u) /2 Eqn. 2.2b
P p(1 + v)(1 2p)
VR = aVs Eqn. 2.2c
where
Vs = shear wave velocity,
Vp = compression wave velocity,
VR = Rayleigh wave velocity,
G = shear modulus,
E = Young's modulus,
P = Poisson's ratio,
p = mass density, and
a is a function of Poisson's ratio and varies from 0.875
for P = 0 to 0.995 for P = 0.5.
In general, R and S waves are not particularly dependent on Poisson's
ratio, but the value of compression wave velocity is strongly dependent
on Poisson's ratio (59,79).
Field test procedures for the wave propagation measurements involve
two general types of tests. Raleigh wave velocities are determined from
steadystate vibratory pavement responses and compression wave veloci
ties are measured from impulse (seismic) tests. The former usually
follows procedures developed by researchers at the Royal Dutch Shell
Laboratory (43,53,93), the British Road Research Laboratory (79), and
the Waterways Experiment Station (79). They utilized a mechanical
vibrator for lowfrequency vibrations (5100 Hz) and a small electro
magnetic vibrator for the highfrequency work (43,53). The general
procedure currently in use is to place the vibrator on the pavement
surface and set the equipment in operation at a constant frequency.
Details of the procedure can be found in Reference 79.
Seismic tests may be conducted to determine the velocity of com
pression waves, which can be used with the shear wave (or Rayleigh wave)
velocity to compute Poisson's ratio. One such method is the hammer
impulse technique in which the pavement is struck with a light hammer
and the resultant ground motion is observed at one or more points with
horizontal motion geophones. However, this method is only good for
soils where the velocity of the materials increases with depth. It is
not applicable to layered pavement systems where strong, high velocity
layers occur at the top and grow progressively weaker with depth. How
ever, Moore et al. (79) report that this procedure has been used to
obtain compression wave velocities of pavement layers during construc
tion.
A method of using surface waves to structurally characterize pave
ments is currently in the research stage at the University of Texas at
Austin (80,81,82). The technique, called Spectral Analysis of Surface
Waves (SASW), determines shear wave velocity at soil or pavement sites.
The elastic shear and Young's moduli profiles are then calculated under
the assumption of homogeneous, isotropic, and elastic medium. The SASW
method is essentially a seismic procedure. An iterative inversion
process is used to interpret the shear wave velocity profiles (81).
Laboratory procedures are available for the determination of the
elastic properties of pavement and soil specimens using wave propagation
techniques. However, the laboratory procedures require that samples of
the pavement material be obtained for testing. Therefore, it may not be
considered as a nondestructive technique. Two laboratory procedures
that parallel the field vibratory procedures and which may be applicable
to pavement design are the resonant column and the pulse methods (79).
The most difficult aspect of the wave propagation techniques is
that of interpretation and analysis of test results. The wave propaga
tion method of testing relies on the ability to interpret the data
obtained in the field so that the characteristics of the structure
beneath the surface may be determined (79). Because of the inherent
complexities involved, such techniques have not gained wide acceptance.
2.3.3 NDT DataInterpretation Methods
2.3.3.1 General. Considerable emphasis has been placed upon
determining the elastic properties of pavement layers using NDT data.
Most of this work has been concentrated on the first three types of NDT
procedures, those associated with deflection measurements. The fourth
category, the wave propagation method, has not gained wide acceptance
because of the relative sophistication required in the field operation
and in the interpretation of test data. However, the interpretation of
measured surface deflection basins has gained widespread popularity with
the advent of NDT procedures. There is a general agreement among
pavement engineers that the measured surface deflection basins from NDT
can provide valuable information for structural evaluation of a pave
ment.
Methods for the interpretation of NDT data can be placed into two
categories: empirical or mechanistic methods. Empirical procedures
directly relate NDT response parameters to the structural capacity of a
pavement. Most of these methods (48,56) do not involve direct or
indirect theoretical analysis. Instead, they are based upon the cor
relation between the maximum deflection under a load (static NDT or
wheel load) and pavement performance.
In an attempt to improve the empirical procedures, other research
ers have relied on the use of deflection basin parameters (90) or semi
empirical correlations (79) for pavement evaluation. Figure 2.7 shows
an example of basin parameters and the criteria used to evaluate a pave
ment. Table 2.1 lists some of the deflection basin parameters that have
been developed for NDT data evaluation of pavements (120). Most of the
basin parameters do not relate directly to the elastic parameters of the
pavement section.
Semiempirical procedures usually involve correlation of modulus
values to other known pavement parameters. For example, Heukelom and
Foster (43) have developed a correlation between modulus E (in psi) from
wave propagation techniques and the California Bearing Ratio (CBR)
value. This correlation, though later refined by WES (79), is of the
form
E = 1500 (CBR)
Eqn. 2.3
ligid Force Wheels
SCI
DMD = Dynaflect Maximum Deflection (Numerical Value of Sensor No. 1)
SCI = Surface Curvature Index (Numerical Difference of Sensor No. 1 and
No. 2)
BCI = Base Curvature Index (Numerical Difference of Sensor No. 4 and
No. 5).
a) Basin Parameters
b) Criteria
Figure 2.7 Empirical Interpretation of Dynaflect Deflection Basin (90)
DMD SCI BCI CONDITION OF PAVEMENT STRUCTURE
GT 0.11 Pavement and Subgrade Weak
GT 0.48
GT 1.25 LT 0.11 Subgrade Strong, Pavement Weak
GT 0.11 Subgrade Weak, Pavement Marginal
LT 0.48
LT 0.11 DMD High, Structure Ok
GT 0.11 Structure Marginal, DMD Ok
GT 0.48
LT 1.25 LT 0.11 Pavement Weak, DMD Ok
GT 0.11 Subgrade Weak, DMD Ok
LT 0.11 Pavement and Subgrade Strong0.48
LT 0.11 Pavement and Subgrade Strong
Table 2.1 Summary of Deflection Basin Parameters
Parameter Definitiona NDT Deviceb
Dynaflect maximum deflection (DMD)
Surface curvature index (SCI)
Base curvature index (BCI)
Spreadability (SP)
Basin slope (SLOP)
Sensor 5 deflection (W )
5
Radius of curvature (R)
Deflection ratio (Qr)
Area, in inches (A)
Shape factors (F F )
1 2
Tangent slope (TS)
DMD = d
1
SCI = d d
1 2
BCI = d d
4 5
SP = ( Idi /5d )x 100
i=1 to 5
SP = ( Idi /4d ) x 100
i=1 to 4
SLOP = d d
1 5
W = d
5 5
R = r2/{2.dm[(dm/dr) 111]}
Qr = r/do
A = 6[1 + 2(d /d ) + 2(d /d ) +
2 1 3 1
F = (d d )/d
1 1 3 2
F = (d d )/d
2 2 4 3
TS = (d dx)/x
Dynaflect
Dynaflect, Road Rater model 400
Dynaflect
Dynaflect
Road Rater model 2008
Dynaflect
Dynaflect
Benkelman beam
FWD, Benkelman beam
(d /d )] Road Rater model 2008
4 1
Road Rater model 2008
a d = deflection; subscripts 1,2,3,4,5 = sensor locations; o = center of load; r = radial distance;
m = maximum deflection; x = distance of tangent point from the point of maximum deflection.
b The NDT device for which the deflection parameter was originally defined.
Source: Uddin et al. (120)
and is the most widely used correlation (133). Other correlations (79)
have been made between E and plate bearing subgrade modulus, K. It
should be recognized that the conditions of dynamic testing generally
yield moduli in the linear elastic range. Conventional tests such as
the CBR and plate bearing tests produce deformations that are not
completely recoverable and, therefore, are partly in the plastic range.
Thus, one would expect some variation in the correlation between E
modulus and pavement parameters, such as K and CBR.
Mechanistic analysis of NDT data is usually performed by one of the
following:
1. Direct relationship between deflection parameters and the
elastic moduli of the pavement layers.
2. Inverse application of a theoretical model by fitting a
measured deflection basin to a deflection basin using an
iterative procedure.
3. A combination of 1 and 2.
The above mechanistic methods employ deflection data from either vibra
tory or impulse loading equipment. While these devices are dynamic in
nature, most of the mechanistic solutions are based on elastostatic
(19,32,74) and viscoelastostatic (7) models. Recently, an elasto
dynamic model (54) has been used to interpret NDT data (66,67,105).
However, the use of dynamic analyses for interpretation of NDT data can
be considered to be in the research stage. Another significant obser
vation is that almost all the mechanistic solutions available employ
layered theory or simplified versions of it. The only exception to this
is the use of a finite element model presented by Hoffman and Thompson
(45). A review of the various solutions is presented below.
2.3.3.2 Direct Solutions. Presently, there are no direct analy
tical solutions that can uniquely determine the elastic moduli for a
multilayered pavement system using surface deflection measurements
alone. The socalled direct solutions have been developed for only two
layer systems which usually involve graphical solutions, nomographs, or
in most cases only provide estimates for the subgrade modulus.
Scrivner et al. (102) presented an analytical technique for using
pavement deflections to determine the elastic moduli of the pavement and
subgrade assuming the structure is composed of two elastic layers.
Based upon the same assumption, Swift (113) presented a simple graphical
technique for determining the same two elastic moduli. Vaswani (124)
used Dynaflect basin parameters to develop charts for the structural
evaluation of the subgrade and its overlying layers for flexible pave
ments in Virginia (see Table 2.1). The methods by Majidzadeh (64) and
Sharpe et al. (107), among others, employ similar basin parameters from
the Dynaflect or Road Rater to estimate the subgrade modulus and develop
charts to assess the overlying layers.
Jimenez (51) described a method for evaluating pavementlayer
modular ratios from Dynaflect deflections. The pavements were
considered to be threelayer systems, and the deflection data were used
to estimate ratios of the elastic moduli of the adjacent layers. The
ratios reduce the system from three values of elastic modulus to two
values of modular ratio. The major limitation of this method is that
the elastic modulus of the asphalt concrete layer must be known.
Wiseman (129) and Wiseman et al. (131) have, respectively, applied
the Hertz Theory of Plates and the Hogg Model to evaluate twolayered
flexible pavements using surface deflection basins. The Hertz theory is
an application of the analytical solution of a vertically loaded elastic
plate floating on a heavy fluid. The solution to this problem was
presented by Hertz in 1884 and was first applied to concrete pavement
analysis by Westergaard in 1926 (79). The Hogg model consists of an
infinite plate on an elastic subgrade. The subgrade can be either of
infinite extent or underlain by a perfectly rigid rough horizontal
bottom at a finite depth. Analysis of this model was reported by A.H.A.
Hogg in 1938 and 1944 (131). In both methods, the flexural rigidity of
the composite pavement which will best fit a measured deflection basin
is calculated.
Lytton et al. (62) and Alam and Little (5) have developed another
method based on elasticlayer theory for prediction of layer moduli from
surface deflections. This method makes use of the explicit expression
for deflection originally postulated by Vlasov and Leont'ev (126). The
major drawback of this technique is the need to develop several con
stants, five in all, for which no analytical or test method exists as
yet. In applying this method, the authors (5,62) resorted to the use of
regression analyses and computer iterative solutions.
Cogill (28) presented a method which provides an estimate of the
stiffness of the pavementlayer materials. The method essentially is a
graphical presentation in which the deflections measured over a parti
cular range of load spacing can be related to the stiffness of the pave
ment material at a certain depth. The relationship is an approximate
one and is expressed with the aid of Boussinesq's formula.
All the methods presented above use deflection measurements
obtained from vibratory loading equipmentsDynaflect and Road Rater.
The only approach for the direct estimation of layer moduli from impulse
loaddeflection response (such as an FWD deflection basin) is the
concept of equivalent layer thickness (121,122) in which the layered
pavement system is transformed into an equivalent Boussinesq (13)
system. This concept, originally proposed by Odemark (87), is based on
the assumption that the stresses, strains, and deflections below a given
layer interface depend on the stiffness and thickness of the layers
above that interface. Although this approach obtains an explicit
solution for the subgrade modulus (121), it relies on estimates of the
asphalt concrete layer modulus and also employs certain modular ratios
to obtain the moduli of the various layers above the subgrade (25). The
method of equivalent thicknesses (MET) has also been incorporated into
some iterative computer programs which are discussed in the next
section.
Several investigators have obtained equations to directly determine
the subgrade modulus from one or more sensor deflections. For example,
Figure 2.8 shows the relationship between the subgrade modulus and the
Dynaflect fifth sensor deflection as summarized by Way et al. (128).
Keyser and Ruth (55) developed a prediction equation from five test road
sections in the Province of Quebec, Canada, by using the BISAR elastic
layer program to match measured Dynaflect deflection basins. The
equation is of the form
1.0006
E = 5.3966(D ) Eqn. 2.4
4 5
where E4 is subgrade modulus in psi, and Ds is Dynaflect fifth sensor
deflection in inches. This equation had an R2 of 0.997 (55), and is
similar to that of Ullidtz (see Figure 2.8). Godwin and Miley (41) have
100,000 I I I 1 1 1 1 1
SMajidazdeh, Ref. 17
S\ Esg =(6115 x D5)0984
c0 ADOT Hyperbolic 
SEsg =(0.000013 + 0.00016 x D )1
ADOT Power Equation
10,000 UlIdtz, Ref. 20 = 8800 x D0.58
Sq xa2 x(1.0 Nu2) 5200xD10
g r x D5 x 0.001
Sq = Load Pressure = 159.16 psi
< a = Load Pressure = 2. Inches
C Nu = Poisson's Ratio = VO
0 r = Fifth Sensor Distance from Load = 49.0 Inches \
D McCullough, Ref. 18
Esg =4000xD51.2
1,000 1 1 1 1 1 1 1 1 1 1 1 1 1 1 I
0.01 0.1 1.0 10.0
DYNAFLECT DEFLECTION, D5, mils
Dynaflect Fifth Sensor DeflectionSubgrade Modulus Relationship (128)
Figure 2.8
developed correlations between the base, subbase and subgrade moduli and
the second, third, and fourth Dynaflect sensor deflections, respec
tively. However, the modulus values used in the correlation were the
surface moduli from plate bearing tests which suffers from the problem
of incorporating plastic and nonrecoverable deformations.
An approach using regression equations to estimate layer moduli has
been attempted by other investigators (83,120,132). This approach
usually involves analysis of computersimulated NDT data using a theo
retical model (usually layered elastic theory). The various investi
gators reported success in the case of the subgrade modulus. To obtain
good correlations for the other layers (surface, base, subbase), certain
assumptions had to be made, such as the base course modulus being
greater than the modulus of the subgrade (83), or they had to resort to
computeriteration programs (83,120).
2.3.3.3 BackCalculation Methods. The method of iteratively
changing the layer moduli in a theoretical model to match the theoreti
cal deflection basin to a measured basin is currently called back
calculation in the literature. Initial developments of this procedure
utilized a trialanderror approach (49,72) using the following steps:
1. Pavementlayer thicknesses, initial estimates of the pavementlayer
moduli, and the loading and deflection measurement configuration are
input into the model (usually a multilayer elastic computer
program).
2. The computed deflections at the geophone positions are compared with
those actually measured in the field.
3. The layer moduli used in the computer program are then adjusted to
improve the fit between the predicted and actual deflection basins.
4. This process is repeated until the two deflection basins are vir
tually the same. The process may have to be repeated several times
before a reasonable fit is obtained.
Because of the time consuming nature (49) of the trialanderror
method, many researchers have developed computer programs to perform the
iteration. Table 2.2 lists some of the selfiterative computer pro
grams. The major differences among the various programs are the differ
ent models, algorithms and tolerance levels used in the iteration pro
cess. A few of these will be discussed here.
Anani (6) developed expressions for surface deflections in terms of
the modulus values of a fourlayer pavement. However, he could not
obtain direct solutions to determine the moduli. Therefore he used an
iterative procedure to obtain the moduli from Road Rater deflection
basins. The computer programs reported by Tenison (114) and Mamlouk
(66) followed the successive approximation method of Anani (6). In the
overlay design program called OAF, Majidzadeh and Ilves (65) employed a
deflection matching technique for determining the in situ layer stiff
nesses. While using field data to substantiate the applicability of the
procedure, they experienced difficulties and commented,
. the computed asphalt layer stiffness shows a
large variation, and in a few cases the asphalt is
stiffer than steel; nevertheless the values are
reasonable in a great majority of the cases . .
(65, p. 85)
The BISDEF computer program (23) is an improvement over the CHEVDEF
(22) to handle multiple loads and variable interface conditions. The
number of layers with unknown modulus values cannot exceed the number of
measured deflections. However, a maximum of four deflections are
Summary of Computer Programs for
of Flexible Pavement Moduli from
Evaluation
NDT Devices
Number Theoretical Applicable
Name Reference of Model Used NDT
Layers For Analysis Device
* Anani (6)
4
Layer
BISARElastic
Road Rater 400
ISSEM4 Sharma and
Stubstad (106)
CHEVDEF Bush (22)
OAF Majidzadeh
and Ilves (65)
ILLIPAVE
Hoffman and
Thompson (45)
Tenison (114)
FPEDD1 Uddin et al.
(118,120)
BISDEF Bush and
Alexander (23)
ELMOD Ullidtz and
Stubstad (123)
IMD Husain and
George (47)
DYNAMIC Mamlouk (66)
* not known or available
4 ELSYM5Elastic
Layer
4(a)
Layer
CHEVRONElastic
3 or ELSYM5 Elastic
4 Layer
3 Finite Element
3 CHEVRON's N
Layer (Elastic)
3 or ELSYMS Elastic
4 Layer
4(a)
2, 3
or 4
BISAR Elastic
Layer
Vibrator, or FWD
METBoussinesq
3 or CHEVRON Elastic
4 Layer
for FWD
4 Elastodynamic
Road Rater 2008
Dynaflect,
Road Rater, or FWD
Road Rater 2008,
or FWD
Road Rater 2000
Dynaflect, FWD
Dynaflect,
Rater, WES
FWD
Road
Dynaflect, but
can be modified
Road Rater 400
(a) not to exceed number of deflections
Table 2.2
targeted during the iteration process, which is also limited to a
maximum of three loops. When applied to field measured deflections on
an airfield pavement in Florida (23), BISDEF predicted unreasonably high
values of the AC modulus for all the different NDT devices used in the
study. Also, Bush and Alexander (23) conceded that the program provides
the best results if the number of unknown layer moduli is three.
The ISSEM4 computer program (106) incorporates the principles of
the method of equivalent thicknesses (MET) into the ELSYM5 multilayered
elastic program to determine the in situ stressdependent elastic
moduli. The parameters for the nonlinear stressdependent relationships
(see Section 2.4.3) are established from FWD tests performed at differ
ent load levels. The iteration process is seeded with a set of Evalues
(106). The ELMOD program (123) also utilizes the MET principle and the
iteration procedure. Both programs provide relatively good solutions if
the asphalt concrete modulus is known.
ILLIPAVE (45), the only program which utilizes a finite element
model, is specifically developed to handle Road Rater deflection data.
However, Road Rater deflection basins must be converted to equivalent
FWD deflection basins prior to being used in the program (45). Also,
the nonlinear stressdependent material models incorporated into the
finite element method utilized relationships established from previous
laboratory material characterization procedures. It is also surprising
that the authors resulted to nomographs for specific applications of the
backcalculation model (45).
Most of the iteration programs listed above require a set of ini
tial moduliseed moduliand therefore are userdependent. Therefore
computational times and cost can be prohibitive. Unique solutions
cannot be guaranteed since an infinite number of layer modulus combina
tions can provide essentially the same deflection basins. Moreover,
most of the iterative programs yield questionable base course and
subbase moduli. In some programs, adjustment of the field data are
required in order to improve the solution (6,47).
2.3.4 Other In Situ Methods
Cogill (27) presented a method involving the use of an ultrasonic
technique. The elastic modulus of the top layer can be accurately
determined; however, the modulus values for the other layers are
questionable. Kleyn et al. (58) and Khedr et al. (57) have developed
different forms of a portable cone penetrometer to evaluate the stiff
ness of pavement layers and subgrade soils. However, these devices do
not provide direct modulus values but rather are based on correlations
with CBR and plate bearing parameters. Similarly, the Clegg Impact
tester, which was developed in Australia in the mid1970s, relies on CBR
correlation for pavement evaluation applications (40). The problems of
the CBR and plate bearing tests have been discussed previously.
Maree et al. (70) presented an approach to determine pavementlayer
moduli based on a device developed to measure deflections at different
depths within a pavement structure. The device, called the multidepth
deflectometer (MDD), is installed at various depths of an existing pave
ment structure to measure the deflections from a heavyvehicle simulator
(HVS) test. Maree et al. (70) suggested that effective moduli for use
in elasticlayer theory can be determined from correction factors esta
blished from field measurements using the MDD at different times of the
year and under different conditions.
Molenaar and Beuving (77) described a methodology in which the FWD
and a dynamic cone penetrometer (DCP) were used to assess stress depen
dent unbound pavement layers and the presence of soft interlayers.
However, the procedure does not provide any direct modulus correlation
but a graphical presentation of FWD surface modulus and DCP profiles.
Geotechnical engineers have, for several years, used various forms
of field tests to assess the engineering properties of soils for con
struction purposes. Recent advances in exploration and interpretation
methods have led to improvements in the determination of important soil
parameters such as strength and deformation moduli. For example, the
following in situ techniques (26,30,50) are suitable for the determina
tion of soil stiffness:
1. Menard Pressuremeter (PMT) and SelfBoring Pressuremeter (SBP)
tests.
2. Cone Penetration Test (CPT), including the mechanical,
electronic, and piezocone penetrometers.
3. Marchetti Dilatometer Test (DMT).
4. Plate Loading Tests (PLT), including Screw Plate Tests (SPL).
Some of these tests have the added advantage of providing detailed
information on site stratification, identification, and classification
of soil types. This is of great appeal since the variation of the
subgrade soil properties with depth can be accounted for rather than
assuming an average modulus value as conventionally used in multilayer
analysis.
Unfortunately, the application of the improved techniques to eval
uate or design pavements has been very limited. As evident from the
previous sections, the material characterization part of a rational
pavement design program, though very important, is often treated with
considerable simplification and empiricism. Geotechnical engineers
often feel that structural engineers have little or no interest in those
parts of their work below the ground level. These feelings are cer
tainly justified in the case of pavements (76). It is therefore not
surprising that most of the in situ geotechnical applications to pave
ments rely on correlations with empirical pavement parameters such as
CBR to validate their proposed methods (40,57,58,77). The other known
applications of geotechnical in situ testing methods to evaluate the
stiffness of pavement structures are discussed below.
Briaud and Shields (14,15) have described the development and
procedure of a special pressuremeter test for pavement evaluation and
design. The pavement pressuremeter consists of a probe, tubing and a
control unit, and works on the same principle as the Menard pressure
meter (30). They illustrated how the modulus values obtained from the
test can be used directly in multilayer mechanistic analysis. In order
to use empirical design charts, however, Briaud and Shields (15) also
developed a correlation between the pressuremeter modulus and the
bearing strength obtained from a Macleod plate test for two airport
pavements in Canada.
Borden et al. (12) have presented an experimental program in which
the dilatometer test (68) was used to determine pavement subgrade sup
port characteristics. A major part of the testing program consisted of
conducting DMT and CBR tests in soil samples prepared in cylindrical
molds and also in a special rectangular chamber. A limited field test
was conducted on a compacted embankment constructed with one of the
soils used in the laboratory investigation. Although they report good
correlations between the dilatometer modulus and CBR value, the use of
the CBR test makes the study empirical, to say the least.
2.4 Factors Affecting Modulus of PavementSubgrade Materials
2.4.1 Introduction
The response characteristics of flexible pavement materials is a
complex function of many variables, which is farfetched from the ideal
materials assumed in classical mechanics. In general, the behavior of
these materials is dependent upon many environmental and load vari
ables. Specifically, the asphalt concrete response is primarily a
function of temperature and rate of loading. Due to its viscoelastic
nature (7,51), asphalt concrete materials become stiffer as the load
rate increases and the temperature decreases. The granular base course
and subgrade characteristics are dependent upon moisture content, dry
density, stress level, stress states, stress path, soil fabric, stress
history, and soil moisture tension (59,78,133).
Several researchers have presented relationships of resilient
modulus as a function of one or more variables, while keeping others
fixed or completely ignored. Most of these relationships were developed
from laboratory studies. A complete review of the relative effects of
the various factors on pavementsoil response, or the relationship
between modulus and other parameters measured in the laboratory can be
found in References 31, 44, 59, 78, 94, and 133. It is not the intent
of this discussion to review the various studies on this topic. The
discussion below will concentrate on two variables that are believed to
be very important in flexible pavement technology, especially when con
sidering NDT and pavement evaluation. These factors are the temperature
of the asphalt concrete layer, and the stress dependency of base/subbase
and subgrade materials. This does not mean that the effects of the
other variables can be ignored or underestimated. For example, moisture
content has a considerable effect on the modulus of flexiblepavement
materials, especially for finegrained subgrade soils (78).
2.4.2 Temperature
Temperature has a very important influence on the modulus of
asphaltbound materials. The modulus of asphalt concrete decreases with
an increase in pavement temperature (51,78,111,133). The temperature of
the pavement also fluctuates with diurnal and seasonal temperature vari
ations. In order to determine the variation of modulus with temperature
for flexiblepavement materials, the mean pavement temperature should be
established. Southgate and Deen (111) developed a method for estimating
the temperature at any depth in a flexible pavement up to 12 inches.
Figure 2.9 shows the graphical solution for the determination of the
mean pavement temperature with depth from the known temperatures. This
relationship has been recommended and in some cases incorporated into
many flexible pavement design procedures (47,65,107). Though, the
curves have been found to be reasonably accurate for other locations
(111), it would be more desirable to make a direct determination of this
temperature.
2.4.3 Stress Dependency
Laboratory studies presented in the literature (31,44,78,94) sug
gest that the moduli of granular base materials and subgrade soils are
stressdependent. The stiffness of the granular base has been found to
be a function of the bulk stress or first stress invariant. A stress
stiffening model in which the modulus increases with the first stress
160
LL
0
a 120
I
W 80
" 40
I
0
0
40 80 120 160 200 240
PAVEMENT TEMPERATURE + 5
DAY MEAN AIR TEMPERATURE
a) Pavements More Than 2 in. Thick
o" (b) Depth in
Pavement, 0
i 0 Inches 1
S120 2

t 80
c:
W 40 Depth in
0
M 2 Pavement, Inches
0o l l I I0I I I
0 40 80 120 160 200 240
PAVEMENT TEMPERATURE + 5
DAY MEAN AIR TEMPERATURE
b) Pavements Equal to or Less Than 2 in. Thick
Figure 2.9 Temperature Prediction Graphs (111)
invariant is generally used to characterize granular base materials.
The relationship is of the form
K
E = K 9 2 Eqn. 2.5
1
where
E = granular base/subbase modulus,
6 = first stress invariant or bulk stress, and
K K = material constants
1 2
The subgrade stiffness, on the otherhand, has been found to be a
function of the deviator stress (stress difference). For finegrained
soils, resilient modulus decreases with increase in stress difference
(78). The mathematical representation of the subgrade stiffness is of
the form
E = AaB Eqn. 2.6
where
E = subgrade modulus
a = stress difference, and
A, B = material constants for the subgrade
The constant B(slope) is less than zero for the stresssoftening model,
while for the stressstiffening model, the slope is greater than zero.
The stressdependency approach of characterizing pavement materials
is of great importance for high traffic loadings. Situations in which
high traffic loadings occur are larger aircraft loadings in the case of
airfield pavements, and when heavy wheel loads and/or single tire
configurations (which result in higher stresses) are applied to flexible
highway pavements. For this reason, some of the NDT backcalculation
procedures have accounted for the stress dependency effect by incorpo
rating Equations 2.5 and 2.6 into their algorithms (45,65,106). How
ever, the problem of determining the material constants, A, B, K and
1
K still remains, especially when NDT deflection basins are used to
2
characterize the pavement. The most common approach is to use labora
tory resilient moduli and regression analysis to determine these para
meters (45,65,72). Thus, the material parameters will depend upon
sample preparation procedures, disturbance, prestressstrain conditions,
etc.
Other researchers (93,106,121) have suggested determining the mate
rial constants from FWD tests conducted at three or more load levels.
However, it is not clear how viable this procedure is since the resul
tant loaddeflection response of a pavement is a combined effect of the
behavior of the individual layers. The relative contribution of each
layer is not clearly known. It is even more complex since the asphalt
concrete layer is dependent on the temperature and agehardening
characteristics of the asphalt cement. Moreover, contrary to previous
belief, Thompson (116) has found that the material parameters are not
independentof each other, especially for granular bases and subbases.
Uddin et al. (118,119,120) have applied the concepts of equivalent
linear analysis developed in soil dynamics and geotechnical earthquake
engineering to evaluate the nonlinear moduli. They concluded that the
in situ moduli derived from an FWD deflection basin (at 90001b. peak
force) are the effective nonlinear moduli and need no further correc
tion. However, an equivalent linear analysis has to be performed to
correct the in situ moduli calculated for nonlinear granular materials
and subgrade soils from a Dynaflect deflection basin. These conclusions
were based on stress analysis comparisons of a singleaxle 18kip design
load, FWD (9000lb. peak force) and Dynaflect loadings simulated in the
ELSYM5 elasticlayer program. An algorithm to perform this equivalent
linear correction has been incorporated into the FPPEDD1 selfiterative
computer program (120). However, results reported by Nazarian et al.
(81) tend to contradict the conclusions of Uddin et al. (120). Their
study involving FWD tests at 5 and 15kip loads indicated that non
linear behavior occurs at higher FWD loads, and is more predominant in
the base course layers than the subgrade.
These results and those from other research work indicate there is
disagreement as to what type of approach should be used when the effects
of nonlinearity and stress dependency are to be considered. There are
at least three schools of thought in this regard. The first group
believes that the use of an equivalent effective modulus in an elastic
layer theory would provide reasonable response predictions. This
approach would eliminate the expense, time and complexity associated
with more rigorous methods such as finite element models (61). The
research works of Maree et al. (70), Roque (96), and Roque and Ruth (97)
on fullscale pavements tend to support this theory.
The second school of thought recommends that the nonlinear stress
dependent models (Equations 2.5 and 2.6) can be incorporated into an
elasticlayer program to predict reasonable response parameters. How
ever, the asphalt concrete layer is treated as linear elastic. This
theory is supported by Monismith et al. (78), among others, and has been
used in iterative computer programs like OAF, ISSEM4, and IMD.
The third school of thought contends that layered elastic theory,
when used with certain combinations of pavement moduli, predicts tensile
stresses in granular base layers, even if gravity stresses are also
considered (16,45,112). Instead of using a layered approach, this group
prefers a finite element model in which the nonlinear responses of the
granular and subgrade materials are accurately characterized. Again,
the asphalt concrete layer is considered to be linear elastic. The
ILLIPAVE finite element backcalculation program (45) is a classic
application of this theory.
In the finite element approach discussed above, researchers have
used, with limited success, various failure criteria and in some cases
arbitrary procedures to overcome the problem of tensile stresses
(16,112). For example, Brown and Pappin (16) used a finite element
program called SENOL with a K0 contour model and found it to be capable
of determining surface deflections and asphalt tensile stresses but
unable to determine the stress conditions within the granular layer.
The asphalt layer was characterized as elastic with an equivalent linear
modulus. They therefore concluded that the simplest approach for design
calculations involves the use of a linear elasticlayered system pro
vided adequate equivalent stiffnesses are used in the analysis. This
conclusion is shared by other investigators (10,61,96,97) and is the
philosophy behind the work presented in this dissertation.
CHAPTER 3
EQUIPMENT AND FACILITIES
Most of the methods available for determining the elastic moduli of
flexible pavements have been outlined in Chapter 2. These include the
use of nondestructive tests (NDT), laboratory methods and other in situ
test methods. The limitations of these methods and the need for a more
simple approach have also been highlighted. An approach which mechanis
tically evaluates pavements with the use of NOT and/or in situ penetra
tion tests is therefore developed in this study. This approach is
developed to simplify the mechanistic analysis and design process, and
to evaluate the effects of important variables involved in the determi
nation of pavement layer moduli. The study consisted of the development
of moduli prediction equations from NDT data, field testing and analyses
of NDT and in situ penetration tests, and finally, comparison and eval
uation of test data. Therefore, this chapter describes the equipment
and facilities used in the study.
The test equipment were either available at the Civil Engineering
Department of the University of Florida or at the Bureau of Materials
and Research, Florida Department of Transportation (FDOT). They are
essentially standard testing devices. This research was concerned with
their optimum use and application for a rational mechanistic design and
evaluation of asphalt concrete pavements.
3.1 Description of Dynaflect Test System
3.1.1 Description of Equipment
The Dynaflect, as previously mentioned, belongs to the dynamic
steadystate forcedeflection group of NDT equipment. It is an electro
mechanical device for measuring the dynamic deflection of a pavement
caused by oscillatory loading. The testing system (84,104,108) consists
of a dynamic force generator mounted on a small twowheel trailer, a
control unit, a sensor assembly and a sensor (geophone) calibration
unit. The Dynaflect can be towed by and operated from any conventional
passenger carrying vehicle having a rigid trailer hitch and a 12volt
battery system.
The oscillatory load is produced by a pair of counter weights
rotating in opposing directions and phased in such a manner that each
contributes to the vertical force of the other, but opposes the horizon
tal force of the other, thereby canceling horizontal forces. The weight
of the unbalanced masses varies sinusoidally from 2500 Ibs. to 1500
Ibs., thereby producing a cyclic force of 1000 Ibs. peaktopeak at a
frequency of 8 Hz (see Figure 3.1). The cyclic force is alternately
added to and subtracted from the 20001b. static weight of the trailer.
The 10001b. cyclic force is transmitted to the pavement through a pair
of polyurethanecoated steel wheels that are 4in. wide and 16in. in
outside diameter. These rigid wheels are spaced 20in. center to center
(see Figure 3.2).
The pavement response to the dynamically applied load is measured
by five geophones located as shown in Figure 3.2. The first geophone
measures the deflection at a point midway between the rigid wheels while
the remaining four sensors measure the deflection occurring directly
11/f A
f = Driving Frequency = 8 Hz
T
PeaktoPeak
 Dynamic Force
I= 1000 Ib
TIME
Typical Dynamic Force Output Signal of Dynaflect (108)
Figure 3.1
Housing and Tow Bar
Loading Wheels \ No.4 No.5
Geophones
(a) The Dynaflect System in Operating Position
Loading
Wheels Geophones
10 No.1 No. 2 No. 3 No. 4 No. 5
^ ~A^*^ ^
10" 12" 4]12" H412" I412" .J
4"
T
hn^
(b) Configuration of Load Wheels and Geophones.
Figure 3.2 Configuration of Dynaflect Load Wheels and
Geophones in Operating Position (108)
beneath their respective locations along the centerline of the trail
er. However, the geophone configuration can be easily changed to a
desired pattern. Each geophone is equipped with a suitable base to
enable it to make proper contact with irregular surfaces (108).
Data are displayed by a digital readout for each sensor on the
control panel which is umbilically attached to the trailer and can be
placed on the seat of the towing vehicle beside the operator/driver.
All operations subsequent to calibration are performed from the control
panel by the operator/driver without leaving the towing vehicle.
3.1.2 Calibration
The Dynaflect unit is calibrated by placing the sensors on a cam
actuated platform inside the calibrator furnished with each unit (108).
This platform provides a fixed 0.005in. vertical motion at 8 cycles per
second. The corresponding meter reading of 5 mils is set in the control
unit by adjustment of an individual sensitivity control for each geo
phone. Subsequent deflection measurements are thus comparisons against
this standard deflection.
3.1.3 Testing Procedure
The normal sequence of operation is to move the device to the test
point and hydraulically lower the loading wheels and geophones to the
pavement surface (84,108). A test is performed and the data of the 5
geophone deflection readings are recorded. At this point the operator
has the option of raising both the sensors and the loading wheels or
only the sensors. With the rigid wheels down and the pneumatic tires
lifted, the trailer may be moved short distances from one measuring
point to another at speeds up to 6 mph on the loading wheels (108).
When the rigid wheels are out of contact with the ground, the
trailer is supported on pneumatic tires for travel at normal vehicle
speeds. The sensors and loading wheels are raised and lowered by remote
control to enable such moves to be made quickly without need for the
operator/driver to leave the towing vehicle (84,104,108).
3.1.4 Limitations
The general limitations of dynamic steadystate NDT devices have
previously been described. In addition to those, the technical limi
tations of the Dynaflect device include (109) peaktopeak loading is
limited to 1000 lbs., load cannot be varied, frequency of loading cannot
be changed, the deflection directly under the load cannot be measured,
and it is difficult to determine the contact area.
3.2 Description of the Falling Weight
Deflectometer Testing System
The Falling Weight Deflectometer (FWD) is a deflection testing
device operating on the impulse loading principle. As described pre
viously, there are various forms of the FWD, with the most widely used
one in the United States being the Dynatest Model 8000 FWD system. This
is the type used by the FDOT and in the study reported herein. There
fore, this section describes the operating characteristics of the Dyna
test FWD test system. Like the Dynaflect, the FWD is also trailer
mounted and can be easily towed by most conventional passenger cars or
vans.
The Dynatest 8000 FWD test system consists essentially of three
main components (34,110), namely
1. a Dynatest 8002 FWD,
2. a Dynatest 8600 System Processor, and
3. a HewlettPackard HP85 Table Top Computer.
3.2.1 The 8002 FWD
The Dynatest 8002 FWD consists of a large mass that is constrained
to fall vertically under gravity onto a springloaded plate, 11.8 in. in
diameter, resting on the pavement surface (see Figure 2.4). A load
range of about 1500 to 24000 Ibs. can be achieved by adjusting the num
ber of weights or height of drop or both. The impulse or impact load is
measured by using a straingaugetype load transducer (load cell). The
impact load closely approximates a halfsine wave (see Figure 2.5), with
a duration of 2530 msec which closely approximates the effect of moving
dualwheels with loads up to 24000 Ibs. (110).
Seven seismic deflection transducers or geophones in movable brack
ets along a 2.25 m raise/lower bar are used to measure the response of
the pavement to the dynamically applied load. The geophones, which are
50 mm in diameter and 55 mm high, operate at a frequency range of 2 to
300 Hz (34). One of the geophones is placed at the center of the plate,
with the remainder placed at radial distances from the center of the
plate (see Figure 3.3). In its present form, the FDOT measures deflec
tions at radial distances of 0, 7.87, 11.8, 19.7, 31.5, 47.2, and 63.0
in. from the plate center. These deflections are respectively called
D D D D D D and D in this study.
1 2 3 4 5 6 7
3.2.2 The 8600 System Processor
The Dynatest 8600 system processor is a microprocessorbased con
trol and registration unit which is interfaced with the FWD as well as
the HP85 computer (34,110). The processor is housed in a 19in. wide
Figure 3.3 Schematic of FWD LoadGeophone Configuration and
Deflection Basin (34)
case which is compact, light weight, and controls the FWD operation. It
also serves as a power supply unit for the HP85 computer.
The system processor performs scanning and conditioning of the 8
transducer signals (1 load + 7 deflections). It also monitors the
status of the FWD unit to insure correct measurements.
3.2.3 The HP85 Computer
The HewlettPackard Model 85 computer is used for input of control
and site/tests identification data as well as displaying, printing,
storing (on magnetic tape), editing, sorting, and further processing of
FWD test data (34,110).
3.2.4 Testing Procedure
The automatic test sequence is identified and programmed from the
HP85 keyboard. This includes the input of site identification, height
and number of drops per test point, pavement temperature, etc. When the
operator enters a "START" command, the FWD loading plate and the bar
carrying the deflection transducers will be lowered to the pavement
surface, the weight will be dropped from the preprogrammed heightss,
and the plate and bar will be raised again. An audible "BEEP" signal
tells the operator that the sequence is complete, and that he/she may
drive onto the next test point. A complete measuring sequence normally
takes about one minute, exclusive of driving time between test points,
for three or four drops of the falling weight (34,110).
The measured set of data (1 load + 7 deflections) will be displayed
on the HP85 for direct visual inspection, and the data will be stored
on the HP85 magnetic tape cartridge, together with site identification
information, etc. The display, printed results, and stored results can
be in either metric or English units (34).
3.2.5 Advantages
The primary advantages of the Dynatest FWD, like many other impulse
deflection equipment, are that the created deflection basins closely
match those created by a moving wheel load of similar magnitude (11,45,
110,123), and the ability to apply variable and heavier dynamic loads to
assess stress sensitivity of pavement materials. The Dynatest FWD test
system has the added advantage that the resulting deflection basin is
constructed from seven deflection measurements compared to five and
three deflections in the KUAB and Phoenix Falling Weight Deflectometers,
respectively.
3.3 BISAR Computer Program
The analyses and evaluation of NDT deflection data in this study
involved the use of BISAR, an elastic multilayered computer program.
BISAR is an acronym for Bitumen Structures Analysis in Roads. The
program, developed by Koninklijke/Shell Laboratorium, Amsterdam,
Holland, is a general purpose program for computing stresses, strains
and displacements in elastic multilayered systems subjected to one or
more uniform loads, acting uniformly over circular surface areas (32).
The surface loads can be combinations of a vertical normal stress and
unidirectional tangential stress.
The use of BISAR to compute the state of stress or strain in a
pavement requires the following assumptions (32):
1. Each layer of pavement acts as a horizontally continuous, isotropic,
homogeneous, linearly elastic medium.
2. Each layer has finite thickness except for the lower layer, and all
are infinite in the horizontal direction.
3. The surface loading can be represented by uniformly distributed ver
tical stresses over a circular area.
4. The interface conditions between layers can vary from perfectly
smooth (zero bond) to perfectly rough (complete bonding) conditions.
5. Inertial forces are negligible.
6. The stress solutions are characterized by two material properties,
Poisson's ratio and Young's modulus for each pavement layer.
BISAR was used over other layeredtheory programs because of its
availability, testedand provenreliability and accuracy (72,91,96),
and, also, its ability to handle variable layer interface conditions.
For example, McCullough and Taute (72) found that the ELSYM5 program (3)
which is based on the CHEVRON program (74) predicts unrealistic deflec
tions in the vicinity of the load. They therefore recommended the use
of BISAR in computing fitted deflection basins, especially if the
deflection measurements are made near the loading point. Also, Ruth et
al. (98) reported correspondence with Mr. Gale Ahlborn, who developed
the ELSYM5 program, that the program is unreliable for certain unpre
dictable combinations of material properties.
3.4 Description of Cone Penetration Test Equipment
The cone penetration test equipment consisted of a truckmounted
hydraulic penetration system, electronic cone penetrometers (95) and an
automated data acquisition system. Detailed descriptions of the truck
and its features have been presented by Davidson and Bloomquist (30).
The hydraulic system serves two functions (29): leveling the truck
and penetrating the cone. The leveling system consists of four
independently controlled jacks. The front two jacks are connected to a
2 ft. x 7 ft. reaction plate; the back two to separate 15in. circular
pads. The vehicle is lifted off the ground and leveling assured by
means of a spirit level. The penetrating system consists of a 20ton
ram assembly located in the truck to achieve maximum thrust from the
reaction mass of the vehicle. Two doubleacting hydraulic cylinders
provide a useable vertical stroke of 1.22 m. Prior to testing, the rams
are used to raise the telescoping roof unit. When locked in the raised
position, the unit allows full travel of the rams (29,30).
The cone penetrometers are of the subtraction type configuration,
with tip and friction strain gauges mounted on the central shaft
(29,95). Cone bearing is sensed by compression in the first load cell,
while the sum of cone plus friction is sensed in the rear load cell.
The friction value is then obtained by subtraction, which is done
electronically (29). The cones used also measure pore water pressure
and inclination. A cable, threaded through the 1meter long push rods,
transmits the field recording signals to the data acquisition system.
The University of Florida currently has three electric cones, with
rated capacities of 5, 10 and 15 metric tons. Each measures tip
resistance, local friction, pore pressure, and inclination. The 5 and
10ton cones are of standard configuration with 10cm tip areas and 150
cm friction sleeves. The larger 15ton cone has the capability of
testing in much stiffer soil materials. All three cones contain
precision optical inclinometers which output the angular deviation of
the cone from the vertical during penetration (30).
The electronic data acquisition system is capable of printing and
plotting penetration data directly on the job site. It consists of a
microprocessor with 128 k magnetic bubble memory, an operator's console
with keypad, an Okidata microline 82A printer and an HP 7470A graphics
plotter. The computer is programmed with preset limits defined to
protect the probe from overloading. If a limit is exceeded, the
computer automatically stops the test and displays the cause of the
abort (29,30).
The electronic cone penetration testing equipment has several
advantages, such as a rapid procedure, continuous recording, high
accuracy and repeatability, automatic data logging, reduction, and
plotting. The CPT provides detailed information on site stratification,
identification, and classification of soil types. Results have also
been correlated with several basic soil parameters, including different
deformation moduli. For example, Schmertmann's method (101) of
computing settlements in sands requires the in situ variation of Young's
modulus. This is obtained from the CPT cone bearing resistance.
3.5 Marchetti Dilatometer Test Equipment
The Marchetti Dilatometer test (DMT) is a form of penetration test
and is fully described in References 17, 29, 68 and 69. Basically, the
test consists of pushing into the ground a flat steel blade which has a
flushmounted thin circular steel membrane on one face. At the desired
depth intervals (usually every 20 cms) penetration is stopped, and
measurements are taken of the gas pressure necessary to initiate
deflection and to deflect the center of the membrane 1.1 mm into the
soil. These two readings serve as a basis for predicting several
important geotechnical parameters, using experimentally and semi
empirically derived correlations (17,50,68). The DMT sounding provides
indications of soil type, preconsolidation stress, lateral stress ratio
at rest (Ko), Young's modulus (E), constrained modulus (M), shear
strength in clays and angle of shearing resistance in sands.
The major components of the dilatometer test equipment are the
dilatometer blade, the gaselectric connecting cable, a gaspressure
source, and the readout (control) unit. Figure 3.4 shows a schematic
diagram of this equipment. In addition there is a calibration unit,
adaptors, electric ground cable and a tool kit containing special tools
and replacement parts. Detailed descriptions and functions of the
various components are presented by Bullock (17) and Marchetti and
Crapps (69).
The dilatometer blade, as shown in Figure 3.4, consists of a stain
less steel blade, 94 mm wide and 14 mm thick, bevelled at the bottom
edge to provide an approximate 16degree cutting edge. A 60 mm stain
less steel circular membrane is centered on and flush with one side of
the blade.
The control unit, housed in an aluminum carrying case, contains
various indicators, a pressure gauge and the controls for running the
test. The control unit gauge used in the current study had a range of
040 bars. Higher and lower range units are also available. This gauge
provides the gas pressure readings for the dilatometer test.
The dilatometer blade is advanced into the ground using standard
field equipment. The blade can be pushed or driven by one of the
following methods (29):
Pressure Gauge
Buzzer
High
Pressure
Tubing 
To Pressure
r Source
Regulator Valve
Drill Rods
SFriction Reducer Ring
(/ I  Diaphragm (60mm diam)
'm ,DMT Blade
14mm
Figure 3.4 Schematic of Marchetti Dilatometer Test Equipment (69)
1. Using a Dutch Cone Penetrometer rig. This method is believed to
yield the highest productivity, up to 250 or more tests per day.
2. Using the hydraulic capability of a drill rig.
3. Using the SPT rig hammer or similar lighter equipment.
4. With bargemounted equipment or by wireline methods for underwater
testing.
3.6 Plate Bearing Test
The plate bearing test conceptually belongs to the static force
deflection group of NDT procedures (79). However, it can also be con
sidered as a destructive field test since the testing requires the
construction and subsequent repair of a trench or test pit. The plate
bearing test consisted of the repetitivestatic type of load test out
lined in ASTM Test Procedure D 119564 (8). The main objective in this
test is to measure the deformation characteristics of flexible pavements
under repeated loads applied to the pavement through rigid, circular
plates. Burmister's twolayer theory (18,19,20) is generally used to
interpret plate load testing results (133).
The test equipment used by the Florida Department of Transportation
consists of a 12inch diameter steel plate, loading system, deflection
gauges and supports (41). A trailer loaded with a huge rubber container
filled with water is used as a reaction. A hydraulic jack assembly is
used to apply and release the load in increments. A detailed descrip
tion of the repetitivestatic plate load test is provided in ASTM test
standards (8, pp. 258260).
CHAPTER 4
SIMULATION AND ANALYSES OF NDT DEFLECTION DATA
4.1 BISAR Simulation Study
4.1.1 General
The Dynaflect and FWD loadinggeophone patterns were simulated in
the BISAR elasticlayer computer program to predict surface deflection
data for fourlayer pavement systems. A flexible pavement structure was
modeled as a fourlayer system with parameters shown in Figure 4.1.
The selection of layer thicknesses and moduli was based on typical
ranges in parameters representative of Florida's flexible pavement
systems. In general, the limerock base and stabilized subgrade thick
nesses were fixed at 8 in. and 12 in., respectively. Table 4.1 lists
the range of layer parameters used in the theoretical analysis. The
subgrade was generally characterized as semiinfinite in thickness with
an average or composite modulus of elasticity. However, the effect of
bedrock at shallow depth was also assessed by varying depth to bedrock
in a fivelayer system. Poisson's ratio was fixed at 0.35 for all the
pavement layers since it has negligible effect on computed deflections.
In using the layered theory to generate and analyze NDT deflection
data certain assumptions had to be made. The following assumptions were
made with the use of the BISAR program:
1. Pavement materials are homogeneous, isotropic, and linearly
elastic. Therefore, the principle of superposition is valid for
calculating response due to more than one load.
I 1
E2 2 H12
E3, 3 H3
E4, P4 3 H3
E4,I H4 = .
FourLayer Flexible Pavement System Model
Figure 4.1
Range of Pavement Layer Properties
Layer Layer Layer Poisson's Layer Modulus
Layer Layer Thickness
Number Type iness Ratio (ksi)
(in.)
1 Asphalt 1.0 10.0 0.35 75 1,200
Concrete
2 Limerock 8.0 0.35 10 170
Base
Stabilized
3 Subgrade 12.0 0.35 6 75
(Subbase)
4 Subgrade Semiinfinite 0.35 0.35 200
(Embankment)
Table 4.1
2. Pavement layers are continuously in contact at the interfaces with
shearing resistance fully mobilized between them, so that the four
layers act together as an elastic medium of composite nature with
full continuity of stresses and displacements.
3. The Dynaflect and FWD dynamic loads are modeled as static circular
loads. Thus, the peaktopeak dynamic force of the Dynaflect is
modeled as two pseudostatic loads of 500 Ibs. each uniformly
distributed on circular areas. The peak dynamic force of the FWD is
assumed to be equal to a pseudostatic load uniformly distributed on
a circular area representing the FWD loading plate.
4. Thickness and Poisson's ratio of each layer are assumed to be known.
4.1.2 Dynaflect Sensor Spacing
In order to determine the optimum locations of the five Dynaflect
sensors, additional ones were included in the BISAR simulation study.
These sensors were placed at intermediate positions near the loaded
wheel and first two (standard) sensors, with hope of detecting the
primary response of the upper pavement layers (surface and base
course). Figure 4.2 illustrates the loading and modified geophone
array.
The Dynaflect was modeled in the BISAR program using two circular
loaded areas, with deflection measurement positions as shown in Figure
4.2. Each load is 500 Ibs. in weight, and the contact area used in this
study was 64 in.2, resulting in an equivalent radius of 4.5 in.
4.1.3 FWD Sensor Spacing
The conventional sensor spacing used by the FDOT and four
additional sensor locations were utilized in the analytical study.
Sensors were placed at radial distances of 0, 7.87, 11.8, 16.0, 19.7,
Spacing
GEOPHONE NO.
Conventional Modified
Figure 4.2 Dynaflect Modified Geophone Positions
@X
24.0, 31.5, 36.0, 47.2, 63.0, and 72.0 in. from the center of the FWD
plate. The deflection measured by the last sensor (at radial distance
of 72.0 in.) is called D in this study. Sensors at 16, 24, and 36 in.
were only used to better define the deflection basin and were not incor
porated in the analysis for prediction equations, which is described
later in this chapter. Thus, eight deflection locations were actually
used in the analysis. These are called D1, D2, D3, D4, D5, D6, D7, and
D8 to represent, respectively, radial distances of 0, 7.87, 11.8, 19.7,
31.5, 47.2, 63, and 72 in. from the center of the FWD plate.
The FWD was modeled in the BISAR program as a circular loaded area
with deflection measurement positions as stated previously. The radius
of the loaded area was 5.91 in., and a load of 9 kips was used which
corresponds to onehalf of an 18kip singleaxle wheel loading.
4.2 Sensitivity Analysis of Theoretical NDT Deflection Basins
4.2.1 Parametric Study
Pavement surface deflection data generated from BISAR were ini
tially evaluated to see the effect of rate of change of each layer
modulus and/or thickness on the NOT deflection basins. This was accom
plished by using the pavement section shown in Figure 4.3 as a typical
Florida pavement under warm temperature conditions. Using the informa
tion in Figure 4.3, each parameter (modulus and thickness) for a given
layer was doubled or halved while the others were kept unchanged. For
example, the E value of 10 ksi was increased to 20 ksi without changing
E E2, E3 and the layer thicknesses. BISAR was then used to calculate
the NDT deflections. The original E value was also halved to 5 ksi and
the theoretical deflections were computed with BISAR. This procedure was
,',',',',ASPHALT CONCRETE' ,'
%%\` '\E1 = 150 ksi ^^^^^
\\%%^/^/^^ *VI\\\\\\^\\\\ \ .^ \\\ *
p p% ,p ,,,,,, I1 pp ~P P K .
*,.b'U 0 /./,y 0 . 0. ,1oo.*o;
.O.' .:O. LIMEROCK BASE COURSE.'..O
C OO0O :E =85 ksi.o.kc;.
.' .. . ..* ....... ::. ..
SSTABILIZED SUBGRADEAY:', Y
(SUBB'AS E(SU B B AS E).
... .......... ....: ...::........
.. .. ........
SemiInfinite
Figure 4.3 Typical FourLayer System Used for the Sensitivity Analysis
repeated for all layer moduli and thicknesses. The NDT device used in
the parametric study was the FWD with a 9kip loading and sensor spacing
as previously described. However, the findings also apply to the
Dynaflect loading system, under the principle of superposition and
linear elastic theory.
Figures 4.4 through 4.10 show the effect of change of either
modulus or thickness on the FWD deflection basins. The rate of change
of deflections is most pronounced with changes in E as compared with
the moduli of the upper pavement layers. In the case of the layer
thicknesses, the effect is most apparent with changes in the base course
thickness, t It is possible that the t2 effect was due to the high E2
relative to E3 and E4.
Table 4.2 shows the percent change in deflections as a result of
doubling or halving each layer modulus while keeping the other para
meters unchanged for the pavement section shown in Figure 4.3. The
table shows that changes in E4 affect the deflections to the greatest
degree. The percent change in deflection is highest for any sensor
position when the E4 value is changed. This change with regard to E4
also increases substantially for the sensors further away from the load
center. The table thus suggests that E4 contributes the most to the FWD
deflections.
Similar comparisons were also made for changes in deflections for
t values of 1.5 and 6.0 in. The effect of layer thicknesses, ti, t2,
and t3, on the theoretical FWD deflections were also studied and the
results are summarized in Table 4.3. In this table, t2 seems to have
the most effect. The effect of tI on the deflections becomes negligible
when the original value (t. = 3.0 in.) is halved.
