ABSTRACT 
 
 
Title of Document: CATALOG OF MATERIAL PROPERTIES 
FOR MECHANISTIC-EMPIRICAL 
PAVEMENT DESIGN   
  
 Rui Li, Master of Science, 2011 
  
Directed By: Dr. Charles W. Schwartz 
Department of Civil and Environmental 
Engineering 
 
 
 
 
The new pavement design methodology is based on mechanistic-empirical 
principles that are expected to be used in parallel with and eventually replace the 
current empirical pavement design procedures. The new mechanistic-empirical 
pavement design guide (MEPDG) requires greater quantities and quality of input 
data. Material characterization for the mechanistic-empirical approach, the focus of 
this thesis, is significantly more fundamental and extensive than in the current 
empirically-based AASHTO Design Guide. The objective of the thesis is to develop 
an organized database of material properties for the most common paving materials 
used in Maryland. A comprehensive material property database in Microsoft Access 
2007 has been developed. The database is initially populated with all information 
received from SHA. It provides complete data management tools for adding and 
managing future data as well as data display screens for MEPDG. Recommendations 
for future material testing for Maryland are also provided. 
 
  
 
 
 
 
 
 
 
CATALOG OF MATERIAL PROPERTIES FOR MECHANISTIC-EMPIRICAL 
PAVEMENT DESIGN    
 
 
 
By 
 
Rui Li 
 
 
 
 
Thesis submitted to the Faculty of the Graduate School of the  
University of Maryland, College Park, in partial fulfillment 
of the requirements for the degree of 
Master of Science 
2011 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Advisory Committee: 
 
  Professor Charles W. Schwartz, Chair 
  Professor M. Sherif Aggour 
  Professor Dimitrios G. Goulias 
 
  
 
 
 
 
 
 
 
 
[THIS PAGE INTENTIONALLY LEFT BLANK] 
 
 
 
 
 
 
 
 
 
 
 
 
 ii 
Dedication 
To my wife, Junjian, and my coming baby. 
 
 
 iii 
Acknowledgements 
I would like to acknowledge Dr. Charles Schwartz for his generosity in 
providing me with a research assistantship and bringing me to the University of 
Maryland. Also, I thank him for his valuable advice and warm encouragement 
throughout this first phase of my graduate studies and for keeping me motivated for 
the challenges ahead. 
I would also like to express my thanks to Dr. Goulias for his valuable advice 
and encouragement. 
I also acknowledge Dr. Aggour for participating in my thesis committee and 
supporting me this last step towards my degree. 
 
 iv 
Table of Contents 
Dedication ..................................................................................................................... ii 
 
Acknowledgements ...................................................................................................... iii 
 
Table of Contents ......................................................................................................... iv 
 
List of Tables ............................................................................................................... vi 
 
List of Figures .............................................................................................................. ix 
 
CHAPTER 1: INTRODUCTION ................................................................................. 1 
 
CHAPTER 2: BINDER DATA .................................................................................... 5 
MEPDG Input Requirements .................................................................................... 5 
Binder Data Received and Preliminary Analysis ...................................................... 5 
Sensitivity Analysis of Level 1/2 vs. Level 3 Binder Property Data ...................... 25 
Summary ................................................................................................................. 26 
Testing Recommendations .................................................................................. 26 
Recommended MEPDG Inputs........................................................................... 27 
 
CHAPTER 3: HMA DATA ........................................................................................ 29 
MEPDG Input Requirements .................................................................................. 29 
New Construction/Reconstruction/Overlays ...................................................... 29 
Rehabilitation ...................................................................................................... 32 
HMA Data Summary and Preliminary Analysis .................................................... 34 
Sensitivity Analyses for HMA Mixture Inputs ....................................................... 50 
Level 1 vs. Level 2 vs. Level 3 Dynamic Modulus ............................................ 50 
Thermal Properties .............................................................................................. 53 
Summary ................................................................................................................. 58 
Testing Recommendations .................................................................................. 58 
Recommended MEPDG Inputs........................................................................... 59 
 
CHAPTER 4: PCC DATA.......................................................................................... 63 
MEPDG Input Requirements .................................................................................. 63 
New Construction/Reconstruction/Overlays ...................................................... 63 
Rehabilitation ...................................................................................................... 67 
PCC Data Summary ................................................................................................ 69 
Sensitivity Analyses for PCC Inputs....................................................................... 71 
Strength and Stiffness Properties ........................................................................ 71 
Thermal Properties .............................................................................................. 92 
Shrinkage Properties ........................................................................................... 96 
Summary ................................................................................................................. 97 
Testing Recommendations .................................................................................. 97 
 
 v 
Recommended MEPDG Inputs........................................................................... 97 
 
CHAPTER 5:  UNBOUND MATERIAL DATA .................................................... 100 
MEPDG Input Requirements ................................................................................ 100 
Summary of Data and Preliminary Analysis ........................................................ 104 
Analyses of Unbound Material Properties ............................................................ 111 
Stiffness Properties ........................................................................................... 111 
Hydraulic Properties ......................................................................................... 123 
Summary ............................................................................................................... 135 
Testing Recommendations ................................................................................ 135 
Recommended MEDPG Inputs......................................................................... 137 
 
CHAPTER 6:  MATERIAL PROPERTIES DATABASE ....................................... 140 
Introduction ........................................................................................................... 140 
Instructions for Using MatProp ............................................................................ 141 
User Interface for Flexible Pavement Material Management ........................... 143 
User Interface for Rigid Pavement Material Management ............................... 153 
User Interface for Unbound Material ................................................................ 154 
Database Structure ................................................................................................ 157 
 
CHAPTER 7:  SUMMARY OF RECOMMENDATIONS ...................................... 168 
Project Summary ................................................................................................... 168 
Testing Recommendations .................................................................................... 169 
Asphalt Binders ................................................................................................. 169 
HMA Mixtures .................................................................................................. 169 
PCC Mixtures.................................................................................................... 171 
Unbound Materials............................................................................................ 172 
 
References: ................................................................................................................ 173 
 
 
 
 
 
 
 vi 
List of Tables 
Table 1. Number of test records received from SHA. .................................................. 6 
Table 2. Legend for supplier code numbers in Figure 1 to Figure 3. ........................... 7 
Table 3. Differences in predicted distresses using MEPDG Level 3 vs. Level 1 binder 
inputs. .......................................................................................................................... 26 
Table 4. Recommended Level 3 binder grade inputs for wearing courses/surface 
layers (OMT, 2006). ................................................................................................... 28 
Table 5. MEPDG thermal conductivity and heat capacity inputs. (NCHRP, 2004). .. 31 
Table 6. Typical coefficient of thermal expansion ranges for common aggregates 
(NCHRP, 2004)........................................................................................................... 31 
Table 7. Typical Poisson?s ratio values for HMA mixtures (from NCHRP, 2004; 
AASHTO, 2008). ........................................................................................................ 31 
Table 8. Asphalt dynamic modulus determination for rehabilitation design at different 
input levels. (NCHRP, 2004). ..................................................................................... 32 
Table 9. Number of mixtures in database for each mixture size and type. Mixtures in 
bold italics were included in the correlation analyses. ............................................... 34 
Table 10. Correlation analysis result for 9.5mm high polish mixes. .......................... 47 
Table 11. Correlation analysis result for 12.5mm virgin mixes. ................................ 48 
Table 12. Correlation analysis for 19.0mm virgin mixes. .......................................... 48 
Table 13. Correlation analysis for 9.5mm RAP mixes. .............................................. 48 
Table 14. Correlation analysis for 19.0mm RAP mixes. ............................................ 49 
Table 15. Definitions of binder and traffic codes. ...................................................... 49 
Table 16. Recommendation material property inputs for new HMA layers for 
Maryland conditions. .................................................................................................. 60 
Table 17. Recommendation material property inputs for existing HMA layers for 
Maryland conditions. .................................................................................................. 60 
Table 18. Recommendation thermal cracking inputs for new HMA layers for 
Maryland conditions. .................................................................................................. 61 
Table 19. Level 3 inputs for Maryland HMA mixtures (based on material properties 
database at time of report). .......................................................................................... 61 
Table 20. SHA historical unit weights for Superpave mixes at 4% air voids (OMT, 
2006). .......................................................................................................................... 62 
Table 21. PCC elastic modulus estimation for new, reconstruction, and overlay design 
(NCHRP, 2004)........................................................................................................... 65 
 
 vii 
Table 22. PCC modulus of rupture estimation for new or reconstruction design and 
PCC overlay design (NCHRP, 2004).......................................................................... 66 
Table 23. Estimation of PCC thermal conductivity, heat capacity, and surface 
absorptivity at various hierarchical input levels (NCHRP, 2004). ............................. 67 
Table 24. Recommended condition factor values used to adjust moduli of intact slabs 
(from NCHRP, 2004). ................................................................................................. 69 
Table 25. Level 3 guidelines for in-place PCC elastic modulus (from NCHRP, 2004).
..................................................................................................................................... 69 
Table 26. Composition of Missouri DOT PCC mixes (ARA, 2009). ......................... 73 
Table 27. Mixture Properties from Missouri DOT. .................................................... 73 
Table 28. Predicted distresses using MEPDG Level 1, 2 and 3 PCC material property 
inputs based on ?Gradation B? in Missouri DOT (ARA, 2009). ................................ 77 
Table 29. Predicted distresses using MEPDG Level 1, 2 and 3 PCC material property 
inputs based on ?Gradation B Opt? in Missouri DOT (ARA, 2009). ......................... 78 
Table 30. Predicted distresses using MEPDG Level 1, 2 and 3 PCC material property 
inputs based on ?Gradation D? in Missouri DOT (ARA, 2009). ................................ 80 
Table 31. Predicted distresses using MEPDG Level 1, 2 and 3 PCC material property 
inputs based on ?Gradation D Opt? in Missouri DOT (ARA, 2009). ......................... 82 
Table 32. Predicted distresses using MEPDG Level 1, 2 and 3 PCC material property 
inputs based on ?Gradation F? in Missouri DOT (ARA, 2009). ................................ 83 
Table 33. Baseline cases in OAT sensitivity analysis. Bold values are measured. .... 90 
Table 34. Generalized sensitivity indices of Ec and MOR at different ages to 
predictions. .................................................................................................................. 90 
Table 35. Recommended PCC thermal and shrinkage property inputs for Maryland 
conditions (all JPCP construction types). ................................................................... 98 
Table 36. Recommended PCC mix property inputs for Maryland conditions. .......... 98 
Table 37. Recommended strength and stiffness input properties for new PCC for 
Maryland conditions (new/reconstruction/rehabilitation designs). ............................. 99 
Table 38. Recommended strength and stiffness input properties for existing PCC for 
Maryland conditions (rehabilitation designs). ............................................................ 99 
Table 39. Ratio of laboratory MR to field backcalculated EFWD modulus values for 
unbound materials (AASHTO, 2008). ...................................................................... 102 
Table 40. Models relating material index and strength properties to resilient modulus 
(NCHRP, 2004)......................................................................................................... 102 
Table 41. MEPDG Level 3 default resilient moduli values at optimum moisture and 
density (AASHTO, 2008). ........................................................................................ 103 
Table 42. Number of test records received from SHA. ............................................ 105 
 
 viii 
Table 43. Recommended moduli for unbound materials from SHA Pavement Design 
Guide. ........................................................................................................................ 111 
Table 44. Suggested bulk stress ? (psi) values for use in design of granular base 
layers (AASHTO, 1993). .......................................................................................... 112 
Table 45. Stress states for various typical Maryland pavement structures. .............. 113 
Table 46. Summary of stress state ranges from Richter (2002). ............................... 116 
Table 47. Typical states of stress for Arizona flexible pavement sections (Andrei, 
2003). ........................................................................................................................ 116 
Table 48. Consolidated estimates of pavement stress states for Maryland conditions.
................................................................................................................................... 117 
Table 49. SHA resilient modulus data evaluated at representative stress states. Shaded 
entries represent values that are anomalously low or high. ...................................... 118 
Table 50. MEPDG values of a, b, and ks for Eq. (6)................................................. 120 
Table 51. Recommendation material property inputs for unbound materials for 
Maryland conditions. ................................................................................................ 138 
Table 52. Typical Poisson?s ratio values for unbound granular and subgrade materials 
(NCHRP, 2004)......................................................................................................... 139 
Table 53. Typical coefficient of lateral pressure for unbound granular, subgrade, and 
bedrock materials (NCHRP, 2004). .......................................................................... 139 
Table 54. Average properties for Maryland unbound materials (based on material 
property database at time of report). ......................................................................... 139 
 
 
 ix 
List of Figures 
Figure 1. High/low/average/volume plots for PG 64-22 binder acceptance properties: 
(a) binder stiffness G*Orig, (b) phase angle ?Orig, (c) ratio of G*Orig/sin(?Orig) at original 
conditions; (d) binder stiffness G*RTFO, (e) phase angle ?RTFO, (f) ratio of 
G*RTFO/sin(?RTFO) at RTFO aged conditions; (g) binder stiffness G*PAV, (h) phase 
angle ?PAV, (i) ratio of G*PAV/sin(?PAV) for PAV aged conditions, (j) BBR Stiffness 
and (k) BBR m value.  Test temperature is 64?C for original and RTFO aged 
conditions, 25?C for PAV aged condition, and -12?C for BBR. ................................ 13 
Figure 2. High/low/average/volume plots for PG 70-22 binder acceptance properties: 
(a) binder stiffness G*orig, (b) phase angle ?Orig, (c) ratio of G*Orig/sin(?Orig) at original 
conditions; (d) binder stiffness G*RTFO, (e) phase angle ?RTFO, (f) ratio of 
G*RTFO/sin(?RTFO) at RTFO aged conditions; (g) binder stiffness G*PAV, (h) phase 
angle ?PAV, (i) ratio of G*PAV/sin(?PAV) for PAV aged conditions; (j) BBR Stiffness 
and (k) BBR m value.  Test temperature is 70?C for original and RTFO aged 
conditions, 25?C for PAV aged conditions and -12?C for BBR. ................................ 18 
Figure 3. High/low/average/volume plots for PG 76-22 binder acceptance properties: 
(a) binder stiffness G*orig, (b) phase angle ?orig, (c) ratio of G*Orig/sin(?Orig) at original 
conditions; (d) binder stiffness G*RTFO, (e) phase angle ?RTFO, (f) ratio of 
G*RTFO/sin(?RTFO) at RTFO aged conditions; (g) binder stiffness G*PAV, (h) phase 
angle ?PAV, (i) ratio of G*pav/sin(?pav) for PAV aged conditions; (j) BBR Stiffness and 
(k) BBR m value.  Test temperature is 76?C for original and RTFO aged conditions, 
25?C for PAV aged conditions and -12?C for BBR. .................................................. 24 
Figure 4. High/low/average/volume plots for 19mm dense graded mixtures. Data 
include all traffic volume categories. .......................................................................... 40 
Figure 5. High/low/average/volume plots for PG 64-22 19mm dense graded mixture. 
Data include all traffic volume categories. ................................................................. 44 
Figure 6. Distribution of binder grades for mixture data sets in correlation analyses. 46 
Figure 7. Distribution of traffic Levels for mixture data sets in correlation analyses. 47 
Figure 8. Comparison of predicted rutting using Level 1 vs. Level 2 vs. Level 3 HMA 
dynamic modulus data. ............................................................................................... 52 
Figure 9. Average normalized sensitivity indices for thermal conductivity, heat 
capacity, surface shortwave absorptivity, and subgrade modulus across all distresses. 
Legend entries from top to bottom correspond to vertical bars from left to right for 
each location. .............................................................................................................. 55 
Figure 10. Normalized sensitivity indices by distress for SSA. Legend entries from 
top to bottom correspond to vertical bars from left to right for each location. ........... 56 
Figure 11. Normalized sensitivity indices by distress for thermal conductivity. Legend 
entries from top to bottom correspond to vertical bars from left to right for each 
distress......................................................................................................................... 57 
 
 x 
Figure 12. Normalized sensitivity indices by heat capacity. Legend entries from top to 
bottom correspond to vertical bars from left to right for each distress. ...................... 57 
Figure 13. Summary of split cylinder tensile strength data provided by SHA. .......... 70 
Figure 14. Summary of slump data provided by SHA. ............................................... 70 
Figure 15. Summary of water-to-cement ratio data provided by SHA. ...................... 71 
Figure 16. Measured Ec for Missouri PCC mixes. ...................................................... 74 
Figure 17. Measured MOR for Missouri PCC mixes. ................................................. 75 
Figure 18. Normalized Ec data for Missouri PCC mixes with default Level 3 aging 
relationship. ................................................................................................................. 75 
Figure 19. Normalized MOR data for Missouri PCC mixes with default Level 3 aging 
relationship. ................................................................................................................. 76 
Figure 20. High/low/average plots of coefficient of variation by distress type. ......... 85 
Figure 21. Predicted slab cracking value for different input levels and slab thicknesses 
for Gradation B mixture. ............................................................................................. 86 
Figure 22. Predicted slab cracking value for different input levels and slab thicknesses 
for Gradation B Opt mixture. ...................................................................................... 87 
Figure 23. Predicted slab cracking value for different input levels and slab thicknesses 
for Gradation D mixture.............................................................................................. 87 
Figure 24. Predicted slab cracking value for different input levels and slab thicknesses 
for Gradation D Opt mixture....................................................................................... 88 
Figure 25. Predicted slab cracking compared to Level 1. Legend entries from top to 
bottom correspond to vertical bars from left to right for each input level option. ...... 89 
Figure 26. Normalized sensitivity of predicted distresses to Ec and MOR values at 
different ages. .............................................................................................................. 92 
Figure 27. Generalized Sensitivity Index of CTE of different Levels in MEPDG ..... 96 
Figure 28. Averages and ranges of resilient modulus values at 95% compaction and 
optimum moisture content (includes all stress states)............................................... 106 
Figure 29. Averages and ranges of optimum water contents. ................................... 106 
Figure 30. Averages and ranges of saturation levels at optimum moisture. ............. 107 
Figure 31. Averages and ranges of maximum dry unit weights. .............................. 107 
Figure 32. Averages and ranges of k2 and k3 for A-2-4 and A-7-5 soils. .................. 109 
Figure 33. Predicted vs. measured resilient moduli for A-2-4 and A-7-5 soils. ....... 110 
Figure 34. Calculated stress states for granular base and subbase layers (Richter, 
2002). ........................................................................................................................ 114 
Figure 35. Calculated stress states for coarse grained subgrade soils (Richter, 2002).
................................................................................................................................... 115 
 
 xi 
Figure 36. Calculated stress states for fine grained subgrade soils (Richter, 2002). 115 
Figure 37. Predicted service life vs. subgrade resilient modulus; base modulus = 
30,600 psi (Schwartz, 2009). .................................................................................... 122 
Figure 38. Predicted service life vs. granular base modulus; subgrade modulus = 5000 
psi (Schwartz, 2009). ................................................................................................ 122 
Figure 39. Examples of SWCC curves from the MEPDG. ....................................... 124 
Figure 40. Predicted distresses for reference conditions (4 in. HMA, A-4 subgrade, 7 
ft GWT depth, medium traffic). Legend entries from top to bottom correspond to 
vertical bars from left to right for each distress. ....................................................... 127 
Figure 41. Sensitivity of distresses to subgrade type at each location. ..................... 129 
Figure 42. Influence of subgrade type on selected predicted distresses at all four 
climate locations (4 in. HMA thickness, 7 ft. GWT depth, medium traffic). ........... 130 
Figure 43. Average modulus of top 2 feet of subgrade vs. time  (4 in. HMA thickness, 
7 ft. GWT depth, medium traffic). ............................................................................ 131 
Figure 44. Sensitivity to MR for all distresses at all locations. .................................. 133 
Figure 45. Average absolute sensitivity to subgrade MR. ......................................... 134 
Figure 46. Average absolute sensitivity to environmental variables. ....................... 134 
Figure 47. Organization of MatProp database. ......................................................... 140 
Figure 48. Security warning. ..................................................................................... 141 
Figure 49. Security alert. ........................................................................................... 142 
Figure 50. Main menu. .............................................................................................. 142 
Figure 51. ?Show MEDPG HMA Input? screen for level 1 Asphalt Mix properties.
................................................................................................................................... 144 
Figure 52. ?Show MEDPG HMA Input? screen for level 2/3 Asphalt Mix properties.
................................................................................................................................... 145 
Figure 53. ?Show MEDPG HMA Input? screen for level 1/2 Asphalt Binder 
properties................................................................................................................... 146 
Figure 54. ?Show MEDPG HMA Input? screen for level 3 Asphalt Binder properties.
................................................................................................................................... 147 
Figure 55. ?Show MEDPG HMA Input? screen for Asphalt General properties (all 
input levels). .............................................................................................................. 148 
Figure 56. ?Manage Binder Data? ? main screen. .................................................... 149 
Figure 57. Add new binder. ...................................................................................... 149 
Figure 58. Look up binder supplier........................................................................... 149 
Figure 59. Add new supplier. .................................................................................... 150 
Figure 60. Look up terminal. .................................................................................... 150 
 
 xii 
Figure 61. Add new terminal. ................................................................................... 150 
Figure 62. Data integrity checking before deleting terminal. ................................... 150 
Figure 63. Saving without completion. ..................................................................... 151 
Figure 64. ID integrity checking. .............................................................................. 151 
Figure 65. ?Edit Binder Property? screen. ................................................................ 151 
Figure 66. Delete without selection. ......................................................................... 151 
Figure 67. ?Manage HMA Data? ? main screen....................................................... 152 
Figure 68. Add HMA mixture. ................................................................................. 152 
Figure 69. Edit dynamic modulus data. .................................................................... 153 
Figure 70. Attempt to add creep data without providing temperature. ..................... 153 
Figure 71. Add creep compliance data. .................................................................... 153 
Figure 72. ?Manage PCC Data? ? main screen. ....................................................... 154 
Figure 73. New PCC mixture.................................................................................... 154 
Figure 74. ?Manage Unbound Data? ? main screen. ................................................ 155 
Figure 75. Add unbound material. ............................................................................ 155 
Figure 76. Edit unbound material. ............................................................................ 156 
Figure 77. New testing condition. ............................................................................. 156 
Figure 78. Edit testing condition. .............................................................................. 156 
Figure 79. Add MR data. ........................................................................................... 156 
Figure 80. Edit MR data. ........................................................................................... 157 
Figure 81. Tables and relations for binder material data. ......................................... 158 
Figure 82. Tables and relations for HMA mixture data. ........................................... 158 
Figure 83. Tables and relations for PCC mixture data. ............................................ 158 
Figure 84. Tables and relations for unbound material data. ..................................... 159 
  
 
 1 
CHAPTER 1: INTRODUCTION 
The new pavement design methodology developed in NCHRP Project 1-37A, 
refined in NCHRP Project 1-40D, and subsequently adopted by AASHTO (AASHTO, 
2008) is based on mechanistic-empirical principles that are expected to be used in parallel 
with and eventually replace the current empirical pavement design procedures derived 
from the AASHO Road Test conducted in the late 1950?s (HRB, 1962). The new 
mechanistic-empirical pavement design guide (MEPDG) requires greater quantities and 
quality of input data in four major categories: traffic; material characterization; 
environmental factors; and pavement performance (for local calibration/validation). 
Material characterization for the mechanistic-empirical approach, the focus of this thesis, 
is significantly more fundamental and extensive than in the current empirically-based 
AASHTO Design Guide (AASHTO, 1993).  
The implementation plan developed by the University of Maryland (UMD) for the 
Maryland State Highway Administration (SHA) recommended a range of research 
projects to be completed in preparation for the MEPDG (Schwartz, 2007). One of the 
higher priority efforts identified in the plan was to catalog and compile existing material 
properties. This thesis addresses this need. 
A hierarchical input data scheme has been implemented in the MEPDG to permit 
varying levels of sophistication for specifying material properties, ranging from 
laboratory measured values (Level 1) to empirical correlations (Level 2) to default values 
(Level 3). It is expected that most states, including Maryland, will begin implementation 
of the new design procedure using Level 3 default inputs or Level 2 correlations that are 
relevant to their local materials and conditions and will, over time, supplement these with 
 
 2 
typical Level 1 measured data for their most common materials. To accomplish this, 
databases or libraries of typical material property inputs must be developed for the 
following categories: 
Binder properties (e.g., binder dynamic modulus G* and phase angle ? or binder 
viscosities ?) 
Hot mix asphalt (HMA) mechanical properties (e.g., dynamic modulus E* master 
curves?either measured directly or predicted empirically) 
Portland cement concrete (PCC) mechanical properties (e.g., elastic modulus Ec, 
modulus of rupture MOR) 
Unbound mechanical properties (e.g., resilient modulus Mr or k1-k3 values) 
Thermohydraulic properties (e.g., saturated hydraulic conductivity ksat) 
The objective of the thesis is to develop this type of organized database of 
material properties for the most common paving materials used in Maryland.  Note that 
this study provides an essential prerequisite for an eventual full local 
calibration/validation of the MEPDG for Maryland conditions. 
The work plan for accomplishing the research objective was organized into seven 
tasks: 
Task 1: Database Design 
Task 2: Binder Properties 
Task 3: HMA Mechanical and Physical Properties 
Task 4: PCC Mechanical and Physical Properties 
Task 5: Unbound Mechanical and Physical Properties 
Task 6: Thermohydraulic Properties 
 
 3 
Task 7: Workshop and Final Report 
The organization of this thesis closely mirrors the work plan. The principal 
difference is that the findings on thermohydraulic properties from Task 6 have been 
merged with the coverage of the mechanical and physical properties for each material 
type. The organization of the chapters of this thesis is thus: 
1: Introduction 
2: Binder Data 
3: HMA Data 
4: PCC Data 
 5: Unbound Material Data 
6: Material Properties Database 
7: Summary of Recommendations 
8: References 
Each of the specific material Chapters 2 through 5 generally follows the same 
consistent organization: 
MEPDG Input Requirements 
  New Construction/Reconstruction/Overlays 
  Rehabilitation (Existing Layers) 
 Data Available from Maryland SHA 
 Analyses of MEDPG Inputs 
  Level 1 vs. Level 2 vs. Level 3 
  Sensitivity Analyses 
 Summary 
  Testing Recommendations 
  Recommended MEPDG Inputs 
The final Chapter 7 compiles in one location all of the detailed testing 
recommendations from each of the specific material Chapters 2 through 5. 
 
 4 
A comprehensive material property database developed in Microsoft Access 2007 
accompanies this thesis. This database is initially populated with all information receive 
from SHA. It provides complete data management tools for adding future data as well as 
data display screens for MEPDG inputs that mirror the input screens in the MEPDG 
Version 1.100 software. Documentation of this database is provided in Chapter 6. 
In addition to this thesis, results from this study have appeared/will appear in part 
in published articles by Schwartz (2009), Schwartz and Li (2010), and Schwartz et al. 
(2011). Complete citations for these articles can be found in the reference list at the end 
of this thesis. 
 
 
 
 5 
CHAPTER 2: BINDER DATA 
MEPDG Input Requirements 
The binder properties required at each of the input levels in the MEPDG are as 
follows: 
? Level 1:  Shear stiffness G* and phase angle ? at multiple temperatures at a 
frequency of ? = 10 radians/sec (AASHTO T315)  
? Level 2:  Same as Level 1 
? Level 3:  Default A-VTS viscosity temperature susceptibility parameters based on 
Superpave Performance Grade (PG) 
The required binder inputs are the same for new construction, rehabilitation, and 
reconstruction. Note that only Superpave binder properties are considered here. The 
conventional softening point, Brookfield viscosity, kinematic viscosity, and penetration 
properties used in the past have not been included in this study since SHA stopped 
measuring these once it had moved to the Superpave mix design system. 
Binder Data Received and Preliminary Analysis 
A large set of binder properties was provided by SHA for initial population of the 
material properties database. The scope of the provided data is described in Table 1. All 
of the SHA testing data was collected for Superpave PG acceptance purposes. The data 
received for SHA represented test results from early 2002 through mid-September 2008. 
 
 6 
Table 1. Number of test records received from SHA. 
PG Grade Number of Test Records 
58-28 15 
64-22 3685 
64-28 150 
70-22 864 
76-22 1540 
 
Considerable effort was devoted to identify incorrect or inconsistent information 
in the data provided by SHA. Incorrect or inconsistent data were corrected when possible 
and eliminated when not.  
The variability of the acceptance test data was also carefully evaluated. Values of 
binder stiffness (G*) and phase angle (?) at original, RTFO and PAV conditions, BBR 
stiffness and BBR m value were reported in the test data received from SHA. Figure 1, 
Figure 2 and Figure 3 summarize the variability of property values by supplier for the PG 
64-22, PG 70-22, and PG 76-22 performance grades, respectively (the PGs most 
commonly used in Maryland). Definitions of the code numbers used in these figures are 
listed in Table 2. The black lines in these figures indicate the minimum, average, and 
maximum property values (left axis), the gray bars summarize the number of test data in 
each category (right axis).  Note that the thick dashed lines in Chart (c), Chart (f) and 
Chart (i) in each figure indicate the specification limits for G*/sin? at each aging 
condition. There are no specification values for BBR stiffness or BBR m value. From 
these figures it can be seen that nearly all data fall within the specification limits for the 
original and RTFO conditions. Some data are significantly above the maximum limit for 
the PAV aged condition (i.e. Supplier 2 and 6 in Figure 2.1(i)). The reasons and 
consequences of these violations of the acceptance specification conditions are unknown. 
However, since the stiffness properties at PAV condition represent binder performance at 
 
 7 
low temperature, this should not have much practical significance in Maryland where low 
temperature cracking is not a problem. Furthermore, since binder data at the PAV 
condition is not an input in MEPDG, it will not affect the MEPDG predictions.  
Table 2. Legend for supplier code numbers in Figure 1 to Figure 3. 
 
Code Figure 1 Figure 2 Figure 3 
1 Associated Asphalt Chevron Associated Asphalt 
2 Chevron Citgo Chevron 
3 Citgo Marathon Ashland Citgo 
4 ESM ASPHALT, LLC NuStar Asphalt Refining, LLC Conoco Phillips 
5 Koch Valero ESM ASPHALT, LLC 
6 Marathon Ashland  Koch 
7 NuStar Asphalt Refining, LLC  Marathon Ashland 
9 United  NuStar Asphalt Refining, LLC 
10 Valero  SEM Materials 
 
 8 
 (a) 
 
 (b) 
 
 9 
 (c) 
 
 
(d) 
 
 10 
 (e) 
 
 
(f) 
 
 11 
 (g) 
 
 
(h) 
 
 12 
 (i) 
 
 
(j) 
 
 13 
 (k) 
Figure 1. High/low/average/volume plots for PG 64-22 binder acceptance properties: (a) 
binder stiffness G*Orig, (b) phase angle ?Orig, (c) ratio of G*Orig/sin(?Orig) at original 
conditions; (d) binder stiffness G*RTFO, (e) phase angle ?RTFO, (f) ratio of G*RTFO/sin(?RTFO) at 
RTFO aged conditions; (g) binder stiffness G*PAV, (h) phase angle ?PAV, (i) ratio of 
G*PAV/sin(?PAV) for PAV aged conditions, (j) BBR Stiffness and (k) BBR m value.  Test 
temperature is 64?C for original and RTFO aged conditions, 25?C for PAV aged condition, 
and -12?C for BBR. 
 
(a) 
 
 14 
 
 
(b) 
 
(c) 
 
 
 15 
 (d) 
 
(e) 
 
 
 16 
 (f) 
 
(g) 
 
 
 17 
 (h) 
 
(i) 
 
 
 18 
 (j) 
 
(k) 
Figure 2. High/low/average/volume plots for PG 70-22 binder acceptance properties: (a) 
binder stiffness G*orig, (b) phase angle ?Orig, (c) ratio of G*Orig/sin(?Orig) at original 
conditions; (d) binder stiffness G*RTFO, (e) phase angle ?RTFO, (f) ratio of G*RTFO/sin(?RTFO) at 
RTFO aged conditions; (g) binder stiffness G*PAV, (h) phase angle ?PAV, (i) ratio of 
G*PAV/sin(?PAV) for PAV aged conditions; (j) BBR Stiffness and (k) BBR m value.  Test 
temperature is 70?C for original and RTFO aged conditions, 25?C for PAV aged conditions 
and -12?C for BBR. 
 
 
 19 
 (a) 
 
 
(b) 
 
 20 
 (c) 
 
 
(d) 
 
 21 
 (e) 
 
 
(f) 
 
 22 
 (g) 
 
 
(h) 
 
 23 
 (i) 
 
 
(j) 
 
 24 
  (k) 
Figure 3. High/low/average/volume plots for PG 76-22 binder acceptance properties: (a) 
binder stiffness G*orig, (b) phase angle ?orig, (c) ratio of G*Orig/sin(?Orig) at original conditions; 
(d) binder stiffness G*RTFO, (e) phase angle ?RTFO, (f) ratio of G*RTFO/sin(?RTFO) at RTFO 
aged conditions; (g) binder stiffness G*PAV, (h) phase angle ?PAV, (i) ratio of G*pav/sin(?pav) 
for PAV aged conditions; (j) BBR Stiffness and (k) BBR m value.  Test temperature is 76?C 
for original and RTFO aged conditions, 25?C for PAV aged conditions and -12?C for BBR. 
 
As documented later in Chapter 6, binder property tables for the MatProp 
database have been designed to accommodate both the current Superpave acceptance 
testing data provided by SHA and to permit future entry of full Superpave 
characterization data?i.e., DSR at multiple temperatures at RTFO conditions, BBR, etc.  
No conventional binder viscosity data (e.g., Brookfield viscosity, penetration, 
etc.) were provided by SHA. Therefore, no provisions for storing these older superseded 
viscosity characteristics have been included in the database design. 
 
 25 
Sensitivity Analysis of Level 1/2 vs. Level 3 Binder Property Data 
Since acceptance testing is performed at a single temperature, it does not provide 
sufficient information for Level 1 or Level 2 Superpave binder characterization in the 
MEPDG. Therefore, only Level 3 inputs?PG grade?can be provided for the binders 
based on the data received from SHA.  
The major question regarding appropriate input levels for binder property data is: 
?Are there significant differences in predicted performance from the MEPDG using 
Level 1, 2, or 3 binder property data?? A review of the literature found no published 
studies that specifically addressed this question. Therefore, a very limited comparison 
analysis using the MEPDG for typical Maryland conditions was conducted. The analysis 
scenario was a simple pavement section consisting of 6 inches of HMA (19mm dense 
graded, PG 76-22) over 15 inches of granular base (A-1-b) over subgrade (A5, upper 12 
inches compacted). The HMA was based on the control mixture at the FHWA ALF test, 
which was designed using aggregates and unmodified binders similar to those commonly 
used in dense graded mixtures in Maryland. Level 2 binder test data was extracted from 
the FHWA ALF research reports. Level 3 defaults were assumed for all other material 
properties. Traffic was set at 950 trucks per day in the design lane (TTC4 for Principal 
Arterials ? Interstates and Defense Highways) and Baltimore (BWI) weather history 
(interpolated with DC and IAD weather history) was taken as the climate input. 
Reliability was set at the MEPDG default 90% level for all distresses. 
The distresses predicted by the MEPDG using Level 3 vs. Level 1 inputs for this 
scenario are summarized in Table 3 (recall that Level 2 binder inputs are the same as 
Level 1). The MEPDG consistently predicts slightly higher distress magnitudes using 
 
 26 
Level 1 than Level 3 inputs for this scenario, but the differences are very small. Although 
this comparison is extremely limited (i.e., just one binder, albeit of a type commonly used 
in Maryland), a reasonable conclusion is that, based just on the binder influence alone, it 
does not seem worthwhile for SHA to embark on a large-scale Level 1/Level 2 binder 
testing program. However, as will be shown in Chapter 3, this conclusion is superseded 
when considering Level 1 vs. Level 3 HMA properties. Level 1 stiffness data for the 
binder must be entered into the MEPDG if Level 1 dynamic modulus data is entered for 
the mixture. The Level 1 binder data is used by the global aging model in the MEPDG 
along with the Level 1 mixture data.  
Table 3. Differences in predicted distresses using MEPDG Level 3 vs. Level 1 binder inputs. 
Distress Type Distress Magnitude Level 3 Inputs Level 1 Inputs ? (%) Level 3 ?1 
Longitudinal Cracking (ft/mile) 470 501 +6.2 
Alligator Cracking(% wheelpath) 2.31 2.43 +4.9 
Transverse Cracking (ft/mile) 0 0 -- 
Subgrade Rutting (in) 0.2655 0.2663 +0.3 
Base Rutting(in) 0.0998 0.1015 +1.7 
HMA Rutting(in) 0.250 0.265 +5.7 
Total Rutting (in) 0.615 0.633 +2.8 
IRI (in/mile) 120.2 121.0 +0.7 
 
Summary 
Testing Recommendations  
The sensitivity of predicted pavement performance to Level 1 vs. Level 2 vs. 
Level 3 binder inputs appears slight. Therefore, based only on this criterion there would 
be little purpose for SHA collection of Level 1 or 2 binder data. As will be shown in 
Chapter 3, however, predicted pavement performance can be substantially different using 
MEPDG Level 1 vs. Level 2 vs. Level 3 HMA mixture inputs. There is consequently a 
 
 27 
motivation for SHA collection of Level 1 HMA dynamic modulus values. However, 
input of Level 1 HMA properties also requires input of Level 1/2 binder data. 
It is recommended that SHA develop a policy of full binder characterization on 
major projects and that the test results be entered into the material property database so 
that typical Level 1/2 properties can be input into the MEPDG in the future. The testing 
frequency for full binder characterization should match the recommendations for HMA 
dynamic modulus testing as detailed in Chapter 3. 
Recommended MEPDG Inputs  
Only binder acceptance data has been collected by SHA to date. This is 
insufficient for Level 1 or Level 2 inputs in the MEPDG. Consequently, only Level 3 
binder data can be input at this time. Until Level 1 binder data become available, it is 
recommended that the PG grade for Level 3 input be selected according to the binder 
recommendations in the SHA/OMT Pavement Design Guide: 
1. All HMA layers other than wearing course/surface layer: PG 64-22 
2. HMA wearing courses/surface layers other than gap-graded: See Table 4. 
3. Gap-graded HMA wearing courses/surface layers: PG 76-22 
 
 
 28 
Table 4. Recommended Level 3 binder grade inputs for wearing courses/surface layers 
(OMT, 2006). 
(a) Wearing surface for all counties except Garrett 
  
(b) Wearing surface for Garrett county 
  
 
 29 
CHAPTER 3: HMA DATA 
MEPDG Input Requirements 
New Construction/Reconstruction/Overlays 
Dynamic modulus is the principal mechanical property input for HMA in the 
MEPDG. The methods for specifying dynamic modulus at each of the input levels in the 
MEDPG are as follows: 
? Level 1: Laboratory-measured dynamic modulus |E*| at multiple temperatures and 
loading frequencies (AASHTO TP62). In addition, Level 1/2 binder stiffness and phase 
angle data are required for the global aging model. 
? Level 2: Gradation and volumetric data for use in the Witczak |E*| predictive model: 
percent retained above the 3/4? sieve; percent retained above the 3/8? sieve; percent 
retained above the #4 sieve; percent passing the #200 sieve; effective volumetric binder 
content (%); and in-place air voids (%). In addition, Level 1/2 stiffness and phase angle 
data are also required for the binder.  
? Level 3: Gradation and volumetric data for use in the Witczak |E*| predictive model. 
Default binder stiffness properties are based on the Superpave Performance Grade for the 
binder.  
Creep compliance and low temperature tensile strength are additional mechanical 
properties required in the MEPDG for predicting thermal cracking distress. The methods 
for specifying these properties at each of the input levels in the MEPDG are as follows: 
? Level 1: Laboratory-measured creep compliance at three temperatures and various 
loading times and laboratory-measured tensile strength at 14oF (AASHTO T322). 
 
 30 
? Levels 2 and 3: Default creep compliance and low temperature tensile strength 
determined from empirical relations built into the MEPDG; empirical relations are 
functions of mix volumetric and binder viscosity properties. 
HMA thermal properties required by the MEPDG include: 
? Thermal conductivity and heat capacity: see Table 5. 
? Surface shortwave absorptivity (SSA), which quantifies the fraction of available solar 
energy that is absorbed by a given surface. Lighter and more reflective surfaces have 
lower SSA values. The recommended methods for determining SSA at each of the input 
levels are: 
o Level 1: Estimate through laboratory testing. However there is no AASHTO certified 
testing standards for this. 
o Levels 2 and 3: Default values based on surface characteristics: 
- Weathered asphalt (gray)  0.80-0.90 
- Fresh asphalt (black)  0.90-0.98 
? Aggregate coefficient of thermal expansion (also sometimes called coefficient of thermal 
contraction): see Table 6.  
Additional physical mixture properties required for all input levels are Poisson?s ratio and 
total unit weight. Both of these properties have relatively small influence on predicted 
pavement performance. There is no national test protocol for measuring Poisson?s ratio for 
HMA; the default Level 3 values recommended in the MEDPG are given in  
Table 7. HMA total unit weight can be measured in the laboratory according to 
AASHTO T166 or estimated based on previous construction records. 
 
 31 
Table 5. MEPDG thermal conductivity and heat capacity inputs. (NCHRP, 2004). 
  
Table 6. Typical coefficient of thermal expansion ranges for common aggregates 
(NCHRP, 2004). 
 
 
Table 7. Typical Poisson?s ratio values for HMA mixtures (from NCHRP, 2004; AASHTO, 
2008). 
Reference 
Temperature     
(oF) 
Dense 
Graded 
Open 
Graded 
< 0 0.15 0.35 
0 ? 40 0.20 0.35 
40 ? 70 0.25 0.40 
70 ? 100 0.35 0.40 
100 ? 130 0.45 0.45 
> 130 0.48 0.45 
 
 32 
Rehabilitation 
The primary difference between characterizing new and existing HMA layers is 
that the dynamic modulus for existing HMA layer must be adjusted for the damage 
caused to the pavement by traffic loads and environmental effects. Table 8 summarizes 
the stated methods for determining dynamic modulus for existing layers at each of the 
input levels in the MEPDG. However, only Level 3 (specification of damage indirectly 
via pavement condition rating) is implemented in the current Version 1.100 of the 
MEPDG software. 
Table 8. Asphalt dynamic modulus determination for rehabilitation design at different 
input levels. (NCHRP, 2004). 
Material 
Group 
Category 
Type 
Design 
Input 
Level Description 
 
Asphalt 
Materials 
(existing 
layers) 
Rehab 1 
? Use NDT-FWD backcalculation approach. Measure 
deflections, backcalculate (combined) asphalt bound 
layer modulus at points along project.  
? Establish backcalculated Ei at temperature-time 
conditions for which the FWD data was collected 
along project.  
? Obtain field cores to establish mix volumetric 
parameters (air voids, asphalt volume, gradation, and 
asphalt viscosity parameters to determine undamaged 
Master curve).  
? Develop undamaged Master curve with aging for site 
conditions by sigmoidal function: 
loglog( * ) 1 rtE e ???? ??? ?
 
? where 
? tr = Time of loading at the reference temperature  
? ? = Minimum value of E*  
? ?+? = Maximum value of E*  
? ?, ? = Parameters describing the shape of the 
sigmoidal function 
? Estimate damage, dj, by:  
? dj = Ei(NDT)/E*(Pred) 
 
 33 
? In sigmoidal function, ? is minimum value and ? is 
specified range from minimum.  
? Define new range parameter ?? to be:  
? ?? = (1-dj) ? 
? Develop field damaged master curve using ?? rather 
than ?  
2 
? Use field cores to establish mix volumetric 
parameters (air voids, asphalt volume, gradation, and 
asphalt viscosity parameters to define Ai-VTSi 
values).  
? Develop by predictive equation, undamaged master 
curve with aging for site conditions from mix input 
properties determined from analysis of field cores.  
? Conduct indirect Mr laboratory tests, using revised 
protocol developed at University of Maryland for 
NCHRP 1-28A from field cores.  
? Use 2 to 3 temperatures below 70?F 
? Estimate damage, dj, at similar temperature and time 
rate of load conditions:  
? dj = Mri/E*(Pred) 
? In sigmoidal function, ? is minimum value and ? is 
specified range from minimum. Define new range 
parameter ?? to be:  
? ?? = (1-dj) ?  
? Develop field damaged master curve using ?? rather 
than ?  
3 
? Use typical estimates of mix modulus prediction 
equation (mix volumetric, gradation and binder type) 
to develop undamaged master curve with aging for 
site layer.  
? Using results of distress/condition survey, obtain 
estimate for pavement rating (excellent, good, fair, 
poor, very poor)  
? Use a typical tabular correlation relating pavement 
rating to pavement layer damage value, dj.  
? In sigmoidal function, ? is minimum value and ? is 
specified range from minimum. Define new range 
parameter ?? to be:  
? ?? = (1-dj) ? 
? Develop field damaged master curve using ?? rather 
than ?  
 
 
 34 
Other existing HMA layer properties are specified in the MEPDG as follows: 
? Creep compliance and low temperature tensile strength: Not required for existing HMA 
layers. 
? Thermal conductivity and heat capacity: Same as for new construction (Table 5). 
? Surface shortwave absorptivity: Not required for existing HMA layers. 
? Aggregate coefficient of expansion: Not required for existing HMA layers. 
Unit weight and Poisson?s ratio: Same as for new construction ( 
? Table 7). 
HMA Data Summary and Preliminary Analysis 
A large set of asphalt mixture design properties was provided by SHA for initial 
population of the material properties database. The scope of the provided data is 
described in Table 9. The date range for these data is unknown, other than that they were 
received from SHA in Fall 2008. As for the binder data in Chapter 2, considerable effort 
was devoted to identifying incorrect or inconsistent information in the data. Incorrect or 
inconsistent data were corrected when possible and eliminated when not. 
Table 9. Number of mixtures in database for each mixture size and type. Mixtures in bold 
italics were included in the correlation analyses. 
HMAS Mix Type Number 
4.75mm  High Polish 1 
 Virgin 25 
9.5mm Gap Graded 10 
 High Polish 122 
 RAP 126 
 Shingle 5 
 Virgin 68 
12.5mm Gap Graded 40 
 High Polish 56 
 RAP 84 
 Shingle 6 
 Virgin 86 
19.0mm Gap Graded 1 
 
 35 
 High Polish 24 
 RAP 122 
 Shingle 8 
 Virgin 37 
25.00mm RAP 50 
 Virgin 20 
37.5mm RAP 6 
 
The HMA mixture information provided by SHA is limited to volumetric and 
gradation data suitable for Level 3 input to the MEPDG. No measured dynamic modulus, 
creep compliance, low temperature tensile strength, or thermal property values suitable 
for Level 1 inputs were provided. Although the volumetric and gradation data provided 
by SHA are sufficient for Level 2 inputs, the required corresponding Level 2 binder data 
are absent. 
The simplest way to categorize typical Level 3 volumetric and gradation MEDPG 
inputs for Maryland materials is to define them as a function only of mix type (e.g., gap- 
vs. dense-graded) and mix size (e.g., 19 mm nominal maximum aggregate size). To 
explore whether this is possible, trends in volumetric and gradation data for a given mix 
type and mix size as a function of binder grade and/or traffic level were examined, as 
illustrated in Figure 4 and Figure 5 respectively for 19 mm dense-graded mixtures. In 
these figures, the grey bars indicate the number of tests included in the database for each 
subset of data (right axis), the heavy black vertical lines indicate the ranges of the data 
(left axis), and the heavy black short horizontal lines indicate the mean values (left axis). 
Noteworthy observations regarding the data in these figures include the following: 
? The PG 64-22 is the most common binder in the data set (Figure 4). This is not surprising, 
as this is the recommended binder for Maryland environmental conditions under all but 
the heaviest traffic conditions. 
 
 36 
? The ranges of the volumetric properties are largest for the PG 64-22 mixtures (Figure 4). 
This is most likely because these are the most common mixtures, and thus the 
opportunity for encountering especially high or low values is large. 
? The ranges of the volumetric properties for the PG 70-22 and PG 76-22 mixtures (Figure 
4), although not as large as for the PG 64-22 data, are still surprisingly large, especially 
given that the number of mixtures using these binders is comparatively small. Air voids 
Va is the only exception to this trend (Figure 4h). Note that the PG 70-22 and PG 76-22 
binders are generally specified by SHA for its premium mixtures?e.g., SMA surface 
mixtures on heavily trafficked interstate highways. 
? The 0.3-3M ESAL traffic category is the most common design condition (Figure 5). Very 
few mix designs fall into the >30M ESAL very high traffic condition. 
? Overall, the range of the volumetric properties is moderate to large for the four lowest 
traffic categories (Figure 5). There are insufficient mixtures in the highest traffic category 
to portray the property ranges accurately. 
There were no consistent overall trends in the mean values for the volumetric and 
gradation properties either with regard to binder grade or traffic level. This is consistent 
with expectations, as the SHA mix design specifications for these properties are not 
functions of binder grade or traffic level.  
 
 37 
 
(a) Binder content by weight of mixture Pb. 
 
(b) Effective binder content by weight of mixture Pbe. 
 
 38 
 
(c) Voids in mineral aggregates VMA. Minimum VMA is 13. 
 
(d) Dust to effective binder ratio D/Pbe. 
 
 
 39 
 
(e) Percent passing 4.75mm sieve size. 
 
(f) Percent passing 2.36mm sieve size. Specific limits are 23 and 49. 
 
 40 
 
(g) Percent passing 0.075mm sieve size. Specific limits are 2 and 8. 
 
(h) Air voids Va. Target value is 4. 
Figure 4. High/low/average/volume plots for 19mm dense graded mixtures. Data include all 
traffic volume categories. 
 
 41 
 
(a) Binder content by weight of mixture Pb. 
 
(b) Effective binder content by weight of mixture Pbe. 
 
 42 
 
(c) Voids in mineral aggregates VMA. Minimum value is 13. 
 
(d) Dust to effective binder ratio D/Pbe . 
 
 43 
 
(e) Percent passing 4.75mm sieve size. 
 
(f) Percent passing 2.36mm sieve size. Specific limits are 23 and 49. 
 
 44 
 
(g) Percent passing 0.075mm sieve size. The specific limits are 2 and 8. 
 
(h) Air voids Va. Target value is 4. 
Figure 5. High/low/average/volume plots for PG 64-22 19mm dense graded mixture. Data 
include all traffic volume categories. 
The large amount of HMA mixture property data provided by SHA can be used to 
develop Maryland-specific average values for use as Level 3 inputs in the MEPDG. In 
 
 45 
order to develop these average properties, however, the appropriate level of data 
aggregation must be determined.  
Clearly, mixture gradation, and possibly volumetric properties, will be direct 
functions of nominal maximum aggregate size (NMAS, termed ?band? in the SHA data 
set). Gradation and volumetric properties will also be functions of mix type (e.g., dense 
vs. gap graded). However, volumetric properties might also vary significantly with 
respect to other categorizations such as binder grade and/or traffic. Although Figure 4 and 
Figure 5 suggest that there were no consistent overall trends in the mean values for the 
gradation and volumetric properties either with regard to binder grade or traffic level, a 
correlation analysis was conducted to examine more thoroughly whether volumetric 
properties are functions of binder grade or traffic. 
The number of mixtures in the MDSHA database corresponding to each mixture 
type is summarized in Table 9. Since there are many different mixture types, only a 
representative subset was considered for the correlation analyses. These, indicated in bold 
italic font in Table 9, were selected to provide a range of mix size and type subsets 
having large numbers of data points for statistical validity. 
In order for the correlation results to be credible, there must be a reasonable distribution of 
binder grades and traffic levels in each analysis data set. As shown in Figure 6 and Figure 7, 
this was achieved in most of the data sets. The results from the correlation analyses for the 
selected mixture types are summarized in  
Table 10 through Table 14. The binder grades and traffic Levels corresponding to 
the binder and traffic code columns are defined in Table 15. The following observations 
can be drawn from these results: 
? The volumetric properties are insensitive to binder grade. Only four correlation coefficients 
were greater than 0.2. The largest coefficient was 0.47 for the correlation of binder grade and 
 
 46 
traffic for 12.5 mm mixtures. This simply reflects the fact that MDSHA uses stiffer binders 
for both dense and gap graded surface mixes on high volume roadways. 
? The volumetric properties are insensitive to traffic Level. Eight correlation coefficients were 
greater than 0.2 but none exceeded 0.35.  
Based on these findings, it was determined that grouping mixtures by NMAS and 
mix type is sufficient for determining average Level 3 input properties. A built-in query 
was implemented in the MatProp database to determine these average values.  
 
Figure 6. Distribution of binder grades for mixture data sets in correlation analyses. 
 
 
 47 
 
Figure 7. Distribution of traffic Levels for mixture data sets in correlation analyses. 
 
Table 10. Correlation analysis result for 9.5mm high polish mixes. 
  BG Code Traffic Code 
BG Code 1.00  
Traffic Code 0.16 1.00 
Gmm 0.23 0.04 
Gmb -0.07 -0.10 
Gse 0.22 0.00 
Pb -0.12 -0.19 
Pba 0.09 -0.03 
Pbe -0.18 -0.11 
Va 0.11 0.12 
Vma 0.10 0.11 
Vfa -0.14 -0.15 
D/Pbe Ratio -0.15 0.01 
D/B Ratio -0.16 -0.03 
 
 
 
 48 
Table 11. Correlation analysis result for 12.5mm virgin mixes. 
  BG Code Traffic Code 
BG Code 1.00  
Traffic Code 0.47 1.00 
Gmm 0.19 0.15 
Gmb 0.16 -0.11 
Gse 0.16 0.17 
Pb -0.20 0.00 
Pba -0.07 0.01 
Pbe -0.14 -0.01 
Va -0.12 0.15 
Vma -0.12 0.15 
Vfa 0.09 -0.14 
D/Pbe Ratio 0.02 -0.05 
D/B Ratio -0.07 -0.09 
 
Table 12. Correlation analysis for 19.0mm virgin mixes. 
  BG Code Traffic Code 
BG Code 1.00  
Traffic Code -0.05 1.00 
Gmm 0.14 -0.10 
Gmb 0.13 -0.09 
Gse 0.14 -0.19 
Pb 0.00 -0.26 
Pba -0.12 0.03 
Pbe 0.09 -0.29 
Va 0.11 -0.13 
Vma 0.14 -0.32 
Vfa 0.03 -0.18 
D/Pbe Ratio -0.03 0.01 
D/B Ratio -0.23 0.06 
 
Table 13. Correlation analysis for 9.5mm RAP mixes. 
  BG Code Traffic Code 
BG Code 1.00  
Traffic Code 0.24 1.00 
Gmm 0.07 0.01 
Gmb 0.09 0.02 
 
 49 
Gse 0.09 0.00 
Pb 0.08 -0.04 
Pba 0.21 0.15 
Pbe -0.17 -0.19 
Va -0.08 -0.04 
Vma -0.08 -0.24 
Vfa -0.02 -0.20 
D/Pbe Ratio -0.19 -0.06 
D/B Ratio 0.04 0.31 
 
Table 14. Correlation analysis for 19.0mm RAP mixes. 
  BG Code Traffic Code 
BG Code 1.00  
Traffic Code 0.11 1.00 
Gmm 0.15 0.15 
Gmb 0.13 0.13 
Gse 0.16 0.09 
Pb 0.00 -0.30 
Pba 0.00 -0.13 
Pbe 0.00 -0.18 
Va 0.09 0.10 
Vma 0.06 0.00 
Vfa -0.01 -0.01 
D/Pbe Ratio 0.00 0.22 
D/B Ratio -0.01 0.21 
 
Table 15. Definitions of binder and traffic codes. 
Binder Code Binder Grade Traffic (MESALs) 
0 PG 58-22 N/A 
1 PG 58-28 <0.3 
2 PG 64-22 0.3 to <3 
3 PG 64-28 3 to < 10 
4 PG 70-22 10 to < 30 
5 PG 76-22 >30 
 
 
 50 
Sensitivity Analyses for HMA Mixture Inputs 
Level 1 vs. Level 2 vs. Level 3 Dynamic Modulus 
The Maryland SHA has not to date collected any Level 1 property data for any of 
its HMA mixtures. The SHA laboratories contain an Asphalt Mixture Performance Tester 
(AMPT) and a UTM-25 general purpose test system, both of which could be employed 
for measuring Level 1 dynamic modulus, creep compliance, and low temperature tensile 
strength properties. The question is whether there is a compelling reason to perform this 
testing.  
Level 1 testing by SHA for creep compliance, low temperature tensile strength, 
and aggregate coefficient of thermal contraction is not recommended. These properties 
are used only for thermal cracking prediction, which is not a major problem in Maryland 
except perhaps for the western mountains in Garrett, Allegany, and Washington counties. 
The MEPDG generally does not predict any significant thermal cracking in Maryland 
provided an appropriate binder grade is specified. Given this, the Level 3 relationships 
built into the MEPDG code for converting dynamic modulus and other mixture properties 
to creep compliance, low temperature tensile strength, and aggregate coefficient of 
thermal contraction are judged sufficient for Maryland purposes. 
The recommendation for Level 1 dynamic modulus testing is different, however. 
Past studies using earlier versions of the MEPDG code have found significant differences 
in predicted performance using Level 1 vs. Level 3 dynamic modulus data (Azari et al. 
2008) and some inability of Level 2/3 inputs to differentiate between different mixes 
adequately (e.g., Flintsch et al., 2008; Ceylan et al., 2009) A limited comparison analysis 
was conducted to confirm these general findings using the current version of the MEPDG 
 
 51 
software for Maryland conditions. The analysis scenario was a simple pavement section 
consisting of 10 inches of HMA (19mm dense graded, PG 76-22) over 20 inches of 
crushed stone base over subgrade (A-7-5). The HMA was based on the control mixture at 
the FHWA ALF test, which uses aggregates and binders similar to those commonly used 
for dense graded mixtures in Maryland. The Level 1 dynamic modulus test data was 
extracted from the FHWA ALF research reports. Level 3 defaults were assumed for all 
other material properties. Traffic was set at 950 trucks per day in the design lane and 
Baltimore (BWI) weather history (interpolated with DC and IAD) was taken as the 
climate input. Reliability was set at the MEPDG 90% default level for all distresses. 
The distresses predicted by the MEPDG using Level 1 vs. Level 2 vs. Level 3 
dynamic modulus inputs for this scenario are summarized in Figure 8. The predicted 
rutting for the HMA layer is slightly larger for the Level 2 and 3 inputs than for the Level 
1 value. However, the predicted total rutting using Level 2 and 3 inputs is significantly 
larger than when using the Level 1 inputs. Although this comparison is extremely limited 
(i.e., just one mixture, albeit of a type commonly used in Maryland), the findings are 
broadly comparable with those by others. 
 
 52 
 
Figure 8. Comparison of predicted rutting using Level 1 vs. Level 2 vs. Level 3 HMA 
dynamic modulus data. 
Past investigations have consistently found that the Witczak predictive model 
used for Level 3 dynamic modulus inputs is dominated by temperature influences and 
does not do a good job of ranking mixtures in terms of their measured stiffness values at 
a given temperature and loading frequency (e.g., Ceylan et al. 2009). In addition, the 
databases used to develop and calibrate the Witczak and other similar dynamic modulus 
predictive models contain very few gap graded SMA mixtures of the type commonly 
used on high volume roads in Maryland. 
Given all of these issues, it is recommended that SHA begin a program of 
measuring Level 1 dynamic modulus data for its more commonly used mixtures. It is 
envisioned that this could be done as part of the project design and/or quality assurance 
activities. The testing, which could be done in-house in the SHA laboratories or 
outsourced to commercial and/or University testing facilities, should focus on larger 
and/or more important projects employing mixtures having the largest tonnage 
 
 53 
production in Maryland or being placed on the highest traffic volume roadways. If this 
type of testing regimen were adopted as routine for large/important paving projects, SHA 
could amass a large body of Maryland specific Level 1 dynamic modulus data in a 
relatively short period of time. 
Note that this recommendation for Level 1 dynamic modulus testing of the HMA 
mixture implicitly includes Level 1 testing of the binder as well. Although it was 
previously concluded in Chapter 2 that Level 3 vs. Level 1/2 binder property data had 
little effect on predicted performance (when coupled with Level 2/3 predicted dynamic 
modulus), Level 1 stiffness data for the binder must be entered into the MEDPG if Level 
1 dynamic modulus data is entered for the mixture. The Level 1 binder data is used by the 
global aging model in the MEPDG along with the Level 1 mixture data. 
Thermal Properties 
The MEPDG requires input values for the HMA thermal conductivity, heat 
capacity and the surface shortwave absorptivity (SSA). HMA thermal conductivity is the 
capability of HMA material to transmit heat, heat capacity is the capability of a HMA 
material to store heat, and SSA is the capability of HMA surface to absorb solar thermal 
radiation.  These HMA thermal properties are expected to have significant effects on 
pavement performance. These properties are not commonly measured in the laboratory, 
and literature data on typical values are sparse. The MEPDG recommends values in the 
range of about 0.4 to 0.8 BTU/hr-ft-oF for HMA thermal conductivity, 0.2 to 0.4 BTU/lb-
oF for HMA heat capacity, and 0.8 to 1.0 (dimensionless) for SSA. The basis for these 
recommended ranges is not known, but the ranges are reasonably narrow. 
 
 54 
In order to evaluate whether more effort needs to be devoted to better quantify 
these properties, a limited sensitivity study was conducted to evaluate the impact of HMA 
thermal conductivity and heat capacity on pavement performance (Schwartz and Li, 
2010). Typical pavement sections were evaluated for College Park MD climate 
conditions as well as for Seattle WA, Caribou ME, and Phoenix AZ in order to evaluate 
more extreme climate cases. Sensitivity of performance to material inputs was quantified 
using the following normalized index Sji 
j iRji
i jR
D XS XD???? ???
??        (1)
 
which quantifies the variation of distress magnitude Dj about some baseline reference 
condition DjR cause by varying an analysis input Xi around its reference condition XiR. The 
normalized sensitivity index Sji can be interpreted as the percentage change in distress Dj 
caused by a given percentage change in input Xi. 
Figure 9 provides an overall summary of the normalized sensitivity indices as 
averaged (in absolute value terms) across all distresses. As shown, SSA has nearly the 
same normalized influence on overall performance at all four sites as does subgrade 
stiffness?i.e., a very high sensitivity. However, the normalized sensitivity indices for 
HMA thermal conductivity and heat capacity are about an order of magnitude lower than 
those for SSA.  
 
 55 
 
Figure 9. Average normalized sensitivity indices for thermal conductivity, heat capacity, 
surface shortwave absorptivity and subgrade modulus across all distresses. Legend entries 
from top to bottom correspond to vertical bars from left to right for each location. 
Figure 10 through Figure 12 summarized the normalized sensitivity indices by 
distress for SSA, HMA thermal conductivity and HMA heat capacity. On the horizontal 
axis of these figures ?LT Crk? means longitudinal cracking, ?All Crk? means alligator 
cracking, ?SG RD? means subgrade rutting, ?GB RD? means granular base rutting, ?AC 
RD? means asphalt concrete rutting, ?Total RD? means total rutting and IRI means 
international roughness index. SSA (Figure 10) has the largest influence on HMA rutting 
and, by extension, on total rutting. It has moderate influence on longitudinal and 
transverse cracking, alligator fatigue cracking, and granular base rutting, most likely due 
to the differences in temperature and thus stiffness/load spreading ability of the HMA 
layer at different SSA values. 
 
 
 56 
 
Figure 10. Normalized sensitivity indices by distress for SSA. Legend entries from top to 
bottom correspond to vertical bars from left to right for each location. 
 
HMA thermal conductivity (Figure 11) has a negligible influence on all distresses. 
HMA heat capacity (Figure 12) also has a small but somewhat larger influence. 
Interestingly, the largest effect of heat capacity is on subgrade and base rutting, with 
rutting in these materials decreasing as HMA heat capacity increases. The mechanism for 
this is unclear. Increased heat capacity increases the thermal inertia of the HMA layer and 
thus smoothes out some of the temperature fluctuations in the layers, which may play a 
role in reducing the rutting in the unbound layers. 
 
 
 57 
 
Figure 11. Normalized sensitivity indices by distress for thermal conductivity. Legend 
entries from top to bottom correspond to vertical bars from left to right for each distress. 
 
 
Figure 12. Normalized sensitivity indices by heat capacity. Legend entries from top to 
bottom correspond to vertical bars from left to right for each distress. 
The overall conclusion that SSA is the only environment-related HMA material 
input showing a strong impact on predicted performance. HMA thermal conductivity and 
heat capacity all show normalized sensitivity indices (averaged across all distresses in 
 
 58 
absolute value terms) an order of magnitude lower than SSA. SSA is not an easily 
measured parameter, however, and it changes significantly over the pavement life (e.g., 
as the asphalt surface bleaches and lightens with time). Consequently, there is no good 
alternative to using the MEPDG Level 3 default values for this input. 
Summary 
Testing Recommendations 
The principal findings and recommendations relevant to HMA material property 
testing by SHA are as follows: 
1. The HMA mixture information provided by SHA is limited to volumetric and gradation data 
suitable for Level 3 input to the MEPDG. No measured dynamic modulus values suitable for 
Level 1 inputs are available. There is the potential for significant differences in predicted 
performance using Level 1 vs. Level 2/3 dynamic modulus data. In addition, the Witczak 
predictive equation used to generate the Level 2/3 dynamic modulus data is not intended for 
SMA mixtures, a common premium mixture type in Maryland, and often does not 
differentiate among different dense graded mixtures adequately. Therefore, SHA should plan 
to begin measuring Level 1 dynamic modulus data over time for the most commonly used 
mixture types in conjunction with major paving projects. Level 1 dynamic modulus testing of 
HMA mixtures will also require companion Level 1 characterization of the asphalt binders. 
 
It is recommended that SHA develop a policy requiring Level 1 HMA dynamic modulus and 
binder characterization testing for all major projects. Major projects could be defined by SHA 
in terms of a minimum placement tonnage, minimum traffic volume, or some other measure 
of project/mix importance. This testing could be done in-house using either the UTM-25 or 
AMPT test systems in the SHA laboratories; however, some equipment repair and/or 
 
 59 
calibration would be required as both of these systems are currently nonoperational. This 
testing could also be outsourced to local commercial testing facilities (e.g., Advanced Asphalt 
Technologies, LLC) and/or the University of Maryland (HMA dynamic modulus testing 
only). 
2. There is no perceived need for measuring Level 1 creep compliance, low temperature tensile 
strength, and aggregate coefficient of thermal contraction properties. These properties are 
used only for predictions of thermal cracking, which is not a major distress type in Maryland. 
The Level 3 relationships built into the MEPDG code for converting dynamic modulus and 
other mixture inputs to creep compliance, low temperature tensile strength, and aggregate 
coefficient of thermal contraction are judged as sufficient for Maryland purposes. 
3. HMA thermal conductivity and heat capacity generally have a very slight influence on 
pavement performance predicted by the MEPDG. Consequently, the Level 3 default values 
built into the MEPDG software are sufficient and laboratory measurement of these properties 
is not warranted.  
Although SSA has a much more significant influence on predicted performance, 
there at present is no easy method for measuring this parameter, either initially after 
construction or over the pavement life. Therefore, the Level 3 defaults values built into 
the MEPDG software should be used. 
Recommended MEPDG Inputs 
The recommended HMA dynamic modulus and asphalt inputs to the MEPDG for 
Maryland conditions are summarized in Table 16 and Table 17 for new and existing 
HMA layers, respectively. Table 18 summarizes the recommendations for creep 
compliance and low temperature tensile strength inputs for both new and existing HMA 
layers.  
 
 60 
Table 16. Recommendation material property inputs for new HMA layers for Maryland 
conditions. 
Property Input 
Level 
Value Comment 
Asphalt material type All Asphalt 
concrete 
Only option available. 
Layer thickness All Project specific  
Asphalt Mix 
Dynamic Modulus 
Table 
1 Mixture 
specific 
Recommended for future collection. 
Aggregate gradation 
and volumetric 
properties1 
2/3 Mixture 
specific 
See Table 19 for typical values for Maryland 
mixtures. 
Asphalt Binder 
Superpave binder 
dynamic stiffness data 
1/2 Binder specific Recommended for future collection. 
Superpave binder 
grade 
3 Mixture 
specific 
See Table 4 for Maryland SHA binder 
recommendations. 
Asphalt General 
Reference temperature All 70 Does not influence predictions. 
Effective binder 
content 
All Mixture 
specific 
See Table 19 for typical values for Maryland 
mixtures. 
In-Place Air Voids All Project specific See Table 19 for typical values for Maryland 
mixtures. 
Total unit weight All Project specific See Table 20 for typical values for Maryland 
mixtures. (Note: Values in table should be 
adjusted for in-place air voids percentage.) 
Poisson?s ratio All 0.35 MEPDG default 
Thermal conductivity All 0.67 MEPDG default (global calibration value). 
Heat capacity asphalt All 0.23 MEPDG default (global calibration value). 
Short wave absorption All 0.85 MEPDG default (global calibration value). 
1Percent retained above the 3/4? sieve; percent retained above the 3/8? sieve; percent retained above the #4 sieve; 
percent passing the #200 sieve; effective volumetric binder content (%); and in-place air voids (%). 
 
Table 17. Recommendation material property inputs for existing HMA layers for Maryland 
conditions. 
Property Input 
Level 
Value Comment 
Asphalt material type All Asphalt 
concrete 
 
Layer thickness All Project specific  
Asphalt Mix 
Aggregate gradation 
and volumetric 
properties1 
All Mixture 
specific 
See Table 19 for typical values for Maryland 
mixtures. 
Superpave binder 
dynamic stiffness data 
1/2 Binder specific Recommended for future collection. 
 
 61 
Superpave binder 
grade 
3 Mixture 
specific 
See Table 4 for Maryland SHA binder 
recommendations. 
Asphalt General 
Reference temperature All 70 Does not influence predictions. 
Effective binder 
content 
All Mixture 
specific 
See Table 19 for typical values for Maryland 
mixtures. 
In-Place Air Voids All Project specific See Table 19 for typical values for Maryland 
mixtures. 
Total unit weight All Project specific  
Poisson?s ratio All 0.35 MEPDG default 
Thermal conductivity All 0.67 MEPDG default (global calibration value). 
Heat capacity asphalt All 0.23 MEPDG default (global calibration value). 
Pavement condition 
rating 
All Project specific  
1Percent retained above the 3/4? sieve; percent retained above the 3/8? sieve; percent retained above the #4 sieve; 
percent passing the #200 sieve; effective volumetric binder content (%); and in-place air voids (%). 
 
Table 18. Recommendation thermal cracking inputs for new HMA layers for Maryland 
conditions. 
Property Input 
Level 
Value Comment 
Average tensile strength at 14 ?C 3 Mixture specific MEPDG calculated 
value  
Creep compliance 3 Mixture specific MEPDG calculated 
value 
Mixture coefficient of thermal 
contraction 
3 Mixture specific MEPDG calculated 
value 
Aggregate coefficient of thermal 
contraction 
3 Project specific See Table 6 
 
Table 19. Level 3 inputs for Maryland HMA mixtures (based on material properties 
database at time of report).  
NMAS 
(mm) 
Mix 
Type 
% 
Retained 
above     
3/4? sieve 
% 
Retained 
above   
3/8? sieve 
% 
Retained 
above      
#4 sieve 
% 
Passing 
#200 
sieve 
Effective 
Volumetric 
Binder 
Content (%) 
In-
Place 
Air 
Voids 
(%) 
4.75 Virgin 0.0 0.0 6.9 7.7 14.06 6.54 
9.5 Shingle 0.0 4.8 39.2 6.4 11.61 6.47 
9.5 RAP 0.0 4.7 38.1 5.5 11.59 6.47 
9.5 Virgin 0.0 3.7 34.5 6.0 11.88 6.47 
9.5 GAP 0.0 10.7 61.3 9.2 14.85 6.47 
9.5 
High 
Polish 0.0 3.4 36.7 5.5 11.76 6.47 
12.5 High 0.0 14.0 49.1 5.1 11.09 6.47 
 
 62 
Polish 
12.5 RAP 0.0 13.0 50.0 5.4 10.70 6.47 
12.5 Shingle 0.0 14.2 50.7 6.1 10.73 6.47 
12.5 Virgin 0.0 15.6 45.3 5.3 11.14 6.47 
12.5 GAP 0.0 21.7 66.8 8.6 14.31 6.47 
19 GAP 5.0 44.0 74.0 8.1 13.84 6.47 
19 RAP 3.5 26.6 57.4 5.0 9.69 6.47 
19 Shingle 4.1 29.5 58.8 5.6 9.72 6.47 
19 
High 
Polish 2.5 30.4 58.3 5.3 10.20 6.47 
19 Virgin 5.5 33.5 55.5 4.9 10.08 6.47 
25 RAP 11.7 40.6 65.2 4.7 9.10 6.47 
25 Virgin 15.0 47.0 63.5 4.2 9.46 6.47 
37.5 RAP 23.0 52.2 70.8 4.4 8.38 6.47 
4.75 Virgin 0.0 0.0 6.9 7.7 14.06 6.54 
 
 
Table 20. SHA historical unit weights for Superpave mixes at 4% air voids (OMT, 2006). 
 
 
 63 
CHAPTER 4: PCC DATA 
MEPDG Input Requirements 
New Construction/Reconstruction/Overlays 
The key project-specific PCC stiffness and strength properties required for 
new construction/reconstruction/overlay designs in the MEPDG are the elastic 
modulus (Ec) and the modulus of rupture (MOR). The methods for determining these 
properties at each of the input levels in the MEPDG are summarized in Table 21 and 
Table 22. The corresponding required user inputs at each level are: 
? Level 1: Ec and MOR at 7, 14, 28, and 90 days and the estimated ratio of 20 year to 
28 day values. 
? Level 2: Compressive strength (fc?) at 7, 14, 28 and 90 days and the estimated ratio of 
20 year to 28 day values. The corresponding Ec and MOR values are estimated using 
the standard empirical relationships shown in Table 21 and Table 22. 
? Level 3: Either the 28-day MOR or the 28-day fc?. The corresponding 28-day Ec 
modulus is then either estimated by the MEPDG software using the standard 
empirical relationship shown in Table 21 or optionally provided by the user. The 
values of Ec and MOR are determined at other time values using the default aging 
relationships shown in Table 21 and Table 22, respectively. 
Additional PCC properties required at all input levels include: 
? Mix properties: Unit weight; Poisson?s ratio; cement type; cementitious material 
content; water cement ratio; aggregate type; curing method.  
 
 
 64 
? Thermal properties: thermal conductivity; heat capacity; surface shortwave 
absorptivity; coefficient of thermal expansion; PCC zero-stress temperature 
? Shrinkage properties: ultimate shrinkage at 40% relative humidity; reversible 
shrinkage; time to develop 50% of ultimate shrinkage. 
The methods for determining thermal conductivity, heat capacity, and surface 
shortwave absorptivity at each of the input levels in the MEPDG are summarized in 
 
 65 
 
Table 23. Coefficient of thermal expansion can be measured using AASHTO 
TP60 (Level 1), approximated using mixture theory (Level 2), or estimated from 
historical values (Level 3). Default values are provided in the MEPDG software for 
all of these additional PCC properties. These default values may be overridden by the 
user if desired. 
Table 21. PCC elastic modulus estimation for new, reconstruction, and overlay 
design (NCHRP, 2004). 
 
 
 
 66 
Table 22. PCC modulus of rupture estimation for new or reconstruction design and 
PCC overlay design (NCHRP, 2004). 
 
 
 67 
 
Table 23. Estimation of PCC thermal conductivity, heat capacity, and surface 
absorptivity at various hierarchical input levels (NCHRP, 2004).  
 
Rehabilitation 
There are two primary differences between characterizing new concrete layers 
and existing layers: (a) the Ec and MOR values for existing PCC slabs to be overlaid 
need to be adjusted for the damage caused to the pavement by traffic loads and 
environmental effects; (b) gains in Ec and MOR over time are not considered for the 
old existing PCC. The material properties required by the MEPDG at each input level 
for existing PCC in rehabilitation projects are as follows: 
? Level 1:  
o Elastic modulus ETEST is measured from cores taken from the existing 
pavement in accordance with ASTM C 46. Alternatively, ETEST can be 
determined via FWD nondestructive evaluation at mid-slab. ETEST is then 
adjusted for pavement condition to determine the Ec value of the existing 
 
 68 
pavement to be used in design: 
 
c TESTE C E??         (2) 
 
in which C is the pavement condition factor given in Table 24. 
o In-place MOR is measured from prismatic beams cut from the existing 
concrete pavement in accordance with AASHTO T97. 
 
? Level 2: 
o In-place fc? is measured from cores taken from the existing pavement in 
accordance with AASHTO T22. 
o In-place fc? is converted to ETEST internally in the MEPDG software using the 
standard empirical relationship (see Table 21). ETEST is then adjusted for 
pavement condition to determine the design Ec value using Eq. (2) and Table 
24. 
o In-place fc? is converted to MOR internally in the MEPDG software using the 
standard empirical relationship (see Table 22). 
 
? Level 3: 
o The in-place Ec of the existing PCC as a function of pavement condition is 
estimated using the guidelines in Table 25.  
o Either 28-day MOR or fc? is estimated based on past historical records or local 
experience; fc? is converted to MOR internally in the MEPDG software using 
the standard empirical relationship (see Table 22). 
 
 
 69 
Table 24. Recommended condition factor values used to adjust moduli of intact slabs 
(from NCHRP, 2004). 
 
 1Table 2.5.15 in PART 2, Chapter 5 of NCHRP (2004) presents guidelines  
to assess pavement condition. 
 
Table 25. Level 3 guidelines for in-place PCC elastic modulus (from NCHRP, 2004). 
 
      1Table 2.5.15 in PART 2, Chapter 5 of NCHRP (2004) presents 
      guidelines to assess pavement condition. 
PCC Data Summary 
The only PCC properties available from SHA were 201 QC/QA data records 
from the Salisbury bypass project on the Eastern Shore of Maryland. These data are 
for a single Mix No. 7 design consisting of #57 limestone coarse aggregate, sand, 580 
lb/cy cementitious material (377 lb/cy Type I cement plus 203 lb/cy ground iron blast 
furnace slag), and a design water-to-cement ratio of 0.44. Material properties included 
in the QC/QA records were the 28-day split cylinder tensile strength, slump, and 
water-to-cement ratio. Summaries of these properties are provided in Figure 13 
through Figure 15. 
Note that the material property data from the Salisbury Bypass project are 
insufficient for direct input to the MEPDG. However, the MOR can be estimated from 
the split cylinder tensile strength ft using a standard empirical relationship (see, e.g., 
Papagiannakis and Masad, 2008):  
 
 70 
  1.35 tMOR f?        (3) 
Using this relation and the data in Figure 13, the average MOR for the 
Salisbury bypass mix is estimated at 685 psi. 
  
2 2
11
15
14
28
20
23 23
28
17
11
4
2
1
0
5
10
15
20
25
30
380 400 420 440 460 480 500 520 540 560 580 600 620 640 660 
F
r
eq
u
en
c
y
S p li t  C y li n d e r  T e n s il e  S t r e n g t h  (p s i)
M e a n :  5 0 7 .7
S t D e v :  5 7 .0 4
 
Figure 13. Summary of split cylinder tensile strength data provided by SHA. 
 
2
11
27
92
38
11
4
2 1
0 1
3
0
3 2 3
0 0 0 1
0
10
20
30
40
50
60
70
80
90
100
0
.2
5
0
.5
0
.7
5 1
1
.2
5
1
.5
1
.7
5 2
2
.2
5
2
.5
2
.7
5 3
3
.2
5
3
.5
3
.7
5 4
4
.2
5
4
.5
4
.7
5 5
F
r
eq
u
en
c
y
S l u mp  (i n )
M ea n :  1 . 2 1
S t D ev :  0 . 7 1
 
Figure 14. Summary of slump data provided by SHA. 
 
 71 
2
4
0
16 16
60 59
22
14
8
0
10
20
30
40
50
60
70
0 . 3 2  0 . 3 4  0 . 3 6  0 . 3 8  0 . 4  0 . 4 2  0 . 4 4  0 . 4 6  0 . 4 8  0 . 5  
F
r
eq
u
en
c
y
W a t e r - to - C e me n t  Ra t i o
M ea n :  0 .4 3
S t D ev :  0 .0 3  
 
Figure 15. Summary of water-to-cement ratio data provided by SHA. 
Sensitivity Analyses for PCC Inputs 
Strength and Stiffness Properties 
Effect of Input Level 
The major question regarding the appropriate input level for PCC strength and 
stiffness data is: ?Are there significant differences in predicted performance from the 
MEPDG using Level 1, 2 or 3 PCC material inputs?? A review of the literature found 
a few relevant studies, but they either did not definitively address the question or they 
were for climate conditions significantly different from Maryland. Therefore, a very 
limited comparison of MEPDG predictions using Level 1 vs. Level 2 vs. Level 3 PCC 
inputs for typical Maryland conditions was conducted. 
Since no Level 1 PCC material properties were available from SHA, several 
alternatives for obtaining the required data were explored. In the first attempt, 28-day 
PCC compressive strength, 28-day modulus of rupture, and 28-day elastic modulus 
data collected by the University of Maryland as part of the Salisbury bypass project 
 
 72 
were used along with the property vs. time relations incorporated in the MEPDG to 
estimate the missing Level 1 and Level 2 PCC properties. No significant differences 
were found among the Level 1, 2, and 3 predicted performances. However, this was 
expected and merely shows that PCC material inputs portion of the MEPDG software 
is bug-free and internally consistent. 
The second attempt capitalized on a study by Hall and Beam (2005) that 
detailed Level 1, 2 and 3 PCC property data for a specific concrete mixture. Using 
these data, the MEPDG predicted performance again showed little difference between 
Level 1, Level 2, and Level 3 inputs. However, the Hall and Beam paper did not 
describe how the Level 2 and Level 3 data were obtained. It is quite likely that these 
data were generated using the same approach as our initial attempt with the Salisbury 
bypass data.  
The third and best attempt was based on measured Level 1, 2, and 3 PCC 
material input data acquired from Missouri DOT for five PCC mixes (ARA, 2009). 
All five mixes used Type I cement, limestone coarse aggregate, and flyash1. The mix 
composition data are described in Table 26 and the corresponding measured stiffness 
and strength data are summarized in Table 27. Plots of Ec and MOR are shown in 
Figure 16 and Figure 17. The Ec and MOR values when normalized by their 28-day 
values are re-plotted in Figure 18 and Figure 19 along with the MEPDG Level 3 
default strength and stiffness gains. Note that although these are the best PCC 
material property data that could be found, there are still some anomalies: 
                                                 
1 Fly ash affects the rate of strength gain. 
 
 73 
? high 14-day Ec and MOR values (or alternatively a counterintuitive dip in the 28-day 
Ec and MOR values) for the Gradation B mixture; 
? an elevated 7-day MOR for the Gradation B Opt mixture at 7 days; 
? anomalously high 28-day Ec and MOR values and/or anomalously low 90-day values 
for the Gradation D Opt mixture; 
? consistently higher MOR values for Gradation F as compared to the other mixtures at 
all ages but particularly at 90 days;  
? measured stiffness and strength changes with time (Figure 18 and Figure 19) that are 
greater than those predicted by the MEPDG Level 3 default aging relations (with the 
exception of the Gradation D Opt mixture, but this may be because of the 
anomalously high 28-day Ec and MOR values used to normalize the trends). 
 
Table 26. Composition of Missouri DOT PCC mixes (ARA, 2009). 
  
Table 27. Mixture Properties from Missouri DOT. 
MoDOT Mix 
Designation 
Sample Age, 
days 
Compressive 
Strength, psi 
Flexural Strength 
(MOR), psi 
Modulus of 
Elasticity, psi 
Gradation B 
3 3343 477 3775772 
7 4001 550 4172195 
14 4390 654 4318238 
28 4902 626 4290195 
90 5421 674 4757531 
Gradation B Opt 3 3472 564 3729516 
 
 74 
7 3936 634 3972549 
14 4474 652 4164558 
28 4857 718 4266237 
90 5606 788 4632843 
Gradation D 
 
3 3756 587 3835707 
7 4472 595 4291245 
14 4848 640 4271614 
28 5082 655 4452082 
90 5875 725 4974852 
Gradation D Opt 
 
3 3884 540 4049615 
7 4382 583 4239712 
14 4810 637 4347735 
28 5120 744 4958388 
90 5970 699 4785520 
Gradation F 
 
3 3243 566 3348184 
7 3847 654 3767819* 
14 4502 739 4101783 
28 4886 772 4320960 
90 5643 897 4635612 
  *Missing data; interpolated via regression analysis from other Gradation F Ec values. 
3000000
3500000
4000000
4500000
5000000
5500000
0 20 40 60 80 100
E
c
(p
s
i)
A g e  (d a y s )
B
B  O p t
D
D O p t
F
 
Figure 16. Measured Ec for Missouri PCC mixes. 
 
 
 75 
400
500
600
700
800
900
1000
0 20 40 60 80 100
MO
R 
(p
s
i)
A g e (d a y s )
B
B  O p t
D
D O p t
F
 
Figure 17. Measured MOR for Missouri PCC mixes. 
0 . 6
0 . 7
0 . 8
0 . 9
1
1 . 1
1 . 2
0 20 40 60 80 100
E
c
A g e  (d a y s )
B
B  O p t
D
D O p t
F
L e v e l 3
 
Figure 18. Normalized Ec data for Missouri PCC mixes with default Level 3 aging 
relationship. 
 
 
 76 
0 . 6
0 . 7
0 . 8
0 . 9
1 . 0
1 . 1
1 . 2
0 20 40 60 80 100
MO
R
A g e (d a y s )
B
B  O p t
D
D O p t
F
L e v e l 3
 
Figure 19. Normalized MOR data for Missouri PCC mixes with default Level 3 aging 
relationship. 
The data in Table 27 were used in analysis scenarios of 10-inch, 9-inch and 8 
inch thick PCC slabs over 6 inches of granular base. Joint spacing was set at 15 ft 
with a 1.25 in dowel diameter and 12 in dowel bar spacing. The measured PCC 
strength and stiffness properties were taken from the Missouri DOT study, and all 
other thermal and mix properties were taken either from the Maryland Salisbury 
bypass project or set equal to the MEPDG Level 3 defaults.2 Initial two-way AADTT 
was set at 4000 (Principal Arterials ? Interstate and Defense, TTC 1), with all other 
traffic variables set equal to the Level 3 default values in MEPDG. Baltimore (BWI) 
weather history was interpolated with Washington Dulles (IAD) data for the climate 
input to the MEPDG. This interpolation was required because the BWI weather 
history has gaps. The MEPDG default reliability of 90% was used for all predicted 
distresses. 
                                                 
2 CTE values were measured for the Missouri mixes, but these data are judged to be unreliable. 
 
 77 
The distresses predicted by the MEPDG using Level 1, 2 and 3 inputs for 
these scenarios are summarized in Table 28 to Table 32. Each table corresponds to 
one set of mix properties in Table 27 as applied to each of the three slab thicknesses 
(only the 8 inch slab thickness was analyzed for Gradation F because of the small 
levels of cracking predicted). The four alternatives for Level 3 are: Level 3a ? 28-day 
fc? only; Level 3b ? 28-day MOR only; Level 3c ? 28-day fc? and Ec; and Level 3d ? 
28-day MOR and Ec. The Missouri DOT report did not specify the estimated 20-year 
to 28-day property ratio. Therefore the Level 3 default value of 1.2 was used. 
Table 28. Predicted distresses using MEPDG Level 1, 2 and 3 PCC material property 
inputs based on ?Gradation B? in Missouri DOT (ARA, 2009). 
Distress 
Type 
Joint 
opening 
(in) 
LTE 
(%) 
Faulting 
(in) 
Cracked 
Slabs 
(%) 
IRI 
(in/mile) 
Input 
Level Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on  
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on  
Level 
1 
Level 1 0.053 0% 91.2 0% 0.068 0% 77.3 0% 165.6 0% 
Level 2 0.052 -2% 91.2 0% 0.062 -9% 47.8 -38% 137.5 -17% 
Level 
3a 0.052 -2% 91.2 0% 0.064 -6% 27 -65% 121.1 -27% 
Level 
3b 0.053 0% 91.2 0% 0.061 -10% 47.7 -38% 137.4 -17% 
Level 
3c 0.053 0% 91.2 0% 0.068 0% 80.9 5% 168 1% 
Level 
3d 0.052 -2% 91.2 0% 0.067 -2% 41.9 -46% 135 -18% 
CoV 0.010  0.000  0.048  0.392  0.129  
Layer thickness = 8 inch 
Distress Joint opening 
(in) 
LTE 
(%) 
Faulting 
(in) 
Cracked 
Slabs 
(%) 
IRI 
(in/mile) 
 
 78 
Type 
Input 
Level Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on  
Level 
1 
Result 
? 
based 
on  
Level 
1 
Level 1 0.052 0% 89.2 0% 0.069 0% 20.1 0% 118.9 0% 
Level 2 0.051 -2% 89.3 0% 0.066 -4% 5.3 -74% 104.4 -12% 
Level 
3a 0.051 -2% 89.3 0% 0.066 -4% 2 -90% 101.9 -14% 
Level 
3b 0.052 0% 89.2 0% 0.065 -6% 5.1 -75% 104.2 -12% 
Level 
3c 0.052 0% 89.2 0% 0.069 0% 22.8 13% 120.8 2% 
Level 
3d 0.051 -2% 89.3 0% 0.068 -1% 4.7 -77% 105.2 -12% 
CoV 0.011  0.001  0.026  0.899  0.076  
Layer thickness = 9 inch 
Distress 
Type 
Joint opening 
(in) 
LTE 
(%) 
Faulting 
(in) 
Cracked Slabs 
(%) 
IRI 
(in/mile) 
Input 
Level Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on  
Level 
1 
Result 
? 
based 
on 
Level 
1 
Level 1 0.049 0% 87.2 0% 0.063 0% 3.5 0% 104.1 0% 
Level 2 0.05 2% 86.8 -1% 0.066 5% 0.6 -83% 100.3 -4% 
Level 
3a 0.05 2% 86.8 -1% 0.066 5% 0.2 -94% 99.9 -4.% 
Level 
3b 0.051 4% 86.7 -1% 0.065 3% 0.5 -86% 100.4 -4% 
Level 
3c 0.051 4% 86.7 -1% 0.068 8% 4.1 17% 104.4 0% 
Level 
3d 0.05 2% 86.8 -1% 0.067 6% 0.6 -83% 100.8 -3% 
CoV 0.015  0.002  0.026  1.095  0.020  
Layer thickness = 10 inch 
Table 29. Predicted distresses using MEPDG Level 1, 2 and 3 PCC material property 
inputs based on ?Gradation B Opt? in Missouri DOT (ARA, 2009). 
 
 79 
Distress 
Type 
Joint opening 
(in) 
LTE 
(%) 
Faulting 
(in) 
Cracked 
Slabs 
(%) 
IRI 
(in/mile) 
Input 
Level Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Level 1 0.052 0% 91.2 0% 0.067 0% 6.8 0% 105.8 0% 
Level 2 0.053 2% 91.2 0% 0.063 -6% 44.4 554% 135.3 28% 
Level 
3a 0.053 2% 91.2 0% 0.064 -4% 28.4 318% 122.5 16% 
Level 
3b 0.052 0% 91.2 0% 0.067 0% 10.2 50% 107.8 2% 
Level 
3c 0.052 0% 91.2 0% 0.066 -2% 9.5 40% 99.5 -6% 
Level 
3d 0.053 2% 91.2 0% 0.067 0% 43.4 538% 136.6 29% 
CoV 0.010  0.000  0.027  0.730  0.135  
Layer thickness = 8 inch 
Distress 
Type 
Joint opening 
(in) 
LTE 
(%) 
Faulting 
(in) 
Cracked Slabs 
(%) 
IRI 
(in/mile) 
Input 
Level Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on  
Level 
1 
Result 
? 
based 
on  
Level 
1 
Level 1 0.051 0% 89.3 0% 0.068 0% 0.3 0% 100.9 0% 
Level 2 0.051 0% 89.2 0% 0.066 -3% 4.7 1470% 104.1 3% 
Level 
3a 0.051 0% 89.2 0% 0.066 -3% 2.1 600% 102 1% 
Level 
3b 0.051 0% 89.3 0% 0.068 0% 0.5 67% 101.1 0% 
Level 
3c 0.051 0% 89.3 0% 0.068 0% 0.5 67% 100.9 0% 
Level 
3d 0.051 0% 89.2 0% 0.068 0% 5 1570% 105.4 4% 
CoV 0.000  0.001  0.015  0.993  0.019  
Layer thickness = 9 inch 
Distress Joint opening LTE Faulting Cracked IRI 
 
 80 
Type (in) (%) (in) Slabs 
(%) 
(in/mile) 
Input 
Level Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on  
Level 
2* 
Result 
? 
based 
on 
Level 
1 
Level 1 0.05 0% 86.9 0% 0.066 0% 0 -100% 99.6 0% 
Level 2 0.05 0% 86.8 0% 0.066 0% 0.5 0% 100.4 1% 
Level 3a 0.05 0% 86.8 0% 0.066 0% 0.2 -60% 99.9 0% 
Level 3b 0.05 0% 86.9 0% 0.066 0% 0.1 -80% 99.6 0% 
Level 3c 0.05 0% 86.9 0% 0.066 0% 0 -100% 99.5 0% 
Level 3d 0.05 0% 86.8 0% 0.067 2% 0.6 20% 100.8 1% 
CoV 0.000  0.001  0.006  1.107  0.005  
*Zero slab cracking predicted using Level 1 inputs.    Layer thickness = 10 inch 
Table 30. Predicted distresses using MEPDG Level 1, 2 and 3 PCC material property 
inputs based on ?Gradation D? in Missouri DOT (ARA, 2009). 
Distress 
Type 
Joint opening 
(in) 
LTE 
(%) 
Faulting 
(in) 
Cracked 
Slabs 
(%) 
IRI 
(in/mile) 
Input 
Level Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Level 1 0.053 0% 91.2 0% 0.07 0% 55.6 0% 148.1 0% 
Level 2 0.052 -2% 91.2 0% 0.063 -10% 39.9 -28% 131.7 -11% 
Level 3a 0.052 -2% 91.2 0% 0.064 -9% 21.6 -61% 117.2 -21% 
Level 3b 0.053 0% 91.2 0% 0.063 -10% 31.3 -44% 124.8 -16% 
Level 3c 0.053 0% 91.2 0% 0.069 -1% 62.5 12% 153.4 4% 
Level 3d 0.052 -2% 91.2 0% 0.069 -1% 39.4 -29% 134 -10% 
CoV 0.010  0.000  0.050  0.363  0.102  
Layer thickness = 8 inch 
 
 81 
 
Distress 
Type 
Joint opening 
(in) 
LTE 
(%) 
Faulting 
(in) 
Cracked 
Slabs 
(%) 
IRI 
(in/mile) 
Input 
Level Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on  
Level 
1 
Result 
? 
based 
on 
Level 
1 
Level 1 0.051 0% 89.2 0% 0.07 0% 9 0 109.7 0% 
Level 2 0.051 0% 89.3 0% 0.066 -6% 3.9 -57% 103.4 -6% 
Level 
3a 0.051 0% 89.3 0% 0.067 -4% 1.5 -83% 101.6 -7% 
Level 
3b 0.051 0% 89.2 0% 0.066 -6% 2.5 -72% 102.3 -7% 
Level 
3c 0.051 0% 89.2 0% 0.07 0% 11.3 26% 111.4 2% 
Level 
3d 0.051 0% 89.3 0% 0.069 -1% 4.5 -50% 105.4 -4% 
CoV 0.000   0.001   0.028   0.708   0.038   
Layer thickness = 9 inch 
Distress 
Type 
Joint opening 
(in) 
LTE 
(%) 
Faulting 
(in) 
Cracked 
Slabs 
(%) 
IRI 
(in/mile) 
Input 
Level Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on  
Level 
1 
Result 
? 
based 
on 
Level 
1 
Level 1 0.05 0% 86.8 0% 0.068 0% 1.4 0% 102.1 0% 
Level 2 0.05 0% 86.8 0% 0.066 -3% 0.5 -64% 100.2 -2% 
Level 
3a 0.05 0% 86.8 0% 0.066 -3% 0.1 -93% 99.8 -2% 
Level 
3b 0.05 0% 86.8 0% 0.066 -3% 0.2 -86% 100 -2% 
Level 
3c 0.05 0% 86.8 0% 0.068 0% 1.7 21% 102.3 0% 
Level 
3d 0.05 0% 86.8 0% 0.067 -2% 0.6 -57% 100.9 -1% 
CoV 0.000  0.000  0.015  0.871  0.011  
Layer thickness = 10 inch 
 
 82 
Table 31. Predicted distresses using MEPDG Level 1, 2 and 3 PCC material property 
inputs based on ?Gradation D Opt? in Missouri DOT (ARA, 2009). 
Distress 
Type 
Joint opening 
(in) 
LTE 
(%) 
Faulting 
(in) 
Cracked 
Slabs 
(%) 
IRI 
(in/mile) 
Input 
Level Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Level 1 0.052 0% 91.3 0% 0.071 0% 18.5 0% 117.3 0% 
Level 2 0.052 0% 91.2 0% 0.064 -20% 36.7 98% 129.3 10% 
Level 
3a 0.052 0% 91.2 0% 0.065 -8% 21 14% 116.8 0% 
Level 
3b 0.052 0% 91.3 0% 0.068 -4% 6.1 -67% 105.7 -10% 
Level 
3c 0.052 0% 91.3 0% 0.072 1% 14.5 -22% 114.9 -2% 
Level 
3d 0.052 0% 91.2 0% 0.073 3% 69.1 274% 160.7 37% 
CoV 0.000  0.001  0.055  0.819  0.157  
Layer thickness = 8 inch 
Distress 
Type 
Joint opening 
(in) 
LTE 
(%) 
Faulting 
(in) 
Cracked 
Slabs 
(%) 
IRI 
(in/mile) 
Input 
Level Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on  
Level 
1 
Result 
? 
based 
on 
Level 
1 
Level 1 0.051 0% 89.3 0% 0.07 0% 1.5 0% 102.7 0% 
Level 2 0.051 0% 89.3 0% 0.067 -4% 3.4 127% 103.1 0% 
Level 
3a 0.051 0% 89.3 0% 0.067 -4% 1.4 -7% 101.5 -1% 
Level 
3b 0.051 0% 89.3 0% 0.068 -3% 0.3 -80% 101 -2% 
Level 
3c 0.051 0% 89.3 0% 0.071 1% 1.1 -27% 103 0% 
Level 
3d 0.051 0% 89.3 0% 0.072 3% 15.9 960% 116.1 13% 
CoV 0.000  0.000  0.031  1.513  0.055  
Layer thickness = 9 inch 
 
 83 
Distress 
Type 
Joint opening 
(in) 
LTE 
(%) 
Faulting 
(in) 
Cracked Slabs 
(%) 
IRI 
(in/mile) 
Input 
Level Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on  
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on  
Level 1 
Result 
? 
based 
on  
Level 
1 
Level 1 0.05 0% 87.0 0% 0.067 0% 0.2 0% 100 0% 
Level 2 0.05 0% 86.8 0% 0.066 -2% 0.4 100% 100.2 0% 
Level 
3a 0.05 0% 86.8 0% 0.066 -2% 0.1 -50% 99.8 9% 
Level 
3b 0.05 0% 87.0 0% 0.066 -2% 0 
-
100.0% 99.5 -1% 
Level 
3c 0.05 0% 87.0 0% 0.068 2% 0.1 -50% 100.3 0% 
Level 
3d 0.05 0% 86.8 0% 0.069 3% 2.8 1300% 103.4 3% 
CoV 0.000  0.001  0.019  1.810  0.014  
Layer thickness = 10 inch 
Table 32. Predicted distresses using MEPDG Level 1, 2 and 3 PCC material property 
inputs based on ?Gradation F? in Missouri DOT (ARA, 2009). 
Distress 
Type 
Joint opening 
(in) 
LTE 
(%) 
Faulting 
(in) 
Cracked Slabs 
(%) 
IRI 
(in/mile) 
Input 
Level Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on 
Level 
1 
Result 
? 
based 
on  
Level 
1 
Result 
? 
based 
on  
Level 
1 
Level 1 0.052 0% 91.3 0% 0.067 0% 0.6 0% 100.3 0% 
Level 2 0.053 2% 91.2 0% 0.063 -6% 41.2 677% 132.9 32% 
Level 
3a 0.053 2% 91.2 0% 0.064 -4% 27.1 4420% 121.5 21% 
Level 
3b 0.052 0% 91.3 0% 0.069 3% 3.4 467% 104 4% 
Level 
3c 0.052 0% 91.3 0% 0.066 -2% 2 233% 101.2 1% 
Level 
3d 0.053 2% 91.2 0% 0.067 0% 44.4 7300% 137.6 37% 
CoV 0.010  0.001  0.036  0.853  0.139  
Layer thickness = 8 inch 
 
 84 
Differences in predicted joint opening displacement and load transfer 
efficiency (LTE) were less than 5% for all input levels, mixtures, and slab 
thicknesses. Differences in predicted faulting were slightly greater, but the largest 
discrepancies between the Level 2/3 and Level 1 were still less than about 10% in all 
cases.  
In contrast, extremely large differences in the percentage if cracked slabs were 
found between the different input levels. As is evident from the data in the tables, 
Level 2 and the four alternatives for specifying Level 3 inputs produced wildly 
varying predictions of slab cracking, all of which were significantly different from the 
predictions using the Level 1 inputs. The differences in predicted slab cracking as 
compared to the Level 1 predictions ranged up to many thousands of percent. 
Differences in IRI predictions using the various input levels were also 
significant, in large part because predicted slab cracking is one of the major inputs to 
the IRI model. The IRI discrepancies among the input levels increase as layer 
thickness decreases. The largest discrepancy in IRI was 74% for Gradation B in an 8-
inch slab.  
The coefficient of variation (CoV) is a good overall measure for the range of 
predicted performance across input levels. CoV was calculated across Level 1, Level 
2, and the four cases of Level 3 for each distress for each mixture; these are 
summarized in the bottom row of Table 28 to Table 32. The ranges of CoV values for 
each distress across all mixtures are summarized in Figure 20. The CoV values for 
LTE, joint opening, and faulting do not exceed 0.06, which means the standard 
deviations are all within 6% of the average. The CoV values for IRI are higher but 
 
 85 
still less than 0.2. In contrast, the CoV values for slab cracking are extremely large, 
with an average value of about 100% and lower and upper bounds of 0.36 and 1.81, 
respectively.  
 
Figure 20. High/low/average plots of coefficient of variation by distress type. 
 
Figure 20 merely shows that predicted slab cracking varies greatly by input 
level. Figure 21, Figure 22, Figure 23 and Figure 24 examine whether there are any 
trends in predicted slab cracking by input level for each slab thickness for the 
Gradation B, Gradation B Opt, Gradation D, and Gradation D Opt mixtures, 
respectively. Several observations can be drawn from the figures. First, for each 
individual mixture the general trends in the variations of predicted slab cracking with 
input level are qualitatively similar for all layer thicknesses, although the absolute 
magnitude of cracking sensibly increases with decreasing slab thickness. Second, 
there are no consistent trends for the variations of slab cracking over different input 
levels. For some mixtures, Level 1 produces the largest amount of predicted slab 
 
 86 
cracking while for others it produces the smallest. Third, predicted slab cracking 
using Level 3c inputs (28-day Ec and MOR) consistently matches the Level 1 
predictions most closely; there is generally poor agreement between Level 2 or the 
other Level 3 predictions and the reference Level 1 values.  
 
 
Figure 21. Predicted slab cracking value for different input levels and slab thicknesses 
for Gradation B mixture. 
 
 
 87 
 
Figure 22. Predicted slab cracking value for different input levels and slab thicknesses 
for Gradation B Opt mixture. 
 
Figure 23. Predicted slab cracking value for different input levels and slab thicknesses 
for Gradation D mixture. 
 
 
 88 
 
Figure 24. Predicted slab cracking value for different input levels and slab thicknesses 
for Gradation D Opt mixture. 
 
Figure 25 summarizes the ratio of predicted slab cracking at Level 2 and 
different Level 3 options to the reference Level 1 predictions for the 8 inch slab 
thickness. The Level 3c predictions are generally the most consistently similar to the 
reference Level 1 values. The only exception to this is the Gradation F mixture, most 
likely because of its exceptionally high 90-day measured MOR strength value. The 
Level 2, 3a, 3b, and 3d predictions are erratic, often giving substantially larger or 
smaller predictions compared to the Level 1 reference depending upon the specific 
PCC mixture; no rational trends are observed. The reasons underlying these 
observations may be that both Level 1 and Level 3c use Ec and MOR data while Level 
2 and Level 3a use only fc? data, Level 3b is missing Ec information, and Level 3d is 
missing MOR information. These results suggest that the strength and stiffness aging 
 
 89 
relationships built into the MEPDG in combination with 28-day measured values of 
both Ec and MOR may be adequate for design. 
 
 
Figure 25. Predicted slab cracking compared to Level 1. Legend entries from top to 
bottom correspond to vertical bars from left to right for each input level option. 
Sensitivity Analysis 
A one-at-a-time (OAT) sensitivity analysis was also conducted to explore the 
relative importance of the various Level 1 strength and stiffness inputs. Base cases of 
7, 14, 28, 90 and 20year/28day Ec and MOR data were generated using 28-day MOR 
and Ec values for Gradation B and D Opt and the Level 3 PCC aging relations, as 
summarized in Table 33. The Gradation B and D Opt mixes were selected because 
they respectively have among the lowest and highest 28-day Ec and MOR values. 
Bold entries in Table 33 are the measured values for Gradation B or D Opt and the 
non-bold values are calculated values using the default Level 3 PCC aging relations 
built into the MEPDG. The values in Table 33 were then used as Level 1 input values 
for the MEPDG. Each Level 1 input value was then increased or decreased by a given 
 
 90 
percentage and the impact on predicted distress was evaluated. Similar to the 
sensitivity analyses in Chapter 3, the normalized sensitivity index for each output 
distress was calculated using Eq. (1) for each input parameter. Table 34 summarizes 
the computed normalized sensitivity indices of predicted distresses to input Ec and 
MOR values at different ages. It was seen that small variations in the 14-day Ec and 
MOR values have little impact on any of the predicted distresses. The sensitivities of 
the distresses to the stiffness and strength values at other ages are larger but more 
variable. The sensitivity indices for the 28-day, 90-day, and 20-year/28-day stiffness 
and strength values are generally larger than those for 7-day stiffness and strength. 
Overall, faulting is least sensitive to all of the stiffness and strength inputs, slab 
cracking is the most sensitive, and IRI exhibits intermediate sensitivity. 
Table 33. Baseline cases in OAT sensitivity analysis. Bold values are measured. 
Age From Gradation B From Gradation D Opt E
c (psi) MOR (psi) Ec (psi) MOR (psi) 
7day 3955929 577 4572061 686 
14day 4129168 602 4772282 716 
28day 4290231 626 4958430 744 
90day 4534010 662 5240177 786 
Ratio of 20year/28day 1.2 1.2 1.2 1.2 
 
Table 34. Generalized sensitivity indices of Ec and MOR at different ages to predictions. 
From Gradation B Normalized Sensitivity to Ec Normalized Sensitivity to MOR 
Age Faulting 
Percent 
slabs 
cracked 
IRI Faulting 
Percent 
slabs 
cracked 
IRI 
7day -0.19 -0.47 -0.23 0.04 1.04 0.42 
14day 0.04 -0.02 -0.01 0.00 0.05 0.02 
28day 0.41 1.70 0.76 -0.11 -2.76 -1.20 
90day 0.45 0.94 0.47 -0.04 -2.23 -0.90 
Ratio of 20year/28day 0.22 -2.09 -0.80 0.22 -2.11 -0.81 
 
 
 91 
From Gradation D Opt Normalized Sensitivity to Ec Normalized Sensitivity to MOR 
Age Faulting 
Percent 
slabs 
cracked 
IRI Faulting 
Percent 
slabs 
cracked 
IRI 
7day -0.14 -1.67 -0.24 0.00 4.25 0.48 
14day 0.00 -0.06 0.00 0.00 0.18 0.02 
28day 0.31 7.42 0.93 -0.10 -15.64 -1.82 
90day 0.38 4.52 0.62 -0.03 -11.03 -1.23 
Ratio of 20year/28day 0.14 -9.17 -0.97 0.14 -9.13 -0.96 
 
To better illustrate these trends, Figure 26 summarizes the normalized 
sensitivity averaged across the Gradation B and D Opt values in Table 34. This figure 
clearly shows that (a) faulting is least sensitive to Ec and MOR; (b) slab cracking is 
most sensitive; and (c) the stiffness and strength properties at 7 and 14 days have less 
influence on predicted distress than do the values at 28, 90, and 20 years. (Note that 
varying the 28-day stiffness or strength values will also change the corresponding 20-
year values for a fixed 20-year/28-day ratio.)  
 
 
 92 
 
Figure 26. Normalized sensitivity of predicted distresses to Ec and MOR values at 
different ages. 
Summary and Recommendations 
The key findings from the analyses reported in this section are as follows: 
1. Predicted slab cracking for jointed plain concrete pavements (JPCP) is highly 
sensitive to the input level for Ec and MOR. IRI is also sensitive to input level, 
primarily because it is a function a slab cracking. Predicted joint faulting and load 
transfer efficiency are essentially the same at all input levels. 
2. Performance predictions using Level 3 inputs of 28-day Ec and MOR closely agree 
with those using the full Level 1 inputs. Therefore, this Level 3 input combination 
should be suitable for most SHA designs. 
3. For full Level 1 inputs, predicted performance is most sensitive to 28-day, 90-day, 
and 20year/28day Ec and MOR inputs and less sensitive to 7-day and 14-day values. 
Thermal Properties 
Previous reviews of the literature (Schwartz and Ceylan, 2010) have documented that 
rigid pavement performance is very sensitive to surface shortwave absorptivity and the 
coefficient of thermal expansion, moderately sensitive to thermal conductivity, and 
insensitive to heat capacity. As described in 
 
 93 
 
Table 23, there is no accepted method for measuring surface shortwave absorptivity, so 
the Level 3 guidelines should be followed for this input. Thermal conductivity can be 
measured in the laboratory, but as indicated in 
 
 94 
 
Table 23 this property is relatively fixed for PCC and therefore the Level 3 
default value should suffice for most designs.  
The strong influence of the coefficient of thermal expansion (CTE) on JPCP 
performance has been demonstrated in several prior studies (Tanesi et al., 2007; Buch 
et al., 2008; Kampmann, 2008; Oh and Fernando, 2008; Haider et al., 2008, 2009; 
Velasquez et al., 2009). Results from these studies suggest that this influence may be 
inconsistent across different climates. Therefore, a limited sensitivity analysis of 
predicted pavement performance to CTE for typical Maryland conditions was 
conducted. 
The baseline inputs for this sensitivity study are the same as for 8 inch thick 
Gradation D Opt PCC calculation (the results are shown in first subtable in Table 33). 
Climate conditions correspond to the Baltimore metropolitan area. A CTE value of 
5.5x106/oF was taken as a baseline, and this value was then adjusted +0.1x106/oF to 
evaluate the sensitivity of predicted distress to CTE. The sensitivity of pavement 
performance to CTE input is defined in terms of a normalized sensitivity index S: 
D CTES CTE D?? ?         (4) 
in which ?D is the change in distress caused by ?CTE and D and CTE are the 
corresponding values for the baseline conditions. 
The results from the CTE sensitivity study are summarized in Figure 27. 
Faulting and IRI were found to have a high sensitivity to CTE, with sensitivity 
indices averaging 2.9 and 1.6, respectively. Slab cracking was found to be extremely 
sensitive to CTE, reaching normalized sensitivity index values of up to 7.3; this 
 
 95 
means that a 10% increase in CTE will cause a 73% increase in the predicted 
percentage of cracked slabs. These findings confirm the literature findings that CTE 
is a critical input for PCC performance predictions in the MEPDG. Consequently, 
accurate values of CTE will be required for design.  
Because CTE is important but difficult to measure, a literature review was 
conducted in an attempt to find good predictive models for estimating CTE. No 
suitable model was found in this search. The weighted average method incorporated 
in the MEPDG appears to be the best model currently available. 
As an added complication, a recent position paper issued by the FHWA 
(2009) cautions that the current AASHTO TP60 test protocol overestimates CTE by 
about 15%. Based on the limited sensitivity analysis in the present study, a 15% 
overestimate of CTE corresponds to about a 25% increase in predicted IRI, an 
approximately 50% increase in faulting, and an over 100% increase in slab cracking. 
The issues raised in the FHWA position paper have serious implications current CTE 
testing and future modifications of the AASHTP TP60 test protocol. They also have 
implications for the global calibration of the rigid pavement performance models in 
the current version of the MEPDG, as these calibrations are based on the erroneously 
overestimated CTE values. This issue is ongoing should be monitored by SHA. In the 
meantime, it is certainly premature to embark on any testing program for CTE for 
local Maryland mixtures. The Level 3 defaults for CTE in the current MEPDG 
software should be used in the interim until these issues are clarified and resolved. 
 
 96 
 
Figure 27. Generalized Sensitivity Index of CTE of different Levels in MEPDG 
 
Shrinkage Properties 
The MEPDG documentation provides little guidance on measurement of 
project-specific shrinkage properties for PCC mixes. For many of these properties 
(e.g., ultimate shrinkage strain, time to 50% shrinkage), there are no acceptable 
practical test protocols. The best recommendation at present is to use the Level 3 
defaults for these properties built into the MEPDG software. 
 
 
 97 
Summary 
Testing Recommendations 
There is very little data on the physical and mechanical properties Maryland 
PCC mixes to be incorporated into the database at this time. Much of the physical 
data required by the MEPDG (e.g., cement type, cementitious material content and 
water/cement ratio) is routinely available for individual projects and should be 
collected and entered into the database. Continued measurement of split cylinder 
tensile strength should be discontinued, as this is not a primary input to the MEPDG 
(or to the 93 AASHTO Design Guide). Instead, 28-day PCC elastic modulus and 
modulus of rupture should be measured for JPCP paving projects in the future, 
incorporated into the database, and used for Level 3 inputs to the MEPDG. There is 
no documented need to perform additional laboratory testing to determine the full 
Level 1 stiffness and strength inputs for PCC. 
Given the lack of practical accepted test standards, ongoing test protocol 
issues, and other reasons, it is recommended that SHA not embark on any additional 
testing for thermal or shrinkage properties at this time. The current version of the 
MEPDG has been calibrated using the default Level 3 values for these properties, and 
these default values should continue to be used until accepted testing standards are 
available. 
Recommended MEPDG Inputs 
The recommended PCC inputs for the MEPDG are summarized in Table 35 
through Table 38 below. 
 
 98 
Table 35. Recommended PCC thermal and shrinkage property inputs for Maryland 
conditions (all JPCP construction types). 
Property Input 
Level 
Value Comment 
General Properties 
PCC Material 3 JPCP Only option available. 
Layer thickness 1 Project specific  
Unit weight 3 150 pcf MEPDG default 
Poisson?s ratio 3 0.2 MEPDG default 
Thermal Properties 
Coefficient of 
thermal expansion 
3 5.5x10-2/oF MEPDG default (global calibration value). 
Thermal 
conductivity 
3 1.25 BTU/hr-ft-
oF 
MEPDG default (global calibration value). 
Heat capacity 3 0.28 BTU/lb-ft MEPDG default (global calibration value). 
 
Table 36. Recommended PCC mix property inputs for Maryland conditions. 
Property Input 
Level 
Value Comment 
Cement type 1 Type 1 Based on Mix 7 data from Salisbury bypass; 
should be replaced by mix-specific value if 
available 
Cementitious 
material content 
1 580 lb/cy Based on Mix 7 data from Salisbury bypass; 
should be replaced by mix-specific value if 
available 
Water/cement ratio 1 0.44 Based on Mix 7 data from Salisbury bypass; 
should be replaced by mix-specific value if 
available 
Aggregate type 1 Limestone Based on Mix 7 data from Salisbury bypass; 
should be replaced by mix-specific value if 
available 
PCC zero stress 
temperature 
3 Project 
specific 
Default calculated by MEDPG as function of 
site weather conditions and cementitious 
material content 
Ultimate shrinkage 2 Project 
specific 
Default calculated by MEPDG as function of 
cement type, cement content, water/cement 
ratio, 28-day compressive strength, and curing 
conditions 
Reversible 
shrinkage 
3 50% Value used in global calibration of distress 
models 
Time to develop 
50% shrinkage 
3 35 days Value used in global calibration of distress 
models 
Curing method 1 Project 
specific 
 
 
 
 
 99 
Table 37. Recommended strength and stiffness input properties for new PCC for 
Maryland conditions (new/reconstruction/rehabilitation designs). 
Property Input 
Level 
Value Comment 
28-day PCC 
modulus of rupture 
3 685 psi Based on Mix 7 split cylinder tensile strength 
from Salisbury bypass and empirical 
conversion; should be replaced by mix-
specific value if available 
28-day PCC elastic 
modulus 
3 4,371,000 psi Based on Mix 7 split cylinder tensile strength 
from Salisbury bypass and empirical 
conversion; should be replaced by mix-
specific value if available 
 
 
Table 38. Recommended strength and stiffness input properties for existing PCC for 
Maryland conditions (rehabilitation designs). 
Property Input 
Level 
Value Comment 
28-day PCC 
compressive 
strength 
2 Project 
specific 
Obtained from cores of existing PCC slab. 
 
 
 100 
CHAPTER 5:  UNBOUND MATERIAL DATA 
MEPDG Input Requirements 
The principal mechanical property for unbound materials is the resilient 
modulus at the reference condition of optimum moisture and in-place density 
(AASHTO T180). The input requirements for resilient modulus vary by input level: 
? Level 1 
o Laboratory Measurement (New Construction/Reconstruction): The 
regression coefficients k1, k2, and k3 for the stress-dependent resilient 
modulus relationship (AASHTO T307 or NCHRP 1-28A): 
23
1 1
kk
o c tRa
aa
M k p pp??? ? ? ???? ? ? ?? ? ? ?     (5) 
in which: 
 MR  = resilient modulus 
 ?  = bulk stress = ?1 + ?2 + ?3 
?oct  = octahedral shear stress 
= ? ? ? ? ? ?2 2 2
1 2 2 3 1 313 ? ? ? ? ? ?? ? ? ? ?
 
          ?1,?2,?3 = principal stresses 
 pa  = atmospheric pressure (normalizing factor) 
o Field Measurement (Rehabilitation/Overlay Design): FWD backcalculated 
EFWD values (AASHTO T256/ASTM D5858). These field backcalculated 
EFWD value must be converted to an equivalent laboratory MR value using the 
adjustment factors in Table 39. 
 
 101 
? Level 2 - MR determined from correlations with California Bearing Ratio, R-value, 
structural layer coefficient ai, or plasticity index and gradation as summarized in 
Table 40. 
? Level 3 - Default MR at optimum moisture and density as a function of AASHTO soil 
type as summarized in Table 41. 
In addition to stiffness, hydraulic properties for the partially saturated 
unbound materials in the base, subbase, and subgrade layers are required as inputs for 
the Enhanced Integrated Climate Model (EICM) built into the MEPDG.3 The 
principal hydraulic properties for unbound materials are the saturated hydraulic 
conductivity (permeability) and the soil water characteristic curve (SWCC). The input 
requirements for these vary by input level: 
? Level 1 ? Measured saturated hydraulic conductivity (AASHTO T215) and measured 
soil water characteristic curve (ASTM D6836) for determining parameters of the 
Fredlund-Xing (1994) model. 
? Level 2/3 ? Default saturated hydraulic conductivity as a function of gradation and 
plasticity index; default Fredlund-Xing SWCC parameters as f unction of gradation 
(nonplastic/coarse soils) or gradation and plasticity (plastic/fine-grained soils). 
Additional mechanical and physical property data required at all input levels 
in the MEPDG include: 
? Gradation (AASHTO T88) 
? Atterberg limits (AASHTO T89, T90) 
                                                 
3 Note that the details of the relations for the EICM inputs in the current version of the MEPDG have 
changed slightly from the descriptions in the NCHPR 1-37A final report. The updated formulation is 
described in Zapata and Houston (2008) and in Zapata et al. (2009). 
 
 102 
? Specific gravity of solids Gs (AASHTO T100) 
? Maximum dry density and optimum moisture content (AASHTO T180) 
? Poisson?s ratio ? 
? Coefficient of lateral earth pressure K0 
Default values for these properties for each AASHTO soil class are provided 
in the MEPDG software. These default values can be replaced by project-specific 
data if available. 
Table 39. Ratio of laboratory MR to field backcalculated EFWD modulus values for 
unbound materials (AASHTO, 2008). 
Layer Type Location MR/EFWD 
Aggregate Base/Subbase Between a stabilized and HMA layer 1.43 
Below a PCC layer 1.32 
Below an HMA layer 0.62 
Subgrade/Embankment Below a stabilized subgrade/embankment 0.75 
Below an HMA or PCC layer 0.52 
Below an unbound aggregate base 0.35 
 
Table 40. Models relating material index and strength properties to resilient modulus 
(NCHRP, 2004). 
 
 
 103 
 
Table 41. MEPDG Level 3 default resilient moduli values at optimum moisture and 
density (AASHTO, 2008). 
 
 
 104 
 
Summary of Data and Preliminary Analysis 
A total of 85 acceptable sets of unbound properties were provided by SHA for 
initial population of the material properties database. Each set of data contained the 
following unbound material properties: AASHTO class and Group Index; Atterberg 
limits (PI and LL); percents passing the No. 4 and No. 200 sieves; moisture content, 
saturation, and dry density at optimum conditions (AASHTO T180 assumed) and at 
other resilient modulus testing conditions; one or more sets of laboratory-measured 
resilient modulus data.  
Considerable effort was devoted to identify incorrect or inconsistent 
information in the data provided by SHA. Incorrect or inconsistent data were 
corrected when possible and eliminated when not.  
The key mechanical property required for unbound base/subbase/subgrade 
materials is the resilient modulus, MR, at optimum moisture content and in-place 
density. For Level 1 inputs, the stress dependence of MR must also be included as 
determined from the AASHTO T307, LTPP P46, or NCHRP 1-28A test protocols. 
The scope of the provided data is described in Table 42. Many of the soils had 
laboratory measured MR data at multiple moisture contents (typically optimum and 
optimum+2%).There are multiple MR testing records for each moisture condition 
since MR is measured at different stress states. Gradation information was limited to 
the percents passing the No. 4 and No. 200 sieves. Other properties received included 
the PI, LL, maximum dry unit weight, optimum gravity water content, saturation 
 
 105 
degree at optimum condition were also received. No hydraulic properties (e.g., 
hydraulic conductivity) were provided. 
Table 42. Number of test records received from SHA. 
Classification 
Number of 
Property 
Sets 
Number of 
Acceptable MR 
Records 
A-1-b 3 44 
A-2-4 17 575 
A-2-6 6 103 
A-3 1 56 
A-4 33 1331 
A-5 1 42 
A-6 13 463 
A-7-5 4 168 
A-7-6 2 84 
Classification Not 
Mentioned 5 45 
TOTAL 85 2911 
 
The mean and ranges of values for the unbound material supplied by SHA are 
summarized in Figure 28 through Figure 31. These summaries include MR at 95% 
compaction and optimum moisture content (Figure 28), optimum moisture content 
(Figure 29), degree of saturation at optimum moisture (Figure 30), and maximum dry 
unit weight (Figure 31) for each soil type. The grey bars (right axis) indicate the 
number of test records included in the database, the heavy black vertical lines (left 
axis) indicate the ranges of the data, and the heavy black short horizontal lines 
indicate the mean values. Noteworthy observations regarding the data in these figures 
include the following: 
? The A-2-4, A-4 and A-6 are the most common unbound material in the data set. 
There is not much data for the A-1-b, A-2-6, A-3, A-5, A-7-5, and A-7-6 soil classes. 
? The ranges of the MR value are large for all soil types. This is because all stress states 
are included in the chart.  
 
 106 
? The ranges of optimum moisture content, saturation at optimum, and maximum dry 
unit weight are within reasonable limits. 
 
Figure 28. Averages and ranges of resilient modulus values at 95% compaction and 
optimum moisture content (includes all stress states). 
 
 
Figure 29. Averages and ranges of optimum water contents.  
 
 
 107 
 
Figure 30. Averages and ranges of saturation levels at optimum moisture. 
 
 
Figure 31. Averages and ranges of maximum dry unit weights. 
 
The k1, k2 and k3 values for each soil property set were calculated using 
nonlinear regression of the laboratory MR test records. Figure 32 shows the minimum, 
maximum, and average values of k2 and k3 for 95% compaction at optimum moisture 
content for A-2-4 and A-7-5 soils, the coarsest and finest grained soils respectively in 
 
 108 
the database. The heavy black vertical lines indicate the ranges of the data, and the 
heavy black short horizontal lines indicate the mean values. The double arrows in the 
figure show the typical expected range for each parameter for each soil type. As 
defined in Eq. (5), k2 is the confining stress stiffening exponent and k3 is the shear 
stress softening exponent. The k2exponent is typically around 0.5 to 0.8 for coarse-
grained soils and near 0 for fine-grained cohesive soils, while the k3 exponent is 
always negative, slightly negative for coarse-grained soils and more strongly negative 
for fine-grained cohesive soils. It is clear from Figure 32 that the k2 and k3 values for 
the coarse-grained A-2-4 soil lie mostly outside their expected ranges; the k3 values 
for the fine-grained A-7-5 are mostly positive, contrary to physical reasoning. These 
material parameters nevertheless provide good predictions of measured MR values; as 
shown in Figure 33, the predicted vs. measured values fall nearly along the line of 
equality with minimal scatter.  
The explanation for these anomalous findings is unclear. Closer examination 
of the data reveals that many of the property sets have some measured MR records that 
do not follow the expected trends of increasing MR as chamber pressure increases or 
decreasing MR as deviator stress increases. It is not known whether these anomalies 
are due to testing or material issues. 
 
 109 
 
 
 
Figure 32. Averages and ranges of k2 and k3 for A-2-4 and A-7-5 soils. 
 
 
 110 
 
 
 
Figure 33. Predicted vs. measured resilient moduli for A-2-4 and A-7-5 soils. 
 
In addition to the laboratory resilient modulus test data, the SHA Pavement 
Design Guide provides recommended moduli for unbound materials. These are 
summarized in Table 43. Note that these values are intended for use with the 1993 
 
 111 
AASHTO Pavement Design Guide and therefore implicitly represent seasonally 
averaged values after adjustment for drainage. 
Table 43. Recommended moduli for unbound materials from SHA Pavement Design 
Guide. 
Material Modulus (psi) Minimum Typical Maximum 
Base/Subbase Materials 
Graded Aggregate Base 15,000 25,000 45,000 
Gravel 10,000 15,000 30,000 
Soil Contaminated Aggregate Base 3,000 10,000 20,000 
Capping Borrow 10,500 10,500 10,500 
Subgrade Soils 
Silts and Clays (w/ high compressibility)  1,000 ? 2000  
Fine Grained Soils with Silts and Clays (w/ low 
compressibility) 
 2,000 ? 3,000  
Poorly Graded Sands  3,000 ? 4,500  
Gravely Soils, Well Graded Sands, and 
Sand/Gravel Mixtures  
 4,500 ? 10,000  
 
Analyses of Unbound Material Properties 
Stiffness Properties 
Input Level 
Level 1 unbound resilient modulus inputs are not recommended for use in the 
MEPDG at this time for the following reasons: 
1. Input of Level 1 properties for unbound materials automatically switches the 
MEPDG structural analysis model from multilayer elastic theory to finite element 
analysis. The execution time for the flexible pavement finite element calculations in 
the current version of the MEPDG is far too long for practical usage.  
2. Performance predictions using Level 1 unbound material properties have not been 
calibrated in the current version of the MEPDG. 
 
 112 
Both of these issues will likely change in future versions of the MEPDG. 
However, to date there have been no published studies using Level 1 unbound 
material inputs to provide any guidance on the sensitivity of predicted performance to 
resilient modulus input level. 
Unlike many agencies, the Maryland SHA is well-equipped to perform Level 
1 laboratory characterization of unbound materials, and the SHA testing effort to date 
has practical value despite the recommendations against Level 1 unbound resilient 
modulus inputs in the current version of the MEPDG. Laboratory measurements can 
be evaluated for expected in-service stress states to develop improved estimates of MR 
values for Level 2/3 input. However, this requires estimates of typical in situ stress 
states for granular base layers and subgrades.   
The 1993 AASHTO Design Guide provides typical values for the bulk stress 
? for granular base layers. These are summarized in Table 44. Since the 1993 
AASHTO Guide does not consider shear stress effects for granular base layers, no 
typical values for the octahedral shear stress ?oct are provided. For typical Maryland 
pavement conditions, the data in Table 44 suggest a bulk stress ?  in the range of 5 to 
10 psi for granular base layers. 
Table 44. Suggested bulk stress ? (psi) values for use in design of granular base layers 
(AASHTO, 1993). 
Asphalt Concrete 
Thickness (inches*) 
Subgrade Resilient Modulus (psi*) 
3,000 7,500 15,000 
< 2 20 25 30 
2 ? 4 10 15 20 
4 ? 6 5 10 15 
> 6 5 5 5 
*1 inch = 25.4 mm; 1 psi = 6.9 kPa 
 
 
 113 
Von Quintus and Killingsworth (1997a, 1997b) provide examples for 
estimating the stress state in granular base and subgrade for an individual pavement 
structure. The representative conditions are taken at one-quarter depth into the 
granular base layer and 18 inches below the top of subgrade for a 9 kip wheel load. In 
situ stresses at these locations based on the material unit weights and the coefficients 
of lateral stress are combined with the load-induced stresses computed using 
multilayer elastic theory and reasonable preliminary estimates of the layer moduli. 
Seasonal effects due to moisture variations can also be included in the calculations. 
Table 45 summarizes for typical Maryland conditions the stress states in the 
granular base and subgrade layers computed using Von Quintus and Killingsworth?s 
approach. The HMA stiffness was assumed as 250,000 psi in all cases and the base 
layer stiffness was estimated as 25,000 psi. The load consisted of a 9000 lb tire 
having a 120 psi pressure. Full-slip conditions were assumed at the layer interface. 
All stress states also include the influence of the in situ stresses. The average 
computed stress states over all HMA and granular base thickness and subgrade MR 
conditions were ? ? 40 psi, ?oct ? 3.5 psi for the granular base layer and ? ? 10 psi, ?oct 
? 2 psi for the subgrade. 
Table 45. Stress states for various typical Maryland pavement structures. 
z  ( i n ) ? v  ( p s i ) ? h  ( p s i ) ?  ( p s i ) ? o ct  ( p s i ) z  ( i n ) ? v  ( p s i ) ? h  ( p s i ) ?  ( p s i ) ? o ct  ( p s i )
4 8 2000 6 3 1 . 0 0 5 2 5 . 4 1 8 8 2 . 8 4 2 . 7 5 30 3 . 8 5 8 0 . 6 0 5 1 0 . 0 7 2 . 1 2
5000 6 3 4 . 0 3 1 2 0 . 6 0 9 7 6 . 2 5 6 . 4 5 30 5 . 3 0 8 0 . 5 8 2 1 1 . 4 7 2 . 8 2
12 2000 7 3 3 . 9 1 9 1 9 . 5 9 8 7 4 . 2 8 6 . 8 9 34 2 . 9 3 2 0 . 5 7 1 9 . 7 4 1 . 7 8
5000 7 3 5 . 2 6 4 1 5 . 9 3 1 6 8 . 2 9 9 . 2 5 34 4 . 2 0 5 0 . 6 0 6 1 1 . 0 8 2 . 3 6
10 8 2000 12 5 . 0 0 7 4 . 7 9 6 1 6 . 6 0 0 . 3 4 36 1 . 4 2 7 0 . 4 1 5 8 . 2 6 1 . 1 8
5000 12 6 . 4 3 4 . 2 8 1 1 6 . 9 9 1 . 2 5 36 2 . 2 1 5 0 . 5 1 7 9 . 2 5 1 . 5 1
12 2000 13 6 . 3 6 4 5 . 1 4 2 1 8 . 8 1 0 . 8 3 40 1 . 3 2 8 0 . 4 0 3 8 . 8 0 1 . 2 2
5000 13 7 . 4 1 6 4 . 4 0 8 1 8 . 4 0 1 . 6 7 40 2 . 0 4 8 0 . 5 0 8 9 . 7 3 1 . 5 1
16 2000 14 7 . 1 6 3 4 . 9 6 9 1 9 . 4 3 1 . 3 1 44 1 . 1 9 6 0 . 3 8 3 9 . 3 0 1 . 2 5
5000 14 7 . 9 1 4 4 . 1 5 2 1 8 . 5 5 2 . 0 5 44 1 . 8 4 3 0 . 4 9 1 0 . 1 6 1 . 5 0
M I N I M U M 1 6 . 6 0 0 . 3 4 8 . 2 6 1 . 1 8
M A X I M U M 8 2 . 8 4 9 . 2 5 1 1 . 4 7 2 . 8 2
A V E R A G E 4 1 . 0 5 3 . 2 8 9 . 7 9 1 . 7 3
B a s e  -  Q u a r t e r  D e p t h S u b g r a d e  -  1 8  i n  D e p t h
D 1  ( i n ) D 2  ( i n )
S u b g r a d e    
M R  ( p s i )
 
 
 114 
 
Richter (Richter, 2002; Richter and Schwartz, 2002) used multilayer elastic 
theory to estimate stress states at various locations within granular base/subbase 
layers and subgrades for field sections in the LTPP Seasonal Monitoring Program. 
These calculations were part of an effort to evaluate stress dependency of 
backcalculated layer moduli. Her calculated stresses are summarized in Figure 34 for 
granular base/subbase layers and in Figure 35 and Figure 36 for coarse and fine 
grained subgrades, respectively. Ranges of stress states encompassing most of the 
data points in these figures are summarized in Table 46 after conversion to U.S. 
Customary units. 
 
 
Figure 34. Calculated stress states for granular base and subbase layers (Richter, 2002). 
 
 
 115 
 
Figure 35. Calculated stress states for coarse grained subgrade soils (Richter, 2002). 
 
 
Figure 36. Calculated stress states for fine grained subgrade soils (Richter, 2002). 
 
 
 
 116 
Table 46. Summary of stress state ranges from Richter (2002). 
Layer/Soil Type ? (psi) ?oct (psi) 
Granular base/subbase 0 - 30 0 ? 15 
Coarse subgrades 7 ? 20 0 - 2 
Fine subgrades 7 ? 20 0 ? 1.5 
 
Andrei (2003) estimated typical stress states for 30 LTPP test sections in 
Arizona. He found that the typical stress states for granular base layers and subgrades 
varied significantly with stiffness of the asphalt layer and thus with season. His values 
for typical stress states summarized in Table 47 are based on the assumption of an 
asphalt stiffness of 50 ksi during the hot summer months (Arizona conditions) and 
1000 ksi during the cold winter. The values in Table 47 corresponding to hot 
conditions should be revised downward slightly for the more temperate Maryland 
summer climate. Adjusting for typical Maryland pavement conditions, the data in 
Table 47 suggest average stress states of approximately ? = 35 psi, ?oct = 10 psi for 
granular base layers and ? = 10 psi, ?oct = 3 psi for subgrades. 
Table 47. Typical states of stress for Arizona flexible pavement sections (Andrei, 2003). 
AC Layer 
Temperature 
Condition 
EAC 
(ksi) 
Granular Base Subgrade 
?  (psi) ?oct (psi) ?  (psi) ?oct (psi) 
Hot 50 44 16 13 5 
Cold 1000 13 7 5 1 
 
Von Quintus et al. (2004) compared backcalculated vs. laboratory measured 
MR values in the LTPP database. In order to make this comparison, they estimated 
typical stress states beneath the FWD to use in determining the correct laboratory 
modulus value. Their estimated stress states for subgrade soils were ?v = 4 psi, ?h = 4 
psi,  ? = 12 psi, and ?oct = 0 psi; the corresponding estimated stresses for granular base 
 
 117 
layers were ?v = 15 psi, ?h = 10 psi,  ? = 35 psi, and ?oct = 2.4 psi. These values are in 
general agreement with those suggested by Andrei (2003). However, no backup 
calculations or other justifications for these values are provided. 
Table 48 consolidates the typical stress states estimated for Maryland 
conditions based on all of the studies described above. In the absence of a detailed 
analysis of a specific pavement structure, the values listed as ?best estimates? in the 
last row of the table can be used to determine an appropriate laboratory MR value. In 
evaluating these ?best estimates,? it is important to remember that most granular base 
materials should be relatively insensitive to ?oct and most Maryland subgrade soils 
(other than on the Eastern Shore) should be relatively insensitive to ?. 
Table 48. Consolidated estimates of pavement stress states for Maryland conditions. 
Source Granular Base Subgrade 
?  (psi) ?oct (psi) ?  (psi) ?oct (psi) 
1993 AASHTO 5 ? 10     
Von Quintus and Killingsworth 
(1997a, 1997b) ? Table 45 
40 3.5 10 2 
Richter (2002) 0 ? 30  0 ? 15  7 ? 20  0 ? 1.5  
Andrei (2003) 35 10 10 3 
Von Quintus et al. (2004) 35 2.4 12 0 
Best Estimate 30 5 12 2 
 
The k1, k2, and k3 resilient modulus parameters for Eq. (5) were evaluated for 
the laboratory MR measurements provided by SHA and then Eq. (5) was evaluated for 
the ?best estimate? stress states in Table 48. The results from this analysis are 
summarized in Table 49 by AASHTO soil class and moisture condition. In cases 
where multiple sets of test records are available, both the mean and the range of 
values are reported. Some soils are typical only of base conditions (e.g., A-1-b), some 
could be either base/subbase or subgrade soils (e.g., A-2-4), and some are 
 
 118 
encountered only in subgrades (e.g., A-7-5). The appropriate MR values at the 
appropriate stress state are given for each case. The shaded entries in the table 
indicate values that appear to be excessively high or low for the given soil class and 
moisture condition. 
Note that the measured MR values in Table 48 for subbase materials (A-2-4, 
A-2-6) are slightly lower on average but within the range of values in Table 43 that 
SHA currently uses in its 1993 AASHTO designs (e.g., soil contaminated granular 
base). Conversely, the measured MR values in Table 48 for fine-grained subgrade 
materials (A-4, A-6) are higher than those in Table 43. However, the values in Table 
43 are intended for use in the 1993 AASHTO Pavement Design Guide and implicitly 
include seasonal effects and drainage influences. 
Table 49. SHA resilient modulus data evaluated at representative stress states. Shaded 
entries represent values that are anomalously low or high. 
Class Condition N 
Average MR 
(Range of MR) 
(psi) 
Granular Base Subgrade 
A-1-b Optimum 1 9851  
Saturated 1 7526  
A-2-4 Optimum -2% 6 9842 
(4894-13308) 
9023 
(5207-12382) 
Optimum 15 9370 
(3870-18146) 
8919 
(4637-15871) 
Optimum+2% 1 7519 8717 
Saturated 3 7922 
(4064-12188) 
7582 
(5088-10687) 
A-2-6 Optimum -2% 1 2737 2438 
Optimum 2 11929 
(8209-15650) 
10022 
(8232-11812) 
A-3 Optimum 1 6670 7410 
A-4 Optimum -2% 13 
 
8258 
(2940-15798) 
Optimum 12 
 
5923 
(2966-8462) 
Optimum+2% 1  15798 
Saturated 12  3964 
 
 119 
(2580-6457) 
A-6 Optimum -2% 3 
 
6688 
(3937-8464) 
Optimum 3 
 
5556 
(3114-8668) 
Saturated 3 
 
3050 
(2134-3653) 
A-7-5 Optimum -2% 2 
 
8180 
(6946-9415) 
Optimum 4 
 
8438 
(3477-13893) 
Saturated 1  5091 
A-7-6 Optimum -2% 1  11555 
Optimum 2 
 
7498 
(7092-7904) 
Saturated 1  5361 
 
Once the laboratory-measured MR value at the appropriate stress state has 
been determined, it can be entered directly as a Level 2 or Level 3 input into the 
MEPDG. There are two options at Levels 2 and 3: ?ICM Inputs,? for which MR is 
entered at optimum moisture content and the EICM adjusts for seasonal moisture 
fluctuations, and ?Representative value (design value),? which bypasses the EICM 
and instead uses an externally-determined seasonally adjusted MR (similar in concept 
to 1993 AASHTO approach). The equation used by the EICM for moisture effects is 
given as (Andrei, 2003): 
 
? ?? ?10 ,l o g 1 s o p tR k S SR o p tM baaM e ? ????? ?
     (6) 
in which: 
 MR = resilient modulus at field saturation S 
 MR,opt = resilient modulus at maximum dry density and optimum 
moisture 
 a = minimum value of the log of the modulus ratio 
 
 120 
 b = maximum value of the log of the modulus ratio 
 ? = location parameter = ln(-b/a) 
 ks = regression parameter 
 (S-Sopt) = deviation of field saturation from optimum (decimal) 
Values for the coefficients in Eq. (6) as implemented in the MEPDG are given 
in Table 50. Equation (6) can also be used for external estimation of a seasonally 
adjusted MR. 
Table 50. MEPDG values of a, b, and ks for Eq. (6). 
Parameter Coarse-Grained 
Materials 
Fine-Grained 
Materials 
a -0.3123 -0.5934 
b 0.3 0.4 
ks 6.8157 6.1324 
 
Yau and Von Quintus (2001) performed an extensive analysis of the LTPP 
database to determine regressions between k1, k2, and k3 in Eq. (5) as a function of 
gradation, Atterberg limits, and other physical properties. However, these regressions 
are more appropriately used when no laboratory resilient modulus test data are 
available. 
Sensitivity Analysis 
The strong impact of base and subgrade stiffness on pavement performance is 
well-known from practical experience, from the 1993 AASHTO Design Guide, and 
from numerous MEPDG sensitivity studies reported in the literature (e.g., Masad and 
Little, 2004; El-Basyouny and Witczak 2005a, 2005b, 2005c; Carvalho and Schwartz, 
2006; Hoerner et al., 2007; Schwartz, 2009; Ayyala et al., 2010). For example, 
Schwartz (2009) examined predicted service life as a function of granular base and 
 
 121 
subgrade properties for typical flexible pavement sections using both the 1993 
AASHTO and MEPDG procedures. Figure 37 and Figure 38 summarize the 
sensitivity of predicted service life to base and subgrade stiffness, respectively, for a 
granular base layer thickness D2 = 12 inches and three HMA layer thicknesses D1 = 3, 
6, and 9 inches. Reliability was set at 50% for both the AASHTO and MEPDG 
procedures. The strong impact of base and subgrade stiffness on predicted service life 
is clearly evident in all cases. There also is good agreement in the trends of the two 
design methods for the thin (3 inch) and medium (6 inch) asphalt layer cases. 
However, for the thick (9 inch) asphalt case, the sensitivity of service life to base 
modulus for the AASHTO design procedure is much greater than that for the 
MEPDG, as indicated by the steeper slope in the curve in Figure 38. Moreover, there 
is a crossing point for the thick asphalt case; the AASHTO procedure predicts much 
longer service life for pavements with high quality thick bases than does the MEPDG, 
but the reverse is true for low quality bases in this scenario. These trends and other 
similar results from the literature emphasize that good estimates of the resilient 
modulus of the unbound layers are important for accurate pavement performance 
prediction. 
 
 122 
0
10
20
30
40
50
0 5000 10000 15000
Su b g r a d e  M r  (p s i )
Se
r
v
ic
e
 L
ife
 N
18
 (x
1
M
)
A A S H T O  D 1 = 9
M E P D G  D 1 = 9
A A S H T O  D 1 = 6
M E P D G  D 1 = 6
A A S H T O  D 1 = 3
M E P D G  D 1 = 3
 
Figure 37. Predicted service life vs. subgrade resilient modulus; base modulus = 30,600 
psi (Schwartz, 2009).  
 
0
10
20
30
40
50
20000 30000 40000 50000
B a s e  M o d u l u s  (p s i )
Se
r
v
ic
e
 L
ife
 N
18
 (x
1
M
)
A A S H T O  D 1 = 9
M E P D G  D 1 = 9
A A S H T O  D 1 = 6
M E P D G  D 1 = 6
A A S H T O  D 1 = 3
M E P D G  D 1 = 3
 
Figure 38. Predicted service life vs. granular base modulus; subgrade modulus = 5000 
psi (Schwartz, 2009). 
 
 
 123 
Hydraulic Properties 
Input Level 
The environmental model in the MEPDG is a one-dimensional formulation for 
vertical heat and partially-saturated moisture flows in the pavement system. Some of 
the assumptions in the MEPDG analyses include: zero rainfall infiltration through the 
pavement surface; no lateral flow to edge drains; liquid flow only?i.e., no vapor 
flow; uncoupled heat and fluid flow; and unbound material thermal conductivity and 
heat capacity values set internally to typical default values. 
The principal hydraulic inputs in the MEDPG are the saturated hydraulic 
conductivity or permeability (ks) and the parameters defining the soil-water 
characteristic curve (SWCC) for unsaturated soil conditions typical of unbound 
materials beneath pavements. Saturated hydraulic conductivity (permeability) is a 
familiar property to most geotechnical and pavement engineers. The SWCC generally 
is not. The SWCC is a fundamental property of unsaturated soils that soil matric 
suction (in conceptual terms, the negative porewater pressure in a partially saturated 
soil) and water content as shown in Figure 39. The SWCC is required for analyses of 
water movement under partially saturated conditions. It is also used in characterizing 
the shear strength and compressibility of unsaturated soils, and the unsaturated 
hydraulic conductivity of soil is often estimated using properties from the SWCC 
together with the saturated hydraulic conductivity. 
 
 124 
The issue of Level 1 vs. Level 2 vs. Level 3 inputs for these properties is less 
significant than for other MEPDG inputs for several reasons: 
? Few geotechnical laboratories, including the one at SHA, are equipped with the 
pressure plate apparatus (ASTM D3152-72) required to determine the level 1 SWCC 
inputs. In addition, this test procedure is exacting, time-consuming, and expensive to 
perform. 
? The developers of the MEPDG expended considerable energy to develop a simplified 
yet accurate approach for specifying the SWCC in terms of the empirical Fredlund 
and Xing (1994) model. This model requires just four parameters to define the SWCC. 
The developers of the MEPDG developed correlations between these four parameters 
and grain size characteristics (for coarse soils) or grain size characteristic s and 
plasticity (for fine-grained cohesive soils). 
? As will be shown in the next subsection, predicted pavement performance is 
relatively insensitive to the SWCC and other hydraulic properties. 
 
Figure 39. Examples of SWCC curves from the MEPDG. 
 
 
 125 
Sensitivity Analyses 
The impacts of thermohydraulic properties for the bound layers on pavement 
performance predictions in the MEPDG have been well document. For flexible 
pavements, for example, the thermal properties of the asphalt concrete layer have a 
direct and pronounced influence on thermal contraction and low temperature cracking 
during sharp temperature drops in the winter and on softening and permanent 
deformations during high summer temperatures (e.g., Masad and Little, 2004; El-
Basyouny et al., 2005a, 2005b; Carvalho and Schwartz, 2006). The sensitivity of 
predicted pavement distresses to variations in the hydraulic properties of the unbound 
pavement materials?or even typical ranges of values for many of these properties?
are much less well established. For example, it is expected qualitatively that 
increasing subgrade moisture content will tend to reduce subgrade stiffness and 
increase rutting and fatigue cracking. What is not known are the quantitative 
magnitudes of these changes with respect to a given percentage change in subgrade 
moisture content or the changes in subgrade moisture for different sets of soil water 
characteristic curve parameters. 
A limited number of sensitivity analyses were performed as part of a 
reconnaissance study to quantify the influence of unbound hydraulic properties on 
predicted pavement performance. Four sites representing different climate extremes 
were considered: Seattle WA (wet-no freeze; PG 52-16), Caribou ME (wet-freeze; 
PG 52-34), Phoenix AZ (dry-no freeze; PG 76-10); and College Park MD (temperate; 
PG 64-22). Three pavement sections were analyzed at each site: 2 inches, 4 inches, 
and 6 inches of hot mix asphalt (HMA) over 12 inches of A-1-a base over a reference 
 
 126 
A-4 subgrade. Traffic levels were adjusted to give a service life of approximately 15 
years for the reference conditions at a 50% reliability level for all distresses. After a 
preliminary study, AADTT values of 300, 1000, and 2000 were used with 2, 4, and 6 
inches of HMA, respectively. HMA material properties were typical for a 19 mm 
dense graded Superpave mixture. All other reference inputs were set equal to the 
Level 3 defaults. Input parameters varied for the sensitivity analyses included 
subgrade type (A-2-4, A-4, and A-7-6), groundwater table (GWT) depth (2, 7, and 12 
feet), SWCC parameters (weighted plasticity index wPI varied by +50% from 
reference), saturated hydraulic conductivity ks (log ks varied by +50% from 
reference), traffic (+50% from reference) and subgrade resilient modulus MR (+50% 
from reference).  
 
Figure 40 summarizes the distress magnitudes predicted by the MEPDG for 
the reference analysis conditions. Because of the disparate units and magnitudes of 
the distress measures, they are expressed as a percentage of the MEPDG default 
design limits: 2000 ft/mi for longitudinal/top-down fatigue cracking; 25% of wheel 
path area for alligator/bottom-up fatigue cracking; 1000 ft/mi for transverse/thermal 
cracking; 0.25 in. for AC rutting; 0.75 in. for total rutting, and 172 in/mi for IRI. As is 
clear from  
Figure 40, rutting was the controlling distress at all locations, followed by IRI. 
Very little fatigue cracking was predicted in any of the analyses and no thermal 
cracking was predicted.  
 
 127 
0
20
40
60
80
100
L T  C rk A l l .  C rk AC  R u t T o t a l  R u t I R I
D
ist
re
ss 
(%
 o
f 
D
e
si
g
n
 L
im
it
) S e a t t l e
C a r i b o u
P h o e n i x
C o l l e g e  P a r k
120 159
 
Figure 40. Predicted distresses for reference conditions (4 in. HMA, A-4 subgrade, 7 ft 
GWT depth, medium traffic). Legend entries from top to bottom correspond to vertical 
bars from left to right for each distress. 
 
A local sensitivity index Si for quantifying the effect of subgrade type on 
performance can be defined in normalized terms as: 
112i DDS D????      (7) 
in which D is the distress magnitude predicted using the reference subgrade type (A-
4) and D+1 and D-1 are the distress magnitude predicted using stronger (A-2-4) and 
weaker (A-7-6) subgrade, respectively. 
Figure 41 summarizes the effect of subgrade type on performance under 
different climate condition in terms of the sensitivity index defined in Eq. (7). Several 
trends can be observed in these results: (a) longitudinal cracking (LT Crk) is most 
sensitive to subgrade type; (b) the sensitivities of alligator cracking (All Crk) and 
subgrade rutting (SG RD) are similarly low in magnitude; (c) granular base rutting 
(GB RD), asphalt concrete rutting (AC RD), and IRI are not sensitive to subgrade 
 
 128 
type; and (d) the sensitivity index values do not appear to be a function of climate 
type. 
 
Seattle 
 
Caribou 
 
Phoenix 
 
 129 
 
College Park 
Figure 41. Sensitivity of distresses to subgrade type at each location.  
 
Figure 42 summarizes the sensitivity of alligator cracking and total rutting to 
subgrade type variations at each of the four climate locations. The trends in Figure 42 
are generally sensible. Alligator cracking decreases when going from a poor (A-7-6) 
to good (A-2-4) subgrade, and the rate of decrease is approximately the same at all 
four locations. Total rutting is less sensitive to subgrade type, although the trends 
from poor to good subgrades are physically reasonable. The fact that subgrade 
deformations are only one part of total rutting may be responsible for the relatively 
lower sensitivities. 
Figure 42 shows that Caribou exhibited both the highest amount of alligator 
cracking and total rutting for all subgrade types among all of the climate locations. 
Examination of the average modulus of the top two feet of subgrade vs. time in 
Figure 43 sheds some insight into this. The annual freeze-thaw cycles at Caribou are 
very much evident for all three subgrade types. When frozen, the subgrade in Caribou 
is vastly stiffer than at any of the other locations. However, during the spring thaw 
and recovery the subgrade at Caribou has only about half the stiffness as at the other 
locations. The influence of subgrade stiffness on performance is not linear; this is true 
 
 130 
even in the AASHTO empirical pavement design procedure. The detrimental effects 
of very soft subgrades far outweigh the beneficial effects of stiff subgrades. In other 
words, the spring thaw at Caribou is more significant for performance than is the 
frozen winter. This certainly conforms to real-world experience?e.g., the posting of 
load limits on roads in northern climes during spring thaw conditions. 
0
1
2
3
4
5
6
7
8
9
S e a t t l e C a r i b o u P h o e n i x C o l l e g e  P a r k
A
ll
ig
a
to
r 
C
ra
c
k
in
g
  
(%
)
0 . 0
0 . 1
0 . 2
0 . 3
0 . 4
0 . 5
0 . 6
0 . 7
0 . 8
S e a t t l e C a r i b o u P h o e n i x C o l l e g e  P a r k
T
o
ta
l 
R
u
tt
in
g
 (
in
)
A - 7 - 6
A - 4
A - 2 - 4
 
Figure 42. Influence of subgrade type on selected predicted distresses at all four climate 
locations (4 in. HMA thickness, 7 ft. GWT depth, medium traffic). 
 
 131 
 
Figure 43. Average modulus of top 2 feet of subgrade vs. time  (4 in. HMA thickness, 7 
ft. GWT depth, medium traffic). 
 
A local sensitivity index Sji for quantifying the effect on performance of 
varying an analysis input Xi  (where X is the vector of analysis inputs) around some 
reference condition XiR can be defined in normalized terms as (Saltelli et al., 2000): 
 
 132 
j iRji
i jR
D XS XD???? ???
??
    (8)  
in which Dj is the magnitude of distress j predicted using the input Xi and DjR 
is the distress magnitude predicted using the reference input XiR. Equation (8) can be 
interpreted as the percentage change in distress Dj caused by a given percentage 
change in input Xi. Figure 40 summarizes for all predicted distresses (except 
transverse thermal cracking) the normalized Sji values computed using Eq. (8) for 
varying subgrade resilient modulus MR +50% from reference. The reference resilient 
modulus values are 11500 psi for A-7-6, 16500 psi for A-4, and 21500 psi for A-2-4. 
Several trends can be observed in these results: (a) longitudinal cracking (LT Crk) is 
most sensitive to MR, and subgrade rutting (SG RD) is second; (b) longitudinal 
cracking is more sensitive in warm climates (e.g., Phoenix) than in cold (e.g., 
Caribou); (c) longitudinal cracking is more sensitive to MR at higher reference moduli 
for all locations except Caribou.; and (d) granular base rutting (GB RD), asphalt 
concrete rutting (AC RD) and IRI are insensitive to MR. 
 
Seattle 
 
 133 
 
Caribou 
 
Phoenix 
 
College Park 
Figure 44. Sensitivity to MR for all distresses at all locations. 
 
Figure 45 and Figure 46 illustrate the influence of climate conditions in terms 
of average absolute sensitivity values?i.e., the average of the absolute values of the 
sensitivity indices across all distresses. The average sensitivity to subgrade modulus 
(Figure 45) shows some variation among sites, but this is sensible. The highest 
sensitivity in Seattle is likely due to the fact that the moisture-adjusted moduli there 
are the lowest of the four sites. Conversely, the relatively low sensitivity of average 
performance to subgrade modulus for Caribou may be attributed to the frozen stiff 
 
 134 
state of the subgrade for much of the year. Subgrade conditions in Phoenix and 
College park are arguably intermediate between these two extremes, which is 
consistent with their intermediate sensitivity values.  
The variations in sensitivity of average performance to environmental inputs 
(Figure 46) are all much smaller than for subgrade modulus (Figure 45). Depths to the 
groundwater table, soil-water characteristic curve parameters, and saturated hydraulic 
conductivity all have minimal influence on the predicted pavement performance.  
 
 
Figure 45. Average absolute sensitivity to subgrade MR. 
 
0 . 0 0
0 . 0 1
0 . 0 2
0 . 0 3
0 . 0 4
0 . 0 5
0 . 0 6
0 . 0 7
0 . 0 8
0 . 0 9
S e a t t l e C a r i b o u P h o e n i x C o l l e g e  P a r k
N
o
rm
a
li
ze
d
 S
e
n
s
it
iv
it
y
G W T  D e p t h
S W C C
l o g  k s
 
Figure 46. Average absolute sensitivity to environmental variables. 
 
 
 135 
The sensitivity study presented here had a very confined scope, and therefore 
one must be cautious in drawing any far-reaching conclusions. However, based upon 
the limited scenarios investigated here, the following observations can be made: 
? MR was the input of those studied that had the largest impact on predicted distresses; 
? The unbound material hydraulic inputs (GWT depth, SWCC parameters, and ks) all 
had slight to negligible influence on the predicted distresses; 
? Variations of performance with climate location and subgrade type were sensible. 
These findings will need to be supplemented by those from other scenarios 
before any truly robust conclusions can be drawn. However, even these limited 
sensitivity studies serve the valuable purpose of confirming that the Level 3 defaults 
for the unbound material hydraulic properties (GWT depth, SWCC parameters, and 
ks) should be suitable for design. This is fortunate, as these properties (except perhaps 
for project-specific GWT depth) are not currently measured by SHA. 
Summary 
Testing Recommendations 
Recommendations for future testing of unbound materials by SHA are as 
follows: 
1. SHA should perform further investigations to determine why the k1, k2, and k3 values 
computed from their laboratory-measured resilient modulus test data do not follow the 
expected physical trends. The causes may be either due to testing issues or unusual 
characteristics of the specific materials included in the database (e.g., cemented sands, 
highly overconsolidated clays, or other extreme /unusual soil conditions). 
 
 
 136 
2. SHA should continue to perform laboratory resilient modulus tests on common unbound 
materials in the state to augment and fill gaps in the database. The current database has a 
reasonable amount of measured resilient modulus data for subbase materials (e.g., A-2-4) 
and some subgrade soils (A-4, A-6). However, it is deficient in measured data for 
granular base materials (e.g., A-1-a and A-1-b) and the poorer subgrade soils (A-7-5, A-
7-6). Similar to the recommendations for asphalt binder and HMA testing, the resilient 
modulus testing for unbound materials could be included as a matter of policy (perhaps as 
part of the contract requirements) for large/important/expensive paving projects in the 
state. This testing could continue to be performed by SHA in its own laboratories or 
outsource to third-party laboratories. 
 
3. There is no need for SHA to begin any testing program for the hydraulic properties of 
unbound materials. These properties have very little impact of predicted pavement 
performance, and the empirical correlations in terms of gradation and plasticity 
parameters built into the MEPDG provide sufficient accuracy. 
Recent results from NCHRP Project 9-23A ?A National Catalog of Subgrade 
Soil-Water Characteristic Curve (SWCC) Default Inputs and Selected Soil Properties 
for Use with the ME-PDG? were made available to the public as NCHRP Web 
Document 153 in September 2010 after the draft final report had been submitted to 
SHA for review (Zapata, 2010). Although the hydraulic properties of unbound 
materials have been found to have little influence on predicted pavement performance 
for Maryland conditions, the information in this thesis could be mined for additional 
default values if desired.  Key primary information in the NCHRP 9-23A database 
include: AASHTO soil class; limited gradation data; Atterberg limits; saturated 
hydraulic conductivity; saturated volumetric water content; and volumetric water 
 
 137 
content at various levels of matric suction (used to compute soil water characteristic 
curve--SWCC). In addition to this primary information, there are also many 
secondary quantities in the database that are derived from the primary information, 
e.g., SWCC parameters and CBR estimated from correlations with gradation and 
plasticity, MR estimated from (estimated) CBR, etc. Data are organized by ?map 
unit,? a geographical area over which the soil properties are assumed roughly 
uniform. There are a total of 568 map units for the state of Maryland; 79 of these have 
no information in the database. Of the remainder, it is expected that many will have 
only partial information. For example, on a national level only 66% of the map units 
have measured soil water characteristic curve data. Unfortunately, detailed evaluation 
of the Maryland data could not be performed in this study because the information 
became available to the public only at the very end of the project, approximately 2 
months after the draft final report was submitted to SHA. 
Recommended MEDPG Inputs 
The recommended unbound material inputs to the MEPDG for Maryland 
conditions are summarized in Table 51. Average gradation, plasticity, and volumetric 
information for Maryland materials as obtained from the data supplied by SHA is 
summarized in Table 54. The limited number of measured values for subgrade MR in 
Table 54 correspond to the data in Table 49 at optimum moisture and 95% 
compaction. Note that these values are almost all significantly lower than the 
MEPDG level 3 defaults and should therefore be used with caution. 
 
 
 138 
Table 51. Recommendation material property inputs for unbound materials for 
Maryland conditions. 
Property Input 
Level 
Value Comment 
Unbound material All Project 
specific 
Material class (e.g., AASHTO) 
Thickness All Project 
specific 
 
Strength Properties (ICM Calculated Modulus) 
Poisson?s ratio 3 Material 
specific 
Use level 3 defaults. Table 52 provides 
additional guidance. 
Coefficient of lateral 
pressure 
3 Material 
specific 
Use level 3 defaults. Table 53 provides 
additional guidance. 
Modulus 2/3 Material 
specific 
Use level 3 defaults or values from Table 
54 where available. 
ICM (Mean Values) 
Gradation 2/3 Material 
specific 
Use values from Table 54 or level 3 
defaults. 
Plasticity Index 2/3 Material 
specific 
Use values from Table 54 or level 3 
defaults. 
Liquid Limit 2/3 Material 
specific 
Use values from Table 54 or level 3 
defaults. 
Compacted All Project/layer 
specific 
 
Maximum dry unit 
weight 
2/3 Material 
specific 
Use values from Table 54 or level 3 
defaults. 
Specific gravity 2/3 Material 
specific 
Use level 3 defaults. 
Saturated hydraulic 
conductivity 
3 Material 
specific 
Use level 3 defaults. 
Optimum gravimetric 
water content 
2/3 Material 
specific 
Use values from Table 54 or level 3 
defaults. 
Degree of saturation 
at optimum 
2/3 Material 
specific 
Use values from Table 54 or level 3 
defaults. 
Soil-water 
characteristic curve 
parameters  
(af, bf, cf, hr) 
3 Material 
specific 
Use level 3 defaults. 
 
 
 139 
Table 52. Typical Poisson?s ratio values for unbound granular and subgrade materials 
(NCHRP, 2004). 
  
Table 53. Typical coefficient of lateral pressure for unbound granular, subgrade, and 
bedrock materials (NCHRP, 2004). 
 
 
Table 54. Average properties for Maryland unbound materials (based on material 
property database at time of report). 
Class N LL (%) PI (%) % < No. 4 
% < 
No. 
200 
OMC 
(%) 
S at OMC 
(%) 
Max Dry 
Unit Weight 
(pcf) 
Subgrade 
MR1 (psi) 
A-1-b 5 23.8 7.0 60 15 6.7 53.6 135.3  
A-2-4 42 24.2 8.7 97 28 8.9 60.3 129.5 10,000 
A-2-6 12 26.0 13.6 89 29 8.9 59.0 128.3 10,000 
A-3 4   99 8 10.8 58.2 111.3  
A-4 96 29.5 8.0 99 48 11.8 67.8 122.5 6,000 
A-5 3 41.0 8.0 99 71 15.6 57.2 112.6  
A-6 34 31.0 12.1 99 54 12.3 76.1 121.5 5,500 
A-7-5 12 46.0 14.5 100 57 16.0 75.1 114.3 8,000 
A-7-6 6 49.5 22.0 100 64 16.8 76.8 110.1 7,500 
  1These values are significantly lower than the MEPDG level 3 defaults in most cases. 
 
 
 140 
CHAPTER 6:  MATERIAL PROPERTIES DATABASE 
Introduction 
MatProp is a MEPDG data management system based on Microsoft Access? 
2007. It incorporates data entry, editing, and storing functionality for the material 
property inputs required by the MEPDG as well as additional data maintained by 
SHA. MatProp displays the data in a format similar to the MEPDG Version 1.100 
data entry screens. The overall organization of MatProp is diagrammed in Figure 47. 
 
Figure 47. Organization of MatProp database. 
 
MatProp is composed of 3 main sections: flexible, rigid, and unbound 
materials. The flexible section includes both binder and HMA related data; the rigid 
section includes PCC related data, and the unbound section includes material property 
data for granular base and subgrade materials. 
 
 141 
Instructions for Using MatProp 
Installation of MatProp consists of simply unzipping the ?MatProp.zip? 
archive.  This creates a folder named ?MatProp System? containing 3 files: 
?MatProp.mdb?, the actual database; ?Mouse Hook.dll,? a utility for increasing 
mouse functionality while within the database; and ?Readme.txt,? which contains any 
release and/or installation notes. Double clicking ?MatProp.mdb? opens MatProp. 
Depending upon the security settings of the host computer, the security warning 
shown in Figure 48 may be displayed (it may be hidden behind the main menu); if so, 
simply click ?Options??, choose ?Enable this content? as shown in Figure 49, and 
click OK. After that, the main menu appears as shown in Figure 50. 
 
 
Figure 48. Security warning. 
 
 
 142 
 
Figure 49. Security alert. 
 
 
Figure 50. Main menu. 
 
 
 143 
There are 3 buttons in the upper rectangle portion of the main menu that open 
screens that display the input data required by the MEPDG. Examples of these are 
shown in Figure 51 to Figure 55 for HMA materials after clicking ?Show MEPDG 
HMA Input.? Note that the data display screens closely mirror the appearance of the 
corresponding data entry screen in the MEDPG Version 1.100. Similar MEPDG input 
screens are provided for the PCC and unbound material categories. 
The 4 buttons in the lower rectangle portion of the main menu are used for 
data entry and management. The data entry and management functionality for the 
different material categories is described in the following subsections. 
User Interface for Flexible Pavement Material Management 
Clicking ?Manage Binder Data? will bring up binder edit form shown in 
Figure 56. Binder data can be added or edited using the form in Figure 57. For 
consistency with the data received from SHA, the ID of binder data is numeric. 
Suppliers and terminals can also be added/edited/deleted as shown from Figure 58 to 
Figure 62. Data integrity checking is enforced as shown in Figure 63 and Figure 64. 
Binders can be edited by clicking ?Edit? besides binder list in Figure 56 as shown in 
Figure 65. If ?Delete? is clicked without selecting a record, a warning (Figure 66) will 
pop up.  
Clicking ?Manage HMA Data? brings up the HMA data management form 
shown in Figure 67. New HMA mixtures can be added as shown in Figure 68. 
Dynamic modulus testing data can be managed as shown in Figure 69. Creep 
compliance data management is shown in Figure 71. If temperature is not specified 
beforehand, a warning will pop up as shown in Figure 70. Note that records can be 
 
 144 
excluded from calculations of average mixture properties (for a given nominal 
maximum aggregate size and mix type) by simply changing the ?excluded? control to 
?Yes?. 
 
Figure 51. ?Show MEDPG HMA Input? screen for level 1 Asphalt Mix properties. 
 
 145 
 
Figure 52. ?Show MEDPG HMA Input? screen for level 2/3 Asphalt Mix properties. 
 
 146 
 
Figure 53. ?Show MEDPG HMA Input? screen for level 1/2 Asphalt Binder properties. 
 
 147 
 
Figure 54. ?Show MEDPG HMA Input? screen for level 3 Asphalt Binder properties. 
 
 148 
 
Figure 55. ?Show MEDPG HMA Input? screen for Asphalt General properties (all 
input levels). 
 
 
 149 
 
Figure 56. ?Manage Binder Data? ? main screen. 
 
 
Figure 57. Add new binder. 
 
 
Figure 58. Look up binder supplier. 
 
 150 
 
Figure 59. Add new supplier. 
 
 
Figure 60. Look up terminal. 
 
 
Figure 61. Add new terminal. 
 
 
Figure 62. Data integrity checking before deleting terminal. 
 
 
 151 
 
Figure 63. Saving without completion. 
 
 
Figure 64. ID integrity checking. 
 
 
Figure 65. ?Edit Binder Property? screen. 
 
Figure 66. Delete without selection. 
 
 
 152 
 
Figure 67. ?Manage HMA Data? ? main screen. 
 
 
Figure 68. Add HMA mixture. 
 
 
 153 
 
Figure 69. Edit dynamic modulus data. 
 
 
Figure 70. Attempt to add creep data without providing temperature. 
 
Figure 71. Add creep compliance data. 
 
User Interface for Rigid Pavement Material Management 
The main PCC data management screen is shown in Figure 72. Integrity 
checking is the same as in other management forms. New PCC material can be added 
and edited as shown in Figure 73. The ID for PCC material is a character string. As 
for binders and HMA mixtures, integrity checking is enforced and no duplicate IDs 
 
 154 
are allowed. Records can be excluded from average property calculations by simply 
changing the ?excluded? control to ?Yes?. 
 
 
Figure 72. ?Manage PCC Data? ? main screen. 
 
Figure 73. New PCC mixture. 
User Interface for Unbound Material 
Figure 74 shows the main data management screen for unbound materials. 
Integrity checking is similar to that in the other data management forms. The stress-
dependent modulus k values can be calculated using the ?Calculate Parameters? 
button.  If the calculated values are not acceptable, they can be modified manually by 
 
 155 
directly inputting the preferred values. Unbound material can be added and edited as 
shown in Figure 75 and Figure 76. Resilient modulus testing conditions (e.g., 
moisture and density as tested) can be added and edited as shown in Figure 77 and 
Figure 78. Resilient modulus data can be added and edited as shown in Figure 79 and 
Figure 80. Individual records can be excluded from the calculation of average 
parameter values for the soil class by changing the ?excluded? control to ?Yes?. 
 
Figure 74. ?Manage Unbound Data? ? main screen. 
 
 
Figure 75. Add unbound material. 
 
 
 156 
 
Figure 76. Edit unbound material. 
 
 
Figure 77. New testing condition. 
 
 
Figure 78. Edit testing condition. 
 
 
Figure 79. Add MR data. 
 
 
 157 
 
Figure 80. Edit MR data. 
 
Database Structure 
The MatProp relational database contains a total of 33 tables. The relations 
among the binder tables are illustrated in Figure 81 to Figure 84. The tblBinderMaster 
and tblBinderProperties tables in Figure 81 store the main information about binder 
material. The tblBinderMaster table stores basic information for the binder while the 
tblBinderProperties table stores the binder testing data. One record in the 
tblBinderMaster table may have many related records in tblBinderProperties table. 
This is shown by the relationship lines that designate one-to-one, one-to-many, or 
many-to-many relationships between tables and the fields (e.g., ID fields) used to 
define these relationships. Deleting a record in tblBinderMaster table will also delete 
all the related record in the tblBinderProperties table. Other tables prefixed with c_ 
are look-up tables used to define the data entry codes in the main tables.  
Figure 82 to Figure 84 similarly show relations among the HMA mixture, 
PCC mixture, and unbound materials tables, respectively. Again, table names having 
c_  prefixes are look-up tables while all other tables are main data tables. The 
relationship lines have the same meanings as in Figure 81.  
 
 
 158 
 
Figure 81. Tables and relations for binder material data. 
 
 
Figure 82. Tables and relations for HMA mixture data. 
 
Figure 83. Tables and relations for PCC mixture data. 
 
 159 
 
Figure 84. Tables and relations for unbound material data. 
 
All tables and fields used in the database are summarized in the listing below: 
Table Name Field Name Data 
Type 
Size Description 
c_tblBinderCondition Condition Text 10  
 Code Byte 1  
c_tblBinderSupplier Supplier Text 35 Supplier name should be 
less than 35 letters 
 
 160 
 Code Byte 1  
c_tblBinderTerminal Terminal Text 30 Terminal name should be 
less than 35 letters 
 Code Byte 1  
c_tblBinderTestKind Test Text 25  
 Code Byte 1  
c_tblHMATraffic Traffic Text 13  
 Code Byte 1  
c_tblHMAType Type Text 15  
 Code Byte 1  
c_tblPCCAggregateTy
pe 
ID Long 
Integer 
4 Auto Number so use Long 
Integer 
 AggregateType Text 10  
 Code Byte 1  
c_tblPCCCementType ID Long 
Integer 
4 Auto Number so use Long 
Integer 
 CementType Text 8  
 Code Byte 1  
c_tblPCCCuringType ID Long 
Integer 
4 Auto Number so use Long 
Integer 
 CuringType Text 16  
 Code Byte 1  
c_tblPCCMixType MixType Text 10  
 Code Byte 1  
c_tblUnboundClassifi
cation 
Class Text 40  
 Code Byte 1  
tblBinderMaster SampleID Long 
Integer 
4 The ID of test sample. 
 PG_Grade Text 6 The PG Grade of test 
sample. 
 Supplier Byte 1 Supplier of the test. See 
tblSupplierCode for detail. 
 LotNumber Text 30 Lot Number of test sample. 
 TankNumber Text 30 Tank Number of test 
sample. 
 Terminal Byte 1 Terminal of the test. See 
tblTerminalCode for detail. 
 SampleDate Date/T
ime 
8 The date when test 
processed. 
 Exclude Yes/N
o 
1  
tblBinderProperties BinderPropertyID Long 
Integer 
4 The ID for the test data. 
Cannot be duplicate. 
 SampleID Long 
Integer 
4 The ID for the test 
sample.CAN be duplicate. 
 Test Byte 1 Test kind of the data. See 
 
 161 
tblTestKindCode for detail. 
 Condition Byte 1 Testing condition. See 
tblCondition Code for 
detail. 
 Temperature Integer 2 The temperature in which 
test proceed. 
 Value1 Single 4 RotationalViscosity in test 
No.1; GStar in test No.2; 
Stiffness in test No.3 
 Value2 Single 4 Spindle size in test No. 1; 
Phase in test No.2; Mvalue 
in test No. 3. 
 Exclude Yes/N
o 
1 Default value is No. If yes, 
the record is excluded from 
being used. 
tblHMACreepData CreepDataID Long 
Integer 
4 The ID number of the data. 
Cannot be duplicate. 
 MixDesignID Text 11 The ID of the sample. CAN 
be duplicate here. 
 LoadingTime Single 4 In the same unit.eg: second. 
 LowTempCompli
ance 
Single 4 Compliance in low 
temperature. 
 MedTempCompli
ance 
Single 4 Compliance in medium 
temperature. 
 HighTempCompli
ance 
Single 4 Compliance in high 
temperature. 
 Exclude Yes/N
o 
1  
tblHMACreepMaster MixDesignID Text 11 The ID number of the 
sample. Cannot be 
duplicate. 
 TensileStrength Single 4 Unit:(psi) 
 AggCoefThermC
ontr 
Single 4 Low temp in which creep 
compliance is tested. 
 LowTemp Single 4 Low temp in which creep 
compliance is tested. 
 MedTemp Single 4 Medium temp in which 
creep compliance is tested. 
 HighTemp Single 4 Hight temp in which creep 
compliance is tested. 
tblHMADynModData DynModDataID Long 
Integer 
4 The ID number of the 
property data. Cannot be 
duplicate. 
 MixDesignID Text 11 The ID number of the mix. 
CAN be duplicate. 
 Temp Single 4 Temperature of the data 
testing.Related to 
tblHMADynModMaster. = 
1,2,3,4,5,6 
 
 162 
 Freq Single 4 Frequency of the data 
testing. Related to 
tblHMADynModMaster. = 
1,2,3,4,5,6 
 Modulus Single 4 Mr of the sample. 
 PhaseAngle Single 4 Phase Angle of the sample. 
 Exclude Yes/N
o 
1  
tblHMADynModMast
er 
MixDesignID Text 11 The ID number of the mix. 
 RefTemperature Single 4 The temperature of testing. 
(F) 
 Temp1 Single 4 Temperature appears at 1st 
place in form 
 Temp2 Single 4 Temperature appears at 2nd 
place in form 
 Temp3 Single 4 Temperature appears at 3rd 
place in form 
 Temp4 Single 4 Temperature appears at 4th 
place in form 
 Temp5 Single 4 Temperature appears at 5th 
place in form 
 Temp6 Single 4 Temperature appears at 6th 
place in form 
 Freq1 Single 4 Frequency appears at 1st 
place in form 
 Freq2 Single 4 Frequency appears at 2nd 
place in form 
 Freq3 Single 4 Frequency appears at 3rd 
place in form 
 Freq4 Single 4 Frequency appears at 4th 
place in form 
 Freq5 Single 4 Frequency appears at 5th 
place in form 
 Freq6 Single 4 Frequency appears at 6th 
place in form 
 TestDate Date/T
ime 
8  
tblHMAGradation MixDesignID Text 11 The ID number of the mix. 
 Targetp50_0 Single 4 Volumetric Property 
 Targetp37_5 Single 4 Volumetric Property 
 Targetp25_0 Single 4 Volumetric Property 
 Targetp19_0 Single 4 Volumetric Property 
 Targetp12_5 Single 4 Volumetric Property 
 Targetp9_5 Single 4 Volumetric Property 
 Targetp4_75 Single 4 Volumetric Property 
 Targetp2_36 Single 4 Volumetric Property 
 Targetp1_18 Single 4 Volumetric Property 
 
 163 
 Targetp0_60 Single 4 Volumetric Property 
 Targetp0_30 Single 4 Volumetric Property 
 Targetp0_15 Single 4 Volumetric Property 
 Targetp0_075 Single 4 Volumetric Property 
tblHMAMaster MixDesignID Text 11 The ID number of the 
record. Unique in this table. 
 Plant Text 3 The code of the Plant. 
 Band Single 4 The band length of the 
sample. 
 Type Byte 1 The Type of the sample. 
See tblType for detail. 
 Traffic Byte 1 The traffic level. See 
tblTraffic for detail. 
 BinderGrade Text 5 The binder grade of the 
Mix. 
 MixNumber Text 2 This is Mix Number. 
 NInitial Text 1 This is N initial. 
 NDesign Text 3 This is N design. 
 NMaximum Text 3 This is N maximum. 
 Exclude Yes/N
o 
1  
tblHMAOther MixDesignID Text 11 The ID number of the 
record. Unique in this table. 
 PoissonRatio Single 4  
 ThermalConducti
vity 
Single 4 unit: BTU/(hr*ft*F) 
 HeatCapacity Single 4 unit: BTU/(lb*F) 
tblHMAVolumetrics MixDesignID Text 11 The ID number of Mix. 
 Gmm Single 4 builk-specific gravity of a 
compacted asphalt mixture 
 Gmb Single 4 theoretical maximum 
specific gravity 
 Gse Single 4 aggregate-effective specific 
gravity 
 Pb Single 4 asphalt content with respect 
to the total mixture weight 
 Pba Single 4 asphalt absorbed with 
respect to the total mixture 
weight 
 Pbe Single 4 effective asphalt content 
with respect to the total 
mixture weight 
 Va Single 4 Air void as tested 
 VaInplace Single 4 Air void as in place 
condition 
 Vma Single 4 Voids in the mineral 
aggregate 
 Vfa Single 4 Voids Filled with Asphalt 
 
 164 
 D/PbeRatio Single 4 The ratio of D over Pbe 
 D/BRatio Single 4 The ratio of D over B 
 Vbe Single 4 Volume of effective asphalt 
binder 
 Gamat Single 4 total unit weight 
tblLevels ID Long 
Integer 
4  
 Field1 Byte 1 Levels used in forms. 
tblPCCAggregateCTE AggregateID Byte 1 Aggregate code, see 
c_tblPCCAggregateType 
 CTE Single 4 CTE of that aggregate 
tblPCCGeneral ID Text 11 PCC ID 
 UnitWeight Single 4 in pcf 
 PoissonRatio Single 4  
 CoefThermalExpa
n 
Single 4 Thermal Conductivity 
 ThermalConduct Single 4 Heat Capacity 
 HeatCap Single 4 Heat Capacity 
 LayerThickness Long 
Integer 
4 May not be used 
 Exclude Yes/N
o 
1  
tblPCCMix ID Text 11 PCC ID 
 PCC0StressTemp Single 4 hard to test 
 ReversibleShrinka
ge 
Single 4 hard to test 
 50UltimateTime Single 4 hard to test 
 Aggregate Byte 1 Aggregate type used in the 
PCC 
 Cement Byte 1 Cement type used 
 CMaterialContent Single 4 Cement content 
 WCR Single 4 Water Cemenet Ratio 
 UltimateShrink Single 4 hard to test 
 Curing Byte 1 Curing Method.  
 PCCType Byte 1 Regular or Recycle 
tblPCCStrengthLv1 ID Text 11 ID of PCC 
 7E Single 4 E at 7 day 
 14E Single 4 Similar to above 
 28E Single 4 Similar to above 
 90E Single 4 Similar to above 
 20a28dE Single 4 E Ratio of 20year over 28 
day 
 7MR Single 4 Modulus of Rupture at 7 
day 
 14MR Single 4 Similar to above 
 28MR Single 4 Similar to above 
 
 165 
 90MR Single 4 Similar to above 
 20a28dMR Single 4 Modulus of Rupture Ratio 
of 20year over 28 day 
tblPCCStrengthLv2 ID Text 11 PCC ID 
 7comp Single 4 Compressive Strength at 7 
day 
 14comp Single 4 Similar to above 
 28comp Single 4 Similar to above 
 90comp Single 4 Similar to above 
 20a28dcomp Single 4 Compressive Strength 
Ratio of 20year over 28 day 
tblUnboundCondition ConditionID Long 
Integer 
4 ID for each unbound 
material condition 
 ID Text 11 ID according to 
tblUnboundMaster(master) 
 ConditionName Text 30 Name of the conditon i.e. 
Optimum 
 DryDensity Single 4 The dry density of this 
condition 
 MoistureContent Single 4 The dry density of this 
condition 
 ICMK1 Single 4 Regression factor K1 
 ICMK2 Single 4 Regression factor K2 
 ICMK3 Single 4 Regression factor K3 
 Exclude Yes/N
o 
1  
tblUnboundGeneral ID Text 11 A unique ID for each 
unbound material sample. 
 PI Single 4 Plastic Index 
 LL Single 4 Liquid limit 
 MaxDryUnitWeig
ht 
Single 4 As field name(pcf) 
 Gs Single 4 Specific Gravity 
 SatHydraulicCon
d 
Single 4 Saturation Hydraulic 
Conductivity. (ft/hr) 
 OptGravWaterCo
ntent 
Single 4 In % 
 SatDegreeOpt Single 4 In % 
 af Single 4 Parameters defining SWCC 
 bf Single 4 Parameters defining SWCC 
 cf Single 4 Parameters defining SWCC 
 hr Single 4 Parameters defining SWCC 
 001mm Single 4 Volumetric Property 
 002mm Single 4 Volumetric Property 
 020mm Single 4 Volumetric Property 
 #200 Single 4 Volumetric Property 
 
 166 
 #100 Single 4 Volumetric Property 
 #80 Single 4 Volumetric Property 
 #60 Single 4 Volumetric Property 
 #50 Single 4 Volumetric Property 
 #40 Single 4 Volumetric Property 
 #30 Single 4 Volumetric Property 
 #20 Single 4 Volumetric Property 
 #16 Single 4 Volumetric Property 
 #10 Single 4 Volumetric Property 
 #8 Single 4 Volumetric Property 
 #4 Single 4 Volumetric Property 
 3/8 Single 4 Volumetric Property 
 1/2 Single 4 Volumetric Property 
 3/4 Single 4 Volumetric Property 
 1 Single 4 Volumetric Property 
 11/2 Single 4 Volumetric Property 
 2 Single 4 Volumetric Property 
 21/2 Single 4 Volumetric Property 
 3 Single 4 Volumetric Property 
 31/2 Single 4 Volumetric Property 
tblUnboundLevel2 ID Text 11 A unique ID for each 
unbound material sample. 
 ICMMr Single 4 Resilient Modulus of 
Integrated Climate Model 
 ICMCBR Single 4 CBR of Integrated Climate 
Model 
tblUnboundMaster ID Text 11 A unique ID for each 
unbound material sample. 
 Classification Byte 1 Code for Classification 
 Group Byte 1 Group number for SHA 
testing convenience. 
 PR Single 4 Poisson's Ratio 
 Ko Single 4 Coefficient of laterial 
pressure 
 TestingDate Date/T
ime 
8  
 Exclude Yes/N
o 
1  
tblUnboundMr SubID Replic
ation 
ID 
16 A unique ID for each Mr 
testing. 
 ConditionID Long 
Integer 
4 ID according to the testing 
condition 
 ChamberPressure Single 4 ChamberPressure (psi) 
 DeviatorStress Single 4 DeviatorStress (psi) 
 MrValue Single 4 Average Resilient Modulus 
 
 167 
(psi) 
 Exclude Yes/N
o 
1  
tblYesNo codeYesNo Yes/N
o 
1 0?No, -1?Yes 
 txtYesNo Text 3 text ?Yes? or ?No? 
 
 168 
CHAPTER 7:  SUMMARY OF RECOMMENDATIONS 
Project Summary 
The objective of the study described in this thesis was to develop this type of 
organized database of material properties for the most common paving materials used 
in Maryland. Separate chapters for each of the major pavement material types 
(asphalt binder, HMA mixtures, PCC mixtures, unbound materials) detail the 
following essential information for understanding and using the MEPDG: 
MEPDG Input Requirements 
 New Construction/Reconstruction/Overlays 
 Rehabilitation (Existing Layers) 
Data Available from Maryland SHA 
Analyses of MEDPG Inputs 
 Level 1 vs. Level 2 vs. Level 3 
 Sensitivity Analyses 
Summary 
 Testing Recommendations 
 Recommended MEPDG Inputs 
A comprehensive material property database developed in Microsoft Access 
2007 accompanies this thesis. This database is initially populated with all information 
receive from SHA. It provides complete data management tools for adding future data 
as well as data display screens for MEDPG inputs that mirror the input screens in the 
MEPDG Version 1.100 software. 
 
 169 
The detailed recommendations from this study for the Maryland-specific 
MEDPG input values are summarized in the last sections of Chapters 2 through 5 for 
asphalt binder, HMA mixtures, PCC mixtures, and unbound materials, respectively. 
These recommendations will not be repeated here. However, the recommendations 
for future MEPDG-related materials testing by SHA are compiled below for the 
convenience of SHA materials engineers and policy makers. 
Testing Recommendations 
Asphalt Binders 
1. Maryland SHA currently has no Level 1/2 input data for asphalt binders. The sensitivity 
of predicted pavement performance to Level 1 vs. Level 2 vs. Level 3 binder inputs 
appears slight. Based only on this criterion there would be little purpose for SHA 
collection of Level 1 or 2 binder data. However, input of Level 1 HMA properties also 
requires input of Level 1/2 binder data. It is therefore recommended that SHA develop a 
policy of full binder characterization on major projects and that the test results be entered 
into the material property database so that typical Level 1/2 properties can be input into 
the MEPDG in the future. The testing frequency for full binder characterization should 
match the recommendations for HMA dynamic modulus testing. 
HMA Mixtures 
2. Maryland SHA currently has no Level 1 input data for HMA mixtures. There is the 
potential for significant differences in predicted performance using Level 1 vs. Level 2/3 
dynamic modulus data. In addition, the Witczak predictive equation used to generate the 
Level 2/3 dynamic modulus data is not intended for SMA mixtures, a common premium 
mixture type in Maryland, and often does not differentiate among different dense graded 
 
 170 
mixtures adequately. Therefore, SHA should plan to begin measuring Level 1 dynamic 
modulus data over time for the most commonly used mixture types in conjunction with 
major paving projects. Level 1 dynamic modulus testing of HMA mixtures will also 
require companion Level 1 characterization of the asphalt binders. 
 
It is recommended that SHA develop a policy requiring Level 1 HMA dynamic modulus 
and binder characterization testing for all major projects. Major projects could be defined 
by SHA in terms of a minimum placement tonnage, minimum traffic volume, or some 
other measure of project/mix importance. This testing could be done in-house using 
either the UTM-25 or AMPT test systems in the SHA laboratories; however, some 
equipment repair and/or calibration would be required as both of these systems are 
currently nonoperational. This testing could also be outsourced to local commercial 
testing facilities (e.g., Advanced Asphalt Technologies, LLC) and/or the University of 
Maryland (HMA dynamic modulus testing only). 
3. There is no need for measuring Level 1 creep compliance, low temperature tensile 
strength, and aggregate coefficient of thermal contraction properties. These properties are 
used only for predictions of thermal cracking, which is not a major distress type in 
Maryland. The Level 3 relationships built into the MEPDG code for converting dynamic 
modulus and other mixture inputs to creep compliance, low temperature tensile strength, 
and aggregate coefficient of thermal contraction are judged as sufficient for Maryland 
purposes. 
4. HMA thermal conductivity and heat capacity generally have only slight influence on 
pavement performance predicted by the MEPDG. Consequently, the Level 3 default 
values built into the MEPDG software are sufficient and laboratory measurement of these 
properties is not warranted.  
 
 
 171 
5. Although SSA has a much more significant influence on predicted performance, there at 
present is no easy widely-used method for measuring this parameter, either initially after 
construction or over the pavement life. Therefore, no testing is recommended for this 
property. The Level 3 defaults values built into the MEPDG software should be used. 
PCC Mixtures 
6. There is very little data on the physical and mechanical properties Maryland PCC mixes 
to be incorporated into the database at this time. Much of the physical data required by 
the MEPDG (e.g., cement type, cementitious material content, water/cement ratio) is 
routinely measured for individual projects. These data should be collected and entered 
into the database on a project-by-project basis.  
7. Maryland SHA has no Level 1 input data for PCC mixtures. However, there is no 
documented need to perform additional laboratory testing to determine the full Level 1 
stiffness and strength inputs for PCC. Instead, 28-day PCC elastic modulus and modulus 
of rupture should be measured for JPCP paving projects in the future, incorporated into 
the database, and used for Level 3 inputs to the MEPDG. Continued measurement of split 
cylinder tensile strength should be discontinued, as this is not a primary input to the 
MEPDG (or to the 93 AASHTO Design Guide).  
8. Given the lack of practical accepted test standards, ongoing test protocol issues, and other 
reasons, it is recommended that SHA not embark on any additional testing for thermal or 
shrinkage properties at this time. The current version of the MEPDG has been calibrated 
using the default Level 3 values for these properties, and these default values should 
continue to be used until accepted testing standards are available. 
 
 172 
Unbound Materials 
9. Maryland SHA already has significant Level 1 resilient modulus data for unbound 
materials. SHA should perform further investigations to determine why the k1, k2, and k3 
values computed from their laboratory-measured resilient modulus test data do not follow 
the expected physical trends. The causes may be either due to testing issues or unusual 
characteristics of the specific materials included in the database (e.g., cemented sands, 
highly overconsolidated clays, or other extreme /unusual soil conditions). 
10. SHA should continue to perform laboratory resilient modulus tests on common unbound 
materials in the state to augment and fill gaps in the database. The current database has a 
reasonable amount of measured resilient modulus data for subbase materials (e.g., A-2-4) 
and some subgrade soils (A-4, A-6). However, it is deficient in measured data for 
granular base materials (e.g., A-1-a and A-1-b) and the poorer subgrade soils (A-7-5, A-
7-6). Similar to the recommendations for asphalt binder and HMA testing, the resilient 
modulus testing for unbound materials could be included as a matter of policy (perhaps as 
part of the contract requirements) for large/important/expensive paving projects in the 
state. This testing could continue to be performed by SHA in its own laboratories or 
outsource to third-party laboratories. 
11. There is no need for SHA to begin any testing program for the hydraulic properties of 
unbound materials. These properties have very little impact of predicted pavement 
performance, and the empirical correlations in terms of gradation and plasticity 
parameters built into the MEPDG Level 2/3 inputs provide sufficient accuracy. 
 
 173 
References: 
AASHTO (1993). Guide for Design of Pavements Structures. American Association 
of State Highway and Transportation Officials, Washington, DC. 
AASHTO (2008). Mechanistic-Empirical Pavement Design Guide: A Manual of 
Practice. American Association of State Highway and Transportation Officials, 
Washington, DC. 
Andrei, D. (2003). Development of a Predictive Model for the Resilient Modulus of 
Unbound Materials. Ph.D. Dissertation, Arizona State University, Tempe, AZ. 
ARA (2009). MODOT Study RI-4-002, Implementing the AASHTO Mechanistic-
Empirical Pavement Design Guide in Missouri. Final Report. Volume I: Study 
Findings, Conclusions, and Recommendations. Applied Research Associates, 
Champaign, IL. 
Ayyala, D., G.R. Chehab, and J.S. Daniel (2010). ?Sensitivity of MEPDG Level 2 
and 3 Inputs using Statistical Analysis Techniques for New England States.? Annual 
Meetings of the Transportation Research Board (CD-ROM), National Research 
Council, Washington, DC. 
Azari, H., A. Mohseni, and N. Gibson (2008). ?Verification of Rutting Predictions 
from Mechanistic-Empirical Pavement Design Guide by Use of Accelerated Loading 
Facility Data.? Transportation Research Record: Journal of the Transportation 
Research Board, No. 2057, pp. 157-167. 
Buch, N., K. Chatti, S.W. Haider, and A. Manik (2008). Evaluation of the 1-37A 
Design Process for New and Rehabilitated JPCP and HMA Pavements. Research 
Report RC-1516, Michigan State University, East Lansing, MI. 
Carvalho, R., and Schwartz, C. W. (2006). ?Comparisons of Flexible Pavement 
Designs: AASHTO Empirical Versus NCHRP Project 1-37A Mechanistic-
Empirical.? Transportation Research Record: Journal of the Transportation 
Research Board, No. 1947, pp. 167-174. 
Ceylan, H., C.W. Schwartz, S. Kim and K. Gopalakrishnan (2009). ?Accuracy of 
Predictive Models for Dynamic Modulus of Hot-Mix Asphalt.? Journal of Materials 
in Civil Engineering, 21(6), 286-293.   
El-Basyouny, M. M., and M.W. Witczak (2005a). ?Calibration of Alligator Fatigue 
Cracking Model for 2002 Design Guide.? Transportation Research Record: Journal 
of the Transportation Research Board, No. 1919, pp. 77-86.   
 
 174 
El-Basyouny, M. M., M.W. Witczak, and S. El-Badawy (2005b). ?Verification of the 
Calibrated Permanent Deformation Models for the 2002 Design Guide.? Journal of 
the Association of Asphalt Paving Technologists, Vol. 74, pp. 601-652.  
El-Basyouny, M. M., and M.W. Witczak (2005c) ?Verification of the Calibrated 
Fatigue Cracking Models for the 2002 Design Guide.? Journal of the Association of 
Asphalt Paving Technologists, Vol. 74, pp. 653-695. 
FHWA (2009). ?Impact of Concrete Coefficient of Thermal Expansion 
Overestimation on the MEPDG Rigid Pavement Design Models and 
Implementation.? FHWA position paper, U.S. Department of Transportation, Federal 
Highway Administration; November. 
Flintsch, G., A. Loulizi, S.D. Diefenderfer, B.K.Diefenderfer and K.A. Gala (2008). 
?Asphalt Material Characterization in Support of Mechanistic-Empirical Pavement 
Design Guide Implementation in Virginia.? Transportation Research Record: 
Journal of the Transportation Research Board, No. 2057, pp. 114-125 
Fredlund, D. G., and A. Xing (1994). ?Equations for the Soil-Water Characteristic 
Curve.? Canadian Geotechnical Journal, Vol. 31, No. 4, pp. 521-532.  
Haider, S. W., N. Buch, and K. Chatti (2008). ?Evaluation of ME PDG for Rigid 
Pavements?Incorporating the State-of-the-Practice in Michigan.? Proceedings of 9th 
International Conference on Concrete Pavement, International Society for Concrete 
Pavement, San Francisco, CA, August, Vol. 1, pp. 111-133. 
Haider, S. W., N. Buch, and K. Chatti (2009). ?Simplified Approach for Quantifying 
Effect of Significant Input Variables and Designing Rigid Pavements using M-E 
PDG.? Annual  Meetings of the Transportation Research Board (CD-ROM), National 
Research Council, Washington, DC. 
Hall, K. and S. Beam (2005). ?Estimating the Sensitivity of Design Input Variables 
for Rigid Pavement Analysis with a Mechanistic-Empirical Design Guide.? 
Transportation Research Record: Journal of the Transportation Research Board, No. 
1919, pp. 65-73.   
Hoerner, T. E., K.A. Zimmerman, K.D. Smith and L.A. Cooley (2007). Mechanistic-
Empirical Pavement Design Guide Implementation Plan. Report No. SD2005-01, 
South Dakota Department of Transportation, Pierre, SD. 
HRB (1962). Special Report 61: The AASHO Road Test. Highway Research Board, 
National Research Council, Washington, DC, pp. 291-306. 
Kampmann, R. (2008). ?Engineering Properties of Florida Concrete Mixes for 
Implementing the AASHTO Recommended Mechanistic-Empirical Rigid Pavement 
Design Guide,? M.S. Thesis, Florida State University, FL 
 
 175 
Masad, S., and D.N. Little (2004). Sensitivity Analysis of Flexible Pavement Response 
and AASHTO 2002 Design Guide for Properties of Unbound Layers. Project No. 
ICAR 504-1, International Center for Aggregates Research, Austin, TX. 
NCHRP, (2004). Guide for Mechanistic-Empirical Design of New and Rehabilitated 
Pavement Structures.Final Report for NCHRP Project 1-37A, National Cooperative 
Highway Research Program, Transportation Research Board, Washington, DC. 
Oh, J., and E.G. Fernando. (2008). Development of Thickness Design Tables Based 
on the M-E PDG. Research Report BDH10-1, Texas Transportation Institute, College 
Station, TX; June. 
OMT (2006). Pavement Design Guide. Office of Materials Technology, Maryland 
State Highway Administration, Hanover, MD. 
Papagiannakis, A.T. and E.A. Masad (2008). Pavement Design and Materials. Wiley. 
Richter, C. A. (2002). Seasonal Variations In The Moduli of Unbound Pavement Layers. PhD 
Dissertation, University of Maryland, College Park, MD, August. 
Richter, C.A., and C.W. Schwartz (2003). "Modeling Stress- and Moisture-Induced 
Variations in Pavement Layer Moduli." Transportation Research Record: Journal of 
the Transportation Research Board, No. 1860, pp. 33-40.  
Saltelli, A., Chan, K., and Scott, E.M. (2000). Sensitivity Analysis. John Wiley and 
Sons Ltd., Chichester, England. 
Schwartz, C.W. (2007). Implementation of the NCHRP 1-37A Design Guide. Volume 
1: Summary of Findings and Implementation Plan. Final Report, MDSHA Project 
No. SP0077B41, Maryland State Highway Administration, Lutherville, MD, 
February. 
Schwartz, C.W. (2009). ?Influence of Unbound Materials on Flexible Pavement 
Performance: A Comparison of the AASHTO and MEPDG Methods.? Eighth 
International Conference on the Bearing Capacity of Roads, Railways, and Airfields, 
Urbana-Champaign IL, July, p. 951 ? 959. 
Schwartz, C.W. and H. Ceylan, (2010) Sensitivity Evaluation of MEPDG 
Performance Prediction: Phase I Interim Report. NCHRP Project 1-47, National 
Cooperative Highway Research Program, Transportation Research Board, 
Washington, DC, March. 
Schwartz, C.W. and Rui Li (2010). ?Sensitivity of Predicted Flexible Pavement 
Performance to Unbound Material Hydraulic Properties.? GeoFlorida 2010: 
Advances in Analysis, Modeling, & Design. Proceedings of the GeoFlorida 2010 
Conference. West Palm Beach, FL, pp. 271-271. 
 
 176 
Schwartz, C.W., R. Li, H. Ceylan, S. Kim, and K. Gopalakrishnan (2011). ?Effect of 
PCC Strength and Stiffness Characterization on MEPDG Predicted Performance of 
JPCP Structures.? submitted to the 2011 Annual Meetings of the Transportation 
Research Board. 
Tanesi, J., M.E. Kutay, A. Abbas, R. Meininger (2007). ?Effect of Coefficient of 
Thermal Expansion Test Variability on Concrete Pavement Performance as Predicted 
by Mechanistic-Empirical Pavement Design Guide.? Transportation Research 
Record: Journal of the Transportation Research Board, No. 2020, pp. 40-44.   
Velasquez, R., K. Hoegh, I. Yut, N. Funk, G. Cochran, M.O. Marasteanu, and L. 
Khazanovich. (2009). ?Implementation of the MEPDG for New and Rehabilitated 
Pavement Structures for Design of Concrete and Asphalt Pavements in Minnesota.? 
Research Report Mn/DOT 2009-06, Universit of Minnesota, Minneapolis, MN, 
January. 
Von Quintus, H.L., and B.M. Killingsworth (1997a), Design Pamphlet for the 
Backcalculation of Pavement Layer Moduli in Support of the 1993 AASHTO Guide 
for the Design of Pavement Structures, Publication No. FHWA-RD-97-076, Federal 
Highway Administration, McLean, VA. 
Von Quintus, H.L., and B.M. Killingsworth (1997b), Design Pamphlet for the 
Determination of Design Subgrade Modulus in Support of the 1993 AASHTO Guide 
for the Design of Pavement Structures, Publication No. FHWA-RD-97-083, Federal 
Highway Administration, McLean, VA. 
Von Quintus, H.L., C.W. Schwartz, R. McCuen, and D. Andrei (2004). Experimental 
Plan for Calibration and Validation of Hot-Mix Asphalt Performance Models for Mix 
and Structural Design, Final Report, NCHRP Project 9-30, National Cooperative 
Highway Research Program, Transportation Research Board, Washington, DC. 
Yau, A. and H.L. Von Quintus (2002). Study of LTPP Laboratory Resilient Modulus 
Test Data and Response Characteristics. Publication No. FHWA-RD-02-051, Federal 
Highway Administration, McLean, VA. 
Zapata, C.E. (2010). NCHRP 9-23A: A National Catalog of Subgrade Soil-Water 
Characteristic Curve (SWCC) Default Inputs and Selected Soil Properties for Use 
with the MEPDG. Final Report, NCHRP Project 9-23A, National Cooperative 
Highway Research Program, Transportation Research Board, Washington, DC 
(submitted to NCHRP in February 2010; made available to the public as NCHRP 
Web Document 153 in September 2010). 
Zapata, C.E. and W.N. Houston (2008). Calibration and Validation of the Enhanced 
Integrated Climatic Model for Pavement Design. NCHRP Report 602, National 
Cooperative Highway Research Program, Transportation Research Board, 
Washington, DC. 
 
 177 
Zapata, C.E., Y.Y. Perera, and W.N. Houston (2009). ?Matric Suction Prediction 
Model Used in the New AASHTO ME-PDG.? Transportation Research Record: 
Journal of the Transportation Research Board, No. 2101, pp. 53-62.