<%BANNER%>

Characterization of Concrete Containing Reclaimed Asphalt Pavement by Superpave Indirect Tensile Test and X-Ray Computed...

Permanent Link: http://ufdc.ufl.edu/UFE0044625/00001

Material Information

Title: Characterization of Concrete Containing Reclaimed Asphalt Pavement by Superpave Indirect Tensile Test and X-Ray Computed Tomography
Physical Description: 1 online resource (133 p.)
Language: english
Creator: Su, Yu-Min
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2012

Subjects

Subjects / Keywords: concrete -- ct -- idt -- rap
Civil and Coastal Engineering -- Dissertations, Academic -- UF
Genre: Civil Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Portland Cement Concrete (PCC) and concrete containing Reclaimed Asphalt Pavement (RAP) as aggregate were evaluated by the Superpave Indirect Tensile (IDT) strength test using a constant force-control and a displacement-control rate.  The tensile strength computed from the maximum load at 28 days curing time from the Superpave IDT strength test as well as the elastic modulus and Poisson’s ratio of the concrete obtained at 40%of its ultimate stress were found correlated well with the convention tests.  When the toughness was calculated by determining the area under the stress-strain plot up to the maximum stress, it can be used to differentiate between concretes containing different percentage of RAP.  The tensile strength of concrete without RAP at early age was seen to increase as the temperature decreased.  However, this effect of temperature was not seen among the concrete at later ages.  In addition, the tensile strength of concrete containing RAP was seen to decrease as the percentage of RAP and temperature increased.  The addition of RAP in concrete noticeably reduced the elastic modulus of the concrete but increased the toughness.  It is recommended to adopt the Superpave IDT strength test using a constant displacement-control rate of 0.00075 in/sec to test concrete mixtures in tension.  Furthermore, the test procedures of X-ray Computed Tomography (CT) in conjunction with the Superpave IDT were developed.  An image-processing technique was established to assess the distribution of air voids and the maximum variation of air voids measured from the central area of the specimen was within plus and minus 0.03%.  The volume of air voids in the concrete was observed to increase significantly when a concrete specimen was loaded to fracture.  As a concrete specimen was loaded and unloaded, air voids which were formed tended to develop in the cement paste.  The path of crack propagation in concrete containing 100% RAP was found to be longer than that in concrete without RAP, as examined on virtual slices from X-ray CT scan.  This explained the higher toughness in the concrete containing RAP.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Yu-Min Su.
Thesis: Thesis (Ph.D.)--University of Florida, 2012.
Local: Adviser: Tia, Mang.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2012
System ID: UFE0044625:00001

Permanent Link: http://ufdc.ufl.edu/UFE0044625/00001

Material Information

Title: Characterization of Concrete Containing Reclaimed Asphalt Pavement by Superpave Indirect Tensile Test and X-Ray Computed Tomography
Physical Description: 1 online resource (133 p.)
Language: english
Creator: Su, Yu-Min
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2012

Subjects

Subjects / Keywords: concrete -- ct -- idt -- rap
Civil and Coastal Engineering -- Dissertations, Academic -- UF
Genre: Civil Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Portland Cement Concrete (PCC) and concrete containing Reclaimed Asphalt Pavement (RAP) as aggregate were evaluated by the Superpave Indirect Tensile (IDT) strength test using a constant force-control and a displacement-control rate.  The tensile strength computed from the maximum load at 28 days curing time from the Superpave IDT strength test as well as the elastic modulus and Poisson’s ratio of the concrete obtained at 40%of its ultimate stress were found correlated well with the convention tests.  When the toughness was calculated by determining the area under the stress-strain plot up to the maximum stress, it can be used to differentiate between concretes containing different percentage of RAP.  The tensile strength of concrete without RAP at early age was seen to increase as the temperature decreased.  However, this effect of temperature was not seen among the concrete at later ages.  In addition, the tensile strength of concrete containing RAP was seen to decrease as the percentage of RAP and temperature increased.  The addition of RAP in concrete noticeably reduced the elastic modulus of the concrete but increased the toughness.  It is recommended to adopt the Superpave IDT strength test using a constant displacement-control rate of 0.00075 in/sec to test concrete mixtures in tension.  Furthermore, the test procedures of X-ray Computed Tomography (CT) in conjunction with the Superpave IDT were developed.  An image-processing technique was established to assess the distribution of air voids and the maximum variation of air voids measured from the central area of the specimen was within plus and minus 0.03%.  The volume of air voids in the concrete was observed to increase significantly when a concrete specimen was loaded to fracture.  As a concrete specimen was loaded and unloaded, air voids which were formed tended to develop in the cement paste.  The path of crack propagation in concrete containing 100% RAP was found to be longer than that in concrete without RAP, as examined on virtual slices from X-ray CT scan.  This explained the higher toughness in the concrete containing RAP.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Yu-Min Su.
Thesis: Thesis (Ph.D.)--University of Florida, 2012.
Local: Adviser: Tia, Mang.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2012
System ID: UFE0044625:00001


This item has the following downloads:


Full Text

PAGE 1

1 CHARACTERIZATION OF CONCRETE CONTAINING RECLAIMED ASPHALT PAVEMENT BY SUPERPAVE INDIRECT TENSILE TEST AND X RAY COMPUTED TOMOGRAPHY By YU MIN SU A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA I N PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2012

PAGE 2

2 2012 YU MIN SU

PAGE 3

3 To my beloved daughter, wife, and parents

PAGE 4

4 ACKNOWLEDGMENTS It is my privilege and honor to study and work under Dr. Man g Tia, my supervisory committee chair, who has been remarkably generous, resourceful, and patient to guide me through many tough moments not only work, but most importantly life. I am indebted on his saving grace, incubation, and encouragements to cultiva te me a better person. I also wanted to thank Dr. Reynaldo Roque for sharing knowledge on his invention toward my work with many innovative ideas Moreover, the gratitude must be conveyed to Dr. Larry Muszynski and Dr. Fazil Najafi for their time and eff orts to advise my work. My appreciations need to be expressed to colleagu es and friends with a n elongated list: male Yu Chen, female Yu Chen, Sanghyun Chun, Julan Feng, Yaming Han, Fa ngyuan Hua, Yen How Huang Marco Isola, K arol Kowalski Qing Liu, Yanjun Liu, Qiang Li, Mitsuhiro Narisawa, Hung An Tsai, Yuan Jien (Jerry) Tung, Tingting Wu, Xiaoguang Xie, Jian Zhou, and Mik e Dreznes with 21 IRF fellows in class of 2008 Special thanks for Patrick (Pat) Carlton, Harvey (Dale) DeFord, Wedd David, Mike Bergin and many other engineers at the State Materials Office of Florida Department of Transportation in helping and funding the project, George Lopp in technical advice and support, and Nabil Hossiney in working on the same research project together. Furthermo re, I wante d to thank additionally to whom ever extended your considerations on helping me in every aspect Lastly, I like to express the most gratefulness to my dear parents who take care of my baby girl since she was b orn, and my beloved wife who scari fies so much to allow me chase the dream. Without a doubt, my humble achievement is only built upon the concrete foundation that my family has poured.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 ABSTRACT ................................ ................................ ................................ ................... 13 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 15 1.1 Background ................................ ................................ ................................ ....... 15 1. 2 Research Objectives ................................ ................................ ......................... 16 1. 3 Hypothesis ................................ ................................ ................................ ........ 17 1. 4 Scope ................................ ................................ ................................ ................ 17 1. 5 Research Approach ................................ ................................ .......................... 17 2 REVIEW OF LITERATURE AND SPECIFICATIONS ................................ ............. 19 2.1 Conventional Tests to Evaluate Concrete Slabs ................................ ............... 19 2.2 Superp ave Indirect Tension Test ................................ ................................ .... 21 2.2.1 Development of Superpave IDT T est A pparatus ................................ ..... 21 2.2.2 Failure Load Detection and Fracture Ene rgy in Superpave IDT .............. 25 2.2.3 Exploratory Work Using Superpave IDT on Concrete ............................. 28 2.3 X Ray Computed Tomography ................................ ................................ ......... 29 2.3.1 Introduction of X ray Computed Tomography ................................ .......... 29 2.3.2 Fundamentals of X ray CT ................................ ................................ ...... 30 2.3.2.1 Radiography ................................ ................................ ................... 30 2.3.2.1 Projection of digital radiography ................................ ..................... 32 2.3.2.2 Geometric positioning and unsharpness issue ............................... 33 2.3.2.3 Process of performing an X ray CT scan ................................ .............. 35 2. 3.3 Application of U sing X ray CT on C oncrete and A sphalt M ixtures. .......... 35 3 EXPERIMENTAL PLAN ................................ ................................ .......................... 40 3.1 Selection of Materials ................................ ................................ ........................ 40 3.1.1 Portlan d Cement ................................ ................................ ..................... 40 3.1.2 Virgin Aggregates ................................ ................................ .................... 40 3.1.3 Reclaimed/Recycled Asphalt Pavement ................................ .................. 41 3. 2 Concrete Mix Proportions ................................ ................................ ................. 44 3. 3 Test P lan for Conventional Concrete Properties Tests ................................ ..... 46 3. 4 Test P lan fo r Superpave IDT Strength Test ................................ .................... 47 3. 5 Test P lan for X ray Computed Tomography ................................ ..................... 47

PAGE 6

6 4 EVALUATION OF PCC AND CONCRETE CONTAINING RAP BY CONVENTIONAL TESTS ................................ ................................ ....................... 48 4.1 Background ................................ ................................ ................................ ....... 48 4.2 Concrete Mixing and Sample Preparation ................................ ........................ 48 4.3 Fresh Concrete Properties and Hardened Concrete Tests ............................... 49 4.4 Results of Fresh and Harden ed Concrete Tests ................................ ............... 50 5 CHARACTERIZATION OF PCC AND CONCRETE CONTAINING RAP USING THE SUPERPAVE IDT STRENGTH TEST ................................ ............................ 52 5.1 Background ................................ ................................ ................................ ....... 52 5.2 Test Sample Preparation ................................ ................................ .................. 52 5.2.1 Slicing of Concrete Cylinder ................................ ................................ .... 52 5.2.2 Air Drying, Surface Cleaning, and Gage Points Attachment .................... 52 5.2.3 Temperature Conditioning ................................ ................................ ....... 53 5.2.4 Installing Strain Gages ................................ ................................ ............ 55 5.2.5 Positioning Concrete Specimen ................................ .............................. 55 5.3 The Superpave IDT Strength Test on Concrete Specimens ............................. 56 5.3.1 Loading Conditio n ................................ ................................ ................... 56 5.3.2 Data Collection and Processing ................................ .............................. 56 5.3.2.1 Data collection ................................ ................................ ............... 56 5.3.2.2 Data processing ................................ ................................ ............. 57 5.3.2.3 Determination of failure stress and peak stress ............................. 58 5.3.2.4 Determination of elastic m .................... 60 5.3.2.5 Determination of fracture energy and toughness ........................... 61 5.3.3 Computations of Concrete Propertie s ................................ ...................... 62 5.3.3.1 Indirect tensile strength ................................ ................................ .. 62 5.3.3.2 Elastic modulus ................................ ................................ .............. 63 ................................ ................................ ................ 63 5.3.3.4 Fracture energy and toughness ................................ ..................... 63 5.4 Results of Superpave IDT Strength Test under Force Control Loading ............ 64 5.4.1 Behaviors of Stress Strain Curve ................................ .......................... 64 5.4.2 Results of Indirect Tensile Strength ................................ ......................... 66 5.4.2.1 Indirect tensile strength (using peak stress) ................................ ... 66 5.4.2.2 Indirect tensile strength (using failure stress) ................................ 68 ................. 70 stress) ................................ ................................ ................................ ..... 70 5.4.3.2 Fracture energy (using failure stress) ................................ ............. 73 5.5 Results of Superpave IDT Strength Test under Displacement Control Loading ................................ ................................ ................................ ................ 75 5.5.1 Behaviors of Stress Strain Curve ................................ .......................... 75 5.5.2 Indirect Tensile Strength (using peak stress) ................................ .......... 76 ................................ ........................ 78 5.5.4 Toughness (using peak stress) ................................ ............................... 78 5.6 Correlations betw een Conventional and Superpave IDT Strength Tests .......... 79

PAGE 7

7 5.6.1 Tensile Strength ................................ ................................ ...................... 79 5.6.2 Elastic Modulus ................................ ................................ ....................... 81 ................................ ................................ ........................ 82 5.7 Research Findings ................................ ................................ ............................ 83 6 APPLICATION of X RAY COMPUTER TOMOGR APHY TECHNIQUE TO CONCRETE ................................ ................................ ................................ ............ 84 6.1 X ray Computed Tomography Equipments at the University of Florida ............ 84 6. 2 Procedures of X ray C omputed Tomography ................................ ................... 86 6.2.1 Calibration of Digital Flat Panel Detector ................................ ................. 86 6.2.2 Position of Test Sample ................................ ................................ ........... 87 6.2.3 Contrast Optimization in DR Images ................................ ....................... 87 6.2.4 Acquisition of DR Images ................................ ................................ ........ 88 6.2.5 R econstruction of 3 D Virtual Model ................................ ........................ 88 6.2.6 Export of 2 D Images of Virtual Slice ................................ ....................... 90 6.3 X ray CT in Conjunction with the Superp ave IDT Strength Test ....................... 92 6.3.1 Test Program ................................ ................................ ........................... 92 6.3.2 Concrete Mixes Evaluated ................................ ................................ ....... 93 6.3.3 Test Procedures ................................ ................................ ...................... 94 6.3.3.1 Geometry of load frame ................................ ................................ 94 6.3.3.2 Positioning of test sample on the load frame ................................ 9 4 6.3.3.3 Performing an X ray CT scan on concrete ................................ ..... 96 6.3.3.4 Exporting virtual slices on concrete ................................ ................ 97 6.3.4 Development of Image Processing Technique of Analyzing Air Voids .... 97 6.3.4.1 Crop the interested area ................................ ................................ 98 6.3.4.2 Apply proper threshold of gray level to isolate voids .................... 100 6.3.4.3 Calculating area of air voids ................................ ......................... 101 6.3.4.4 Performing the analysis of air voids distribution ........................... 101 6. 4 Analysis of Air Voids Distribution in Different Load Levels .............................. 102 6.4.1 Effects of Loading and Unloading on Concrete Mixes ........................... 102 6.4.2 Average Air Voids under Different Loading Level ................................ .. 103 6.4.3 Average Air Voids in Concrete containing RAP during Fracture ............ 104 6. 5 Visualization of Microcracking in Concrete ................................ ..................... 114 6.5.1 Visual ization of Concrete Microcracking under Load/Unload Pattern .... 114 6.5.2 Visualization of Concrete Microcracking during Fracture ....................... 122 6. 6 Research Findings ................................ ................................ .......................... 125 7 CONCLUSIONS AND RECOMMENDATIONS ................................ ..................... 127 7.1 Conclusions ................................ ................................ ................................ .... 127 7.2 Recommendations ................................ ................................ .......................... 129 LIST OF REFERENCES ................................ ................................ ............................. 130 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 133

PAGE 8

8 LI ST OF TABLES Table page 2 1 Proposition for using Superpave IDT to determine strength, strain, and fracture energy for PCC ( modified after Zheng, 2007) ................................ ........ 29 3 1 Gradation of coarse aggregate (#57) ................................ ................................ .. 41 3 2 Gradation of fine aggregate (#89) ................................ ................................ ....... 41 3 3 P rop erties of recovered binder from RAP ................................ ........................... 42 3 4 Gradation s of fractionated coarse and fine RAP ................................ ................. 43 3 5 Concrete mixes containing RAP t o be evaluated ................................ ................ 45 3 6 Mix proportions of PCC and c oncrete mixtures containing RAP ......................... 45 4 1 Standard tests on fresh and hardened co ncrete ................................ ................. 50 4 2 Results of hardened conventional concrete tests ................................ ............... 51 4 3 Results of fresh concrete tests ................................ ................................ ........... 51 5 1 Results of Indirect tension test (using peak stress) ................................ ............ 67 5 2 Results of Indirect tension test (using failure stress) ................................ .......... 69 5 3 toughness (using peak stress) ................................ ................................ ................................ ........................... 71 5 4 Results fracture energy (using failure stress) ................................ ..................... 74 6 1 Load/unload pattern used in Superpave IDT strength test ............................... 102 6 2 Average air voids under different load levels ................................ .................... 103

PAGE 9

9 LIST OF FIGURES Figure page 2 1 ( A ) Jig for aligning concrete cylinder of ASTM C496 splitting tensile test, ( B ) ASTM C78 three point flexural strength test ................................ ....................... 20 2 2 Tensile stresses in a diametrically loaded specimen ( after Buttlar et al., 1996 ) 25 2 3 Example of strength results by using Superpave IDT ( after Buttlar et al., 1996 ) ................................ ................................ ................................ ........................... 26 2 4 Schematic of dissipated creep strain energy (after Roque et al., 2002) ............. 27 2 5 Schematic of D R image with flaw (modified after NSI training materials, 2009) ................................ ................................ ................................ .................. 33 2 6 Schematic of geometric relations (modified after NSI training materials, 2009) 34 2 7 Typical load displacement plot for specimen (after Landis, 2000) ...................... 36 2 8 Typical steps involved in image acquisition processing and analysis ( a fter Masad and Kutay, 2012) ................................ ................................ .................... 39 3 1 Combined g radation chart for coarse aggregate s ................................ ............... 45 3 2 Combined g radation chart for fine aggregate s ................................ .................... 46 5 1 Gage point mounting system (Photo courtesy of Yu Min Su) ............................. 54 5 2 ESPEC environment chamber (Photo courtesy of Yu Min Su) ........................... 54 5 3 Thin concrete specimen on the load frame with strain gages (Photo courtesy of Yu Min Su) ................................ ................................ ................................ ...... 56 5 4 E xample of determining the failure load o n PCC with w/c=0.50 ......................... 59 5 5 Example of ( A ) secant modulus and ( B ) stress strain curve of PCC and concrete containing 70% RAP ................................ ................................ ............ 61 5 6 Examples of st r ess strain plots with two horizontal strain gages in the Superpave IDT test under force control load rate. ................................ .............. 65 5 7 Comparison of Indirect tensile s trength results in PCC contr ol mixes (using peak stress) ................................ ................................ ................................ ........ 67 5 8 Comparison of Indirect tensile s trength results in concrete mixes containing RAP (using peak stress) ................................ ................................ ..................... 68

PAGE 10

10 5 9 Comparison of Indirect tensile s trength results in PCC control mixes (using failure stress) ................................ ................................ ................................ ...... 69 5 10 Comparison of Indirect tensile s trength results in concrete mixes containing RAP (failure stress) ................................ ................................ ............................ 70 5 11 Comparison of elastic modulus in PCC and concrete containing RAP ............... 72 5 12 atio in PCC and concrete containing RAP ................ 72 5 13 Comparison of toughness in PCC and concrete containing RAP (force control) ................................ ................................ ................................ ............... 73 5 14 Comparison of Fracture energy in PCC and concrete containing RAP (using failure stress) ................................ ................................ ................................ ...... 75 5 15 Examples of st r ess strain plots with two horizontal strain gages in the Superpave IDT test un der displacement control load rate ................................ .. 76 5 1 6 C omparison of tensile strength obtained from force control versus displacement control tests ................................ ................................ .................. 77 5 17 versus force control and displacement control in the Superpave IDT strength tests ................................ ................................ ................................ .................... 78 5 1 8 Comparison of t oughness obtained from force control versus displacement control tests ................................ ................................ ................................ ........ 79 5 19 C orrelation of tensile strength between splitting and Superpave IDT strength test under 23C ................................ ................................ ................................ ... 80 5 20 Correlation of elastic modulus between conventional and the Superpave IDT strength tests ................................ ................................ ................................ ...... 81 5 21 entional and the Superpave IDT strength tests ................................ ................................ ................................ ...... 82 6 1 Major components of X ray CT facility at UF ................................ ...................... 85 6 2 D iagonal plan view of CT scan chamber ................................ ............................ 85 6 3 Acquisition of spatial information by 3 D calibration tool ................................ ..... 89 6 4 3 D reconstructed model of a concrete spe cimen ................................ .............. 90 6 5 (A) A cross sectional image of reconstructed 3 D model in concrete after process of window leveling on Z axis (B) Histogram of the 3 D model ............... 91

PAGE 11

11 6 6 The load frame set up under Superpave IDT apparatus (Photo courtesy of Yu Min Su) ................................ ................................ ................................ .......... 95 6 7 DR image of the Superpave IDT concrete specimen with strain gages .............. 96 6 8 Cropping the interested area from a virtual slice ................................ ............... 98 6 9 Cropped image 1.5 inches square on each slice. ................................ ............... 99 6 10 Cropped image after binary process (threshold=87). ................................ ........ 100 6 11 Air void distribution of w/c=0.45 PCC mix. ................................ ........................ 106 6 12 Air void distribution of w/c=0.55 PCC mix. ................................ ........................ 107 6 13 Air void distribution of concrete mix containing 20% RAP. ............................... 108 6 14 Air void distribution of concrete mix containing 40% RAP. ............................... 109 6 15 Air void distribution of concrete mix containing 70% RAP. ............................... 110 6 16 Air void distribution of concrete mix containing 100% RAP. ............................. 111 6 17 Voids distributions after fracture of w/c=0.45 and 0.55 PCC mixes. ................. 112 6 18 Voids distributions after fracture of concrete mixes containing 40% and 100% RAP. ................................ ................................ ................................ ................. 113 6 19 Air voids distribution in w/c=0.45 PCC with location s of interest. ...................... 115 6 20 Air voids at the location of 0.069 inches. ................................ .......................... 116 6 21 Air voids at the location of 0.15 inches. ................................ ............................ 117 6 22 Air voids at the location of 0.32 inches. ................................ ............................ 118 6 23 Air voids at the location of 0.52 inches. ................................ ............................ 119 6 24 Air voids at the location of 0.72 inches. ................................ ............................ 120 6 25 Air voids at the location of 1.00 inches. ................................ ............................ 121 6 26 Air voids at the location of 0.60 inches of 100% RAP. ................................ ...... 122 6 27 Air voids at the location of 0.09 inches of 100% RAP (voids reduction). .......... 122 6 28 Air voids at the location of 0.60 inches of 100% RAP (voids dull). .................... 123 6 29 Air voids at the location of 1.00 inches of 100% RAP (voids growing). ............. 123

PAGE 12

12 6 30 Fracture pattern in w/c=0.45 PCC mix. ................................ ............................. 125

PAGE 13

13 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 CHARACTERIZATION OF CONCRETE CONTAINING RECLAIMED ASPHALT PAVEMENT BY SUPERPAVE INDIRECT TENSILE TEST AND X RAY COMPUTED TOMOGRAPHY By Yu Min Su December 2012 Chair: Mang Tia Major: Civil Engine ering Portland Cement Concrete (PCC) and concrete containing Reclaimed Asphalt Pavement (RAP) as aggregate were evaluated by the Superpave Indirect T ensile (IDT) strength test using a constant force control and a displacement control rate The tensile st rength computed from the maximum load at 28 days curing time from the Superpave IDT strength test as well as t obtained at 40% of its ultimate stress were found correlated well with the convention tests When the toughness wa s calculated by determining the area under the stress strain plot up to the maximum stress, it can be used to differentiate between concretes containing different percentage of RAP. The tensile strength of concrete without RAP at e arly age was seen to increase as the temperature decreased. However, this effect of temperature was not seen among the concrete at later ages. In addition, t he tensile strength of concrete containing R AP was seen to decrease as the percentage of RAP and temperature increased. The addition of RAP in concrete noticeably reduced the elastic modulus of the concrete but increased the toughness. It is recommended to adopt the Superpave IDT strength test using a constant displacement control rate of

PAGE 14

14 0.00075 in /sec to test concrete mixture s in tension. Furthermore, the test procedures of X ray Computed Tomography (CT) in conjunction with the Superpave IDT were developed. An image processing technique was established to assess the distribution of air voids and t he maximum variation of air voids measured from the central area of the specimen was within The volume of air voids in the concrete was observed to increase significantly when a concrete specimen was loaded to fracture. As a con crete specimen was loaded and unloaded, air v oids which were formed tended t o develop in the cement paste The path of crack propagation in concrete containing 100% RAP was found to be longer than that in concrete without RAP, as examined on virtual slice s from X ray CT scan. This explained the higher toughness in the concrete containing RAP.

PAGE 15

15 CHAPTER 1 INTRODUCTION 1.1 Background Reclaimed asphalt pavement (RAP) materials have been used in asphalt pavement mixtures in Florida, resulting in substantial savings in cost and in conservation of aggregates and asphalt. However, with the adoption of the more stringent Superpave mix design method in Florida in recent years, a smaller percentage of RAP is now being used in the recycled asphalt mixtures. This has resulted in an excess of RAP which should be put into good use. A previous FDOT research study (Tia et al 2009) demonstrated the pro cess of blending RAP in concrete for pavement slabs. Another study (Huang et al 2006) also showed that the toughn ess and crack resistance of concrete could be improved by the addition of RAP. Flexural strength of concrete is normally used as an input parameter in the analysis of concrete pavement. The specific test ASTM C78 is used to determine the flexural streng th of concrete. However, the flexural strength test requires a relatively large beam specimen. It is often inconvenient to cut such a large specimen fr om a concrete slab. It is therefore necessary to develop a more convenient test method to characterize the tensile property of concrete in place of the flexural strength test. One promising test for this purpose is the Superpave Indirect t ensile test. The Superpave Indirect tensile testing (IDT) developed under the Strategic Highway Research Program ( SHRP) is capable of determining resilient modulus, creep and indirect tensile strength of an asphalt mixture. A well defined failure plane located in the near central area of specimen can be assessed by measuring the plane deformations of two edges on a thin circular specimen, which can offer the stress

PAGE 16

16 strain curve for further assessment A previous exploratory work (Zheng, 2007) was done to evaluate concrete specimens by the Superpave IDT strength test and described a test protocol to determine concre te properties such as tensile strength, failure strain, and fracture energy. This protocol may enable testing of concrete properties conveniently o n thin concrete specimen s slic ed from the cylinder prepared in the laboratory or cored from the field The n ovel X Ray Computed Tomography (CT) facility installed at the Advanced Materials Characteristics Laboratory (AMCL) in the Department of Civil and Costal Engineering can receive two dimensional (2 D) digital radiography image s and accumulated 2 D images ca n be reconstructed as a virtual three dimensional (3 D) model for further analysis. The re are two X ray tubes: a micro focus 225 kV ; and a high power 450 kV, which could offer high resolution and high penetration capabilities to inspect the test object. For this study, i t was necessary to study the failure mechanism of concrete containing RAP since concrete mix containing RAP has a higher ductility than concrete without RAP The Superpave IDT test with well defined failure plane which is capable of de termining tensile strength, vertical and horizontal strain can approximate the characteri stic properties for concrete slabs. In addition, the non destructive X ray CT scan can potentially be a very useful instrument for visualizing the internal elements of concrete mixtures, observing the initiation of microcracking, and identifying crack propagation. 1. 2 Research Objectives The objective was to assess the Superpave IDT strength test to evaluate Portland Cement Concrete (PCC) and concrete containing RA P by refin ing test procedures,

PAGE 17

17 study ing stress s train behavior and find ing the relat ionship between strength test results and conventional concrete properties as well as using X ray CT in conjunction with the load frame to visualize internal microstruct ure in concrete mixes 1. 3 Hypothesis The Superpave IDT strength test can effectively characterize concrete stress strain behavior of PCC and concrete mixes using RAP in tension ; and the X ray CT in conjunction with the load frame can be used to assess mi crostructure in concrete mixes 1. 4 Scope Three FDOT approved PCC mixes with w/c=0.4 5, 0.50, and 0.55 and four other concrete mixes containing 20%, 40%, 70%, and 100% of RAP with w/c=0.5 were produced and evaluated in this study. Conventional fresh concre te properties were tested immediately after production. Harden ed concrete properties were obtained by different moisture curing age s, including 7, 14, 28, and 90 day s and tested con ventional ly and evaluated by the Superpave IDT strength test X ray Comp uted Tomography was adopted in conjunction with the Superpave IDT s trength test to assess distribution of air voids in concrete mixes cured in 90 days 1. 5 Research Approach The objectives were accomplished through the following tasks in this study Task 1: Specifications and literature revie w were conducted, including convention al design and test of concrete slab, Superpave IDT test on both PCC and HMA mixes, and current applications of X ray CT shown in Chapter 2. Task 2: C oncrete mix design s and sam ple preparation of PCC and concrete containing RAP were developed and performed including selection of materials determination of mix designs, production of concrete mixes, and preparation of test samples, presented in Chapter 3.

PAGE 18

18 Task 3: Conventional co ncrete t esting program of PCC and concrete containing RAP were developed and performed including fresh concrete tests and harden ed concrete tests. The test results are show n in Chapter 4. Task 4: The use of Superpave IDT strength test in concrete was re fined and performed to examine PCC and concrete mixes containing RAP. The test results and analysis are indicated in Chapter 5. Task 5: Procedures for performing a n X ray CT scan in conjunction with the load frame and image processing technique were deve loped. Air void distribution and crack initiation, and propagation were assessed. The test procedures analysis of air voids, and cracks in concrete mixes are presented in Chapter 6.

PAGE 19

19 CHAPTER 2 REVIEW OF LITERATURE AND SPECIFICATIONS 2.1 Conventional Tests to Evaluate Concrete Slabs It is well recognized that the tensile strength of concrete is much less than the compressive strength and con sequently is a c ritical factor for designing concrete slab s of rigid pavement. There are two regular tests for evaluating tensile strength in concrete: ( ( (ASTM C78). The former is used to measure the tensile strength induced by the axial loading applied vertically in order to create nearly uni form stress on this vertical failure plane. The latter is employed to measure the tensile property induced by three point bending on the section where only pure movement is present, with zero shear force. I t has to be noted that the three point test is pr eferable to the center point test (ASTM C293), because it is difficult to define clearly the cause of failure either from movement, shear stress or undetermined stress concentrations in the splitting tensile stress test. The flexur al strength test has bee n commonly used for designing concrete slab s because the scheme of three point bending enables the measurement of flexural strength without the effects of shear stresses. It is easier to prepare concrete cylinders for the splitting tensile t est than for t he flexur al strength test in the laboratory, due to the size and weight of the required cylinder It requires either four inch by eight inch (diameter by length) or six inch by twelve inch concrete cylinders for the splitting tensile test, shown in Figure 2 1 ( A ); while the flexural strength test normally requires a beam of six inch by six inch by twenty inch (width by depth by span length), shown in Figure 2 1 ( B ). As to field samples, it is necessary to drill the circular cylinders or saw suitable concre te beams for both tes ts,

PAGE 20

20 respectively. The size and weight of the concrete beam makes the flexural strength test rather difficult to perform. In addition, the period of drying during processes of handling and transportation may cause a significant reduct ion of values in tensile strength. On the other hand, the sample for the splitting tensile apparatus needs only to apply axial loading on the laid down cylinder with identical sample size as that for the compressive strength test and there is a better de fined failure plane to be analyzed. However, the crack initiation may occur in any place along the length of concrete cylinder where it may be distant from both end surfaces, while performing the splitting tensile test. This locational uncertainty creat e s difficulty in accurately detecting the deforma tion when microcracks and subsequent fractures are initiat e d Additionally, Florida Department of Transportation requires a standard value of 635 psi in flexural strength (FDOT, 2009) and in modulus elasticity over the mix subject to 28 d ay curing time. The compressive strength must reach 2200 psi before the pavement is open ed for traffic. ( A ) ( B ) Figure 2 1. ( A ) Jig for aligning concrete cylinder of ASTM C496 splitting tensi le test, ( B ) ASTM C78 three point flexural strength test

PAGE 21

21 2.2 S uperpave Indirect Tension Test 2.2.1 Development of Superpave IDT T est A pparatus The Superpave indirect tensile test (IDT) was developed to satisfy the requirement of the Strategic Highway Re search Program (SHRP) to determine properties of hot mix asphalt (HMA) including: resilient modulus; creep modulus; and tensile strength ( Roque and Buttlar, 1992, 1994 ) A later study ( Roque et al., 1997 ) was utilized to establish a data reduction system for the IDT test ; this system can automatically compute resilient modulus, creep, and indirect tensile strength as well as providing an assessment of fracture energy. AASHTO T322 has detailed descriptions of the layout of the load frame, gage point mount ing, a nd modulus calculations for asphalt concrete. The common procedure is that the asphalt specimen is sliced from a cylinder s ix inches in diameter Specimen thickness can be determined by o ne and one half inches for dense graded mixes and two inche s for open graded friction course (OGFC) mixes, respectively. F our linear variable differential transformers (LVDTs) are employed in total to measur e deformations in horizontal and vertical directions on both s ides of a specimen. The load frame is design ated a servo hydraulic and closed loop feedback control device which can constantly control either force control or displacement control rate by 20 Hz (tw enty measurements per second) of record ing testing data. A pre load of ten to fifteen pounds is rout inely applied, before starting the test, to avoid impact effect on testing samples. A series of testing data including testing time, axial load, axial deformation, and de formations on four gages are recorded and exported to a text file, and as such can d erive a stress strain curve for further analysis.

PAGE 22

22 The analysis of the indirect tensile test can be traced back to an early study ( Hondros, 1959, ) that derived a full analytical solution of stress analysis for evaluating Portland Cement Concrete (PCC). For the Superpave IDT test, the distribution of tensile stress along the vertical and horizontal axis is modified and the plane stress condition near the center is as follows ( Lee et al., 2011 ) : = ( 2 1 ) = ( 2 2 ) where, = applied load, lbf. a= rim width of loaded section d= diameter of specimen x= radial dist ance from the center R= radius of specimen = angular displacement from the vertical axis Therefore, the average stresses with the gage length were defined as: ( 2 3 ) ( 2 4 ) And the horizontal and vertical strain s can be derived from the strain gage measurements as:

PAGE 23

23 ( 2 5 ) ( 2 6 ) where = horizontal strain and =vertical strain over the gage length Three major tests: resilient modulus; creep ; and indirect tensile tests are routinely performe d by Superpave IDT in sequence as follows: Resilient modulus test This test is performed by applying the repeated load resulting in horizo ntal deformation within the range of 200 300 microstrain. The repeated load was applied in the form of a 0.1s loading followed by a 0.9s rest period. The resilient modulus or MR can be calculated as follows: ( 2 7 ) In which stands for a non dimensional factor that varied linearly with ratio of horizontal to vertical deformation as follows: ( 2 8 ) whe re = ratio of horizontal to vertical deformation. Creep test The creep compliance test is performed following the resilient modulus test on the specimen. The constant load is chosen to keep horizontal deform ation within the ra n ge of 200 750 microstrain after 1000s of loading. Hence the creep compliance is computed as follows:

PAGE 24

24 ( 2 9 ) w here is defined as the creep comp liance at the given time. Indirect t ensile s trength t est A con stant displacement control rate of 0.5 in/min is applied to the test specimen in order to generate the stress and strain curve. In order to account for the three dimensional effect of the stress state, the stress and strain must be multiplied by the correction factors, CSX and CBX, respectively. Both correction modulus test. ( 2 10 ) where, = average specimen thickness = average specimen diameter Hence, the indirect tensile strength is determined as follows: ( 2 11 ) ( 2 12 ) where,

PAGE 25

25 2.2.2 Failure Load Detection and Fracture Energy in Superpave IDT (a) Diametrally loaded specimen (b) Tensile stress distributions along axis of symmetry Figure 2 2 Tensile stresses in a diametr ic ally loaded specimen ( after Buttlar et al., 1996 ) The stress state within Superpave IDT was found not to be a plane stress (Heinicke et al., 1988, Roque et al., 1992, and Buttlar et al., 1996) as is shown in Figure 2 2 In plan e stress state (2 D), the tensile strength is constant, while the strength in the 3 D analysis varies along the axis of symmetry and will reach maximum at the vicinity of the end surf ace on both side s of a specimen. It was suggested that the near surface failure load is less than the load that can break a specimen apart. Hence, the failure load wa s defined and determined by the stress level at the failure instant of the specimen edge The Superpave IDT strength test along with vertical and horizontal displacement measurements was found to be capable of ascertaining the failure load. Figure 2 3 has detailed steps for finding the failure load. F ailure load is shown as a rapid incr ease in the rate of horizontal deformation at the precise instant of failure occurring on the edge of a specimen.

PAGE 26

26 Figure 2 3 Example of strength results by using Superpave IDT ( after Buttlar et al., 1996 ) Figure 2 3 ( A ) presents a constant ly increas ing rate of applied load, Figure 2 3 ( B ) indicates deformations of two vertical (Y1&Y2) and two horizontal (X1&X2) gages shown against time, and Figure 2 3 ( C ) presents two relationship lines by plotting the differences between vertical and horizontal defo rmations (Y X). As the applied load

PAGE 27

27 increases over time, deformations of four LVDTs are gaining. It was found at the imminent occurrence of cracking ; one horizontal deformation r is es more than the other rates, which also causes the significant reduction on the (Y X) line Thus, the instant of failure is identified at the peak of (Y X), and the failure load will be observed in the same second. Figure 2 4 Schematic of dissipated creep st r ain energy (after Roque et al., 2002) Roque et al ( Z hang, 2000, Zhang et al., 2001, Roque et al., 2002 ) concluded and proposed the HMA fracture mechanics by performing the Superpave IDT test to determine the stress strain response in order to generate the dissipated creep strain energy threshold, or : The definition is as follows: ( 2 13 ) where FE stands for the fracture energy which is the integral area under the stress strain curve over the failure strain and EE stands for the elastic energy. The schematic of DCSE is shown in Fi gure 2 4

PAGE 28

28 2.2.3 Exploratory Work U sing Superpave IDT on Concrete A pilot study ( Zheng, 2007 ) attempted to ada pt the Superpave IDT strength test with an acous tic emission (AE) device to evaluate concrete in terns of indirect tensile strength, microdamage, and fracture properties. The researcher defined the failure load in HMA fracture mechanic concrete The secant modulus from the stress strain curve was found to identify the onset of microdamage well, considering the integrity of concrete being affected negatively by damage. In conjunction with AE, the study also verified the onset of first fracture initiating at approximately se venty to ninety percent and the onset of microcracking starting at about forty percent of the peak load from the Superpave IDT strength test. A trial determining prop er loading rate was suggested to adopt a displacement control rate of 0.00075 in/sec (1. 14 mm/min) when performing the Superpave IDT strength test in concrete. The researcher proposed a protocol for the Superpave IDT strength test in concrete, including specimen preparation, test procedures, and analysis consonant with A ASHTO T332. Three specimens are need ed and the failure load for each specimen is the value corresponding to the inst ant when first peak of (Y X) has been reached. The fai lure strength and strain are identified individually from each spe cimen, and average values ar e reporte d. The det obtained at hal f of the average failure load from three specimens. However, it is the stress strain curve up to the instant of failure load that is chosen for calculating the fracture energy for each specimen. It is shown in Table 2 1 that the average failure strength, strain, and fracture energy are calculated from three replicate specimens in this protocol

PAGE 29

29 Table 2 1. Proposition for using Superpave IDT to determine strength, strain, and fracture energy for PCC ( modified after Zheng, 2007 ) Specimens Strength F ailure strain Fracture energy (psi) (in/in) (lb*in/in 3 x10 6 ) Replicate 1 Replicate 2 Replicate 3 Average T here are merits in the proposed protocol of using the Superpave IDT strength test on concrete as follows, compared with ASTM C496 and ASTM C469: The protocol suggests a displacement control loading rate 0.00075 in /sec, while loading rate is 100 200 psi/min of pressure c ontrol in ASTM C496 instead. The failur e lo ad is determined by th e instant aneous loads of first (Y X) strain peak while the peak load of the splitting tensile test is chosen for determining the indirect tensile strength in ASTM C496 conventionally. An equation to i can be calculated directly from vertical and horizontal displa ce m e nt measurements when the load level is at half of the failure load. No suggestion on obtaining elastic modulus is made in t his protocol. Th e modulus of elasticity ( or ) peak load in compression on a concrete cylinder in ASTM C469 Th e fracture energy is proposed to calculate the area under stress str ain curve up to the failure s tress in the protocol. There is no similar standard to calculate the facture energy of PCC in indirect tension The fractur al toughness determination in PCC is regularly tested on a three point bending beam with a notch, which is a completely different s etup than the concept of Superpave IDT It is apparent that th e discussions above may need to be validated in order to warrant the use of the Superpave IDT strength test on concrete 2.3 X Ray Computed Tomography 2.3.1 Introduction of X ray Computed Tomo graphy X ray C omputed T omography (CT) is a nondestructive testing (NDT) to investigate test object s without physically cutting or slicing In nuclear medicine, Computerized

PAGE 30

30 Axial Tomography, or CAT scan, is also often referred to as a CT scan (Buzug, 2008 ). It is common to employ an industrial X ray source or an X ray synchrotron facility to perform the CT scan. Other medical techniques such as Magnetic Resonance Imaging (MRI) and Positron Emission Tomography (PET), are also frequently mentioned. MRI i s the technique of detect ing the body magnetization of atomic nuclei using a power ful magnet to generate the magnetic field, while a PET scan detect s the gamma ray s generated by positron emission s PET scan may be incorporated with X ray CT for a better assessment. In medical CT or CAT scan, the equipment has evolved from the first pencil beam to a wider aperture fan beam while using the same principle of rotating the aligned x ray source and detector scanning around the patient, while industrial CT sca nner s tend to be fixed to scan the moving or rotating object. The X ray scanner also has developed from the fan (line), to spiral ( he lical) and cone beam type to acquire the imag e 2.3.2 Fundamentals of X ray CT 2.3.2.1 Radiography Radiography includes u sing of X ray to penetrate the target object and the radiation is absorbed or attenuated by the target materials. The unabsorbed X ray beams then pass through the object and the projection of attenuated X ray can be captured by the photographic film conve ntionally. The Computed Radiography (CR) was developed to digitalize the traditional X ray film into digital image. Moreover, it was even more convenient to utilize the flat panel detector (FPD) to acquire the digital radiography (DR) directly from proje ction of X ray. Due to the different density and thickness of materials, it presents variations of gray scale on the projected DR image to reflect difference of X ray attenuation. Hence,

PAGE 31

31 it is able to examine the internal structure by evaluate changes in gray scale on a DR image. ASTM E1411 (ASTM, 2011) presents a detail description about the mathematical and quantitative expressions in attenuation. It describes that the X ray attenuation is governed by Lambert s law of absorption and can be expressed a s follows: (2 14) where, intensity of transmitted X ray flux after it has traversed a material with thickness and intensity of emitted X ray linear absorption ( attenuation ) coefficient. If a non homogenous material is penetrated by the X ray, the equation (5 3) can be modified into (2 15) where, linear absorption coefficient at each point on the path of X ray. It has also to be noted that a term commonly known in nuclear medicine, the Hounsfield unit (HU) or so called CT number. It is the measurement of linear attenuation coefficient (or a radiodensity ) ratio compared with distilled water tested under STP. It is defined the Hounsfield unit of water as zero and subsequently the radiodensity of air is 1000. The defini tion of Hounsfield unit presents as follow: (2 16) where, average linear attenuation coefficient of target material,

PAGE 32

32 average linear attenua tion coefficient of distilled water tested under standard temperature and pressure. 2.3.2.1 Projection of digital radiography It has been said that the projected DR image composed a map of pixels with varied gray scale values digitally is the essence of NDT for X ray CT technique Figure 2 5 shows a schematic of projection of flaw on a DR image. The flaw within a specimen normally describes a different material density, such as an air void or a different substance, which is going to absorb the radiatio n differently than the surrounding material. Thus, the projected flaw of DR image will reflect the difference of traversing radiation, namely HU value or gray levels. In terms of DR image, air void can be noticeable significantly than others, since the s ubstance of void is more and less air, showing approximately with 1000 HU value or appearing brighter in gray levels. It has to be noted that the subject thickness plays an important part for the X ray attenuation. The thicker the subject is, the more a bsorption of radiation, which will increase the HU value or appearing darker in gray levels. It is now convenient to adopt the digital detector, for instance FPD, record DR image, and immediately show it on the computer monitor for further examinations.

PAGE 33

33 Figure 2 5 Schem atic of DR image with flaw (modified after NSI training materials, 2009) 2.3.2.2 Geometric positioning and unsharpness issue There is the unsharpness (U g ) issue with regards to the quality of projection DR images. Geometric positioning influence s the clarity and contrast of projected DR image, due to different focal spot sizes of X ray sources. Figure 2 6 shows that X ray beams pass through the test object and as such results the unsharpness around edges of targeted object. When the size of focal spot is small, it normally can produce quite clear image with minor unsharpness around the edge. On the contrary, X ray source with larger focal spot size offers DR image with the less unsharpness only if does o bject move close to FPD (eg. away from X ray source). If the targeted subject moves forward to the X ray source with large focal spot, the diffraction and interference of traversing X ray causes the unsharpness.

PAGE 34

34 Figure 2 6. Schem atic of geometric relations (modified after NSI training materials, 2009) It takes time and efforts to find the best fit in geometric positioning of the targeted object, associated with the use of which X ray sources. Additionally, geometric magni fication can be also achieved by manipulating the location of test object, namely the positioning of turn table or load frame in this facility. The geometric relations of unsharpness (U g ) and magnification (M) are as follows: (2 17)

PAGE 35

35 (2 18) where, U g =unsharpness M= magnification, F= focal spot size of X ray source, t = distance between targeted objec t and detector, and D= distance between targeted object and X ray source. 2.3.2.3 Process of performing an X ray CT scan The process of completing a n X ray CT scan normally requires two major steps: the acquisition of two dimensional ( 2 D ) digital radiog raph ic (DR) images and the reconstruction of three dimensional ( 3 D ) model by stacking DR images of the targeted object. After the completion of the reconstruction, the reconstructed 3 D model can be virtually rotating zooming in / out, or slice specimen a part to characterize structural properties w ithout intrusion or physical slicing to the test subject The virtual 3 D model can be the source of exporting virtual sliced 2 D images on the designated axis for further image analysis. 2. 3.3 Application of U s ing X ray CT on C oncrete and A sphalt M ixtures. A series of International Workshop s on X Ray CT for Geomaterials have been held, initiating in Kumamoto, Japan (GeoX 2003), Aussois, France (GeoX 2006), and New Orlean s USA (GeoX 2010). Each workshop conc luded and published the latest developments on the application of X ray CT for investigating geotechnical materials, including minerals, aggregates, soils, concrete and asphalt specimen, and on damage assessments in the internal structural properties of ma terials.

PAGE 36

36 Figure 2 7 Typical load displacement plot for specimen (after Landis, 2000) In terms of research in Portland Cement Concrete, a micromechanical view of cement mortar (Landis et al 2000, and Landis, 2006) was made to reveal the internal d amage of a four by four millimeter mortar specimen within a custom made in situ loading frame. A load and unload protocol was applied to the concrete cylinder. The CT sca ns w ere performed when the prescribed deformation was reached. F our c y cles of lo ad/unload with the subsequent CT scans were performed in this study shown in Figure 2 7 B oth load/deformation and CT scan were assessed to evaluate the crack and fracture properties. This study shed light on integrating load data with CT image analysis to make quantitative microstructure measurements. Another research study (Landis, 2010) showed good agreement with 3 D images obtained by the X ray synchrotron collaborating with D iscrete Element Modeling (DEM) L attice topology was applied to define t hree phases as applied to hardened cement paste, cement aggregate interface, and aggregates. Although this was a preliminary study, the lattice formulated

PAGE 37

37 and simulated excellent illustrations of positioning aggregates and fracture pattern s The author a lso published a tutorial review article on X ray microtomography to explain the fundamental knowledge and technique in CT, especially on synchrotron based system s (Landis and Keane, 2010). Another study ( P romentilla and Sugiyama, 2010) examin ed the frost induced damage process in PCC. A twelve inch by twenty four inch PCC core was obtained and placed in various condition s between 20 C and 18 C in water. It was observed from CT slices that most cracks occurred around aggregates. The author suggested tha t the internal cracks tended to develop in the region of the interfacial transition zone, rather than in the frost damaged mortar. This study also successfully adopted global thresholding to separate void space from cement paste. A plot was made to prese nt the void fraction at different freezing thawing cycles along with the segmented CT slice number, which is in essence the location within the cylinder. Moreover, a measure of internal crack width, connectivity, and tortuosity had been evaluated. The 3 D model of crack tortuosity visualized and characterized the connected crack network and distribution. Another similar study to investigate the damage within the freeze thawed concrete (Suzuki et al., 2010) used a helical CT scan on concrete field cores s ubject ed to the freeze thawing process for damage analysis. The damage analysis observe d the loss of attenuation along the cylinder height. The loss of attenuation value was determined from the occurrence of cracks induced by the freeze and thaw process. In addition, other research studies ha ve been performed to quantify the air void distribution inside concrete and asphalt specimen s Masad et al. (Masad et al., 1999 & 2002) described the air void distribution inside asphalt specimen s and tr ied to relat e th at

PAGE 38

38 distribution to the force and efficiency of compactors. A CT scan was performed on t he asphalt specimen and then a 3 D model was reconstructed. 2 D images extracted from the 3 D model were sliced and each image was processed by a chosen gray in tensity threshold of in order to isolate voids from internal structure. A plot between percent age of air voids and specimen depth was made to represent the void distribution internally. Wang et al (Wang et al., 2001) employed a similar technique to inves tigate the void distribution of the WesTrack mixes. Figure 2 8 shows typical steps involved in image acquisition processing for HMA (Masad and Kutay, 2012). This flow chart summarize s the latest development in using X ray CT to evaluate internal properti es pertaining to aggregate contact points, orientation, and segregation, voids distribution in specimen compacting efforts, permeability analysis of open graded friction course, and the modeling of mixture response in DEM. Although there were many well kn own advantages, authors listed limitations on manually selected threshold value s based on visual inspection in image analysis disturbance of artifacts, and optim al cho ice of proper resolution during the reconstruction, that would need to be addressed in t he future.

PAGE 39

39 Figure 2 8 Typical steps involved in image acquisition processing and analysis ( a fter Masad and Kutay, 2012)

PAGE 40

40 CHAPTER 3 EXPERIMENTAL PLAN 3.1 Selection of Materials 3.1.1 Portland Cement Materials selected in this study were approved by the State Materials Office (SMO) of FDOT. T ype I/II Portland cement, purchased by the plant of Florida Rock industries, Inc. in Gainesville, Florida was used for this work. The chemical composition approved by the Florida D epartment of Transportation was: 54% Tricalcium Silicate; 20.5% Silicon Dioxide ; 5.2% Aluminum Oxide ; 3.8% Ferric Oxide ; 2.8% Sulfur Trioxide ; 0.6% Magnesium Oxide ; and 0.25% total a lkali as Sodium Oxide with 0.30% loss of ignition by mass. 3.1.2 Virgin A ggregates A combination of #57 Florida limestone and # 89 silica sand were chosen as the coarse and fine aggregates in concrete respectively. The #57 Florida limestone was purchased from a plant with FDOT quarry number #87 090. The physical properties of coarse aggregates were tested and were provided from SMO. The bulk specific gravity of dry, SSD, and apparent specific gravity were 2.33, 2.41, and 2.53, correspondingly The absorption was 3.45%. On the other hand, the #89 silica sand was supplied and its physical properties were tested by SMO. The bulk specific gravity of dry, SSD, and apparent specific gravity were 2.62, 2.63, and 2.65, in th at order. The absorption was 3.45%. It must be noted that the percent of fine aggregates passing #30 was a little beyond the maximum grad ing limits of ASTM specification and the fineness modulus was a slightly lower than specification Table 3 1 and Table 3 2 show gradations of these aggregates.

PAGE 41

41 Table 3 1 Gradation of coarse aggregate (#57) Sieve Size Cumulative Retained (%) Passing (%) Grading Limits US inch mm Min (%) Max (%) 11/2" 1.476 37.5 0 100 100 100 1" 0.984 25 1 99 95 100 1/2" 0.492 12.5 80 20 25 60 #4 0.187 4.75 98 2 0 10 #8 0.093 2.36 98 2 0 5 Table 3 2 Gradation of fine aggregate (#89) Sieve Size Cumulative Reta ined (%) Passing (%) Grading Limits US inch mm Min (%) Max (%) #4 0.187 4.75 0 100 95 100 #8 0.093 2.36 1 99 85 100 #16 0.046 1.18 7 93 65 97 #30 0.024 0.60 25 75 25 70 #50 0.012 0.30 69 31 5 35 #100 0.006 0.15 97 3 0 7 #200 0.003 0.075 100 0 0 2 Fineness modulus 1.99 3.1. 3 Reclaimed/Recycled Asphalt Pavement According to the geological survey published by Florida Bureau of Geology (Lane, 1987), limestone or so types of lime stone that have been categorized i n this report, such as Key Largo Miami Oolite, Coquina, Ocala, Suwannee, Tampa. Dolomite, a somewhat harder rock containing CaMg(Co3)2, has also been found in several counties. Florida limerock has been included and cru shed for building roadways. In asphalt pavement industries, due to the fact that most limestone tend s to be soft er than that obtained in other states, contractors import granite from adjacent states such as Alabama and Georgia, or from places as far as No va Scotia, Canada and Mexico, in order to produce a stronger and more durable surface mixture. A majority of Open Graded Friction Course (OGFC)

PAGE 42

42 paved on the surface s of Interstates to prevent hy droplaning of vehicle tires has been built with granite, inst ead of Florida limestone. However, it is sometimes difficult to acquire Miami oolite near the Florida panhandle due to the higher expense of transportation from S outh Florida or other agency certified sources. Regional contractors tend to use aggregate s limestone from Alabama. Hence, it is Miami oolite, Alabama dense limestone, and granite that can be found mostly on roadways in Florida and as such these materials would most likely be found within RAP. The source of RAP in this study was selected through the help of Mr. David Webb the manager of the b ituminous laboratory at SMO. It was transported from the Community Asphalt Corporation (FDOT source number A0697 ) located in Vero Beach, Florida, and con tain ed mostly Florida limestone in mineral composition rather than granite. The RAP was sent to the asphalt plant of V.E. Whitehurst & Sons Inc. in Gainesville Florida for the fractionating process. T he #4 sieve (4.75mm) was employed to separat e coarse and fine stones. The fractionated coarse and fine RAP were then hauled back to SMO and stored in separate bins. Properties regarding recover ed binder were tested by the bituminous lab of SMO, shown in Table 3 3. Table 3 3 P roperties of recovered bi nder from RAP RAP Source Recovered Aggregate Type Recovered Viscosity (poises) Recovered Asphalt Content (%) Penetration of Recovered Asphalt Fine RAP Coarse RAP Fine RAP Coarse RAP Fine RAP Coarse RAP Vero Beach, FL Florida l imestone 422769 112847 4. 9 3.9 8 15

PAGE 43

43 In addition, the absorption and bulk specific gravity were examined in house at the concrete laboratory at the Department of Civil and Coastal Engineering at the University of Florida, under the procedures of AASHTO T84 and T85. For coarse RA P, the bulk specific gravity of dry, SSD, and apparent specific gravity were in order, 2.29, 2.33, and 2.38, with fine RAP, in the same order 2.30, 2.31, and 2.32. Absorptions of coarse and fine RAP were 1.72% and 0.51%, respectively. Similar bulk spec ific gravity values of fractionated RAP, compared with virgin coarse aggregate may suggest the same mineral composition as Florida limestone. B oth RAP aggregates were much lower in absorption compared with virgin rocks This may be explained by the coatin g of asphalt binder preventing moisture from penetrating and being absorbed by Florida limestone. Table 3 4 Gradation s of fractionated coarse and fine RAP Sieve Size Passing (%) US inch mm Coarse RAP Fine RAP 1" 0.984 25 100 3/4" 0.75 19 100 1/2" 0.492 12.5 96 3/8" 0.375 9.38 64 100 #4 0.187 4.75 4 85 #8 0.093 2.36 1 62 #16 0.046 1.18 47 #30 0.024 0.60 34 #50 0.012 0.30 16 #100 0.006 0.15 4 #200 0.003 0.075 0.5 Fineness modulus 3. 52 Table 3 4 presents gradations of fr actionated coarse and fine RAP. T he gradation of fractionated coarse RAP was recognized to be finer than the chosen coarse aggregates, which can be attributed to nominal maximum aggregate size being generally finer when adopted in an asphalt surface layer as well as to the possible

PAGE 44

44 losses occurring during the fractionation processes of RAP. The fineness modulus of fine RAP was obtained and was much higher than that of the silica sand chosen for this study. 3. 2 C o ncrete Mix Proportions Three FDOT approve d concrete mixes with w/c=0.45, 0.50, and 0.55 PCC and four other concrete mixes containing 20%, 40%, 70%, and 100% of RAP with w/c=0.50 were u sed in this study, as shown in Table 3 5 Concrete was mixed with the help of FDOT personnel at SMO. The cement content was fixed at 500 lbs/yd 3 of w/c=0.50 and could be modified for w/c=0.45 and 0.55 mixes. The ratio of total fine was designed to fall between 38% and 43% of total aggregate by volume. Tr ia l mix es w ere batched for every PCC and RAP concrete mi x A low range water reduction (WRDA60) at the ratio of 6 ounce s per 100 pound s of cement, and of the air entr ainment agent (Dar avair AEA) at the ratio of 1 ounce per cubic yard by volume were added to control the desired slump ( one to three inches ) and imp rove the workability for each mix For concrete mix containing RAP, for instance 20% RAP, the design called for replac ing 20% of coarse RAP with virgin #57 Florida limestone and the same percentage of fine RAP with virgin #89 silica sand. Similar replac ement s we re employed for other designated mixes of 40% and 70%. Experimental design called for c oncrete mix containing 100% RAP us ing only fractionated RAP mixed with cement and chemical admixtures. Table 3 6 exhibits the mix proportions of PCC and RAP c oncrete mixes that were evaluated in this study. The combined gradations for total coarse and fine aggregates are show n in Figure s 3 1 and 3 2. It can be seen that concrete mixes containing 70% and 100% fell out side either coarse or fine gradation limits resulting in concrete mixes with finer gradation.

PAGE 45

45 Table 3 5 Concrete mixes containing RAP to be evaluated % RAP 0% 20% 40% 70% 100% w/c = 0.45 X + + + + w/c = 0.50 X X X X X w/c = 0.55 X + + + + X Seven mixes to be evaluated. Table 3 6 M ix proportions of PCC and c oncrete mixture s containing RAP Mix RAP (%) W/C Water (lbs/yd 3 ) Cement (lbs/yd 3 ) Coarse Aggregate (lbs/yd 3 ) Fine Aggregate (lbs/yd 3 ) AEA Daravair (oz) WRDA 60 (oz) Virgin RAP Virgin RAP PCC 0 0.45 250 556 1816 0 1195 0 1 15 0 0.50 250 500 1845 0 1212 0 1 12 0 0.55 250 454 1865 0 1228 0 1 9 RAP 20 0.50 246 491 1403 330 1049 226 1 30 40 0.50 246 491 1047 660 821 441 1 30 70 0.50 250 499 511 1137 415 804 1 30 100 0.50 250 500 0 1562 0 1241 1 30 Figure 3 1 C ombined g radation chart for coarse aggregate s

PAGE 46

46 Figure 3 2 Combined g radation chart for fine aggregate s After the trial batch was determined, the production mix would then be mixed to fabricate concrete cylinders and bending beams for the subsequent te st plans. The fresh concrete properties including slump, air content, unit weight, and mixture temperature were examined by FDOT personnel after the production mixes were performed. Fresh concrete properties are reported in Chapter 4 3. 3 Test P lan for Conventional Concrete Propert ies Tests i nch concrete cylinders and inch bending beam s for e ach concrete mix were fabricated for conventional concrete properties tests. The s e were : compressive strength (ASTM C39) ; splitting tensile strength (ASTM C496) ; and modulus of elasticity (ASTM C469) in this plan. However, only specimens with 28 d ay

PAGE 47

47 and 90 day curing periods were tested for results with regards to conventional concrete properties. Results of conventi onal concrete proper ties are reported in Chapter 4 3. 4 Test P lan for Superpave IDT Strength Test inch concrete cylinders were fabricated f rom same concrete mixes in conventional tests ; curing periods were 7, 14, 28, and 90 days for the Superpave IDT test Cylinders of 1.5 inches in thickness and 6 inches in diameter were prepared The thin concrete specimen s w ere sliced by the large masonry saw at SMO. One cylinder could supply up to six or seven thin concrete slices. The con crete slices were subsequently transported to the asphalt laboratory o n campus for further surface cleaning and fan drying. The Superpave IDT strength test with constant force control and displacement control were used. Results and analysis of the Superp ave IDT strength test s on concrete mixes are reported in Chapter 5 3. 5 Test P lan for X ray Computed Tomography Concrete cylinders cured for 90 days for each mix were prepared as with the S u perpave IDT strength test. PCC specimens were designed with w/c= 0.45 and 0.55 and concrete containing 20%, 40%, 70%, and 100% RAP mixes with w/c=0.50 to be evaluated in t he X ray CT plan. The X ray Computed Tomography scan required placing a specimen on the load frame in order to perform a revolution scan to acquire a sufficient number of two dimensional (2 D) digital radiography images to allow the reconstruction of the three dimensional (3 D) model. This 3 D model subsequently was able to export 2 D virtual slices on the Z axis for more analysis. Details and analys is are reported in Chapter 6

PAGE 48

48 CHAPTER 4 EVALUATION OF PCC AN D CONCRETE CONTAININ G RAP BY CONVENTIONA L TESTS 4.1 Background This chapter presents the results obtained from the performance of fresh concrete and hardened concrete tests on PCC containing onl y virgin aggregates and concrete mixes containing RAP Those concretes containing RAP were also evaluated after curing using the Superpave IDT strength test and were assessed by X ray CT scan. 4.2 Concrete Mixing and Sample Preparation Concrete mixtures w ere prepared at the concrete laboratory at the State Materials Office of the Florida Department of Transportation in Gainesville, Florida. Trial batch mixes were tested to ensure the slump and workability before the real production mixes began Steps inv olved in the preparation of the concrete mix es include the follow ing: The fine aggregate was dried for at least 24 hours in the oven at 230 F, and then allowed to cool for another 24 hours inside the lab. The coarse aggregate was soaked for at least 48 hou rs and then allowed to dry outside the tank for 30 minutes before weighing. All the RAP material was stored inside the lab in bags and weighed in amounts to be used for each mix Other materials were weighed in amounts to be used in batches, and bagged. Al l the aggregate for a mix was placed in the drum mixer, and mixed for 30 seconds. All of air entraining agent for a mix was added to half of the water required for that mix. The mixing water containing the air entraining agent was added into the drum mixer and the batch was mixed for 1 minute. The required amount of water reducer was mixed with the remaining half of the mixing water.

PAGE 49

49 The measured amount of cement was placed into the mixer, and the mixture of water with the water reducer was added. The batch was then mixed for 3 minutes, followed by a 2 minute rest period, followed by 3 minutes of mixing. Fresh concrete property test s were performed and the workability of the batch was ensured. If the workability had not been achieved, more water reducer was added to the mix until the condition of the batch was correct. After each production mix was produced the proper amount of concrete mixture was placed in molds to cast either concrete cylinders or simple beams for subsequent curing and testing. A suffici ent portion of each concrete mixture was reserved for the fresh concrete t ests. C oncrete sample fabrication involved the following steps: Molds were filled with concrete in three successive layers, each layer being vibrated for approximately 30 seconds on a vibrator table. If the consistency of the concrete was not correct, the vibration period was extended until consistency was appropriate. Concrete specimens were covered with polythene sheets to avoid loss of moistur e Specimens were removed from the mol ds after 24 hours and place d in the cur ing room at 100% humidity while curing. 4.3 Fresh Concrete Properties and Hardened Concrete Test s Three PCC mixes using virgin aggregates exclusively, and four concrete mixes containing 20, 40, 70 and 100% of RAP re placement were evaluated in this study. The 28 day and 90 day moisture curing periods were applied to cylinders of PCC and of concrete containing RAP for tests of compressive strength, splitting tensile strength, elastic modulus, and Poisson s ratio, whi le only PCC beams were evaluated for flexural strength at 28 days and 90 days curing. Fresh and h ardened concrete tests considered

PAGE 50

50 as conventional tests to evaluate concrete mixture properties in this study included a list of experiments as follows: Fresh concrete properties: 1) Slump; 2) Unit weight; 3) Air content; 4) Concrete temperature Hardened concrete properties: 1) Compressive strength; 2) Modulus of elasticity; of thermal expansion; 7) Drying shrinkage Table 4 1 Standard tests on fresh and hardened concrete Concrete Test Standard Slump ASTM C 143 Unit weight ASTM C 138 Air content ASTM C 173 Fresh concrete temperature ASTM C 1064 Compressive strength AST M C 39 Flexural strength ASTM C 78 Splitting tensile strength ASTM C 496 ASTM C 469 ASTM C 469 4.4 Results of Fresh and Harden ed Concrete Tests Th e results from the testing are shown in Table 4 2 Except in the case of the flexural strength test o n PCC with w/c=0.50 and 28 day moisture cur ing concrete mixes with lower water to cement ratio, lower RAP replacement, and longer curing time yielded higher strength than the inverse in the compressive and splitting tensile tes t. T he modulus of elasticity generally decreased in the cases with the higher water to cement ratio and higher RAP replacement. T 0.29 for all PCC and RAP concrete mixes. Table 4 3 shows t he results of the fre sh concrete properties tests performed by FDOT personnel. It can be seen that the unit weight decreas ed with the addition of RAP replacement, which may be at tribut able to the lower density of RAP as compared with that of virgin aggregates. The convention al hardened concrete properties wer e

PAGE 51

51 later correlated with the results of the Superpave IDT strength test, as described in Chapter 5. Table 4 2 Results of hardened conventional concrete tests Water to cement ratio for concrete containing RAP was 0 .50 Table 4 3 Results of fresh concrete tests Mix Type Number RAP (%) W/C Slump (in) Air Content (%) Unit Weight (lbs/ft 3 ) Temperature ( F) PCC 1 0 0.45 3.50 4.50 140 73 2 0 0.50 4.25 3.20 140 70 3 0 0.55 5.00 3.30 140 74 RAP 4 20 0.50 1.25 4.25 139 75 5 40 0.50 1.75 4.25 138 75 6 70 0.50 1.00 3.50 137 75 7 100 0.50 0.50 3.50 132 77

PAGE 52

52 CHAPTER 5 CHARACTERIZATION OF PCC AND CONCRETE CON TAINING RAP USING TH E SUPERPAVE IDT STRENG TH TEST 5.1 Background This chapter presents the sample prepar ation, test procedure, results, and analysis for the Superpave IDT strength test using force control and displacement control loading rates. Correlations between results from Superpave IDT strength test and conventional test were also presented in this ch apter. 5.2 Test Sample Preparation 5.2.1 Slicing of Concrete Cylinder Inch c oncrete cylinders were fabricated from each concrete mix C uring periods of 7, 14, 28, and 90 days were used Specimens with 1.5 inches in thickness and s ix inches in diameter were prepared for the Superpave IDT strength test. The thin concrete specimens were sliced by a large masonry saw at the concrete laboratory of the State Materials Office in Gainesville One cylinder could supply up to six or seven thin concrete slices. The sliced concrete specimens were subsequently transported to the asphalt laboratory at the University of Florida for further surface cleaning and fan drying. 5.2.2 Air Drying, Surface Cleaning, and Gage Points Attachment The slic ed concrete specimens were covered by moisture and dusts through the process of slicing. The specimens needed to be dried and cleaned before gage points could be attached. After they were taken to the asphalt laboratory, the thin concrete specimens were placed in front of a fan to be dried. The process of air drying took approximately one hour to complete. Concrete specimens were labeled using a permanent marker after air drying.

PAGE 53

53 In addition, the surface of specimens were cleaned to sweep out dusts that were generated from the slicing process when the masonry saw cut through the aggregates and cement paste. The surface cleaning process was started by using steel brushes to sweep around central area of specimen in which gage points were going to be glued Steel brushes can remove dusts or broken particles out of surface and it was followed by having paper towels soaked with acetone to wipe the same central area. The use of acetone can also take away the left over moisture After the surface cleaning pr ocess was finished, the thin concrete specimens were ready to be attached with strain gage points. Four strain gage points for each surface of the thin concrete slice were attached by using Loctite P rism 454 adhesive and aligned by the gage point mou nting system, shown in Figure 5 1 The mounting system can place and align eight gage point s (two for vertical and two for horizontal for faces ) in the center of each specimen spaced at one and one half inches on center for each direction. The purpose of attaching gage points was to provide a fixed and aligned position for later process of installing strain gages. 5.2.3 Temperature C onditioning Th ree temperatures were selected : 10, 23, and 60 C for Superpave IDT strength test s representing the ambient t emperature in winter, standard laboratory testing temperature, and the pavement temperature in summer in Florida, respectively. After the adhesive was dried for two hours concrete slices with installed gage points were stored in the Advanced Materials Ch aracterization Laboratory (AMCL) for temperature conditioning. After having been positioned in the ESPEC environment chamber (shown in Figure 5 2) for low temperature conditioning, in a Fisher Scientific oven for high

PAGE 54

54 temperature conditioning, and in ambi ent temperature overnight, concrete slices were ready for the Superpave IDT strength test. T hree concrete slices were prepared for each testing condition ; for example three specimens for a mix with 7 day s curing and being tested at 23 C Figure 5 1 Gage point mounting system (Photo courtesy of Yu Min Su) Figure 5 2 ESPEC environment chamber (Photo courtesy of Yu Min Su)

PAGE 55

55 5.2.4 I n stalling Strain Gages The Superpave IDT strength test was performed using a n MTS load frame. The displacement meas urements from strain gages were recorded simultaneously along with the loading information. Moreover, clip on caps were needed to be installed on gage points and subsequently aligned in the same orientation. The strain gages (or LVDTs) were hereafter att ached on the aligned caps and spacing adjustment between caps had to be calibrated as close to zero as possible based on strain gage readings on the screen. This spacing adjustment was often a time consuming process to perform. For low and high temp erature conditioning situations, the specimen would be returned to the weather chamber or oven for thirty minutes reconditioning after installing caps. Gage points attachment, clip on caps alignment, and strain gages adjustment were reported in details in a FDOT final report (Roque et al., 1997). 5.2.5 Positioning Concrete Specimen The diameter and thickness of concrete thin specimen were measured and recorded after the temperature reconditioning. Strain gages (or LVDTs) were accurately placed on the alig ned and adjusted caps before the concrete specimen was put on to the load frame. The concrete specimen had to be carefully positioning on the center of loading head in order to warrant precise loading position. Figure 5 3 shows a concrete slice situated i n the load frame with vertical and horizontal strain gages attached. A pre load about ten pounds was set to hold the specimen in position without any movement. Additionally, cables connected with strain gages were arranged to stay away from the concrete specimen to avoid damage caused by explosive concrete blast of fracture.

PAGE 56

56 Figure 5 3 Thin c oncrete specimen on the load frame with strain gages (Photo courtesy of Yu Min Su) 5.3 The Superpave IDT Strength Test on Concrete Specimens 5.3.1 Loading Con dition In this study, both force control and displacement control loading conditions were used. In force control mode, a constant load rate of 35.343 lbf/sec was used. This load rate was chosen so that it would be similar to the one used in the standard splitting tensile strength test. In the displacement control mode, a constant displacement rate of 0.00075 in/sec was used. 5.3.2 Data Collection and Processing 5.3.2.1 Data collection Two horizontal, two vertical and axial displacements, axial force, time of loading were captured in twenty data points per second. In force control mode, it needed about 150 to 300 seconds to complete a Superpave IDT strength test on concrete specimens,

PAGE 57

57 while it required only 30 to 40 seconds running in the displacement control mode. Enormous data including displacements, loads, and time were recorded to a text file which can be used for further data processing. 5.3.2.2 Data processing Horizontal and vertical displacements were used to compute vertical and horizontal str ains on both side of the specimen with known spacing between gage points ( 1.5 inches). The corresponding indirect tensile stress also was calculated using the measured thickness (approximately 1.5 inches) and diameter ( approximately 6 inches) of concrete specimen as follows: (5 1) where, indirect tensile stress at time psi, ax ial load at time lbf, thickness of concrete thin specimen, inches, diameter of concrete thin specimen, inches. Two horizontal strains and the calculated tensile stress could be plo tted together to generate the individual stress strain plot containing two curves indicating the strain developments for each concrete specimen. Moreover, t h ere were three replicate specimens evaluated for each designed mix prepared and tested by each sel ected temperature condition. The mean tensile stress was calculated from three axial loads and a trimmed mean strain was taken by computing the average strain from six faces of strains from three replicate specimens. The trimmed mean strain was obtained by ranking the highest and lowest strains but discarding the highest and lowest one s. Hence, a mean stress strain curve was generated to each designated concrete mix for computing concrete mixture properties.

PAGE 58

58 5.3.2.3 Determination of failure stress and pe ak stress The failure stress is a vital pa rameter to be evaluated for adopting the Superpave IDT strength test to concrete Conv entionally, the peak stress of splitting tensile strength test is considered as the arrival of concrete failure and used for de riving the indirect tensile strength in ASTM C4 96 T h e definition of failure in HMA fracture mechanics is rather different than that in concrete. The failure of HMA is defined at the load level when the first peak of (Y X) has been reached ( Roque et al., 1996 ) described in C hapter 2.2. The ten sile strength is calculated from the failure load and corrected by a correction factor which is affected by computed from the repeated loading condition as the Superpave IDT resilient modulus ( ) test. acture energy can be identif ied and calculated conveniently by a n automation software ITLT (Roque et al, 1997). In additional, ITLT requires consecutive Superpave IDT resilient modulus, creep, and tensile strength tests for conducting the automatic analysis. The proposed protocol (Zheng, 2007) of the Superpave IDT strength test in testing concrete has been discussed in C hapter 2.2. Failure stress, or first fracture in concrete was proposed to be determined in the same fashion as for HMA T here was no definitive description on determining the elastic modulus, while was proposed to be determined by strains at the half o f the average failure stress from three replicate specimens. Addi tionally, the fracture energy was determined by calculat ing the area under the stress s train curve up to th e designated failure stress Figure 5 4 shows the determin ation process of failure load on one PCC example with w/c=0.50 tested at 23 C in this study by using the proposed protocol.

PAGE 59

59 ( A ) ( B ) ( C ) Figure 5 4 E xample of determining the failure load on PCC with w/c=0.50 F ailure at 192 sec A bout 467 psi

PAGE 60

60 The determination started from Figure 5 4(C) by identify ing the first peak of Y X from two curves. After the first peak was indentified, the load could be determined for the corresponding loading time. For this example, it can be seen that the f ailure occurred at 19 2 seconds of loading as shown in Figure 5 4 ( B), and the corresponding failure stress was determined to be 467 psi. Additionally, the peak stress was found to be 534 psi which was the last data reported in the stress strain plot in Figure 5 4(A). By observing the stress strain curve, the peak str ess by definition of conventional failure in concrete occurred clearly at the end of plastic deformation. On the other hand, the failure by definition of HMA or the proposed protocol was located about 88% of peak stress in this case. 5.3.2.4 Determinati on of elastic modulus and Poisson s ratio ASTM C4 69 provides the access to obtain elastic modulus ( ) st up to 40% of peak stress to a concrete cylinder. The elastic modulus in concrete is a chord modulus and t heoretically such modulus can be identical with secant modulus in elastic region. Figure 5 5 (A) shows an example of secant modulus plotted with stresses under the force control loading rate. There were two curves derived from the PCC with w/c=0.50 and a concrete containing 70% RAP and dash lines presented the occurrences of 40% of peak stress in these mixes. The secant modulus of PCC indicated a seemingly flat plateau which may imply the elastic characteristic at 40% of ultimate stress. However, the secant modulus of concrete containing 70% RAP showed a continuous declining curve that may indicate the non elastic or plastic characteristic at this loading level. Therefore, i t may be reasonably suggested that it is elasti c within the 40% of peak load

PAGE 61

61 for PCC, though it might not be genuine for concrete containing RAP with regards to the determination of elastic modulus ( A ) ( B ) Figure 5 5 Example of ( A ) secant modulus and ( B ) str ess strain curve of PCC and conc rete containing 70% RAP Similar considerations occurred for determining the Poisson s ratio whether or not the 40% of peak stress is elastic for concrete mixes. It was selected to use vertical and horizontal ratio at 40% of peak s tress to calculate the Poisson s ratio. It has to be noted that both elastic modulus and Poisson s ratio were computed from the mean stress strain curve of each concrete mixture in this study. 5.3.2.5 Determination of fracture energy and toughness Fractur e energy was defined in HMA fracture mechanics and proposed protocol as the area integration of stress strain curve up to the failure stress. Figure 5 4 (A) shows fairly clear that this area integration is more and less presenting the energy up to about e lastic limit. Figure 5 5 (B) shows the example of stress strain curves of PCC

PAGE 62

62 with w/c=0.50 and concrete containing 70% RAP. The stress strain curve of concrete containing R AP indicated a much long er plastic deformation up to 0.015 in/in horizontal strai n, while the strain of PCC with w/c=0.50 only extended up to 0.001 in/in. The ductility for concrete containing RAP may be accessed effectively to include the area of plastic deformation. The toughness of concrete mix that includes overall area covering elastic and plastic regimes was used to assess the ductility of concrete containing RAP in this study. 5.3.3 Computations of Concrete Properties In this study, individual stress strain curve was plotted for each concrete specimen to access the failure str ess, peak stress, and fracture energy ( calculated up to the failure stress). In addition, the mean stress strain curve of each concrete mix was also obtained to evaluated the elastic modulus, Poisson s ratio, and toughness (calculated up to the peak stres s). 5.3.3.1 Indirect tensile strength The average failure stress was obtained from individual stress strain curves of three replicate concrete specimens, while the average peak stress was calculated from three axial peak loads. T he indirect tension streng th of concrete tested by the Superpave IDT strength test ca n be evaluated as follows: ( 5 2 ) ( 5 3 ) where, = indi rect tensile strength of failure stress in HMA fracture mechanics psi

PAGE 63

63 = indi rect tensile strength of peak stress in conventional concrete spitting tensile test psi axial load at time of fa ilure stress, lbf, and axial load at time of peak stress, lbf. 5.3.3.2 Elastic modulus The mean stress strain curve was obtained for each concrete mix. It was selected to use stress and strain at the 40% of ultimate stress to com pute the secant modulus. T he determination of elastic modulus by Superpave IDT strength test for concrete was defined as follow: ( 5 4 ) where, = Elastic modulus of Superpave IDT test on concrete = Mean t ensil e stress at the 40% of peak stress psi = Mean h orizontal stra in at the 40% of axial peak stress in/in 5.3.3.3 Poisson s r atio Po i was computed from corresponding vertical to horizontal strain ratio ( ) at 40% of peak stress associated with the e quation 2 10. 5.3.3.4 Fracture energy and toughness T he fracture energy of Superpave strength I DT test on concrete denoted as was calculated the area under stress strain curve up to the failure stress as follows:. ( 5 5 ) ( 5 6 )

PAGE 64

64 where, = a segment of fracture energy per one unit time = fracture energy ( calculated up to the failure stress) The toughness of Superpave IDT str ength test on concrete was calculated the area under stress strain curve up to the peak stress as follows: ( 5 6 ) where, = toughness ( calculated up to the peak stress) In addition, the fracture energy were averaged from three replicate concrete specimens and the toughness was computed from the mean stress strain curve for every concrete mixes. 5.4 Results of Superpave IDT Strength Test under Force Control Loading 5.4.1 Behaviors of Stress Strain Curve I ndividual stress strain plot of PCC and concrete containing RAP at 28 and 90 days curing time were evaluated. Fi gure 5 6 presents typical stress strain plots for dense graded HMA, PCC, and concrete containin g RAP. Figure 5 6 ( A ) is a typical example of dense graded HMA with PG 64 22 asphalt binder and tested at 10 C under a displacement control rate 0.05 in/sec for the purpose of comparison The plots of t wo horizontal strains were shown u p to the peak load After the peak load was reached, one strain was extended longer in development than the other. However, both strains were developing in the same direction. Figure 5 6 ( B ) shows a rather different behavior of a PCC specimen with w/c=0.45 and tested at 2 3 C under a force control rate Two horizontal strain s were increasing more and less at the same rate, but went apart when the peak load was approached One strain w ould develop further more in the positive direction but the other would turn

PAGE 65

65 in the nega tive direction Different behavior of recorded strains on both faces of a specimen suggested that one side of specimen was in tension, while the other side was in compression, due to the fact that specimen was not perfectly symmetrical. ( A ) Dense Gr aded HMA with PG 64 22, tested at 10 C ( displacement control) ( B ) PCC with w/c=0.50, tested at 23 C ( C ) Concrete containing 70% RAP, tested at 10 C ( D ) Concrete containing 70% RAP, tested at 10 C F igure 5 6 Examples of st r ess strain plots with two horizontal strain gages in the Superpave IDT test under force control load rate

PAGE 66

66 For concrete containing RAP mixes, the stress strain cu rve behav iors are shown in Figure 5 6 ( C ) and (D) F i gure 5 6(C) shows h orizontal strain s on both sides of specime n were growing uniformly in most cases for concrete mixes with higher RAP replacement such as 70% and 100%. However, there were still cases of concrete mixes containing 20% and 40% RAP where the recorded horizontal strains moved in opposite directions s imilar with PCC, shown in Figure 5 6 ( D ). Additionally, clear boundary limit between elastic and plastic regimes could be seen on those plots 5.4.2 Results of Indirect Tensile Strength 5.4.2.1 Indirect tensile strength (using peak stress) Concrete specime ns of 7, 14, 28, and 90 days curing time were evaluated by the Superpave IDT strength test at 10, 23, and 60 C Table 5 1 shows indirect tensile strength r esults of the Superpave IDT strength tes ts calculated from peak loads at three different testing te mperatures and four different curing periods. Generally, indirect tensile strength of all PCC and concrete containing RAP were higher along with the curing time and the addition of RAP considerably reduced the tensile strength. Figure 5 7 shows the indire ct tensile strength of all the PCC mixes. It was found that t he thermal effect was indeed affecting the tensile strength of PCC mixes. The frozen internal moisture in low temperature seemed to support the early tensile strength On the other hand, the possible accelerated drying in high temperature cause d the loss in early strength. Mature concrete m ixes with 28 and 90 days curing time were seemingly not to be affected by thermal effect in terms of tensile strength

PAGE 67

67 Table 5 1 Results of Indirect ten sion test (using peak stress) IDT Strength test on PCC of w/c=0.55 and 14 d cured was in complete in 60C Water to cement ratio was 0.50 for every for concrete containing RAP. Figure 5 7 Comparison of Indirect tensile s trength results in PCC c ontrol mixes (using peak stress)

PAGE 68

68 Figure 5 8 Comparison of Indirect tensile s trength results in concrete mixes containing RAP (using peak stress) Figure 5 8 presents the indirect tensile strength of concrete mixes containing RAP tested. It was found that t he thermal effect clearly influenced the tensile strength of concrete mixes containing RAP due to the presence of RAP Th e asphalt binder played the similar role as the frozen moisture to support the tensile strength in low temperature especially the mixes containing 70% and 100% RAP. However the tensile strength loss tested in high temperature was significant on these mixes Mature concrete mix containing 20% RAP was the least one to be affected by loss of tensile strength h igh temperature, si milarly with the observation in PCC mixes. 5.4.2.2 Indirect tensile strength (using failure stress) Table 5 2 shows r esults of the Superpave IDT strength tes ts computed from failure loads at three different testing temperatures and four different curing pe riods. Failure stresses of concrete specimens with 28 and 90 days were evaluated from the individual stress strain curves The tensile strength calculated from failure stress was about 74 94% of that computed from peak stress for PCC mixes, and was about 57 87% of that derived from peak stress for concrete containing RAP mixes.

PAGE 69

69 Table 5 2. Results of Indirect tension test (using failure stress) Stress strain curves of c oncrete specimens with 7 and 14 days curing time were not included to be tested for computing failure stress. Figure 5 9 Comparison of Indirect tensile s trength results in PCC control mixes (using failure stress) Figure 5 9 and Figure 5 10 show similar comparison as Figure 5 7 and Figure 5 8. For PCC mixes, the tensile strengt h computed from failure stress on mature concrete was not dramatically affected by the either low or high temperature which was similar to the results calculated from peak stresses. The tensile strength of concrete mixes with

PAGE 70

70 low water to cement ratio of 0.45 tested in high temperature appeared to have noticeable strength loss which may be attribute d to the accelerated drying shrinkage. For concrete containing RAP, tensile strength computed from failure stress presented similar tread of low temperature. The reduction of tensile strength tested at high temperature for concrete containing higher RAP was significant. Nevertheless, the failure stress was generally reduced with the addition of RAP. Figure 5 10 Comparison of Indirect tensile s trength r esults in concrete mixes containing RAP (failure stress) 5.4.3 Results of Elastic Modulus, Poisson s Ratio, and Toughness 5.4.3.1 Elastic modulus, Poisson s ratio, and toughness (using peak stress) Results of elastic modulus, Poisson s ratio and fracture energy are shown in Table 5 3. T h ese parameters were computed from the mean stress strain curve of each concrete mix.

PAGE 71

71 Table 5 3. Results of elastic modulus, Poisson s ratio, and toughness (using peak stress) Water to cement ratio was 0.50 for eve ry for concrete containing RAP.

PAGE 72

72 Elastic m odulus Figure 5 11 shows the elastic modulus of PCC and concrete mixes containing RAP evaluated. It was found that the addition of RAP in concrete significantly reduce d the elastic modulus. In general, elastic modulus of PCC mixes were slightly reduced with longer curing time but only increased a bit along with higher test temperature. On the other hand, neither curing time nor test temperature was affecting elastic modulus of concrete containing RAP significa ntly Figure 5 11 Comparison of elastic modulus in PCC and concrete containing RAP Figure 5 12 Comparison of Poisson s ratio in PCC and concrete containing RAP Poisson s Ratio Figure 5 12 show s that the Poisson s ratio of PCC and concrete mixes containing RAP examined by Superpave IDT test as well as the conventional test results. It was found that Poisson s ratio stayed about 0.30 for both PCC and RAP

PAGE 73

73 mixes in conventional test For results tested under force control loading rate, Poisso n s ratio appeared to be abnormal except for results of PCC mixes tested in 23 C Toughness Figure 5 13 shows toughness of PCC and concrete mixes containing RAP evaluated It was found the toughness was generally reduced along with the curing time, w hich may attribute to the increasing concrete brittleness from the hydration of cement. Figure 5 13 Comparison of toughness in PCC and concrete containing RAP (force control) Concrete mixes containing 40%, 70% and 100% RAP developed higher toughne ss in high temperature than PCC and concrete containing 20% RAP The concrete containing 100% RAP mixes on both 28 and 90 days cur ing were of highest toughness, except for the low temperature test condition Higher toughness may attribute to the ductilit y improved by RAP. It has to be noted that the toughness is calculated as the area under mean stress strain curve up to the peak stress. 5.4.3.2 Fracture energy (using failure stress) Fracture energy is defined differently than the toughness. The fractur e energy is the area under the stress strain curve up to the failure stress as defined in HMA

PAGE 74

74 fracture mechanics in this study. Table 5 4 shows the fracture energy of concrete mixes of 28 and 90 days curing time. Table 5 4. Results fracture energy (us ing failure stress) Water to cement ratio was 0.50 for every for concrete containing RAP. When compared with the toughness computed from the area of stress strain curve up to the peak stress, all the fracture energy were less than the toughness, espec ially for concrete mixes containing RAP. Generally speaking, fracture energy of PCC mixes was higher than all concrete mixes containing RAP. Fracture energy of PCC mixes was not affected by testing temperatures, but higher testing temperature seemed to r educe the fracture energy of concrete containing RAP.

PAGE 75

75 Figure 5 14 Comparison of Fracture energy in PCC and concrete containing RAP (using failure stress) 5.5 Results of Superpave IDT Strength Test under Displacement Control Loading 5.5.1 Behaviors o f Stress Strain Curve Stress strain curve of PCC and concrete containing RAP tested under displacement control are shown in Figure 5 12. Figure 5 15(A) shows the same dense graded HMA plot (same with Figure 5 6(A)) for the purpose of comparison and Figure 5 15(B) to (D) present example behaviors of PCC and concrete containing RAP. Figure 5 15(B) and (D) indicated similar stress strain behaviors with results under force control loading rate of PCC and concrete containing RAP. One side of horizontal strai n appeared the similar negative turning, but it was shorter than examples in force control loading rate. Meanwhile, concrete containing 70% and 100% had less to none negative turning in one side of horizontal strain, shown in Figure 5 15(C). The stress -strain plots of concrete containing higher RAP developed similarly with the plot of HMA. All stress strain plots had clear boundary limit in terms of elastic and plastic regimes, which was again similar with curves in force control loading rate.

PAGE 76

76 ( A ) Dense Graded HMA with PG 64 22, tested at 10 C ( B ) PCC with w/c=0.50, tested at 60 C ( C ) Concrete containing 70% RAP, tested at 60 C ( D ) Concrete containing 20% RAP, tested at 60 C F igure 5 15 Examples of st r ess strain plots with two horizontal strain gages in the Superpave IDT test under displacement control load rate 5.5.2 Indirect Tensile Strength (using peak stress) Further effort was made to validate compare the indirect tensile strength under between the force control rate of 35.343 lbf/ sec and displacement control rate of

PAGE 77

77 0.00075 in/sec. Three PCC and four concrete contai ning RAP mixes of 90 days curing had been tested by the Superpave IDT under 23C. The test result s, shown in Figure 5 16 appeared no significant difference for PCC an d RAP mixes in terms of the way of loading. It may suggest that either loading rate may yield similar indirect tensile strength. This comparison reflected the well known understanding about force control and displacement control loading rates. Mier (Mi er, 1997) commented that specimen will fail in a n uncontrolled manner by applying the load control while the displacement control with the closed loop servo controlled equipments can measure the post peak softening curve of stress strain plot. However, t he load control tests traditionally gave materials, such as concrete, ceramics, and rock, sufficient pre peak information about the initial Young s modulus and maximum strength. Figur e 5 1 6 C omparison of tensile strength obtained from force control versus displacement control tests

PAGE 78

78 5.5.3 Poisson s Ratio (using peak stress) A nother endeavor was made to compare the ratios that were determined with the displacement c ontrol rate shown in Figure 5 1 7 Poisson s ratio s of PCC and concrete cont aining RAP tested under 23 C in displacement control mode were about 0.30, similar with conventional results In computation of tensile strength under force control loading rate, the correction factor for correcting tensile strength (CSX, described in cha pter 2.2) was omitted because of irregular P o isson s ratios. If Poisson s ratio can be reasonably assessed under the displacement control load rate, the correction factor could be considered to correct the tensile strength. In this study, when correctio n factor of CSX were to be applied, it would cause a range of 2.1% to 5.5% reduction in tensile strength from results tested by the displacement control loading rate. Figure 5 17 Comparison of P oisso ratio between conventional test in compression versus fo rce control and displacement control in the Superpave IDT strength tests 5.5.4 Toughness (using peak stress) Figure 5 18 shows toughness for concrete specimens of 90 days cur ing time tested in 23C by force control and displacement control loadi ng rates

PAGE 79

79 Figure 5 1 8 Comparison of toughness obtained from force control versus displacement control tests The failure detection in displacement control mode was set to stop the test when strength reduced to 5 0% loss of peak load, while the previ ous failure detection in force control mode was set at 20%. The displacement control mode seemed to capture noticeable toughness on 40% and 70% RAP that were not captured in the force control mode tested in 23C which may proxy a better assessment in te r ms of toughness. 5.6 Correlations between Conventional and Superpave IDT Strength Tests 5.6.1 Tensile Strength A further analysis ha d endeavored to establish correlations of splitting, flexural, and Superpave IDT strength, corresponding to identical cur ing periods, shown in Figure 5 19 It was found that the split ting tensile strength correlate d well with th e Superpave IDT strength of 28 days curing time computed from peak stress

PAGE 80

80 Figure 5 19 C orrelation of tensile strength between splitting and Supe rpave IDT strength test under 23C The linear regression between the splitting tensile ( ) and the tensile strength ( ) computed from peak stress from the Superpave IDT strength test in this study, under 23C, is as follows: ( 4 5 ) = 0.9455 It was reasonable to see a good correlation between conventional splitting tensile and the Superpave IDT strength tests since both tests were simil ar, except for the thickness of specimens. The st ress state and analysis can be considered reasonably close to each other. However, none of the flexural strength data were close to the tensile strength from the Superpave IDT strength test, although this relationship may be able to be established through other well documented correlations between splitting tensile and flexural strength.

PAGE 81

81 5.6.2 Elastic Modulus Another attempt to correlate elastic modulus obtained from the conventional test ASTM C469 in com pression and that from the Superpave IDT strength tests of concrete mixes with 28 days curing time is shown in Figure 5 20. Both PCC and concrete containing RAP mixes showed a high correlation between results from these two tests The relationships of PC C and RAP mixes are as follows: PCC mixes: (4 6) = 0.9 6 55 C o ncrete containing RAP mixes: (4 7) = 0.9 8 55 where, = chord modulus of elasticity at 40% of peak stress in compression = secan t modulus of elasticity from stress strain curve a t 40% of peak stress in Superpave IDT strength test. Figure 5 20 Correlation o f elastic modulus between conventional and the Superpave IDT strength tests

PAGE 82

82 5.6.3 P o isson s Ratio Additional correlation between the Poisson s ratios obtained from conventional and Superpave IDT strength tests on 28 d cured mixes under 23 C are shown in F igure 5 21 Poisson s ratio of PCC mixes obtained from these two tests related well with one another. However, the same cannot be found for concrete containing RAP mixes. The correlation of Poisson s ratios from ASTM C469 and the Superpave IDT strength test of PCC mixes is as follows: ( 5 4 ) = 0.9 702 w here, = Poisson s ratio of ASTM C469. = Poisson s ratio of the Superp ave IDT strength test. Figure 5 21 Correlation of Poisson s ratio between conventional and the Superpave IDT strength tests

PAGE 83

83 5.7 Research Findings The main findings from the investigation on the use of Superpave IDT strength test on concrete specimens with and without incorporation of RAP are as follows: When the maximum load obtained from the Superpave IDT strength test was used to calculate the tensile strength of the concrete at 28 days curing time the computed tensile strength from the Superpave I DT strength test correlated well with the corresponding splitting tensile strength from the conventional splitting tensile strength test. The elastic modulus and Poisson s ratio of the concrete obtained at 40% of its ultimate stress in the S u perpave IDT st rength test cor related well with the corresponding values obtained from the conventional compressi ve test. When the toughness is calculated by determining the area under the stress strain plot up to the maximum stress, it can be used to differentiate betwe en con cretes containing different percentage of RAP. Superpave IDT strength test using a displacement control mode were very close to those obtained from the Superpave IDT strength t est using a force control mode In running the Superpave IDT strength test on concrete specimens, when the concrete specimen was loaded beyond its elastic limit, the horizontal strain on one of the specimen face tended to go in the negative direction while the horizontal strain on the other face increased in the positive direction. However, this problem was not observed when concrete containing high percentage of RAP was tested. The tensile strength of concrete without RAP at early age was seen to increa se as the temperature decreased. However, this effect of temperature was not seen among the concrete at later ages. The tensile strength of concrete containing RAP was seen to decrease as the %RAP increased. The tensile strength of concrete containing RAP was seen to decrease as the temperature increased. The addition of RAP in concrete noticeably reduced the elastic modulus of the concrete. Toughness of concrete was seen to increase as the %RAP in concrete increased.

PAGE 84

84 CHAPTER 6 APPLICATION OF X RAY COMPUTER TOMOGRAPHY TECHNIQUE TO CONCRET E 6 .1 X ray Computed Tomography Equipments at the University of Florida At the University of Florida, the X ray CT facility is located in the Advanced Materials Characterization Laboratory (AMCL) in the Departme nt of Civil and Costal Engineering where it was installed in 2009. It is equip ped with 250 kV and 450 kV X ray sources, a turn table, and the digital flat panel detector ( FPD) inside the scanning chamber as shown in Figure 6 1 and Figure 6 2 in schemati c and plan views There is also an in situ MTS load frame that can be arranged so as to replace the turn table. Meanwhile, t he 450 kV X ray source provide s a higher power to pe netrate a larger object, while the 250 kV source with M icro focu s offers finer spatial resolution in scanning T he cone beam X ray system for both sources and the projected subject image or so called digital radiography (DR) image, will be record ed by the FPD The 14 bit ( i.e. 2 14 = 16384 gray levels) FPD is made of amorphous silic on (a Si) with a pixel display, and 127 micron s in pixel pitch Both X ray sources and FPD can be adjusted on the Y axis The definition of orientations is shown in the lower left corner of Figure 6 1. Additionally, t he turn tabl e or the load frame controlled by joysti cks or the proprietary software, can provide accurate segment turns for acquiring DR images. The resolution in voxel size is determined not only by the pixel pitch of the FPD, but also by the geometr y between loc ations of the X ray source, the test object on either the turn table or load frame, and the FPD. The geometric positioning of turn table, in situ load frame, or FPD to X ray sources can be arranged to optimize the contrast of DR images as well as achieve the geometric magnification. Details of geometric positions and the loss of definition issue is discussed in chapter 2.3.2.2.

PAGE 85

85 Figure 6 1 Major components of X ray CT facility at UF Figure 6 2 D iagonal plan view of CT scan chamber

PAGE 86

86 6 2 Procedur es of X ray Computed Tomography To start with, both X ray sources require a warm up routine every other day. A large lead block is place d in front of the X ray source to prevent the emission of X ray s If the X ray CT scan is to be employed on a given da y the warm up process on the designated X ray source must take place. The warm up process is controlled by the FXE Control software for 225 kV X ray source, while it is necessary to operate the warm up process manually for 445kV X ray source. Several steps involved in performing an X ray CT scan at UF are as follows. 6.2.1 Calibration of Digital Flat Panel Detector The manufacturer suggests that the flat panel detector be calibrated every time that a new subject is to be scanned. There are three steps : dark field l i ght field and middle field to be performed. Dark field calibration ensure s the condition of the detector and the suggested target value is approximately 16000, which is about the limit of 14 bit gray levels. There is no X ray emi ssion when this calibration is running. The dark field calibration also offers a clean background and compensates the malfunctioning pixels on the detector. Th e light field calibration works to make certain that the detector could capture sufficient X ra y flux of the lowest gray levels 1500 to 3500. The least desir able gray level on the acquired DR image requires 2500, which means that the gray level on the darkest area of the targeted sample need s to have enough X ray transverse to ensure the details that the DR image may carry. T he middle field calibration requires medium gray levels from 4000 to 8000. Both calibrations need the X ray to be powered on with the filter placed in front of the X ray source, if the following X ray CT scan may require t hat

PAGE 87

87 The voltage and amp erage will be recorded for future reference, in case of some abnormality occurring in the detector. 6.2.2 Position of Test Sample Either turn table or in situ load frame is cap able of being arranged between the X ray source and th e detector. Test sample needs to be fixed in its position to prevent undesired movement while the turn table is turning the segment during scanning. Additionally, the manufacturer suggests that it is convenient to scan a raised sample on the turn table. A plastic foam or low density material is suggested to help elevate the test sample. Additionally, duc t tap e is usually used to hold the test sample and plastic foam together. It is also important to place the test sample with the plastic foam carefull y in the center of turn table. The duc t tap e can also help in fixing the specimen position on the load frame without the hydraulic attenuator. When using the load frame, it is also a good idea to set a pre load to hold the specimen still. 6.2.3 Cont rast Optimization in DR I mages The manufacturer suggests adjusting the voltage (kV) and amperage ( ) to optimize the contrast of DR images The combination of kV and for CT scan can be varied. The rule of thumb (NSI training materials 2009) suggests to firstly ramp ing up the voltage high enough, while not saturating the circular edge of concrete sample. Second ly, adjust amperage to determine a sufficient gray level (more than 2500) by observing the longest route (thickest portion) of the specimen w here the X ray s penetrate completely Selection of X ray source, geometric positioning, and the use of filter may influence the clarity of DR image. A trial run obtaining quali ty DR images is ne cessary

PAGE 88

88 6.2.4 A cquisition of DR I mages The proprietary soft ware, X View IW CT is used for acquiring DR images at UF. I t is necessary to have the scanning degree s of angle (number of view steps) assigned for each revolution scan of 360 degrees. For instance, if six degree s per segment turn was assigned, th e re a re sixty DR images to be scanned and acquired The larger the number of DR images scanned, the more deta ils of the target ed subject can be captured for later 3 D reconstruction. Common number s of view steps in the X ray CT facility at UF are 120, 360, 72 0, and 1440 However, it is time consuming to acquir e more DR images For example, a 120 im age scan on a concrete slice with 1.5 inches in thickness and 6 inches in diameter will consume approximately one hour or about three hours pe r 360 image scan to collect DR images It is ne cessary to determine how many view steps are sufficient for the test sample. The manufacturer suggests view steps of 720, or a half degree per segment turn, for a revolution scan. 6.2.5 Reconstruction of 3 D Virtual Model The c ollected DR images will be transferred to the workstation computer with the proprietary software, efX CT that is used to gather DR images, acquire the spatial information, and stack DR images into the 3 D virtual model. A special medium calibration rod with spheres spac ed at 5 mm is designed to capture the spatial information where the test specimen was positioned. The purpose of obtaining this spatial information is to provide a volume reference for software to establish the reconstruction. It is no rmal to capture 60 images with the same voltage used for a test specimen uses, but the amperage needs to be tu r ne d down in order to capture DR images on the calibration tools with sufficient contrast. It is easier to complete this task with the 225 kV X r ay source, and it may take many trials to finish this calibration with

PAGE 89

89 the 450 kV X ray source. The definition issue is usually the main reason associated with the geometric positioning of the test sample. An adjustment using other amperages or an enhanc ement of contrast on acquired DR images may sometime s facilitate the process. A repetition of the scan from the beginning to obtain a better geometric positioning may be needed. It has been said that this 3 D calibration is crucial but may take many att empts to complete the process. Figure 6 3 shows a completed process of obtaining the cylindrical spatial information. Figure 6 3. A cquisition of spatial information by 3 D calibration tool After acquiring 60 DR images of the calibration tool, the s ystem is ready to assemble the collected DR images of test sample and of th e calibration images to reconstruct the 3 D virtual model. Similar to the process of collecting DR images, the

PAGE 90

90 system requires about one hour for a 120 image scan or three hours fo r a 360 image scan for reconstruction process. However, it must be noted that the overall time span for completing an X ray CT scan depends on the specimen thickness, the pixel size of the DR image, and the number of view step s Figure 6 4 depicts a 3 D virtual reconstructed model of a concrete specimen. Figure 6 4. 3 D reconstructed model of a concrete specimen 6.2.6 Export of 2 D Images of Virtual Slice When the process of reconstruction is finished, a virtual 3 D will be virtually available with in the window created by the software efX CT It has been mentioned that the advantage of X ray CT is the ability to observe and to investigate the properties or phenomen a of interest within the test specimen. The software is c ap able of

PAGE 91

91 recording video clips of the user s manipulation on a test sample or of exporting the virtual sliced images on the desired axis or angle. (A) (B) Figure 6 5. (A) A cross sectional image of reconstructed 3 D model in concrete after process of window leveling on Z axis (B) Histogram of the 3 D model A histogram is built associated with the reconstructed 3 D model. The histogram is a distribution or presentation that plots the accumulated number of pixels by specific gray levels. A technique of so called by confining the histogram to

PAGE 92

92 certain gray levels, can result in segmentation or a divide d view of each element separately For example, a reconstructed model of concrete can apply window leveling to separate aggregates, cement paste, and void s. Figure 6 5 (A) shows a cross sectional image of a 3 D reconstructed model in concrete after the process of window leveling. Figure 6 5 (B) presents the histogram of the 3 D reconstructed model. The observer pull s the left and right bars to adjust the central bar so that it fall s more or less on the trough between two crests on the histogram. The whole cross sectional virtual slice is visible on the computer monitor by carefully tuning these bars. Once the area of interest is clearly observed the process of exporting 2 D images of virtual slice s can be performed. The software designates the default three axes X, Y, and Z to export such cross sectional slices. The spacing between slices is based on the resolution of the 3 D reconstructed model. For example, if the reconstructed model has the resolution of 71 micron s in voxel size, the distance between two virtual slices is 71 micron s or 0.003 inches. Hence, the location of a specific virtual slice may be located for further investigations. In s hort, it is feasible to perform X ray CT scan to acquire 2 D DR images and to subsequently reconstruct the virtual 3 D model. By performing the window leveling technique on the histogram enable s the observer to reveal the area of interest The 2 D images of virtual slice s on the test sample can be exported with the spacing distance identical with the resolution size for further analysis 6 3 X ray CT in Conjunction with the Superpave IDT Strength Test 6.3.1 Test Program The Superpave IDT Strength Test h as been adopted to test concrete specimens as described in C hapter 5. A displacement control loading rate of 0.00075 inches/sec

PAGE 93

93 with the closed loop servo controlled was selected for this task. It was decided to employ t he loading program to apply an inc remental load/unload pattern using the concrete mixes with 90 days curing by the Superpave IDT strength test, namely from 0% (before loading), 10%, 20%, and up to the initiation of crack or failure. The test was performed in ambient temperature approxim ately 23 C The X ray CT scan was performed after applying ten pounds of pre loading to hold the test specimen. After the first scan was completed, a 10% load was then applied to the specimen and sustained for ten seconds. After ten second s the force w as adjusted back to pre load and another X ray CT scan was performed. It was assumed that the specimen had been damaged as much as necessary in ten second s by the 10% incremental load. The loading and unloading pattern was to prevent the creep effect fr om adversely affecting the cause of microcrack initiation. The loading and unloading pattern continued until a failure was detected. 6.3.2 Concrete M i xes Evaluated Two PCC mixes (w/c=0.45 and w/c=0.55) and four w/c=0.50 concrete mixes containing differe nt levels of RAP replacement (20%, 40%, 70%, and 100%) were tested under the load/unload loading pattern, while X ray CT scans were p erformed at the unloading mode RAP material was obtained from Vero Beach, Florida, where its aggregate composition had be en verified as mostly Florida limestone Concrete cylinders were subjected to 100% moisture for 90 days of curing and concrete specimens with 1.5 inches in thickness were prepared. The preparation of concrete specimen s was identical with the description s of preparation for the Superpave IDT strength test described in Chapter 5.

PAGE 94

94 6.3.3 Test Procedures 6.3.3.1 Geometry of load frame The in situ load frame with force range of 22 kip (100 kN) is located inside the chamber and can be arranged in a chosen posit ion between X ray sources and FPD. Under such an arrangement, it is possible to apply loads to a specimen without moving its position, before and after a test. The geometrical positioning was set where the center of specimen was about three feet away fro m the X ray source and two feet away from the FPD. This geometry offered resolution to 71 micron in voxel size. The X ray source of 225 kV with M i crofocus was selected for this task. The s maller focal spot (< 6 micron) of the 225 kV tube can help minimi ze the unnecessary loss of definition and facilitate the efficiency of the later process of 3 D reconstruction. It was decided to use 120 kV and 550 for scanning the concrete specimen, and 120 kV and 130 for 3 D calibration of reconstruction under this geometry throughout this task. For 3 D calibration of reconstruction, the kV had to remain the one used for scanning concrete (i.e ., 120 kV). Th e adjustment of amperage was used to observe black dots along the calibration rod clearly and to saturate the wrapped plastic materials. 6.3.3.2 Positioning of test sample on the load frame An aligning steel bar was required to be attached to the turn tabl e of the loading frame, as shown in Figure 6 6 in order to synchronize the movement between the top hydraulic actuator and the turn table on the bottom when the X ray CT scan was performed.

PAGE 95

95 (A) Hydraulic actuator (B) Aligning steel bar (C) V ertical (V) and horizontal (H) strain gages on both edges of specimen. (D) Turn table of load frame Figure 6 6 The load frame set up under Superpave IDT apparatus (Photo courtesy of Yu Min Su) Additionally, there is a limitation in synchronizing bot h compartments. It was found that the revolution X ray CT scan can be complete d when than 2500 pounds is applied

PAGE 96

96 at the same time. However, a pre load of ten to fifteen pounds was chosen to secur e the concrete specimen in the same location before and aft er the test, which was far less than this limitation of appl ied load. 6.3.3.3 Performing an X ray CT scan on concrete The X ray CT scan set to acquir e 120 DR images was selected in this task. It has to be noted that the CT scan in conjunction with the loa d frame took a longer time for acquiring each DR image, because shad ows of the top hydraulic actuator and the bottom turn table appear ing nearly black as shown in Figure 6 7, comprised more volume information for the acquisition system to handle. It took 45 minutes to complet e an acquisition of 120 DR images. Figure 6 7 DR image of the Superpave IDT concrete specimen with strain gages

PAGE 97

97 Moreover, if the scan of the 3 D calibration tool for reconstruction could be acquired properly, the reconstruct ion could be established within about an hour on the workstation computer by 120 DR images. Detailed procedures for obtaining DR images, reconstructing 3 D virtual model, and exporting virtual slices have been discussed in C hapters 6.2.4 to 6.2.6. 6.3.3. 4 Exporting virtual slices on concrete The reconstructed 3 D model can be manipulated to export virtual slices at the designated angle. In this study, hundreds of virtual slices were exported on the Z axis ( i e., along the axis of sample thickness). The spacing between two virtual slices was 71 micron s (0.003 inches), identical with the voxel size, resulting in about 530 effective slices for a concrete specimen 1.5 inches in thickness. However, the usable slices usually were lower. The position, orienta tion, and surface condition of a specimen as well as artifacts of a reconstructed 3 D model can reduce us able images fo r analysis. There were either 411 or 441 effective virtual slices of any mix found suitable for further analysis. 6.3.4 Development o f Image Processing Technique of Analyzing Air Voids A n image process ing protocol f or per ceiving air void s ha d been developed and will be introduced in this chapter By examining the exported virtual slices, the cupping or so called beam hardening eff ect had been found, which cause d the area close to the circular edges that appeared to be higher density. However, the attenuation in the central area was rather uniform. Since the concrete specimen was tested under the indirect tension mode, the most in terest ing area was in the cent e r. Theoretically the crack would initiate approximately in the central area under this loading condition. Thus,

PAGE 98

98 a 1.5 inch square area of interest w as determined for this task, where it also w ould more or less fit in with the stain gage measurement of the Superpave IDT test apparatus used to measure horizontal and vertical displacements on both surfaces of concrete specimen. Several image processing techniques were introduced as follows: Cropping the interest area ; Apply ing proper threshold of gray level to isolate voids ; Calculating area of air voids ; Performing the analysis of air void distribution. 6.3.4.1 Crop the interested area Figure 6 8 shows a real virtual slice cut from a reconstructed 3 D model by a quick CT scan with 120 DR images. Figure 6 8 Cropping the interested area from a virtual slice

PAGE 99

99 Several artificial defects were found on both top and bottom and also near central area. The central artifacts were induced by four gage points designated f or measuring vertical and horizontal displacements in Superpave IDT. S everal image processing software s exist that can crop images to the desired area and i t was decided to use Adobe Photoshop CS5 in this study. Firstly, the blue square was designate d manually to include the whole visible concrete spec imen. After selecting this six inch area, a similar way of cropping one segment one quarter that size from the center can be chosen to accurately receiv e the central 1.5 inch square area. The unit can be manipulated by pixel numbers or by operator s preferences The cropped area of interest can thereafter be saved into a new image file, shown in Figure 6 9. Figure 6 9. Cropped image 1.5 inches square on each s lice

PAGE 100

100 6.3.4.2 Apply pr oper threshold of gray level to isolate voids The next step was to isolate voids from aggregates and cement paste. A binary process had been adopted by taking advantage of considering voids and pores as air having a density of zero. Therefore, by adjusting the his togram of image, or so called window leveling the operator can obtain this information M ost image analysis software has the image adjusting function of threshold which can provide access to the binary process. It was determined to use threshold of gray levels of 87 in this study which could fit well to isolate voids f rom other elements, as shown in Figure 6 10. Figure 6 10. Cropped image after binary process (threshold=87) The area s appearing as black or near black w ere voids and pores and it can be seen that the rest of the elements, such as aggregates, cement paste, and certain

PAGE 101

101 artifacts were white d out. However, it must be remarked that the voids inside Florida limestone, artifacts from the reconstruction process, and noise signals from gage points were inevitabl y included. A tri a l and error process was extensively performed to avoid these issues and the threshold was used throughout this analysis. Moreover, a fixed position of the load frame without changing the X ray emission power ( i.e. kV and ) was suggested in order to constantly produce similar DR images for 3 D reconstruction as well as to maintain nearly identical contrast of export ed slices. The resulting image was saved in an independent folder for the next process. An automatic order batch with selected actions was compiled to find, crop, binary, and save resulting images in the chosen folder up to this step in the software. 6.3.4.3 Calculat ing area of air voids M ore than four hundred images for each mix were needed for further ana ly sis to calculate the void or area of nearly black pixels. There was a need to develop an automatic process to calculate the area of voids. The MATLAB R2011b with signal process and image analysis toolboxes was selected for this study. The MATLAB code was developed and compiled to calculate total pixel numbers determine numbers of nearly white pixels, and save these data in a M i crosoft Excel file in this cropped binary image. The concept can be reverse d, calculating numbers of quasi black pixels as well. 6.3.4.4 Perform ing the analysis of air void s distribution The air void s distribution from all virtual slice s can be estimated and recorded in the same Excel file. A plot of percentage of air voids versus the location along the Z axis ( i.e., axis of sample thickness) for each mix was established

PAGE 102

102 6. 4 Analysis of Air Voids Distribution in Different Load Levels 6.4.1 Effects of Loading and Unloading on Concrete Mixes Table 6 1 shows the loading/unload pattern of concrete mixes. The load level was designated and incrementally increased by ten percent based on the peak load of a concrete specimen after 90 day s of curing in the Superpave IDT strength test. In Table 6 1, for instance the w/c=0.45 mix with 10% load level, it was stopped at a load 12 23 pounds, which reflected 13.2% (1223/9268) of final failure load. Table 6 1. Load/unload pattern used in Superpave IDT strength test Load Level lbf PCC w/c=0.45 PCC w/c=0.55 Concrete 20%RAP Concrete 40%RAP Concrete 70%RAP Concrete 100%RAP 10% 1223 ( 13.2%) 980 (15.3%) 683 (12.8%) 600 (12.0%) 600 (17.0%) 480 (11.1%) 20% 2097 (22.6%) 1751 (27.3%) 1350 (25.3%) 1000 (21.8%) 1180 (33.5%) 860 (25.5%) 30% 3100 (33.4%) 2474 (38.6%) 1960 (36.7%) 1750 (35.0%) 1550 (44.0%) 1350 (40.1%) 40% 4300 (46.4%) 3500 ( 54.5%) 2750 (51.5%) 2300 (45.9%) 2020 (57.3%) 1620 (48.1%) 50% 4956 (53.5%) 4217 (65.7%) 3280 (61.4%) 2800 (56.0%) 2500 (71.0%) 2050 (60.9%) 60% 5979 (64.5%) 5094 (79.4%) 3890 (72.8%) 3500 (69.9%) 3000 (85.2%) -70% 6762 (71.7%) 5742 (89.5%) 4520 (83.3 %) 4050 (80.9%) --80% 7778 (83.9%) -5200 (97.4%) ---Failure Load (Peak Load) 9268.9 (9489.6) 6417.7 (8544.5) 5341.4 (6500.3) 5006.8 (5958.6) 3523.1 (5094.9) 3368.6 (4311.8) Almost every mix was tested up to 80 percent of the failure load w ith the only exception being the concrete mix containing 100% RAP. Every tested concrete m ix had a lower final failure load, compared with the 90 d strength test, with the exception of w/c=0.45 mix. This may suggest that the 80% load level may not initia te damage to

PAGE 103

103 w/c=0.45 mix. However, a rather clear strength reduction can be observed for w/c=0.55 PCC and for all four mixes of concrete containing RAP. 6.4.2 A verage Air Voids under Different Loading Level The air void distribution had been used to analyze two PCC mixes and four concrete mixes containing RAP, as shown in Figure 6 11 to Figure 6 1 6 and Table 6 2. Each curve presented an internal distribution of air voids through the Z axis under corresponding load level. Each figure contains the dis tribution curve before loading in gray color, namely zero percentage of loading, and other color curves with four load levels for the purpose of observation. It was assumed that the variation of void distribution would be manifest as the indication of mic rocracking occurrence internally. Table 6 2. Average air voids under different load levels F or analyzing void differential among all tested mixes, the maximum variation did not excee d of air voids. Void distribution of w/c=0 .45 PCC, concrete mixes containing 40% and 70% RAP were more or less in a n increasing t rend A relative large number of voids growing at 60% load level (corresponding with 64.5% of peak load) in PCC with w/c=0.45 PCC, 50% to 70% load level (corresponding with 56.0% of peak load) in concrete containing 40% RAP, and 40% to 60% load level (corresponding with

PAGE 104

104 57.3% of peak load) in concrete containing 70% RAP were noticeable. Void distribution in 100% RAP had great void variation at 30% load level (correspond ing with 40.1% of peak load), while air void distributions in w/c=0.55 PCC and 20% RAP mixes appeared to be irregular, or not noticeably exact in any manner A drop in air voids was also observed in the early stage of loading, such as 10% load level at w /c=0.55 (corresponding with 15.3% of peak load) and 20% load level at w/c=0.45 PCC (corresponding with 22.6% of peak load), 20% RAP (corresponding with 25.3% of peak load), 70% RAP (corresponding with 33.5% of peak load), and 100% RAP (corresponding with 2 5.5% of peak load). This reduction can be attributed to the well known debonding of concrete layers caused by relatively low loads. Such debonding or microcracks will not propagate and the microstructure is considered intact (Mier 1997). 6.4.3 Average Air Voids in Concrete containing RAP during Fracture There were two concrete containing 40% and 70% RAP that proved subject to the condition whe rein the failure was detected without breaking the specimen apart. The distributions of air voids at the point of fracture were more or less captured without moving the sample from the load frame. Two attempts were made to recover PCC specimens after fracture ; the specimens were duc t taped to perform the CT scan. However, the loss of concrete materials during rup ture increased the average percentage of air voids. It was rather difficult to match up the distribution of air voids after the specimen was remov ed from the original place. Nevertheless, it was fairly noticeable that the fracture would cause a dramatic jump in air voids, shown in Figure 6 17 and Figure 6 18.

PAGE 105

105 Another attempt was made to evaluate the void variation during fracture. There was a dot curve with orange color shifted by the magnitude of difference in percentage of air voids before loading an d in fracture. It was found in the shape of an Arabic numeral 8 or a double S that voids on one side grew more than on the other on the concrete specimen during fracture, compared with the shifted distribution. PCC mixes had more app arent crack s opening than was the case with RAP mixes, which may also suggest the explosive nature of failure in PCC. This phenomenon may be re lated to the similar observation on stress strain curve in this study, in chapter 5. The crack initiation and g rowing horizontal displacement occur red on the same side of specimen, while a snap or a negative displacement on the other side of specimen happen ed simultaneously The near fracture bending of concrete specimen s under the Superpave IDT strength test w ork ed differently than regular asphalt specimen s

PAGE 106

106 (A ) ( B ) Figure 6 11. Air void distribution of w/c=0.45 PCC mix

PAGE 107

107 ( A ) ( B ) Figure 6 12. Air void distribution of w/c=0.55 PCC mix

PAGE 108

108 (A ) ( B ) Figure 6 13. Air void distribution of concrete mix containing 20% RAP

PAGE 109

109 ( A ) ( B ) Figure 6 14. Air void distribution of concrete mix containing 40% RA P

PAGE 110

110 ( A ) ( B ) Figure 6 15. Air void distribution of concrete mix containing 70% RAP

PAGE 111

111 ( A ) ( B ) Figure 6 16. Air void distr ibution of concrete mix containing 100% RAP

PAGE 112

112 ( A ) w/c=0.45 PCC ( B ) w/c=0.55 PCC Figure 6 17. Voids distributions after fracture of w/c=0.45 and 0.55 PCC mixes

PAGE 113

113 (A ) concrete containing 40% RAP (B ) concrete containing 100% RAP Figure 6 18. Voids distributions after fracture of concrete mixes containing 40% and 100% RAP

PAGE 114

114 6. 5 Visualization of Microcracking in Concrete 6.5.1 Visualization of Concrete Microcracking under Load/Unload Pattern A detailed side by side comparison in virtual slices wa s performed for w/c=0.45 PCC mix to approximate visualiz ation of microcracking. The intent was to examine how distributions of air void varied along with the incremental load level. Figure 6 19 shows identical air void distribution, as Figure 6 11, with several locations of interest from one surface. By observing the curves compared with the distribution before loading, there were locations at 0.069 and 0.52 inches with higher average air voids, while locations of 0.32 and 1.0 inches developed lower air voids. Locations at 0.15 and 0.72 inches were in a transition zone of void reduction Figure 6 20 to Figure 6 25 provided a side by side comparison of slices under different load levels. For lower load levels, the variation of void distribution seemed t o be insignificant except for the one of 22% ( i.e., debonding of concrete layer) However, for higher load levels, it appeared that the internal structure with large air voids, such as 0.069 and 0.52 inches, and with lower air voids, for instance 0.32 an d 1.00 inches, had the tendency of enlarging the average air voids. By carefully examining the slices on these locations ( marked in orange circles ), the new microcracks were occurring on the cement paste. Some existing voids were seemingly enlarging the size of voids or the shape of voids had been deformed, due to the loading. At some narrow tips of cement paste in between aggregates (marked in blue circles) of locations of 0.15 and 0.72 inches, it was found that previous discrete microcracks became cha nnelized to be connected to each other. The tiny new microcracks or flow movement in terms of voids can be arguably detected under this quick X ray CT scan protocol However, it was not found that the microcracks were

PAGE 115

115 specifically generat ed around aggre gates in this mix nor upon additional examination of concrete mix containing 100% RAP, shown in Figure 6 26. ( A ) ( B ) Figure 6 19. Air voids distribution in w/c=0.45 PCC with locations of interest

PAGE 116

116 ( A ) Cropped image ( B ) before loading ( C ) 13.2% of peak load ( D ) 22.6% of peak load ( E ) 33.4% of peak load ( F ) 46.4% of peak load ( G ) 53.5% of peak load ( H ) 64.5% of peak load ( I )71.7% of peak load ( J ) 83.9% of peak load Figure 6 20. Air voids at the location of 0.069 inches

PAGE 117

117 ( A ) Cropped image ( B ) before loading ( C ) 13.2% of peak load ( D ) 22.6% of peak load ( E ) 33.4% of peak load ( F ) 46.4% of peak load ( G ) 53.5% of peak load ( H ) 64.5% of peak load ( I )71.7% of peak load ( J ) 83.9% of peak load Figure 6 21. Air voids at the location of 0.15 inches

PAGE 118

118 ( A ) Cropped image ( B ) before loading ( C ) 13.2% of peak load ( D ) 22.6% of peak load ( E ) 33.4% of peak load ( F ) 46.4% of peak load ( G ) 53.5% of peak load ( H ) 64. 5% of peak load ( I )71.7% of peak load ( J ) 83.9% of peak load Figure 6 22. Air voids at the location of 0.32 inches

PAGE 119

119 ( A ) Cropped image ( B ) before loading ( C ) 13.2% of peak load ( D ) 22.6% of peak load ( E ) 33.4% of peak load ( F ) 46.4% of peak load ( G ) 53.5% of peak load ( H ) 64.5% of peak load ( I )71.7% of peak load ( J ) 83.9% of peak load Figure 6 23. Air voids at the location of 0.52 inches

PAGE 120

120 ( A ) Cropped image ( B ) before loading ( C ) 13.2% of peak load ( D ) 22.6% of peak load ( E ) 33.4% of peak load ( F ) 46.4% of peak load ( G ) 53.5% of peak load ( H ) 64.5% of peak load ( I )71.7% of peak load ( J ) 83.9% of peak load Figure 6 24. Air voids at the location of 0.72 inches

PAGE 121

121 ( A ) Cro pped image ( B ) before loading ( C ) 13.2% of peak load ( D ) 22.6% of peak load ( E ) 33.4% of peak load ( F ) 46.4% of peak load ( G ) 53.5% of peak load ( H ) 64.5% of peak load ( I )71.7% of peak load ( J ) 83.9% of peak load Figure 6 25. Air voids at the location of 1.00 inches

PAGE 122

122 ( A ) Cropped image of before loading ( B ) Before loading ( C ) 60.9 % of peak load Figure 6 26. Air voids at the location of 0.60 inches of 100% RAP 6.5.2 Visualization of Concrete Microcracking during Fracture Moreover, a further assessment of void distribution was attempted to evaluate the air void distribution. As has been discussed in Chapter 6.4, the percentage of air void s on one surface was growing positively more than the other side, during frac ture. Voids on the other side actually grew less than average, for instance in Figure 6 16 of 100% RAP mix. Three locations at 0.09, 0.60, and 1.00 inches slices, before loading and fracture, had been pulled out for the side by side comparison, shown in Figure 6 2 7 to 6 2 9 ( A ) Cropped image of before loading ( B ) Before loading ( C ) Break Figure 6 27. Air voids at the location of 0.09 inches of 100% RAP (voids reduction)

PAGE 123

123 ( A ) Cropped image of before loading ( B ) Before loading ( C ) Break Figure 6 28. Air voids at the location of 0.60 inches of 100% RAP (voids dull) ( A ) Cropped image of before loading ( B ) Before loading ( C ) Break Figure 6 29. Air voids at the location of 1.00 inches of 100% RAP (voids growing) The lo cation at 0.09 inches found the air void distribution, compared with the shifted average, showing a reduction of voids; the location at 0.60 inches showed no voids variation; and the location at 1.0 indicated a slight grow th in voids. The fracture pattern and voids variation can be seen on all ( C ) images with red marks. Air voids around both cliffs of fracture were visually less at location of 0.09 inches, seemingly no activities at the location of 0.60 inches, but slight enlargement at the location of 1. 00 inches. It was obvious that the fracture itself had created a relatively enormous amount

PAGE 124

124 of voids and also pushed both plateaus away from each other. It has to be assumed here that this separation only cause d a trivial derivative error of air voids di stribution. Since it could be now recognized that the crack started from the vicinity of surface farthe r from the location of zero propagat ing through the other end, the closing cracks may be expl icable by the well known crack analysis first proposed by Hillerborg (Hillerborg et al., 1976) It was described for the mode I crack the opening due to the tensile stress. Due to the plasticization or tensile softening around the crack tip, cracks around the tip may be closing by the lateral force generated a long with the crack propagati o n. O n the other hand the rupture behavior of concrete may rip the microstructure apart instantaneously without softening both regions, hence no closing voids. In the concrete mix containing 100% RAP, the reduction in air vo ids around the fracture plane can be found between locations of zero to 0.60 inches, which may suggest the location of onset crack tip. Figure 6 30 presents the fracture pattern of w/c=0.45 PCC, which exhibited a rather straight vertical fracture plane, w hich also resembled similar observations of fracture plane in testing thin specimen s with Superpave IDT strength test or flexural beam test of conventional ASTM C78. Additionally, the right bottom air void appeared to siphon the fracture plan e toward its vicinity and eventually th at plane connected with void. When compared with Figure 6 27 (A) and (C), fracture in RAP concrete mixes developed a longer route than PCC and also noticeably propagated around aggregates. The more meanders in the route of the fracture plane, the more energy dissipates which can support the result of higher toughness in concrete mixes containing RAP.

PAGE 125

125 (A ) ( B ) Figure 6 30 Fracture pattern in w/c=0.45 PCC mix 6. 6 Research Findings The main findings from the investiga tion on the use of X ray CT on concrete are as follows: An image processing technique for analyzing air voids was developed. For the test configuration used, the use of a threshold value of 87 gray levels (out of 2 8 =256) was found to give good results in differentiating between air voids and non air voids in the X ray images of the concrete surface. The computer code for automatic processing of the X ray image for calculation of air voids was successfully developed. It was found that there was very smal l variation in air voids for concrete with or without RAP. The maximum variation of air voids measured from the central are a of the specimen was within for all the concrete mixes tested. The volume of a ir v oids in the concrete was observed to increase significantly when a concrete specimen was loaded to fracture The air voids developed more on the specimen face where a crack initiated, but grew less on the other face. As a concrete specimen was loaded and unloaded, air v oids wh ich were formed tended t o develop in the cement paste This observation applies to both concrete with or without RAP.

PAGE 126

126 The path of crack propagation in concrete containing 100% RAP was found to be longer than that in concrete without RAP, a s examined on virtual slices from X ray CT scan. This explained the higher toughness in the concrete containing RAP.

PAGE 127

127 CHAPTER 7 CONCLUSIONS AND RECO MME N DATIONS 7.1 Conclusions Portland Cement Concrete (PCC) and concrete incorporating RAP as aggregate were evaluated b y the Superpave Indirect T ensile (IDT) strength test in this study. Concrete mixture properties were obtained from the IDT test using a constant force control and a displacement control rate The properties obtained included tensile strength, elastic mod ulus, Poisson s ratio, and fracture energy. S tress strain behavior of the concrete were also studied. The effects of temperature and different curing time to concrete properties were also investigated. Relationships between the conventional and Superpav e IDT tests were studied using the limited data obtained Major research findings with regards to use the Superpave IDT strength test on concrete mixes are as follows: When the maximum load obtained from the Superpave IDT strength test was used to calcula te the tensile strength of the concrete at 28 days curing time the computed tensile strength from the Superpave IDT strength test correlated well with the corresponding splitting tensile strength from the conventional splitting tensile strength test. The elastic modulus and Poisson s ratio of the concrete obtained at 40% of its ultimate stress in the S u perpave IDT strength test cor related well with the corresponding values obtained from the conventional compressi ve test. When the toughness is calculated by determining the area under the stress strain plot up to the maximum stress, it can be used to differentiate between con cretes containing different percentage of RAP. Superpave IDT s trength test using a displacement control mode were very close to those obtained from the Superpave IDT strength test using a force control mode In running the Superpave IDT strength test on concrete specimens, when the concrete specimen was loaded beyond its elastic limit, the horizontal strain on one of the specimen face tended to go in the negative direction while the horizontal

PAGE 128

128 strain on the other face increased in the positive direction. However, this problem was not observed when concrete containing high percentage of RAP was tested. The tensile strength of concrete without RAP at early age was seen to increase as the temperature decreased. However, this effect of temperature was not seen among the concrete at later ages. The tensile strength of concrete containing RAP was seen to decrease as the %RAP increased. The tensile strength of concrete containing RAP was seen to decrease as the temperature increased. The addition of RAP in concrete noticeably reduced the elastic modulus of the concrete Toughness of concrete was seen to increase as the %RAP in concrete increased. Furthermore, the test procedures of X ray Computed Tomography (CT) in conjunction with the S u perpave IDT were developed. An image processing technique was established t o assess the distribution of air voids and to visualize the microcracking in concrete. The following are the main findings from this investigation: An image processing technique for analyzing air voids was developed. For the test configuration used, the use of a threshold value of 87 gray levels (out of 2 8 =256) was found to give good results in differentiating between air voids and non air voids in the X ray images of the concrete surface. The computer code for automatic processing of the X ray image for calculation of air voids was successfully developed. It was found that there was very small variation in air voids for concrete with or without RAP. The maximum variation of air voids measured from the central are a of the specimen was within for all the concrete mixes tested. The volume of a ir v oids in the concrete was observed to increase significantly when a concrete specimen was loaded to fracture The air voids developed more on the specimen face where a crack initiated, bu t grew less on the other face. As a concrete specimen was loaded and unloaded, air v oids which were formed tended t o develop in the cement paste This observation applies to both concrete with or without RAP.

PAGE 129

129 The path of crack propagation in concrete containing 100% RAP was found to be longer than that in concrete without RAP, a s examined on virtual slices from X ray CT scan. This explained the higher toughness in the concrete containing RAP. 7.2 Recommendations Based on the research findings the fol lowing recommendations are made : It is recommended to adopt the Superpave IDT strength test using a constant displacement control rate of 0.00075 in/sec to test concrete mixture s in tension. More replicate tests are needed to obtain sufficient data to est ablish the relationships between the properties obtained by conventional tests and those by the Superpave IDT strength test. A protocol of fatigue test using the Superpave IDT test on concrete mixtures needs to be evaluated and established. The image proc essing software to indentify internal structural properties on the 3 D reconstructed model or the exported 2 D virtual slices needs to be developed A technique to compensate for the beam hardening effect around the edges of a specimen in a X ray CT sca n needs to be developed, so that t he distribution of air voids on a whole concrete specimen could be properly determined More trials on different combination s of geometry voltage, and amperage are needed. The use of the Superpave IDT strength test in co njunction with the X ray CT technique for the study of the stress strain behavior of concrete beyond the elastic limit and crack initiation needs to be further studied.

PAGE 130

130 L IST OF REFERENCES AASHTO T322 (2007). Standard Method of Test for Determining the C reep Compliance and Strength of Hot Mix Asphalt (HMA) Using the Indirect Tensile Test Device American Association of State Highway and Transportation Officials Washington DC Alshibli, K and Reed A. (2010) Advances in Computed Tomography for Geomateri als Geo X 2010 28 February 3 March, 2010, New Orlean, United States of America. London: ISTE, 2010. Print. ASTM C469 (2010). Standard Test Method for Static Modulus of Elasticity and ASTM International West Conshohocken PA. ASTM C496 (2011). Standard Test Method for Splitting Tensile Strength of Cylindrical Concrete Specimens ASTM International West Conshohocken PA. ASTM E 1411 (2011). Standard Guide for Computed Tomography (CT) Imaging ASTM Internationa l, West Conshohocken, PA. Strategic Highway Research Program Measurement and Analysis for Indirect Tensile Testing at Low Temperature Trans Res Rec 1454 Transportat ion Research Board, National Research Council, Washington, D.C., 163 171. Buzug, T M. (2008). Computed Tomography: From Photon Statistics to Modern Cone Beam CT Berlin: Springer Desrues, J, Viggiani G. and Besuelle P. (2006) Advances in X Ray Tomogra phy for Geomaterials. GeoX 2006 4 7 October, 2006, Aussois, France London: ISTE, Print. Farca F. A. (2012). Evaluation of Asphalt Field Cores with Simple Performance Tester and X ray Computed Tomography Licentiate Thes Div. of Highway and Railway Eng., Dep Tran Sci., KTH Royal Institute of Technology, SE 110 44 Stockholm, Sweden Hillerborg, A., Moder, M., and Petersson P. E. (1976). Analysis of crack formation and crack growth in concrete by means of fracture mechanics and finite elements Cem. Concr. Res. 6, 773 782. a Low Tensile Resistane by the Brazilian (Indirect Tensile) test with particular reference to Concrete Australian J App Sc i 10 ( 3 ) 243 268. J. Mater. Civ. Eng. 17(6), 632 639.

PAGE 131

131 Landis, E N.. (2006). ray Tomography as a Tool for Micromechanical Investigations of Cement and Mortar GeoX 2006 4 7 October, 2006, Aussois, France. London: ISTE, Print 79 93. Landis, E. N. and Bolander J. E. (2010). Element Modeling for Concrete Fracture Problems GeoX 2010 28 February 3 March, 2010, New Orlean, United States of America. Lo ndon: ISTE, Print 117 123. Landis, E. N. and Keane, D. T. (2010), X ray Microtomography. Mater Charac. 61, 1305 1306. Landis, E. N. and Nagy, E. N. (2000). Three Dimensional Work of Fracture for Mortar in Compression. Eng. Frac. Mech. 65, 223 234 Lane, E. (1987). Guide to rocks and minerals of Florida Florida Geological Survey Special Publications 8, Tallahassee, Florida, 61. Curves Using Resilient Modulus and Cree p Test Data Res Rep Num FL/DOT/SMO/11 544 State Material Office, Florida Department of Transportation, Gainesville, Florida. Masad E. and Kutay, M. E. (2012). Characterization of the Internal Structure of Asphalt Mixtures. Trans. Res. Circ. E C161 Transportation Research Board, Washington D.C., 2 16 Masad, E. Jandhyala, V. K., Dasgupta, N., Somadevan, N., and Shashidhar, N. (2002). Ray CT J. Mater., Civ. Eng. 14(2), 122 129. Ma sad, E., Muhunthan, B., Shashidhar, N. and Harman, T. (1999). Laboratory Compaction Effects on the Interna l Structure of A s phalt Concrete. Trans. Res. Rec. 1681, 179 185 Mier, J G. M. (1997). Fracture Processes of Concrete: Assessment of Mat erial Parameters for Fracture Models Boca Raton: CRC Press North Star Imaging (2009). Real Time X ray, Digital Radiography & Computed Tomography Training Materials and Notes, Rogers, Minnesota. and O bara, Y. (2004). X ray Ct for Geomaterials: Soils, Concrete, Rocks : Proceedings of the International Workshop on X Ray Ct for Geomaterials Geox2003 6 7 November, 2003, K u mamoto, Japan Lisse, the Netherlands: A.A. Balkema, Print.

PAGE 132

132 P romen tilla, M. A. B. and S ugiyama, T. (2010). ray CT to Investigate the Frost induced Damage Process in Cement based Materials GeoX 2010 28 February 3 March, 2010, New Orlean, United States of America. London: ISTE, Print 124 1 31. Mechanics: A Fundamental Crack Growth Law for Asphalt Mixtures J Asso Asp h. Pav Tech 71, 816 827. ent and Analysis System to Accurately Determine Asphalt Concrete Properties Using the Indirect Tension Mode Asso Asph Pav Tech 61, 304 332. SHRP Indirect Tensi on Tester to Mitigate Cracking in Asphalt Concrete Pavements and Overlays ., FDOT Project #0510755 (B9885), Department of Civil Engineering, Gainesville, Florida, 32611 6580 Suzuki, T. Ogata, H., Takada, R., Aoki, M., and Ohtsu, M (2010). f A coustic E mission and X ray C omputed T omography for D amage E valuation of F reeze T hawed C oncrete Constr. Build. Mater. 24, 2347 2352. Using the Superpave Indirect Tensi le Test (IDT) Ph. D. dissertation Department of Civil Engineering, University of Florida, Gainesville, Florida 32611 6580 Verification of a Suitable Crack Growth Law J Asso Asph Pav Tech 70, 206 241.

PAGE 133

133 BIOGRAPHICAL SKETCH Yu Min Su came from Taiwan and was born in 1975 He received the Bachelor of Science in 1997 and Master and S cience in 2002 at the Department of Civil Engineering at the National Central Univers ity, Chung Li, Taiwan. Yu Min Su served in the Coast Guard Administration of Taiwan in fulfillment of his 24 month military service Yu Min Su joined the Ph. D program of the materials group at the Department of Civil and Co a stal Engineering of the Univer sity of Florida in 2008, and worked as a graduate research assistant under the supervisory of Dr. Mang Tia. He received the fellowship of Inte rnational Road Federation in 2008 and attended the IRF meeting in Washington, D.C. in January of 2009. He also a ttended a technical training o n X ray computed tomography in Rogers, Minnesota and became a certified operator since then. Academically, Yu Min Su received three consecutive outstanding achievement awards from 2010 to 2012 at UF. After graduation, he pla ns to work in academia or public sector in c ivil engineering.