National Cooperative Highway Research Program

RESEARCH RESULTS DIGEST

September 2006Number xxx

Subject Area: IIB Pavement Design

Management, and Performance

Responsible Senior Program Officer: E.T. Harrigan

Changes to the Mechanistic-Empirical Pavement Design Guide

Software Through Version 0.900, July 2006
This digest summarizes key findings from NCHRP Project 1-40D, Technical Assistance to NCHRP and NCHRP Project 1-40A: Versions 0.9
and 1.0 of the M-E Pavement Design Software, conducted by Applied Research Associates, Inc. and Arizona State University. The digest was
prepared by Michael M. Darter, Ph.D., Jag Mallela, Leslie Titus-Glover, Chetana Rao, Ph.D., Gregg Larson, Alex Gotlif, Ph.D., and Harold Von
Quintus, Applied Research Associates, Inc., Lev Khazanovich, Ph.D, University of Minnesota, and Matthew Witczak, Ph.D., Mohamed ElBasyouny, Ph.D., Sherif El-Badawy, Ph.D., Aleksander Zborowski, and Claudia Zapata, Ph.D., Arizona State University.

INTRODUCTION
This digest is a summary of the many changes made to the Mechanistic-Empirical Pavement
Design Guide software under NCHRP 1-40D between August 2005 and June 2006. As such, it
describes both corrections incorporated in version 0.8 issued in November 2005 and mainly
technical improvements and enhancements included in version 0.9 issued in July 2006.
Changes in version 0.9 include software changes in general (including changes to traffic and
other general topics), changes in the integrated climatic model, changes in flexible pavement
design and analysis, and changes in rigid pavement design and analysis. These changes are in
response to the recommendations of the NCHRP 1-40A independent review team, the NCHRP 140 panel, the general design community, various other researchers, and the Project 1-40D team
itself. Most of these changes have been tracked and summarized at www.ara-tracker.com and
the parenthetical item # is provided for each of these. Some were more technical including
definition changes and were not included in the bug tracker system.
Users of the new version 0.9 should recognize two important facts. First, the climatic database
used in the latest calibration and available for design is now substantially larger and enhanced.
Old databases downloaded with versions 0.7 and 0.8 should be scrapped. Second, design
solutions and associated files generated with versions 0.7 and 0.8 require some additional inputs
and may experience some difficulty if simply reloaded and run. It is recommended that the user
verify each input (clicking ok) prior to re-running an old file. It is especially important to review
the inputs for the unbound materials (base, subbase, subgrade) as several are new inputs. If a run
still experiences difficulty, it is recommended that the user reenter the project inputs as a new
file and then rerun. The research team recommends that problems of interest be run again with
the new version for comparison with earlier solutions.

SOFTWARE IN GENERAL INCLUDING TRAFFIC

1. (#430) Add CRCP and JPCP plots to ACC/PCC outputs: Missing plots for LTE, Punchout,
Crack width, Cracking, and Cracking Damage. Work completed.
2. (#427) Import/Export capability for unbound material grain size distribution, PI, LL: Created
import / expert capability for unbound materials so that a default grain size and LL/PI values
for soils of interest can be created and used when needed.
3. (#426) Calculate ESALs for flexible pavement: The desire of the panel was to have ESALs
calculated and available to the user but since the procedures is NOT based on ESALs, they
would only appear in intermediate files, not on the official output for a project. Calculation
of ESALs for JPCP has been included in an intermediate file. A similar calculation for
flexible pavements is needed and was completed. These are found in the intermediate files
under the project folder.
4. (#414) Include proper default values for grain size distribution and Atterberg limits for
unbound materials. Unbound materials data from several hundred LTPP sections were
obtained for all available AASHTO soil classifications. Means and standard deviations were
computed for the range of grain size distribution and Atterberg limits and other properties
and incorporated into the defaults of the Design Guide.
5. (#259) When reducing the number of years model traffic still calculated for longer period.
This effects run time negatively. It may cost an extra 20-30 seconds. This problem was
resolved.
6. (#252) Change default design period start dates for all analyses. Used September 2006 as the
start date for all new analysis.
7. (#244) Create traffic export/import capabilities. Allow the user to import/export all of the
data need for the traffic files within the interface. Work completed.
8. (#232) Multiple file select for batch mode. Currently the user has to select each file to add to
batch mode individually. Make it possible to use multiple file select to add files to batch
mode. Work completed.
9. (#231) Increase the width of batch mode file entry screen. Currently the width of the batch
file input screen is not wide enough to fully display files that are stored in long directories.
Work completed.
10. (#125) Double clicking on cone file (*.dgp) crashes. The ability to open a ME-PDG project
file by clicking on the file has never worked properly. The only current method for opening a
ME-PDG project file is to open the program first and then selecting File|Open. This problem
was fixed.
11. (#116) Incorporate NCHRP 1-39 TrafLoad files into the ME-PDG. Modify the code to allow
the direct use of the outputs of 1=39 software in the ME-PDG. Work completed.
12. (#79) Batch Mode Option of Software. Software does not allow runs in batch mode although
this option is provided. The batch mode option must be removed or activated. Work
completed.
13. (#78) Inadequate amount of climatic data for given file. If climatic file has less than 12
months of climatic the data program will crash. An improved warning message should be
provided. Also, several procedures to estimate missing data were included in software.
Work completed.
14. (#77) Base type default values. When the base type is changed some default input values are
not changed. Work completed.

15. (#76) Problems with Save-As option. On a few occasions, two input files containing the
same data can produce different outputs. This occurs when file is changed and then saved
using "Save-As" command in the file menu of the software, and then rerun. The changed
inputs may not be actually saved. This problem was corrected.
16. (#70) Summation of Axle Load Distribution Factors. The Axle Load Distribution Factors in
the traffic analysis does not appear to count in the last load level in the summation column.
Examples that have been used in other problem studies, showing 25% in the last 5 columns
cause the program to show 100%. This problem was fixed.
17. (#66) Remove single tire configuration for traffic input. The General Traffic Input screen
currently shows a Dual and Single tire pressure option. The single tire option should be
removed as the software only uses dual tire axles. Work completed.
18. (#65) Special axle configuration problems. Several problems have been uncovered with the
Special Axle Configuration portion of Version 1.0. They are:
a. A significant amount of information for the Special Vehicle Help option must be
added
b. The asphalt. Td file must be corrected to insure that the correct wheel load coordinate
locations are input into the file. The program currently allows the user to input wheel
load coordinate locations to entered. Unfortunately, the program will read all
coordinates at an x=0; y=0 location for each tire of the special vehicle.
c. The General traffic output summary information should be eliminated when the
Special Vehicle Option is selected. This output table should be replaced with the key
input traffic wheel load properties of the special vehicle.
d. There is a repetitive error message that should be eliminated by correcting the code.
e. There is a problem associated with the frequency input associated with the use of the
Special Vehicle in a Rehabilitation scenario.
All of these problems were fixed.
19. (#50) Changing pavement types after completed run. Changing the type of pavement
analysis once run has been completed causes the program to hang. (Example-Changing
CRCP to HMA Overlay of CRCP. Problem was fixed.
INTEGRATED CLIMATIC MODEL
1. Addition of more climatic data for each weather station. New set of weather station files
with up to 9 years of hourly data provided for 851 stations. It is recommended that any old
weather stations be deleted and not used with the new version. The new weather stations
should be downloaded and used with the 0.900 version of the software.
2. (#360) Review and correct Hourly Climatic Database. Version 0.800 hourly climatic data
base had some errors in the precipitation. This was reviewed and repaired in December 2005.
Unfortunately the fixed files were not widely disseminated.
3. (#276) Failure in climatic model building. When trying to build a climatic model from
weather stations around Delaware, Iowa, every combination of the six weather stations near
the site (4228' N 9121'W, elev = 1053') reported missing data for month 200101. However,
I was able to use the Dubuque, Iowa station directly. This problem was resolved and fixed.
4. (#253) Update ICM to allow use of the NCHRP 9-23 Models. Created grain size distribution
plot. Updated grain size distribution to index property correlation. Updated base moisture
model to Thornthwaite Moisture Index (TMI) method.

5. Defaults provided for all unbound materials based on measured properties from several
hundred LTPP sections across the U.S.
6. (#229) Failed ICM stability check. Other project files created in the past run fine. The
current one (an Unbonded PCC Overlay) does not (I uploaded it to you). After pressing Run
Analysis, the Traffic module completes, then the Climatic module tries to load but stops with
error "Failed ICM stability check". Problem fixed.
7. (#122) ICM crashes with file and directory names of over 80 characters. ICM crashes with
file and directory names of over 80 characters. Problem fixed.
8. (#74) The Enhanced Integrated Climatic Model (EICM) has been modified and enhanced
based on the results of the NCHRP 9-23 project. See under Flexible Pavements.
FLEXIBLE PAVEMENTS
The following changes were made to the software to enhance the design of flexible pavements:
1. (#413) Granular layer placed over stabilized base crashes for flexible pavement. Modulus for
granular layer was not being output in _space.dat. Problem fixed. This makes it possible to
include for example a lime stabilized layer below an unbound granular base/subbase layer.
2. (#74) The Enhanced Integrated Climatic Model (EICM) has been modified and enhanced
based on the results of the NCHRP 9-23 project. Several models have been revised and
incorporated into the M-E PDG code. These models include the following:
a. New Suction Models
b. New ASU - TMI Models
c. New SWCC Models
d. Moisture Content Models with p200-w%
e. Revise Compaction Models
f. Revised Specific Gravity Models
g. New Saturated Hydraulic Conductivity (ksat) Model
3. The current program runs in batch mode. This tool is a great for research as well as design as
it allows the user to load several DG files and run them in a batch mode.
4. (#124) The analysis period (design life) is increased from 25 to 50 years.
5. The run time of the program is decreased.
6. The calibration models in the settings screens are updated.
7. The national calibration factors of the distress models are updated based on the
enhancements done to the models. In addition these calibration factors are based on the most
up to date database. Additional information provided under Re-Calibration of Flexible
Pavements.
The following changes were made in the software for the design of new and rehabilitated flexible
pavements:

1. (#56) Reflective Cracking was not reported correctly to the output summary sheet. Problem
was corrected and the reflective cracking is now reported correctly under the column with the
heading Reflective Cracking from the Existing Layers.
2. The methodology used to report area cracking in the HMA rehabilitation design is based on
Reflective Cracking reflected from the existing pavement through the overlay to the surface
of the pavement in addition to the amount of fatigue cracking occurring in the new layer.
The way the results should be interpreted is by looking at the time the cracking reached the
surface not by the amount of cracking at the end of the design life.
3. (#57) In the distress summary output sheet, for the Rehab Analysis, the Alligator Cracking
output column name is changed to Alligator Cracking in New Overlay.
4. (#58) A new output column named Total Cracking at Surface is added to the output. This
column represents the sum of the Alligator Cracking in New Overlay and the Reflective
Cracking from the Existing Layers.
5. (#99) In some cases of AC over AC Rehab analysis the software does not accept more than
one AC layer. Problem was corrected and the program now accepts up to three new AC
layers over the existing flexible pavement.
6. The software wrongly allows the user to input any material over the existing layers. Problem
was corrected and the software is now allowing the user to input either AC materials or base
courses (unbound or CTB).
7. (#54) In Level 1 Rehab analysis if the user did not input the FWD modulus the software
crashes. Problem was corrected by incorporating a warning message to warn the user to input
the FWD modulus or the software will crash.
8. (#64) In Level 1 and 2 Rehab analyses if the user did not input the actual rutting and cracking
for the existing pavement the software crashes or yields unreasonable results. Problem was
corrected by incorporating a warning message to warn the user to input the actual rutting and
cracking for the existing pavement or the software will crash or yield unreasonable results.
9. (#55) The damage function in level 2 rehab yields incompatible with Level 3. Problem was
corrected.
10. The input summary worksheet in the output file for the Rehab analyses Levels 1, 2, and 3 is
missing a lot of information. Problem was corrected.
11. Use of any new or overlay AC sublayers with thicknesses greater than 9-in may lead to
sublayering yielding AC thicknesses greater than the original input thickness of the layer.
Problem was corrected.
12. (#158) The program crashes if the user has a limited access account. Problem was corrected.
The following changes and corrections were made in the treatment of E* in the design of flexible
pavements:
1. (#67) An error message usually appears when using Level 1 E* characterization in HMA
analysis. Problem was corrected.

2. Several Level 1 E* input data for AC mixtures causes program crashes. Problem was
corrected. The seed values for the optimization process used for the development of the E*
master curve coefficients (Time-Temperature Superposition Optimization) were reviewed
and updated based on historical database in order to prevent the program from crashing.

3. (#72) In Level 1 input analysis for HMA, inputting E* values at small range of temperatures
may result in incorrect master curve which may cause program to crash or yield very
incorrect results. Problem was corrected. Program restricted the E* input values to be at
least at three temperatures in a wide range (Temperature > 125 oF, Temperature between 60
and 90 oF, and Temperature < 45 oF).

4. In level 1 input analysis for HMA, limits were imposed on the Temperature for the G*
testing to provide a reasonable range of temperatures. The G* temperature is cannot be less
than 40 oF and not more than 300 oF.
The following errors that led to discontinuities in the calculation of rutting in the flexible
pavement design were corrected:
1. (#68) A discontinuity occurs in the AC rut depth with very small changes in AC layer
thickness due to the sublayering scheme used in the software. This discontinuity error may be
as high as 20 to 25%. Problem was corrected and the error was reduced to less than 10%.
2. (# 68) A discontinuity occurs in the subgrade rut depth with small changes in the AC layer
thickness. Problem was corrected the error is totally eliminated.
The following improvements and corrections in the treatment of unbound materials were made:
1. The typical default values and ranges for the unbound materials resilient modulus at optimum
moisture condition have some wrong values. Problem was corrected. New guidelines
provided with program for flexible pavements.
2. The software mistakenly uses the default values for the unbound materials resilient modulus
at optimum moisture condition when the user chooses Representative Value (Design Value).
Problem was corrected and a new set of typical default values and ranges for the
Representative Value (Design Value) for the unbound materials resilient modulus has been
incorporated.
3. (#157) The software crash when running an analysis with unbound materials Level 2
seasonal modulus. Problem was corrected.
The following improvements and corrections in the treatment of thermal fracture were made:
1. (#71) In Level 3 predictive system for thermal fracture; as Vbeff (effective bitumen content)
increases, prediction if thermal cracking increases. Problem was corrected and the level 3
models for the creep compliance and the tensile strength prediction were modified based on a
larger database. Separate creep compliance predictive equations were developed for every
test temperature (0, -10, -20oC) to replace one, universal equation for all temperatures used in

the old version of the Level 3 analysis. These new models led to thermal fracture predictions
that follow the engineering reasonableness.
The new models are as follows:
Tensile Strength at -10oC
Revised Level 3 Prediction Model for Combined Old and New Data
St = 4976.34 -42.49*Va -2.73*Va2 -80.61*VFA +0.465*VFA2 +174.35*log(Pen77) -1,217.54*log(ARTFO)

MEASURED VS. PREDICTED TENSILE STRENGTH

REVISED OLD PREDICTION MODEL FOR COMBINED DATA

Predicted Tensile Strength [psi]

800

700

600

y = 0.4668x + 229.19
R2 = 0.4668
Se/Sy = 0.730
n = 42

500

400

300

200

100
100

200

300

400

500

600

Measured Tensile Strength [psi]

700

800

IDT Creep Compliance

Separate Prediction Models For Every Temperature
"m" as a function of Va and Pen77
log(D 1 )

= -11.92540+1.52206*log(Va)+4.49876*log(VFA)-3.81320*log(A RTFO )

-20C

log(D 1 )

-10C

= -10.76560+1.51960*log(Va)+3.49983*log(VFA)-2.99870*log(A RTFO )

log(D 1 )

0C

= - 9.80627+1.50845*log(Va)+2.99000*log(VFA)-2.90157*log(ARTFO )

m -20C = -1.75987+1.78187*Va 0.02030 +0.00089*Pen77 0.96870

m -10C = -1.82690+1.94218*Va 0.01600 +0.00098*Pen77 0.96857
m

0C

D(t) = D 1 *t m

= -2.41043+2.59093*Va 0.01547 +0.00199*Pen77 0.97247

MEASURED VS. PREDICTED CREEP COMPLIANCE

SEPARATE MODELS FOR EVERY TEMPERATURE
"m" as a function of Va and Pen77

Predicted D(t) [1/psi]

1.0E-04
All Temps
y = 1.0059x 0.9983
R 2 = 0.8034
R 2 adj = 0.7998
Se/Sy = 0.449
n = 939

1.0E-05

1.0E-06

1.0E-07
1.0E-07

1.0E-06

1.0E-05

1.0E-04

Measured D(t) [1/psi]

MEASURED VS. PREDICTED CREEP COMPLIANCE

SEPARATE MODELS FOR EVERY TEMPERATURE
"m" as a function of Va and Pen77

Predicted D(t) [1/psi]

1.0E-04

1.0E-05

1.0E-06

Temp = -20C
y = 0.2583x 0.9101
R 2 = 0.5637
R 2 adj = 0.6204
Se/Sy = 0.624
n = 313
p=9

1.0E-07
1.0E-07

Temp = -10C
y = 0.2869x 0.9095
R 2 = 0.6397
R 2 adj = 0.5775
Se/Sy = 0.659
n = 313
p=9

Temp = 0C
y = 0.9468x 0.9913
R 2 = 0.7768
R 2 adj = 0.7737
Se/Sy = 0.482
n = 313
p=9

1.0E-06

1.0E-05
Measured D(t) [1/psi]

1.0E-04

The following issues related to special axle configurations were resolved:

1. (#65-2) Asphalt.td file does not read the wheel load coordinates inputted by the user. Instead
it sets all the coordinates to zeros. Problem was corrected.
2. (#65-3) The output summary for the special axle configuration shows wrong data belongs to
the general traffic. Problem was corrected.
3. (#65-4) An error message appears when performing an analysis using the special axle
configuration. Problem was corrected.
4. (#65-5) In Rehab analysis, the software requires the user to input the frequency for the cases
of AC, granular base and subgrade layers. However the frequency should be a required input
only for the AC layers. Problem was corrected.
5. (#65-5) The program does not save the frequency values inputted by the user. In other words,
once the user inputs the frequency and clicks OK, then comes back to the same screen
again, he/she will find that the previously inputted frequency values disappeared. Problem
was corrected.
The following traffic related issues in flexible pavement design were addressed:
1. (#60) Potentially significant problem may occur with the use of the current Representative
Load Analysis Procedure especially when evaluating overloaded vehicular traffic scenarios.
Problem was corrected and a fixed tire load of 4500 lb is used in as a representative load
instead of the variable 95 percentile. This new procedure eliminated the potential problems in
the distress prediction that might occur with the use of the old procedure.
2. (#70) Axle load distribution factors in traffic analysis does not count in the last load level in
summation (Total) column. Problem was corrected.
3. (#66) General traffic input screen shows Dual and Single tire pressure options. However, it
should only show the Dual tire pressure option. Problem was corrected.
The following items summarize the major recalibration of the flexible pavement distress models
that was carried out:
1. A major effort was expended to recalibrate all of the flexible pavement models including:
bottom up fatigue cracking, top down fatigue cracking, permanent deformation, transverse
cracking, and IRI. The final newly calibrated models had (1) a lower model error, (2)
reasonable sensitivity to changes in inputs, and (3) improved reliability than those originally
developed under NCHRP 1-37A. Major steps in this effort included the following.
a. Improvement of the database.
i. Updating all existing sections with 4-5 additional years of performance data,
traffic data, materials data, climatic data, and rehabilitation data.
ii. Incorporation of the weather stations that included 9 years of historic hourly
data.

b.
c.
d.
e.

iii. Establishment of the proper input subgrade resilient modulus through an

iterative process that included backcalculation of in situ moduli for the
sections.
Re-establish the model coefficients for all of the models using the expanded database.
Model coefficients were selected that minimized the residual error of prediction.
Re-established the reliability model coefficients for all of the models using the
expanded database.
Conduct of limited sensitivity analyses to validate the software and changes to the
software.
Documentation of work accomplished (to be published).

2. Some of the calibration curves with data plotted on them along with statistics are shown
below.
Bottom-Up Fatigue (alligator) Cracking Calibration

Alligator Cracking (% of Total Lane Area)

100
Se = 5.01%
Se/Sy = 0.815
N = 405
R2 = 0.275

90
80
70
60
50
40
30
20
10
0
-4

-3

-2

-1

Log Damage (%)

SeFCBottom = 1.13+13/(1+e7.57-15.5*logD)
New Fatigue cracking calibration factors are
Bf1 = 0.00432 / 0.571 = 0.007566
Bf2 and Bf3 stay the same.

10

Top-Down Fatigue Cracking Calibration

7000
R2 = 0.544
Se = 582.8 ft /mile
Se/Sy = 0.688
N = 312

6000

Predicted Cracking (ft / mile)

5000

4000

3000

2000

1000

0
0

1000

2000

3000

4000

5000

6000

7000

Measured Cracking (ft / mile)

10000

Longitudinal Cracking (ft/mile)

9000
8000
7000
6000
5000
4000
3000
2000
1000
0
-5

-4

-3

-2

-1

log Damage (%)

Measured Longitudinal Cracking

SeFCtop = 165.68*Log D + 542.53

Predicted Top-Down Cracking

Damage in % and limit Se to 700 ft/mile

11

Permanent Deformation Calibration

Asphalt Layer - Rutting
1

Predicted AC Rutting (in)

0.9
R2 = 0.64
N = 334
Se = 0.045
Se/Sy = 0.713

0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Average Estimated Measured AC Rutting (in)

Predicted vs Ave. Estimated Measured AC Rutting

Equality Line

Granular Base - Rutting

Predicted GB Rutting (in)

0.5

0.4
R2 = 0.785
N = 334
Se = 0.026
Se/Sy = 0.502

0.3

0.2

0.1

0
0

0.1

0.2

0.3

0.4

Average Estimated Measured GB Rutting (in)

Predicted vs Est. Measured GB Rutting

12

Equality Line

0.5

Subgrade - Rutting
0.6

Predicted SG Rutting (in)

0.5
0.4
0.3
R2 = 0.708
N = 334
Se = 0.045
Se/Sy = 0.576

0.2
0.1
0
0

0.1

0.2

0.3

0.4

0.5

0.6

Average Estimated Measured SG Rutting (in)

Predicted vs Est. Measured SG Rutting

Equality Line

Total Pavement - Rutting

1.2

Predicted Total Rutting (in)

R2 = 0.577
N = 334
Se = 0.107
Se/Sy = 0.818

0.8

0.6

0.4

0.2

0
0

0.2

0.4

0.6

0.8

1.2

Average Measured Total Rutting (in)

Predicted vs Measured Total Rutting

Equality Line

SeRDAC = 0 .24 ACrut^0.8026

SeRDGB = 0.1477 GBrut^0.6711
SeRDSG = 0.1235 SGrut^0.5012

limit to a max value of 0.1

limit to a max value of 0.055
limit to a max value of 0.065

New Rutting calibration factors are

Br1 = -3.35412, Br2 and Br3 stay the same.
BrGB = 2.03, BrSG = 1.67

13

3. Recalibration of the thermal fracture models.

Modification of the thermal fracture predictive equations was followed by the necessary process
of recalibration of the level 3 thermal cracking model. Since there were no new sections
available, the 32 sections from the original calibration developed under NCHRP 1-37A were
used for the recalibration. The final newly calibrated level 3 model had, as an overall, much
lower residual error of prediction compared to the old model, and what is more important led to
thermal fracture predictions that follow the engineering reasonableness.
Comparison of predicted versus observed thermal cracking for newly calibrated level 3 model, as
well as, comparison of standard deviation for new and old model is presented below.
New Thermal Fracture Level 3 calibration factor and a corresponding Standard Deviation is:
Bt3 = 6.0
Std.Dev (Thermal) = 0.0869 * Thermal + 453.98
TCMODEL CALIBRATION
DG2002 Level 3
Calibration Factor = 6.0
2500

Predicted Cracking [ft/mile]

2000

1500
y = 0.2279x + 1330.8
R2 = 0.064

1000

500

0
0

500

1000

1500

2000

2500

Observed Cracking [ft/mile]

Level 3 - Old Vs New Calibration

1,600.0
1390

1,200.0
1050

1,000.0

400.0
200.0

493

747

Old Calib
662

600.0

578

709

800.0

369

Std. Deviation

1,400.0

0.0
500

1,000
1,500
2,000
Thermal Cracking [ft/mile]

14

New Calib

Since no new sections were available when compared to the original calibration of the thermal
fracture models for level 1 and level 2 analysis, the main objective of the new process was to
calibrate the thermal cracking models with expanded (9 years of hourly data) EICM climatic
database.
The original calibration of the thermal fracture models developed under the NCHRP 1-37A was
done using the real climatic data corresponding to the time of exploitation of the test sections (in
some cases period of 20 years). However, this process was done without the use of EICM and its
database (due to the limited amount of data) that is the integral part of the MEPDG software. It is
recommended that level 1 and 2 factors from the original calibration task should be used in the
forensic studies when true climatic data are available. However, it was more reasonable to
calibrate the models using the same climatic database that will be further used in the
performance prediction. It is then recommended that whenever the Enhanced Integrated
Climatic Model is used to predict the thermal fracture performance of the pavement, the new
calibration factors should be used. It was observed, that recalibration process resulted with
higher standard deviations when compared to the original calibration which is attributed to the
fact that EICM data were not as precise as true climatic data used in the original calibration.
Comparison of predicted versus observed thermal cracking for recalibrated level 1 and 2 models,
as well as, comparison of standard deviation for new and old models is presented below.

New Thermal Fracture Level 1 and level 2 calibration factors and corresponding Standard
Deviations are:
Bt1 = 1.0
Std.Dev (Thermal) = -0.0899 * Thermal + 636.97
Bt2 = 0.5
Std.Dev (Thermal) = -0.0169 * Thermal + 654.86

Level 1 - Old Vs New Calibration

1,600.0
1,200.0
1,000.0
Old Calib

800.0

0.0

500

1,000
1,500
Thermal Cracking [ft/mile]

15

457

505

382

258

200.0

547

592

400.0

502

New Calib

600.0

134

Std. Deviation

1,400.0

2,000

TCMODEL CALIBRATION
DG2002 Level 1
Calibration Factor = 1.0
2500

Predicted Cracking [ft/mile]

2000

1500

1000
y = 0.7001x + 73.142
R2 = 0.405
500

0
0

500

1000

1500

2000

2500

2000

2500

Observed Cracking [ft/mile]

TCMODEL CALIBRATION
DG2002 Level 2
Calibration Factor = 0.5
2500

Predicted Cracking [ft/mile]

2000

1500

1000

y = 0.5202x + 314.64
R2 = 0.2219

500

0
0

500

1000

1500

Observed Cracking [ft/mile]

16

Level 2 - Old Vs New Calibration

1,600.0
1,200.0
1,000.0
Old Calib

200.0
0.0

500

1,000
1,500
Thermal Cracking [ft/mile]

New Calib
621

689

520

352

400.0

638

646

600.0

630

800.0

183

Std. Deviation

1,400.0

2,000

Recalibration permitted the following updating of the flexible pavement IRI models:
1. A large amount of additional flexible pavement distress and IRI data has become available
since the original model was developed in 1999. This was extracted from LTPP and used in
the analysis.
2. Two models were developed:
a. New flexible pavements and overlaid flexible pavements combined.
b. HMA overlays of jointed concrete pavements
3. The updated model included the key M-E predicted distress types as follows:
a. Fatigue cracking: bottom up and top down, all severities.
b. Permanent deformation, mean values.
c. Transverse cracking (all severities).
4. The updated models also included site factors that contain: mean annual air temperature,
freezing index, annual precipitation, percentages of silt, clay, and sand in the subgrade soil
and age of the section.
5. The re-calibrated IRI models were an improvement on the existing models in the following
ways:
a. Functional form is more similar to IRI for rigid pavements
b. Removal of problematic non-predicted distress models
c. Based a large number of additional LTPP sections.
d. Based on sections with additional distress and IRI development over time (19982005).
e. Directly considers the effect of each ME predicted distress on IRI:
i. Permanent deformation
ii. Wheelpath cracking: Alligator and Longitudinal fatigue
iii. Transverse cracking
f. Considers effect of site conditions
i. Subgrade: percent fine sand, silt, clay, & PI
ii. Climate: Freezing index, precipitation
iii. Age: represents cycles of hot/cold, wet/dry, freeze/thaw
17

g. Sensitivity shows reasonableness of effects of distress on IRI.

h. Far larger database of sections
i. Flexible pavement IRI: 1950 vs 1578
ii. Five years of additional distress development (better determine effect on IRI)
iii. Standard error of prediction slightly lower
iv. 18.9 vs 24.5 in/mi
6. Variance models were developed for use in reliability design with IRI. They produce
realistic results and are similar to those developed for IRI of rigid pavements.
7. Sensitivity analyses were run to verify the reasonableness of the IRI models.
8. Plot showing the predicted versus measured IRI for new flexible pavements and HMA
overlaid flexible pavements are shown below.
200
N = 1926
2
R = 56 percent
SEE = 18.9 in/mi

Predcited IRI, in/mi

150

100

50

0
0

50

100

150

200

Measured IRI, in/mi

Plot of measured versus predicted IRI for new HMA and HMA over HMA flexible pavements.

18

150

N = 244
2
R = 51 percent
SEE = 9.6 in/mi

Predicted IRI, in/mi

125

100

75

50

25

0
0

25

50

75

100

125

150

Measured IRI, in/mi

Plot of measured versus predicted IRI for HMA overlaid PCC pavements.

19

Recommended Resilient Modulus Input (at optimum density and moisture) for Flexible
Pavements and Rehabilitation of Flexible Pavements.
[Do not use these resilient modulus values for compacted base or subbase course. Use appropriate table
for base/subbase course resilient modulus]

AASHTO
Soil Class

A-1-a
A-1-b
A-3
A-2-4
A-2-5
A-2-6
A-2-7
A-4
A-5
A-6
A-7-5
A-7-6

Mean LTPP NDT

Moduli
(Std. dev.)
(psi)*
35,397
(20,115)
33,270
(19,292)
35,413
(19,652)
33,463
(16,700)
NA
30,220
(20,132)
NA
33,124
(17,709)
7,337
25,969
(5,937)
24,644
11,360
(8,106)

Mean LTPP NDT

Moduli for
Subgrades
(Std. dev.)
(psi)*
46,764
(15,950)
39,368
(20,489)
32,047
(14,251)
38,821
(16,735)

Recommended
Input Optimum
Resilient
Modulus
(psi)**
36,000
(25,203)
31,000
(22,869)
28,500
(12,849)
42,500
(25,281)

35,775
(18,294)

44,500
(32,381)

30,423
(14,143)
7,337
24,665
(5,518)
24,644
12,638
(8,758)

50,000
(27,316)
20,100
40,500
(8,429)
38,000
17,500
(12,898)

*Information provided in this table was obtained from the LTPP database.
**Information obtained after correcting the NDT values by environmental conditions, using the
Environmental factors given by the Design Guide.
These results are based on about 166 soils used in the calibration of the Design Guide for flexible
pavements.

20

RIGID PAVEMENTS
The following modifications were made in response to items identified in the Bug-Tracker
database (www.ara-tracker.com):
1. (#434) Concrete pavement restoration (CPR) model for slab cracking: The CPR algorithm
for damage and slab cracking (after CPR) was found to be deficient in that it utilizes percent
cracked/replaced slabs to obtain an estimate of overall accumulated fatigue damage. This
damage was neither top or bottom fatigue but a composite of both that was not very accurate.
Top down and bottom up damage needs to be directly correlated with measured cracking and
an improved methodology installed. This upgrade was completed and verified. The CPR
cracking model is now more theoretically correct and provides a better projection of future
cracking after diamond grinding and other repairs.
2. (#433) Widened slab computation deficiency: A deficiency in the fatigue damage algorithm
was identified for JPCP. Program does not provide for proper damage calculation for
widening less than 24-in. Problem was fixed so that user may enter 3-in, 6-in, 12-in etc. up to
24-in and the program will calculate fatigue damage properly. After approximately 15-in the
critical fatigue location shifts from the outer slab edge of the widened portion to the inner
slab edge near the lane to lane longitudinal joint. The algorithm makes this switch properly.
3. (#428) HMA overlay of JPCP and CRCP includes several major deficiencies that require
updating: The modeling of HMA over JPCP and CRCP in the existing version has serious
deficiencies in how the overlay and concrete slab and base course are transformed into an
equivalent section for stress calculation purposes. Major modifications are required for both
HMA over JPCP and HMA over CRCP to make this a more effective overlay design
procedure. [note: these modification have not been completed in the 0.9 version yet. It is
recommended to not use this overlay design procedure until this is completed in late July
2006]
4. (#425) Level 1 inputs for PCC incorrect for HMA over JPCP or HMA over CRCP: Program
is using the xx year (example 30 year) flexural strength instead of the input (long term)
existing strength. Corrected.
5. (#393) CRCP shoulder load transfer factor (Js) reporting incorrectly. Problem was corrected.
6. (#386) When designing AC overlay of JPCP, the design screen displays the AC overlay
instead of the PCC thickness. Problem was corrected.
7. (#385) CRCP crashes on exit for analysis more than 500 months. Problem was corrected.
8. (#382) Open to traffic date and construction dates that differ by more than 11 months crashes
for JPCP. Problem was corrected.
9. (#364) Remove drainage inputs from interface. Use of TMI model makes entry of drainage
path and infiltration unnecessary.
10. (#362) Erosion models for CRCP were revised to show more impact on performance as per
review comments. Major revision made to models, incorporated into program. Tests
indicated further bugs that were fixed.
11. (#359) JPCP Cracking model stopped at 40 Years. Problem was corrected. Design life can
be over 80 years.
12. (#345) Addition of algorithm to compute equivalent temperature gradient for bonded
slab/base conditions. Bonded herein means full contact friction, no slippage between
layers at the interface. The existing JPCP model includes a procedure for computing the

21

equivalent linear temperature gradient through the concrete slab given a non-linear gradient.
The procedure assumes that the slab and base are unbonded. A similar procedure is needed to
compute an equivalent linear temperature gradient through the concrete slab for bonded (no
slippage, full contact friction) conditions. The significance of the bonding condition has been
determined to be very high. This change required a change to the linearization and damage
calculation algorithms of the M-E PDG cracking model. Note: it was discovered that an
error existed in the 2004 Version 0.700 of the Design Guide wherein bonding of slab and
base only lasted for 12 months, regardless of the input by the user. This was fixed under
#343 and contact friction or bonding now works properly.
13. (#344) Modification of JPCP computational algorithm that includes "design periods" due to
various problems. JPCP computational algorithm includes periods of time (1 yr, 1 yr, 1 yr, 2
yrs, 3 yrs, 5 yrs, etc.) over which fatigue damage is being computed throughout the design
analysis period. These were introduced to reduce the run time. An error has recently been
identified in conjunction with the "bonding" slab/base algorithm and extrapolation
procedures over the design periods that is not easily solved. The design period procedure
may not have as much effect on computer run time as previously thought and it was modified
so that the damage accumulation will be done month by month and year by year using exact
strength, modulus, contact friction, k-values etc. This approach will also help the
implementation of other features into the JPCP design procedure. CRCP already works in
this way.
14. (#343) Contact friction between concrete slab and base course (commonly called bonding)
has an error in computational algorithm. User has option of selecting the number of months
over which the slab and base will remained "bonded" (this is really contact friction). Some
sensitivity analysis indicated this input to have small effect. Investigation showed that there
is an error in the coding that is causing this result. Subsequent runs and the calibration
process indicated that most slabs and base courses show full contract friction over their lives
than previously believed. Design input recommendations were revised.
15. (#241) Difference in CRCP results between program estimated crack spacing and user input
crack spacing. There was a difference in CRCP output results when the program uses the
internal model to predict crack spacing and when the user inputs the exact same crack
spacing. For example, if the program model predicts 31.7 in, and this is input and the
program re-run, a different output may result. Problem fixed.
16. (#240) Permanent curl/warp input needs further examination to determine improved
estimation procedures. Currently, a -10F is recommended for design. This is inadequate as
the permanent curl/warp is known to depend on several key factors. Develop procedures to
estimate the permanent curl/warp input for JPCP and CRCP separately. Through the
calibration process, attempt to identify values or relationships that will minimize error of
prediction for all JPCP and CRCP distress models. There was not sufficient time and
resources to solve this problem at this time. It is recommended that it be addressed in the
next program version.
17. (#239) Base erosion index does not appear to have adequate sensitivity to performance.
Evaluated the impact of the base erosion index on performance of CRCP and JPCP and
determined that it needed improvement. New erosion models were derived for each type of
base course and subgrade type to that the loss of support along the edge is computed
automatically, not dependent on user input of erosion factor.

22

18. (#238) Add CRCP design criteria to Output Reliability Screen: Crack Width, Crack LTE,
Crack Spacing. CRCP design criteria include crack width, crack spacing, and crack LTE in
addition to punchouts. These were added to the Reliability output screen to emphasize that
they are just as important as punchouts and IRI (even though there is not a reliability level
associated with them).
a. Crack Width < 0.02 in
b. Crack LTE > 95%
c. Crack spacing 3 to 6 ft
The following modifications were completed, but were not included in the Bug Tracker database:
1. The definition of a CRCP punchout was revised based on review comments to include only
medium and high severity punchouts and y-cracks. Distress maps for all CRCP sections
were reviewed and the correct number of punchouts (including y-cracks) was included for
each time frame. This change is reflected in the design input recommendations for critical
levels for design.
2. The AC/JPCP or AC/CRCP overlay design procedure was found to contain various technical
deficiencies. One problem was that the 2004 version did not fully consider the width of
transverse crack (after many years of aging), their load transfer efficiency (which may have
deteriorated), and the extent of erosion along the slab edge that exists in the field at the time
of placement of a new overlay. In addition, procedures to calculate the equivalent slab
thickness (where the overlay is combined with the CRCP slab) with proper full friction
included errors. These are being fixed in the software and this type of overlay tested to
ensure reasonableness. This fix is not yet in 0.900 version but will be in the next version.
3. A major effort was expended to recalibrate all of the JPCP, CRCP, and rehabilitation distress
models including: joint faulting (new and rehab), slab top down and bottom up cracking (new
and rehab), punchout (new and rehab), and crack spacing. The final newly calibrated models
had (1) a lower model error even with an expanded data set, (2) reasonable sensitivity to
changes in inputs, (3) more robust due to an expanded data set, and (4) improved reliability
than those originally developed under NCHRP 1-37A. Major steps in this effort included the
following.
a. Improvement of the database.
i. Updating all existing sections with 4-5 additional years of performance data,
traffic data, materials data, climatic data, and rehabilitation data.
ii. Identification of additional LTPP and other sections for inclusion in the
calibration.
iii. Incorporation of the weather stations that included 9 years of historic hourly
data.
iv. Establishment of the proper input subgrade resilient modulus through an
iterative process that included backcalculation of in situ moduli for the
sections.
b. Re-established the model coefficients for all of the models using the expanded
database.
c. Re-established the reliability model coefficients for all of the models using the
expanded database.

23

d. Conducted limited sensitivity analyses to validate the software and changes to the
software.
e. Documented work accomplished (not yet completed)
f. Plots of some of the calibration results are shown below.
4. The following are plots of measured and predicted JPCP cracking distress for new JPCP,
unbonded JPCP overlays, and JPCP subjected to restoration.

100
R2 = 0.86
SEE = 5.02 percent
N = 1585

Predicted percent slabs cracked

90
80
70
60
50
40
30
20
10
0
0

10

20

30

40

50

60

70

80

90

100

Measured percent slabs cracked

Plot of measured versus predicted transverse cracking for new JPCP calibration.

24

100

R = 0.72
SEE = 3.95 percent
N = 60

Predicted percent slabs cracked

90
80
70
60
50
40
30
20
10
0
0

10

20

30

40

50

60

70

80

90

100

Measured percent slabs cracked

Plot of measured versus predicted transverse cracking for unbonded JPCP overlay calibration.
100
2

R = 0.90
N = 94
SEE = 6.5 percent

Predicted percent slabs cracked

90
80
70
60
50
40
30
20
10
0
0

20

40

60

80

100

Measured percent slabs cracked

Plot of measured versus predicted transverse cracking for restored JPCP (CPR) calibration.

25

The following are plots of measured and predicted JPCP faulting distress for new JPCP,
unbonded JPCP overlays, and JPCP subjected to restoration.
0.3
2

R = 0.62
SEE = 0.0276 in
N = 1260

Predicted mean transverse

joint faulting, in

0.25

0.2

0.15

0.1

0.05

0
0

0.05

0.1

0.15

0.2

0.25

0.3

Measured mean transverse joint faulting, in

Plot of measured versus predicted transverse joint faulting for new JPCP calibration.

26

0.30
2

R = 0.74
SEE = 0.025 in
N = 43

Predicted mean transverse

joint faulting, in

0.25

0.20

0.15

0.10

0.05

0.00
0.00

0.05

0.10

0.15

0.20

0.25

0.30

Measured mean transverse joint faulting, in

Plot of measured versus predicted transverse joint faulting for unbonded JPCP overlays
calibration.
0.14
Predicted transverse joint faulting, in

R = 0.61
N = 40
SEE = 0.02 in

0.12
0.1
0.08
0.06
0.04
0.02
0
0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

Measured transverse joint faulting, in

Plot of measured versus predicted transverse joint faulting for restored JPCP (CPR) calibration.

27

Measured punchouts, #/mile

The following are plots of measured and predicted CRCP punchout distress for new CRCP.

60

y = 0.9993x
R2 = 0.7395

50
40
30
20
10
0
0

10

20

30

40

50

60

Predicted punchouts, #/mile

Plot of predicted versus measured punchouts for new CRCP calibration.
The following improvements were made to the input guidelines:
1. Input resilient modulus, Mr, for SUBGRADE. The calibration process required a complete
re-establishment of the proper input subgrade Mr at optimum moisture content that would
ultimately provide an in situ modulus that matched the FWD backcalculated value at each
test section site. The modification of the ICM and subdrainage models resulted in a change
of in situ Mr resulting in this required work. As a result of this, the JPCP and CRCP section
provide a large database from which to obtain typical values for Mr for each AASHTO soil
class. It was discovered that the in situ Mr varied greatly for any given LTPP classification
of subgrade soils and thus the selection of typical values for use as defaults or recommended
values for Level 3 was difficult. The values obtained were averages from all of the data and
represent the best Level 3 source of information available. Separate recommendations for
rigid pavements were made based on these results. The table is included.
2. Input resilient modulus, Mr, for BASE/SUBBASE. The values recommended in the 2004
original version of the design guide are recommended for the base and subbase.
3. Updated grain size distribution, Atterberg limits, and other soil properties based on extensive
data from LTPP test sites across the US.
4. Contact friction between slab and base (formerly called bonding). This value was found to
be far more significant than before due to an error in the software. Guidelines on how many
months of full friction to select for different base courses were based on best fit to match
cracking in the field. Results are as follows which are recommended for input:
o Asphalt stabilized base: range of 60 to 360 months with an average of 229 months.
o Cement stabilized or lean concrete base: range of 0 to 360 months with an average of 136
months.
28

o Unbound material base: range of 0 to 360 months with an average of 245 months.
o Lime stabilized base: range of 0 to 360 months with an average of 176 months.
o Unbonded overlay (with HMA separation layer): zero months of full friction (bond) to
match cracking.
o These values were used in the calibration and are recommended for design unless specific
efforts are made for full or reduced friction at the interface.
5. Recommendations for the modulus of existing PCC slabs for use in unbonded overlay design
were updated.
6. Permanent curl/warp gradient through slab. A value of -10 F was found in the original
calibration in 2002-2004 to provide the lowest error in slab cracking prediction. This value
was used for nearly all projects of JPCP and CRCP in the 2006 calibration. However, those
projects cured with water or constructed at night required a lower value (-3 F) and those built
under harsh curing conditions (morning paving, sunshine, windy) often required greater than
-10 F, even up to -25 F. Obviously, additional research is greatly needed to more clearly
quantify this important input.
7. Coefficient of thermal expansion recommendations were upgraded to the latest LTPP data
analysis of this parameter as shown below.
PCC CTE results sorted by aggregate origin and classification
(units are in in/in per F x10-6). Mallela, J, Abbas, A., Harman, T, Rao, C, Liu, R, Darter, M
(2005) Measurement and Significance of the Coefficient of Thermal Expansion of Concrete in
Rigid Pavement Design. Transportation Research Record 1919, Transportation Research
Board, Washington, DC.
Primary Origin
Igneous (Extrusive)
Igneous (Extrusive)
Igneous (Extrusive)1
Igneous (Plutonic)
Igneous (Plutonic)1
Igneous (Plutonic)
Metamorphic
Sedimentary
Sedimentary
Sedimentary
Sedimentary
Sedimentary
Lightweight1

Primary
Aggregate Class
Andesite
Basalt
Diabase
Diabase
Gabbro
Granite
Schist
Chert
Dolomite
Limestone
Quartzite
Sandstone
Expanded shale

Average
CTE
5.3
5.2
4.6
5.2
5.3
5.8
5.6
6.6
5.8
5.4
6.2
6.1
5.7

Standard
Deviation (s)
0.5
0.7
0.5
0.5
0.6
0.6
0.5
0.8
0.8
0.7
0.7
0.8
0.5

Sample
Count (n)
23
47
4
17
4
83
17
28
124
236
69
18
3

Long-Term Pavement Performance (LTPP) test section results.

Testing conducted by Federal Highway Administration at the Turner-Fairbank Highway Research Center.
1
Results based on very limited testing.

29

8. Improved estimates of wheel base percentages were obtained from two states. They were
somewhat different than the 33, 33, and 34% assumed for short, medium, and long used in
the original design guide calibration. Further study is needed to more firmly establish the
wheel base percentages and then they will be used in a JPCP calibration.
Summary of major technical improvements in 0.9 version for rigid pavements:
1. Addition of algorithm to compute equivalent temperature gradient for bonded slab/base
conditions (unbonded only was available). This was used in calibration whenever full
contact friction was specified.
2. Modification of JPCP computational algorithm so that it now includes month by month
damage accumulation rather than accumulating over a multiyear period.
3. Fixed a bug in the algorithm for contact friction between concrete slab and base (time to debond). The full contact friction was only working for 12 month period in the 2004 version.
The term bond or bonding was changed to contact friction to better describe the amount
of slippage between the PCC slab and the base coarse. Recommendations were provided.
4. Erosion prediction for CRCP was improved.
a. Calculates loss of support along edge over time as a function of:
i. Base type and quality
ii. HMA: asphalt content
iii. CTB: Ec
iv. Granular: fines content
v. Annual precipitation
vi. Type and quality of subbase/subgrade (strength, fines)
b. Erosion calculated for 10 years but uniformly accumulated year by year with practical
cap.
5. Re-calibrations of all JPCP and CRCP models were successful. Partial sensitivity indicates
the calibrated models are reasonable. Further sensitivity analyses are needed.
6. Re-calibration of all rehabilitation models for JPCP and CRCP were successful.
7. Results confirm that original 2004 models were valid over a much wider range of design
parameters and larger number of additional sections throughout the U.S.
8. Joint spacing for skewed joints was increased by the amount of skew in a 12-ft wide slab
(normally 2-ft) to account for increased curl/warp stresses over that of a perpendicular joint.
9. Improved recommendations for subgrade resilient modulus inputs are provided.
10. Slab widening algorithm for JPCP was improved to allow smaller widening values (e.g., 3 to
24-in) with proper calculation of fatigue damage at each edge of slab.
11. Concrete pavement restoration algorithm for cracking was upgraded to more accurately
predict top down and bottom up damage and cracking.

30

Recommended Subgrade/Embankment Resilient Modulus Input (at optimum density and

moisture) for Rigid Pavements and Rehabilitation of Rigid Pavements.
[Do not use these resilient modulus values for compacted base or subbase course. Use appropriate table
for base/subbase course resilient modulus]

Subgrade
AASHTO
Soil Class
A-1-a
A-1-b
A-3
A-2-4
A-2-5
A-2-6
A-2-7
A-4
A-5
A-6
A-7-5
A-7-6

Optimum
Dry
Density
(mean, std.
dev.)*
128 pcf,
17 pcf
122,
9
NA
119,
7
NA
120,
6
NA
119,
7
NA
114,
5
103,
19
102,
8

Optimum
Moisture
Content
(mean)*

11 %
11%
NA
11%
NA
12
NA
12
NA
14
19
20

Design Guide
Input Resilient
Modulus at
Optimum
Density/Moist.
(mean, std.
dev.)**
13,228 psi,
3,083 psi
14,760,
8,817
NA
14,002,
5,730
NA
16,610,
6,620
NA
17,763,
8,889
NA
14,109,
5,935
7,984,
3,132
13,218,
322

Design Guide
Backcalculated
Output
Dynamic
k-value
(mean, std.
dev.)**
322 psi/in,
68 psi/in
335,
92
NA
256,
79
NA
289,
51
NA
270,
88
NA
211,
54
148,
32
203,
53

Recommended
Input Subgrade
Resilient
Modulus
(Opt.
Density/Moisture
Content)
18,000 psi
18,000
16,500
16,000
16,000
16,000
16,000
15,000
8,000
14,000
10,000
13,000

*Information provided in this table were obtained from the LTPP database (optimum density and
moisture).
**Information obtained from Design Guide backcalculation and from use of the Design Guide (input
subgrade resilient modulus, Mr, at optimum density and moisture).
These results are based on about 250 JPCP and CRCP pavements located across the U.S. and used in the
calibration of the Design Guide rigid pavements.
Use of resilient modulus input (at optimum density and moisture) for a project that is significantly
different than these test results and recommendations may result in erroneous model prediction.
Specifically, input of higher resilient modulus that results in significantly higher output dynamic k-values
may result in erroneous model prediction as very few LTPP sections across the country showed higher
values. Note that bedrock close to the surface would be an exception to this guideline.

31