Falling Weight Deflectometer (FWD)

Testing and Analysis Guidelines Volume II:

Supporting Documentation
Publication No. FHWA-FLH-07-001 February 2007

U.S. Department
of Transportation
Federal Highway
Administration

Federal Lands Highway Division

400 7th Street, Southwest, Room 6311
Washington, DC 20590
 FOREWORD

The Federal Lands Highway (FLH) promotes development and deployment of applied research and
technology applicable to solving transportation related issues on Federal Lands. The FLH provides
technology delivery, innovative solutions, recommended best practices, and related information and
knowledge sharing to Federal agencies, Tribal governments, and other offices within the FHWA.

The objective of this study was to produce a guide for project development and design personnel
that clearly defined Falling Weight Deflectometer (FWD) testing requirements, data analysis
approach and reporting requirements.

The study included a review of backcalculation computer programs, a review of prominent State
DOT data collection and analysis procedures, as well as a review of the current FLH FWD
testing and analysis approach. Based on the information collected and through consultation with
a FLH technical working group (TWG), recommendations for a best practice FWD testing and
analysis procedure were made. The guidelines developed will better assure that FLH collects
quality FWD data and that the data is appropriately analyzed.

The contributions and cooperation of the CFLHD personnel is gratefully acknowledged.

Butch Wlaschin, P.E., Director of Program Delivery

Federal Highway Administration
Federal Lands Highway Division

Notice

This document is disseminated under the sponsorship of the U.S. Department of Transportation
in the interest of information exchange. The U.S. Government assumes no liability for the use of
the information contained in this document. This report does not constitute a standard,
specification, or regulation.

The U.S. Government does not endorse products or manufacturers. Trademarks or

manufacturers' names appear in this report only because they are considered essential to the
objective of the document.

Quality Assurance Statement

The Federal Highway Administration (FHWA) provides high-quality information to serve

Government, industry, and the public in a manner that promotes public understanding. Standards
and policies are used to ensure and maximize the quality, objectivity, utility, and integrity of its
information. FHWA periodically reviews quality issues and adjusts its programs and processes to
ensure continuous quality improvement.
 Technical Report Documentation Page

1. Report No. 2. Government Accession No. 3. Recipient's Catalog No.

FHWA-FLH-07-001
4. Title and Subtitle 5. Report Date
February 2007
Falling Weight Deflectometer (FWD) Testing and Analysis
Guidelines Volume II: Supporting Documentation (U.S. Customary 6. Performing Organization Code
Units Version)

7. Author(s) 8. Performing Organization Report No.

P.N. Schmalzer, G.R. Rada, J.S. Miller
9. Performing Organization Name and Address 10. Work Unit No. (TRAIS)
MACTEC Engineering and Consulting, Inc.
1575 Delucchi Lane, Suite 201 11. Contract or Grant No.
Reno, NV, 89502 DTFH68-05-P-00118
12. Sponsoring Agency Name and Address 13. Type of Report and Period Covered
Federal Highway Administration Final Report, October 2005 –
Federal Lands Highway Division February 2007
400 7th Street, Southwest, Room 6311 14. Sponsoring Agency Code
Washington, DC 20590 HFL-1
15. Supplementary Notes
COTR: Mike Voth, FHWA-FLH. Technical Working Group: Chris Chang, FHWA-EFLHD, Steve
Deppmeir, FHWA-CFLHD, Gary Evans, FHWA-WFLHD. This project was funded as a FLH functional
discipline leader (FDL) initiative.
16. Abstract

This report contains supporting documentation for the Falling Weight Deflectometer (FWD) Testing and
Analysis Guidelines Volume I. This report includes a review of previous FWD data collection on eight FLH
development projects. The FWD data collection and analysis practices of nine state DOTs is also
summarized. The methodology and capability of nine backcalculation programs is discussed.

17. Key Words 18. Distribution Statement

FALLING WEIGHT DEFLECTOMETER, No restriction. This document is available to the

MODTAG, BACKCALCUALTION, public from the sponsoring agency at the website
MODULUS, SURFACE DEFLECTIONS http://www.cflhd.gov.

19. Security Classif. (of this report) 20. Security Classif. (of this page) 21. No. of Pages 22. Price
Unclassified Unclassified 37

Form DOT F 1700.7 (8-72) Reproduction of completed page authorized

 SI* (MODERN METRIC) CONVERSION FACTORS
APPROXIMATE CONVERSIONS TO SI UNITS
Symbol When You Know Multiply By To Find Symbol
LENGTH
in inches 25.4 millimeters mm
ft feet 0.305 meters m
yd yards 0.914 meters m
mi miles 1.61 kilometers km
AREA
in2 square inches 645.2 square millimeters mm2
ft2 square feet 0.093 square meters m2
yd2 square yard 0.836 square meters m2
ac acres 0.405 hectares ha
mi2 square miles 2.59 square kilometers km2
VOLUME
fl oz fluid ounces 29.57 milliliters mL
gal gallons 3.785 liters L
ft3 cubic feet 0.028 cubic meters m3
yd3 cubic yards 0.765 cubic meters m3
NOTE: volumes greater than 1000 L shall be shown in m3
MASS
oz ounces 28.35 grams g
lb pounds 0.454 kilograms kg
T short tons (2000 lb) 0.907 megagrams (or "metric ton") Mg (or "t")
TEMPERATURE (exact degrees)
°F Fahrenheit 5 (F-32)/9 Celsius °C
or (F-32)/1.8
ILLUMINATION
fc foot-candles 10.76 lux lx
fl foot-Lamberts 3.426 candela/m2 cd/m2
FORCE and PRESSURE or STRESS
lbf poundforce 4.45 newtons N
lbf/in2 poundforce per square inch 6.89 kilopascals kPa
APPROXIMATE CONVERSIONS FROM SI UNITS
Symbol When You Know Multiply By To Find Symbol
LENGTH
mm millimeters 0.039 inches in
m meters 3.28 feet ft
m meters 1.09 yards yd
km kilometers 0.621 miles mi
AREA
mm2 square millimeters 0.0016 square inches in2
m2 square meters 10.764 square feet ft2
m2 square meters 1.195 square yards yd2
ha hectares 2.47 acres ac
km2 square kilometers 0.386 square miles mi2
VOLUME
mL milliliters 0.034 fluid ounces fl oz
L liters 0.264 gallons gal
m3 cubic meters 35.314 cubic feet ft3
m3 cubic meters 1.307 cubic yards yd3
MASS
g grams 0.035 ounces oz
kg kilograms 2.202 pounds lb
Mg (or "t") megagrams (or "metric 1.103 short tons (2000 lb) T
TEMPERATURE (exact degrees)
°C Celsius 1.8C+32 Fahrenheit °F
ILLUMINATION
lx lux 0.0929 foot-candles fc
cd/m2 candela/m2 0.2919 foot-Lamberts fl
FORCE and PRESSURE or STRESS
N newtons 0.225 poundforce lbf
kPa kilopascals 0.145 poundforce per square inch lbf/in2
*SI is the symbol for the International System of Units. Appropriate rounding should be made to comply with Section 4 of ASTM E380.
(Revised March 2003)

ii
 Table of Contents

1 Introduction................................................................................................................. 1
2 FWD Data Collection and Analysis on FLH Projects ................................................ 1
2.1 Project Summaries .............................................................................................. 1
2.2 FWD Equipment ................................................................................................. 3
2.3 FWD Test Points................................................................................................. 4
2.4 Data Analysis ...................................................................................................... 5
2.5 Summary ............................................................................................................. 7
2.6 Recommendations............................................................................................... 8
3 FWD Analysis and Data Collection Procedures used by Selected State DOTs ......... 8
3.1 Analysis Procedures............................................................................................ 9
3.2 Data Collection ................................................................................................. 13
4 Review of Backcalculation Programs....................................................................... 15
4.1 Software Investigated........................................................................................ 16
4.2 File Formats and Units...................................................................................... 17
4.3 Theory and Modeling........................................................................................ 18
4.3.1 Pavement Response Models ..................................................................... 18
4.4 Program Limitations ......................................................................................... 20
4.5 Additional Features........................................................................................... 21
4.5.1 Non-Linear Materials................................................................................ 21
4.5.2 Depth to Bedrock ...................................................................................... 24
4.5.3 Asphalt Concrete Temperature Correction ............................................... 24
4.5.4 Asphalt Concrete Mid-depth Temperature Prediction.............................. 25
4.5.5 Raw Data QC Tools .................................................................................. 25
4.5.6 Load Transfer Efficiency .......................................................................... 27
4.5.7 Void Detection .......................................................................................... 28
4.5.8 Segmentation Tools .................................................................................. 28
4.6 Recommendations............................................................................................. 29
References......................................................................................................................... 33

List of Figures

Figure 1. Equation 1.......................................................................................................... 11

Figure 2. Equation 2.......................................................................................................... 22
Figure 3. Equation 3.......................................................................................................... 23
Figure 4. Equation 4.......................................................................................................... 23
Figure 5. Equation 5.......................................................................................................... 23
Figure 6. Equation 6.......................................................................................................... 24
Figure 7. Equation 7.......................................................................................................... 25
Figure 8. Equation 8.......................................................................................................... 27
Figure 9. Equation 9.......................................................................................................... 27

iii
 List of Tables

Table 1: Projects Reviewed ................................................................................................ 1

Table 2: Pavement Sections ................................................................................................ 2
Table 3: FWD Equipment and Setup .................................................................................. 3
Table 4: Test Point Summary ............................................................................................. 4
Table 5: Data Analysis Methodologies............................................................................... 5
Table 6: Analysis Theories and Software Used.................................................................. 9
Table 7: Seed values used by Nevada DOT...................................................................... 10
Table 8: Parameters used by Texas DOT ......................................................................... 11
Table 9: Base, Subbase and Stabilized Materials Parameters Used by Washington DOT11
Table 10: Subgrade Parameters by Climate Type Used by Washington DOT................. 12
Table 11: Correction Factors ............................................................................................ 12
Table 12: Equipment Type and Setup............................................................................... 13
Table 13: Test setups used by Maryland DOT ................................................................. 14
Table 14: Test Point Spacings........................................................................................... 15
Table 15: Test Point Spacings Used by Virginia DOT..................................................... 15
Table 16: Software Investigated ....................................................................................... 16
Table 17: File Formats and Units...................................................................................... 17
Table 18: Theory and Modeling ....................................................................................... 18
Table 19: Additional Features........................................................................................... 21

List of Acronyms and Symbols

AASHTO - American Association of State Highway and Transportation Officials

AC - asphalt concrete
AGDPS - AASHTO Guide for the Design of Pavement Structures, 1993
edition
C - subgrade resilient modulus correction factor
DOT - department of transportation
FAA - Federal Aviation Administration
FHWA - Federal Highway Administration
FEM - finite element model
FLH - Federal Lands Highway
FWD - falling weight deflectometer
HWD - heavy weight deflectometer
LTE - load transfer efficiency
LTPP - Long Term Pavement Performance Program
MEPDG - mechanistic-empirical pavement design guide
MET - method of equivalent thickness
Mr - Resilient Modulus
OWP - outer wheel path
PCC - portland cement concrete
PDDX - pavement deflection data exchange
RMSE - root mean squared error
SNeff - effective structural number

iv
1 INTRODUCTION

This report is Volume II of the guidelines for the collection and analysis of Falling
Weight Deflectometer (FWD) data for use on Federal Highway Administration (FHWA)
Federal Lands Highway (FLH) projects. It contains supporting documentation gathered
during the development of the guidelines.

2 FWD DATA COLLECTION AND ANALYSIS ON FLH PROJECTS

This section summarizes our findings of the FWD testing and analysis procedures used
on eight projects performed by or on behalf of FLH. This investigation was performed to
augment the information on current FLH procedures and project types gathered during
the initial project meeting and teleconference.

2.1 Project Summaries

We obtained eight reports involving FWD testing performed on behalf of Federal Lands
Highway Offices. These reports are summarized in Table 1.
Table 1: Projects Reviewed

FLH Project Test Extent Prime FWD

Region Date (miles) Contractor Contractor
Central General’s Highway (CA) 9/19/05- 15.91 Stantec Stantec
9/21/05
Forest Highway 120 (CA) 11/19/03 9.25 Kleinfelder Dynatest
Trail Ridge Road (CO) 7/17/03 6.9 CFLHD Ground
Engineering
Eastern Newfound Gap Road 5/07/02 – 15 EFLHD EFLHD
(NC) 5/14/02
Natchez Trace Parkway 3/?/042 34 ELFHD ELFHD
(MS)
Western Cascade-Warm Lake 6/18/01 – 26.8 Terracon Terracon
Road (ID) 6/19/013
North Umpqua Highway 9/14/04 14.4 NTL Pavement
(OR) Engineering Services,
& Inc.
Geosciences
McKenzie Highway (OR) 9/22/04 14.7 Terracon Terrracon

Notes: 1: Includes five spur roads

2: Report states field work performed in March 2004
3: Dates determined from FWD data files; report is dated 4/2/2004

In Table 1, the term “Prime Contractor” is used for the organization that produced the
pavement analysis report, and “FWD Contractor” is used for the organization that

1
 performed the FWD testing. In all cases the organization that performed the FWD testing
also performed the FWD data analysis.

The report for Forest Highway 120 notes that “freezing conditions were encountered
during the work.” FLH staff involved in the project noted that although there was some
snow on the ground, freezing conditions below grade were unlikely. The reported
modulus of the base layer ranged from 27 to 59 ksi, and the reported subgrade modulus
ranged from 8.6 to 15.3 ksi. These moduli ranges are typical, and do not indicate that
either layer was frozen at the time of testing.

All of the pavement sections in these projects were flexible. The reported layer
thicknesses and general pavement condition are summarized in Table 2.

Table 2: Pavement Sections

Project Thickness Condition Notes

AC Aggregate
I I I 1 Base I I

General’s 2.5 – 7.8 8

Highway
Forest 2.5 – 4.5 6 - 10.25 Good to fair with isolated fatigue, transverse
Highway 120 and block cracking
Trail Ridge 3 – 162 ?3 Chip seal may be masking pavement distress
Road in lower layers. Frost heaves in areas.
Stripping in many cores. Shallow depth to
bedrock in areas.
Newfound 3.75 – 8.5 8 Low to medium severity longitudinal
Gap Road cracking, isolated heaves and high severity
cracking
Natchez 5–7 5-8 Low to medium transverse, block and fatigue
Trace cracking. Isolated potholes and high severity
Parkway cracking
Cascade- 2.25 – 0–9 Mostly poor condition, some portions fair.
Warm Lake 5.75 Areas with frost heaves
Road
North 5.5 – 9.5 9.1 – 24.1
Umpqua
Highway
McKenzie 3.5 – 6 0 - 22 Little cracking. Asphalt cores mostly show
Highway heavy stripping.

Notes: 1: Base thickness was assumed, boring results were unsatisfactory

2: Includes material recycled with foamed-asphalt in some areas
3: Aggregate base layer could not be determined through coring; a base layer was used in data
analysis, but information on the thickness used is not included in the report.

2
 The thickness values presented in Table 2 are those used in backcalculation where
possible. In cases where the report does not state the layer structures used in
backcalcualtion, the data is determined from the boring logs or core reports. The base
thickness also includes a subbase, where one was noted. The Trail Ridge Road report
does not include any information on the base layer thickness, although a base layer was
included in backcalcualtion. FLH project personnel have stated that the base layer could
not be differentiated from the subgrade based on coring.

Determination of the thickness of the base and subbase layers was a problem on several
projects. The large range of base layer thicknesses reported for these projects is in part
due to difficulty in differentiating between the base and subgrade materials, although
construction variability may also play a part. As many FLH projects are located in
mountainous regions the subgrade may be naturally granular, or may consist of granular
fill which contributes to the difficulty of differentiating the base material from the
subgrade material.

2.2 FWD Equipment

The types of FWD equipment used on the projects reviewed along with their test setups
are summarized in Table 3.

Table 3: FWD Equipment and Setup

Project FWD Type Seating Recorded Sensor Spacing (in.)

I I I Drops (kips) I Drops (kips) I I
General’s Dynatest FWD 9 6,9,12 0,8,12,18,24,36,48,60,-12
Highway
Forest Dynatest FWD None 6,9,16 0,8,12,18,24,36,60
Highway 120
Trail Ridge JILS Trailer ?2 6,9,12 -151 0,8,12,18,24,36,60
Road
Newfound KUAB 2m 12,12,12 6,6,6,6,9,9,9, 0,-12,8,12,18,24,36
Gap Road 9,12,12,12,12,
16,16,16,16
Natchez Dynatest FWD 5,7 9,9,9 0,8,12,18,24,36,48
Trace
Parkway
Cascade- Dynatest HWD None 9,9 0,8,12,18,24,36,60
Warm Lake
Road
North KUAB 2m ?,?3 9,9 0,8,12,18,24,36,48,60
Umpqua
Highway
McKenzie Dynatest HWD None 6,9,12 0,11.8,24,36,48,60,72
Highway

3
 Notes: 1: 12 kip nominal load was increased to 15 kips during testing due to low deflections at outer
sensors.
2: One seating drop was used; the load level was not reported
3: Two seating drops were used; the load levels were not reported

Four different types of FWD equipment were used on the eight projects reviewed. A
different test setup was used on each project. Only half used seating drops. The recorded
load levels also varied, although all included at least one 9 kip load.

All FWDs tested with deflection sensors located at 0, 8, 12, 18, 24 and 36 inches. On
seven of the eight projects the FWD also mounted a sensor at 60 inches. Two of the
FWDs tested with a sensor mounted in the -12 inch position, which is generally used only
for load transfer testing on PCC pavements.

The drop sequence used in testing on Newfound Gap Road includes 19 drops at each test
location, and is a definite outlier. This drop sequence is that used by the Long Term
Pavement Performance Program (LTPP) for research-grade testing on test sections up to
1,000 feet in length, and is not normally used for project-level testing.

2.3 FWD Test Points

Test point locations are summarized in Table 4. The total number of test points was
calculated by dividing the project length (from Table 1) by the spacing and multiplying
by two. This may not exactly match the actual number of tests performed. Likewise, the
number of drops was calculated by multiplying the number of drops at each location
(including seating drops) by the number of test points.

Table 4: Test Point Summary

Project Spacing (ft) Lane Test Drops Testing

I I 1 I Position I Points I I Days I
General’s Highway 328 (100 m) ML 435 1740 3
Forest Highway 120 3281 (100 m) OWP 298 894 1
Trail Ridge Road 3282 (100 m) OWP 223 892 1
Newfound Gap Road 2,000 OWP 80 1520 4
Natchez Trace 2,640 (0.5 miles) OWP 136 680 3
Parkway
Cascade-Warm Lake 1,0001 OWP 284 568 2
Road
North Umpqua 6001 ? 247 988 1
Highway
McKenzie Highway 1,0561 (0.2 miles) ? 147 441 1

Notes:
1: Report specifies that test points in opposing lanes were staggered
2: Additional data points collected in areas with high deflections

4
 The test point spacing varied by almost an order of magnitude, from 328 feet to 2640
feet. The transverse position of the FWD was only specified in three reports; two
specified outer wheel path (OWP). The printed text of the General’s Highway report
states that the OWP was tested, however that is crossed in the version supplied to
MACTEC and the statement “No, between wheel lines” is written in by hand. According
to FLH staff, the shoulder was too narrow to allow the FWD to be positioned in the
OWP.

The report on the Trail Ridge Road project states that the test point spacing was reduced
from 100 meters to 25 or 50 meters in areas with high deflections or variable test results.
The results from these weak or variable areas were averaged along with the data collected
at fixed spacings. None of the remaining project reports indicate that testing was targeted
in specific areas.

2.4 Data Analysis

Table 5 summarizes the data analysis methods used on the selected projects.

Table 5: Data Analysis Methodologies

Highway Software Mr No. AC Temp. Mid-depth

Correction Layers Correction Temperature
I I I
factor I I I
Source I

General’s Highway AGDPS1 0.33 2 No Measure

Forest Highway 120 ELMOD ? 3 Yes BELLS
Trail Ridge Road DAPS 1.5.1 0.33 3 Yes Measure, ~ 1.5
mile interval
Newfound Gap Road DarWin 0.33 2 Yes Measure, each test
point
Natchez Trace DarWin 0.33 2 Yes Measure
Parkway
Cascade-Warm Lake Evercalc 5.0 ? 3 Yes ?
Road
North Umpqua PAVBACK 1.0 3 Yes ?
Highway
McKenzie Highway DarWin 0.33 2 ? ?

Notes:
1: Report does not specify software, only that AASHTO 1993 method was followed.

DARWin was used for data analysis on three of the projects. The report for a fourth
project, General’s Highway, states that the “AASHTO 1993 Design Guide
backcalculation analysis” method was used. While this does not rule out use of
DARWin, from the plots included in the report it seems likely that the analysis was
performed in a spreadsheet.

5
Of the remaining four projects, two were analyzed using layer-elastic backcalculation
programs (DAPS, EVERCALC), and two were analyzed using Odemark-Boussinesq
equivalent layer backcalculation programs (PAVBACK, ELMOD). In each case, the
pavement was modeled as a three layer system, comprised of an AC surface layer, an
aggregate base layer and a subgrade layer. None of the reports listed the seed modulus
values used in the analysis. The report for the Forest Highway 120 project did not
include root mean squared errors (RMSEs) or any other criteria for evaluating the quality
of the analysis.

The three reports based on DARWin all state that a subgrade resilient modulus correction
factor of 0.33 was used. The report based on PAVBACK states that “because the
program accounts for the potential of non-linear, finite subgrade depth, direct calculation
of the subgrade moduli can be performed without the need for a correction factor
recommended by the procedure outlined in the 1993 AASHTO Guide.” Of the remaining
projects, it appears that only Trail Ridge Road used a correction factor. None of the
analyses addressed seasonal variation of subgrade modulus.

The four reports that are not based on the AGDPS all include a temperature correction for
the asphalt concrete layer. For the Forest Highway 120 project, the asphalt mid-depth
temperature was estimated using the BELLS equations. For the Trail Ridge Road project
the mid-depth temperature appears to be measured based on comments in the FWD data
file, however the methodology used is unspecified. For the Cascade-Warm Lake Road
and North Umpqua Highway projects, a temperature correction was applied to the AC
layer; however the source of the pavement temperature data is unspecified.

None of the reports documented any quality control checks or processes performed on
raw FWD data.

Two of the projects included FWD analyses that were highly problematic. On the
Cascade-Warm Lake Road project, very high RMSEs were reported. The minimum
RMSE was 3.3%, and the maximum was 96.46%. Typically, an RMSE of 2% is
considered high. The report describes this situation as follows:

“When reviewing the Evercalc data, the RMS error (how well the collected
deflection match the predicted deflections) is higher than normal. In many cases,
there is very little perceivable difference between the deflections of the sixth and
seventh sensors. Stiffness and thickness of the pavement layers affect the shape
of the stress bowl and the magnitude of the stress in each individual layer.
Because the pavement section is relatively thin and not stiff, Evercalc might be
modeling the subgrade resilient modulus using the last sensor which may be too
far away from the load plate to give a result with low errors. Modeling the
pavement as a two-layer system or analyzing a narrower bowl (5- or 6- sensor
analysis) will tend to improve the error.”

Without having analyzed this data ourselves, it is difficult to speculate on the source of
the error. Certainly removing the outer sensor from the analysis is a reasonable approach.

6
As the pavement structure at this section is very thin, perhaps two or three outer sensors
should be removed. A review of the surface modulus plot would be helpful in
determining the validity of data from the outer sensors. Combining the base layer and the
subgrade layer also seems reasonable, especially in light of the difficulty in
differentiating the base layer from the subgrade layer on this project. Also, it appears that
the minimum values for the seed moduli were set too low for both the AC layer and the
base layer, which limits the ability of Evercalc to find a good fit.

On the Trail Ridge Road project the reported base and subgrade moduli are unreasonable
for the section “From Entrance to Moraine Park Museum, North on Bear Lake Road”. In
this area, the average moduli for the AC, granular base layer and subgrade are 499 ksi, 25
ksi and 999 ksi respectively. The base layer is highly variable, with a low of 1 ksi and a
high of 487 ksi. The subgrade is also highly variable, with a low of 32 ksi and a high of
7,817 ksi. Extreme high values of the subgrade modulus are associated with extreme low
values of the base modulus, and vice versa. Interestingly, the RMSEs were not extreme
for this section, averaging 0.92%, with a high of 1.94% and a low of 0.41%. The report
addresses high subgrade modulus values as follows:

“The average resilient modulus of the base material was often lower than that of
the subgrade. Many times, the modulus of the subgrade exceeded one million psi.
This is likely due to rock cuts and dense cobbles/rock fragments which compose
the subgrade throughout many of the roadway alignments. In addition, it may be
the result of the gradual contamination of the base course with fine material that
has propagated upward.”

Again, without having analyzed this data ourselves, it is difficult to speculate on the
source of the error. One technique often used in this situation is creating a bedrock layer
with a fixed stiffness, usually 1,000 ksi. This may yield more reasonable base layer
moduli. However, it is likely that the problem is highly variable sub-surface conditions
as the report stated. FWD testing is not capable of determining individual layer moduli
unless the layer structure is reasonably well known for each FWD test point.

2.5 Summary

The variation of FWD testing and analysis procedures in the eight projects reviewed
reinforces the need for standardization. In addition, many of the reports failed to
adequately document the testing and analysis procedures used. Basic information such as
deflection sensor spacing and drop sequences varied widely across projects and could
only be found by reviewing the included raw data in many cases, and could not be
determined at all in a few cases.

Three different analysis techniques were utilized: AASHTO 1993 (4 projects), Odemark-
Boussinesq equivalent layer theory (2 projects) and layered elastic theory (2 projects).
Those analyses performed using DARWin were well documented because that program
intrinsically reports its input parameters. The remaining analyses were less well
documented, and in general could not be repeated based on the information included in

7
the reports. Temperature correction of AC modulus was addressed in the four analyses
that were not based on the AGDPS. None of the analyses addressed seasonal variation of
subgrade modulus.

The pavement conditions varied from good to poor. The pavement sections were highly
variable, but generally consisted of a thin layer of AC over an aggregate base. Shallow
depth to bedrock was noted in one report. Frost heaves and areas of high distress were
noted in five of eight reports. Unreasonable analysis results were noted in two projects.
In one of these projects the poor results were attributed to weakness of the pavement
structure, in the other they were attributed to shallow depth to bedrock.

2.6 Recommendations

Above and beyond the general need for standardization, several needs were noted.

• Clear and consistent reporting of equipment setup, including deflection sensor

spacing and drop sequence (including seating drops).

• Clear and consistent reporting of analysis parameters, including layer

thicknesses, seed moduli, resilient modulus correction factors and AC
modulus temperature correction methodology.

• Quality control of raw data, including deflections, loads and temperature

measurements.

• Quality control of analysis results. Reports should include a factor such as

RMSE which indicates the goodness of fit between computed and measured
deflections. Reasonableness ranges of reported moduli should be established.
Procedures should be established for cases where the RMSE or reported
moduli are outside acceptable ranges.

• Guidelines on combining layers. On several projects, the base and subgrade

layers were difficult to differentiate, not only through analysis of FWD data
but also with other techniques such as coring and ground penetrating radar.
The guidelines should address situations when combining layers leads to
better analysis results.

3 FWD ANALYSIS AND DATA COLLECTION PROCEDURES USED BY

SELECTED STATE DOTS

This section summarizes the FWD data collection and analysis practices of nine selected
state DOTs. A wide range of practices were found. The analysis procedures in use by
the state DOTs investigated are described first, as these procedures in large part
determine the operational procedures used.

8
3.1 Analysis Procedures

Table 6 lists the theoretical basis of the analysis procedures used by the state DOTs
investigated, as well as the computer software utilized to carry out those procedures.

Table 6: Analysis Theories and Software Used

State DOT Analysis Theory Software

California Deflection N/A
Colorado AGDPS DARWin
Maryland AGDPS MODTAG
Minnesota Deflection “FWD DATA ANALYSIS”
Nevada AGDPS/Layer Elastic DARWin/MODULUS
Oregon AGDPS/Deflection DARWin
Texas Layer Elastic MODULUS
Virginia AGDPS MODTAG
Washington Layer Elastic AREA/EVERCALC

3.1.1.1 Surface Deflection Based Methods

Two of the state DOTs, California and Minnesota, use surface-deflection based analysis
procedures. Oregon also uses the California method for some overlay projects. These
methods use only the deflection recorded in the center of the load plate, and are derived
from methods originally developed for the analysis of Benkleman Beam data.

California uses a surface-deflection based FWD analysis procedure, which uses only the
deflections measured at the center of the load plate. The mean and 80th percentile
deflections are calculated for the test section. These deflections are then converted to
equivalent “California Deflectometer” measurements using an empirical calibration
curve. Tolerable surface deflections are determined for the section based on the
pavement thickness and traffic volume using a compiled chart. If the measured 80th
percentile deflection is more than the tolerable surface deflection, rehabilitation is
required. The percentage by which the deflection needs to be reduced is calculated, and
the required structural enhancement, expressed as feet of “gravel equivalence”, is
determined from a table. This thickness of gravel equivalence can then be converted to
thickness of various types of overlay materials using further conversion factors.

Minnesota DOT uses an approach originally developed for Benkelman Beam data.
Center deflections are normalized to 9,000 pounds, and then converted to an equivalent
Benkelman beam defection using a regression equation. Structural sufficiency of an in-
service pavement is determined by allowable deflection – a table has been compiled with
allowable deflections for different traffic volumes and pavement thicknesses. An
equation is provided to estimate the reduction in deflection based on the thickness of an
asphalt concrete overlay.

9
3.1.1.2 AGDPS Based Methods

Colorado DOT, Maryland DOT, and Virginia DOT all use the deflection analysis
procedures described in the AGDPS for all analysis. Nevada and Oregon use this
procedure for new construction only. Washington uses this procedure, with some
modifications, on the majority of projects. Maryland and Virginia use the MODTAG
computer program to carry out this analysis. The remaining states use DARWin.

3.1.1.3 Linear Elastic Based Methods

Nevada, Texas and Washington perform analysis of FWD data with backcalculation
using linear-elastic theory. Nevada uses this method for the analysis of overlays on
flexible pavements only. Texas and Washington use this method for the analysis of all
FWD data. Nevada and Texas use the MODULUS computer program for
backcalculation. Washington uses the EVERCALC program for backcalculation and
AREA as a screening tool to identify areas in need of specific attention.

None of these states have strictly-defined procedures for performing backcalculation. In

Nevada and Washington, backcalculation is performed by a limited number of
individuals in a centralized materials office. In Texas, backcalculation is performed by
individuals in the regional offices and Texas DOT has developed a series of training
courses to educate these individuals in backcalculation techniques.

Nevada uses the seed values shown in Table 7 for backcalculation.

Table 7: Seed values used by Nevada DOT

Material Type Modulus (ksi)

Lower bound I Upper bound
Asphalt Concrete 100 1,000
Stabilized Base 100 400
Granular Base 30 I 200
I
Subgrade I
8 I

Texas uses the parameters shown in Table 8 for backcalculation.

10
 Table 8: Parameters used by Texas DOT

Material Type Modulus (ksi) Poisson’s Ratio

Lower bound Upper bound
Portland Cement Concrete 2000 7000 0.20
Asphalt Concrete 500 850 0.35
Asphalt Stabilized Base 250 400 0.35
Cement Treated Base 80 150 0.25
Granular Base 40 70 0.35
Lime Stabilized Subgrade 30 45 0.30
Good Subgrade 16 20 0.40
Normal Subgrade 8 12 0.40
Poor Subgrade 2 4 0.40

Washington has an in-depth method for determining seed moduli for various material
types. The seed modulus of asphalt concrete is determined using Equation 1.

E = 10 (6.4721−0.000147362×T )
2

Figure 1. Equation 1.
where: E = asphalt concrete modulus, psi
T = mid-depth asphalt concrete temperature, Fahrenheit

Washington DOT’s guidelines note that the modulus of fatigued asphalt concrete is
typically between 100 ksi and 200 ksi. The Poisson’s ratio of asphalt concrete is assumed
to be 0.35.

Seed moduli and Poisson’s ratios for base, subbase and stabilized materials are as shown
in Table 9.

Table 9: Base, Subbase and Stabilized Materials Parameters Used by Washington DOT

Material Type Seed Poisson’s

I I Modulus, ksi I Ratios I
Granular Base 35 0.35
Granular Subbase 30 0.35
Sand Base 20 0.35
Sand Subbase 15 0.35
Cement Stabilized Material 1000 0.25-0.35
Lime Treated Subgrade 50 0.25-0.35

The seed moduli for subgrade materials are determined based on material type and
climate, as shown in Table 10.

11
 Table 10: Subgrade Parameters by Climate Type Used by Washington DOT

CJ
Material

Clay
Silt
Dry

15
15
I
Seed Modulus, ksi
Wet -
No Freeze
6
10
I
Wet-Freeze
Unfrozen
6
5
I Frozen
50
50
D Poisson’s
Ratio

0.45
0.45
Silty or Clayey 20 10 5 50 0.35-0.40
Sand
Sand 25 25 25 50 0.35-0.40
Silty or Clayey 40 30 20 50 0.35-0.40
Gravel
Gravel 50 50 40 50 0.35-0.40

3.1.1.4 Correction Factors

Beyond the basic analysis methodology, we also investigated the use of correction
factors. These correction factors include laboratory/field resilient modulus correlation
factor, asphalt concrete temperature correction and seasonal correction of results. Table
11 shows the usage of these factors by the state DOTs investigated.

Table 11: Correction Factors

State DOT Subgrade Modulus AC Temperature Seasonal

Correction Correction Correction
California NA No No
Colorado No policy (typically No policy No
0.33)
Maryland 0.33 Yes No
Minnesota NA Yes Yes
Nevada 0.331/No2 Yes No
Oregon 0.331/NA3 Yes1/No3 No
Texas No Yes No
Virginia 0.33 Yes No
Washington 0.331/No4 Yes No

Notes 1: For AGDPS analysis

2: For MODULUS analysis
3: For Surface deflection based analysis
4: For Evercalc analysis

All of the state DOTs use a subgrade resilient modulus correction factor of 0.33 when
using the AGDPS analysis procedure. None of the state DOTs use a resilient modulus
correction factor when using a surface deflection or linear-elastic analysis procedure. As
surface deflection based analysis procedures do not directly calculate the subgrade
resilient modulus, such a correction is not applicable. None of the three states that use

12
 linear-elastic methodologies for backcalculating layer moduli use those moduli in the
AASHTO 1993 pavement design procedure. Therefore although they do not use a
correction factor, this is not relevant to the use of a subgrade resilient modulus
backcalculated using a layered elastic program in AGDPS procedure.

All of the states investigated performed temperature correction of FWD results, with the
exception of California. Minnesota, which uses a similar procedure, applies a
temperature correction directly to measured deflections. The remaining states with the
exception of Nevada use the temperature correction procedures provided by the analysis
software used. Nevada uses an external spreadsheet to perform temperature correction of
asphalt concrete.

Only Minnesota DOT applies a seasonal adjustment as part of the FWD data analysis
procedure. Tables of seasonal adjustment factors have been compiled by Minnesota
DOT for various test calendar dates, pavement thicknesses and subgrade types. The
measured surface deflections are multiplied by the appropriate seasonal correction factor
to determine the equivalent “spring deflection”.

3.2 Data Collection

Our investigation of state DOT FWD operational procedures focused on equipment setup
and test point spacing.

3.2.1.1 Equipment and Setup

The types of FWD equipment used by the nine states investigated along with information
on how that equipment is configured are presented in Table 12.

Table 12: Equipment Type and Setup

State FWD Number Load Levels, kips Sensor Spacing, inches

I I Types I Owned I Seating I Recorded I I
California KUAB/JILS 2/2 9 9,9,9 0
Colorado JILS 1 None 9,9,9 0,8,12,18,24,36,60
Nevada Dynatest 2 9 9 0,12,24,36,48,60,72
Maryland Dynatest 1 Varies, see Table 13
Minnesota Dynatest 4 None 9,9,12,12,15,15 0,8,12,18,24,36,48,60,72
Oregon Dynatest 1 None 6,9,12 0,8,12,18,24,36,60
Texas Dynatest 15 None 9 0,12,24,36,48,60,72
Virginia Dynatest 1 12,12 6,6,6,9,9,9, 0,8,12,18,24,36,48,60,72
12,12,12,15,15,15 (0,8,12,18,24,36,60
acceptable)
Washington Dynatest 1 9,9 16,12,9,6 -12,0,8,12,24,36,48

Of the states investigated, only California owns FWDs built by more than one
manufacturer. At this time however, the KUAB FWDs are not in active use.

13
Five of the nine state DOTs investigated include at least one seating drop in their drop
sequence. Every state except Maryland includes at least one recorded drop at the 9 kip
load level. Maryland includes a 9 kip drop in most, but not all test setups.

Maryland is unique in tailoring the drop sequence and sensor spacing to the specific
project. Drop sequences are selected using the following table, where 1,2,3,4 are
recorded drops at 6,8,12,16 kips, respectively, and A,B,C,D are seating drops at those
load levels. Sensor spacings other than those in this table are allowed at the discretion of
the engineer. Table 13 presents the test setups used by Maryland DOT.

Table 13: Test setups used by Maryland DOT

Pavement HMA Test Type Sensor Spacing, Drop

Type Thickness, inches Sequence
inches
Flexible <4 Basin 0,8,12,24,26,48,60 BA1B2
Flexible 4–8 Basin 0,8,12,24,26,48,60 BB2C3
Flexible >8 Basin 0,8,12,24,26,48,60 BB2D4
Rigid N/A Basin 0,8,12,24,26,48,60 BC3D4
Rigid N/A Joint -12,0,12,24,26,48,60 BB2C3D4
Composite N/A Basin 0,8,12,24,26,48,60 BC3D4
Composite N/A Joint -12,0,12,24,26,48,60 BB2C3D4
Composite N/A Joint/Basin 0,12,18,24,26,48,60 BB2C3D4
Subgrade N/A Basin 0,8,12,24,26,48,60 AA1B2

3.2.1.2 Test Point Spacing

The test point spacings used by the nine DOTs investigated are presented in Table 14.
These test point spacings are on a lane basis, however all of the DOTs perform testing
only in the outer lane on highways with more than one lane per direction except in special
circumstances.

14
 Table 14: Test Point Spacings

State DOT Test Point Spacing, feet Lane Pos.

California 262 (80 m, minimum 21/project) OWP
Colorado 528 (0.1 mile) OWP
Nevada 1056 (0.2 mile) OWP
Maryland Varies OWP
Minnesota 528 (0.1 mile, minimum 5/section) OWP
Oregon 250 OWP
Texas 528 (0.1 mile, minimum 30/project) OWP
Virginia Varies, see Table 8 OWP
Washington 250 OWP

All of the state DOTs test in the outer wheel path (OWP).

California tests at 80 meter interval, with a minimum of 21 tests in a project.

Alternatively, one 300 meter section tested at a 15 meter interval may be selected out of
every 1600 meters.

Maryland leaves the test point spacing up to the pavement engineer, based on the
expectation of 150 test points to be performed in a day. A minimum of thirty test points
per direction should be collected. Two days of testing are recommended for projects over
3 centerline miles. For a two lane, five mile section this would equate to a test point
spacing of 176 feet, which would be rounded up to 200 feet in practice.

Virginia varies the test point spacing according to the project length, as shown in Table
15.

Table 15: Test Point Spacings Used by Virginia DOT

Project Length, miles Test point spacing,

feet
0 – 0.5 25
0.5-2 50
2-4 100
4-8 150
>8 200

4 REVIEW OF BACKCALCULATION PROGRAMS

This section summarizes our investigation of FWD backcalculation programs with

regards to their suitability FLH usage.

15
 4.1 Software Investigated

We investigated nine FWD backcalculation programs. In addition, DARWin has been

included in the analysis as a reference, although it is not under consideration for future
FLH usage. Summary information on these programs in contained in Table 16.

Table 16: Software Investigated

Program Latest Developer Environment Availability

I I Version I I I I
BAKFAA 1/3/2006 FAA Windows Free, source
code available
Bousdef 2.01 Haiping Zhou Windows Free
(1/2/2001)
DAPS 1.5.1 Abatech Windows $600
ELMOD 5 Dynatest Windows $5000 site
(12/8/2005) license,
additional
license $575
Evercalc 5.0 Washington DOT Windows Free
(3/2001)
MICHBACK 1.0 Michigan State DOS Free
(1995) University
MODCOMP 6 Cornell DOS Free
(9/26/2004) University
MODULUS 6.01 Texas Windows Free
(2/3/2003) Transportation
Institute
MODTAG 3.1.68 Virginia Windows Free
(10/12/2005) DOT/Cornell
University
DARwin 3.1 AASHTO Windows $2000

Of these nine programs, seven are available for free. The other two, DAPS and ELMOD
have substantial licensing fees, however they are commercially supported. The others are
nominally supported by their authors in the cases of errors and bugs, however prompt
support for installation and other user-associated problems is not likely.

Seven of these programs have a MS Windows-based graphical user interface. The

remaining two, MICHBACK and MODCOMP are DOS programs which can run under
DOS or in a DOS shell under MS Windows. All of these programs will run on MS
Windows 98, 2000 and XP.

BAKFAA is unique among the programs investigated in that its source code is freely
available. This would allow an interested user to add or modify its functionality,
especially with regards to adding support for new FWD file formats.

16
DARWin was included in this investigation as a reference point only, as DARWin is not
capable of backcalculating individual layer moduli.

4.2 File Formats and Units

Table 17 summarizes the file formats and units used in the nine backcalculation
programs.

Table 17: File Formats and Units

Program File Formats Program Units

BAKFAA F20, PDDX US
Bousdef User input, program US
specific format
DAPS F20 SI
ELMOD F9, F10, F20, F25, US or SI
FWDWin
Evercalc F20, F25 US or SI
MICHBACK KUAB, program US or SI
specific format
MODCOMP User input, program US or SI
specific format
MODULUS F9, F10, F20, US
program specific
format
MODTAG F20, F25, PDDX US (SI support
under development)
DARWin F20, KUAB, PDDX US or SI

Of the nine backcalculation programs, only two accept data from more than one brand of
FWD equipment. These are MODTAG and BAKFAA, both of which accept data in the
Pavement Deflection Data Exchange (PDDX) format. PDDX is a file format specified by
the AASHTO publication “Pavement Deflection Data Exchange Technical Data Guide,
Version 1.0” dated April 1998. All four FWD manufacturers have recently added
support for this format in their data collection software, however it is often not supported
by older equipment.

The F9, F10, F20, F25 and FWDWin formats were created by Dynatest. Support for
these formats by equipment made by other manufacturers is rare. The KUAB format was
created by KUAB, although some FWDs manufactured by JILS can also export data in
that format. Two of the programs, BOUSDEF and MODCOMP can not directly import
data from any standard FWD data format.

17
Four programs have dual-unit capability, while four use the US Customary system only,
and one uses the SI system only. SI support for MODTAG is currently under
development.

4.3 Theory and Modeling

The basic theories and methodologies used by the backcalculation programs in modeling
the pavement system are presented in Table 18.

Table 18: Theory and Modeling

Program Theory Forward Engine Max. Max.
I I I I Sensors I Layers I
BAKFAA layer-elastic LEAF 7 10
Bousdef MET internal 9 5
DAPS layer-elastic ELSYS 9 41
ELMOD MET2 internal 15 51
Evercalc layer-elastic Weslea 10 5
MICHBACK layer-elastic Chevlay2 10 41
MODCOMP layer-elastic Chevlay2 9 12
MODULUS layer-elastic Weslea 7 41
MODTAG layer-elastic Chevlay2 (via 9 7
MODCOMP)
DARWin MET/DL internal 9 2

Notes: 1 Program can additionally include a stiff layer (e.g. bedrock)

2 ELMOD can also use Weslea or a finite element model through add-in modules

4.3.1 Pavement Response Models

There are two different models commonly used for calculating surface deflections in
flexible pavement systems: Layer-elastic theory and the Method of Equivalent Thickness
(MET). Seven of the programs use the former, and two use the latter. The two programs
based on MET both have internal forward-calculation engines that are specific to them.
Of the linear-elastic based programs, three use Chevlay2 as a forward calculation engine,
which is an enhanced version of the CHEVRON program originally developed by the
Chevron Corporation. Two use the Weslea program, which was developed by the Army
Corps of Engineers Waterways Experiment Station. One uses the layered-analysis
program LEAF which was recently developed by the Federal Aviation Administration.
The remaining program uses ELSYS, the origin of which is not known to us, but it may
be related to ELSYM5 which was originally developed at the University of California
Berkley and modified by FHWA.

Layered-elastic theory was first developed by Burmister in 1945.[1] This theory is

mathematically derived from the theory of elasticity, and it is correct for materials that

18
are linear-elastic, isotropic, homogenous and of infinite extent in the horizontal
directions. It is, however, computationally-intensive. The first computer program
capable of solving arbitrary problems using this theory was CHEVRON in 1962. These
programs were not ported to personal computers until the mid-1980s.

In the interim, the method of equivalent thickness, which is not theoretically correct but is
capable of being solved by hand, was developed by Odemark in 1949.[2] Odemark’s
basic assumption is that the influence of a layer on the layers below it is dependent on the
stiffness of that layer only. Therefore a layer with a given elastic modulus and thickness
can be equivalent to another layer with a different elastic modulus and thickness but the
same overall stiffness for the purpose of determining stresses, strains and deflections in
the layer below it. If all the layers in the system are transformed into equivalent layers
with the same elastic modulus as the subgrade, the problem is now a semi-infinite half-
space and the deflection of the subgrade is now solvable using Boussinesq theory.
Deflections in the layers above the subgrade are calculated as the difference in deflection
at the top of the layer and the bottom of the layer in a transformed system with the
modulus equal to the modulus of that layer.

For most problems layered-elastic theory and the method of equivalent thickness do not
yield exactly the same answers. For problems where the layer moduli decrease with
depth, and the thickness of each layer is greater than the load plate radius both methods
tend to give similar answers. For these situations a simple calibration factor of 0.8 or 0.9
is typically applied to moduli backcalculated using the method of equivalent thickness.
For other situations more complicated calibration equations must be used.

Thin layers are a problem for all backcalculation methods and programs. This problem is
inherent in the nature of backcalculation. Because there are no closed-form solutions for
calculating layer moduli based on surface deflections, backcalculation involves iterative
methods of varying layer moduli to match the measured surface deflections. Thin layers
contribute little to the surface deflections regardless of their layer moduli. Therefore the
thin layer may have a wide range of moduli without a significant effect on surface
deflections. Solutions to the thin layer problem would require collecting data other than
surface deflections, such as sub-surface stresses or strains.

4.3.1.1 Rigid Pavement Models

DARWin performs backcalculation of flexible and rigid pavements in different ways.

Flexible pavements are analyzed using the method of equivalent thickness, however only
two layers are considered. These layers are the subgrade and pavement layer (including
all bound and unbound pavement layers). Rigid pavements are analyzed using
Westergaard’s theory for the interior loading of a plate on a dense liquid (DL)
foundation.[3] The plate is assumed to be infinite in the horizontal directions. The
subgrade is modeled as a dense liquid with no shear strength. Both these assumptions
greatly simplify the calculations.

19
None of the nine computer programs investigated are specific to rigid pavements,
although such programs exist. These programs use either slab on dense liquid, slab on
elastic sold or finite element models. Some of these programs, such as ILLI-SLAB, can
correct for slab size. These corrections are complicated, as the load transfer between
adjacent slabs and the shoulder must be known.

Historically, there has been little consensus in the pavement engineering community as to
whether special-purpose rigid pavement backcalculation programs yield better results
than general purpose programs such as those investigated here. The Long Term
Pavement Performance Program (LTPP) avoided this issue by performing
backcalculation on rigid pavements using the layered-elastic model, the slab on dense
liquid model and the slab on elastic solid model.

However, two recent and important rigid pavement design methodologies use layered
elastic theory. LEDFAA, which was developed by the Federal Aviation Administration
for airfield pavement design, uses the computer program LEAF to calculate stresses and
strains in both flexible and rigid pavements. LEAF is based on layered elastic theory,
and is also used in the BAKFAA backcalculation program.

The mechanistic-empirical design guide being developed under NCHRP 1-37A

(commonly known as the “MEPDG”) uses JULEA, which is a layered elastic program,
for computing the effective dynamic modulus of subgrade reaction (effective k-value) for
rigid pavements with multiple subgrade layers. This reduces the problem to three layers
(slab, base and subgrade), which is further analyzed using ISLAB2000, which is based on
plate theory.

4.4 Program Limitations

Two programs, MODULUS and BAKFAA are limited to a maximum of seven deflection
sensors. Many FWDs currently mount more then seven deflection sensors, with nine
being a typical number. Some new FWDs are capable of mounting up to 15 deflection
sensors, but such a configuration is rare.

MODULUS, MICHBACK and DAPS have a pavement model that is limited to a

maximum of four layers. This is typically sufficient for project-level pavement analysis.
These four layers are typically interpreted as a bound surface layer, unbound base layer,
unbound subbase layer and subgrade. These three programs can also include a stiff layer
(e.g. bedrock) which is treated separately from the four explicit layers. For programs that
do not include a separate stiff layer, the lowest layer can be explicitly modeled as
bedrock.

MODTAG uses MODCOMP as its backcalculation engine. Its theory of operation is

therefore identical to that of MODCOMP, subject to an additional limitation on the
number of layers. Alternatively, MODTAG can export data to MODULUS. MODTAG
is unique among the investigated backcalculation programs in that deflection basins at the
same location and target load are averaged prior to analysis.

20
4.5 Additional Features

The additional features of the backcalculation programs investigated are summarized in

Table 19.
Table 19: Additional Features

Depth-to-bedrock Estimation

AC Temperature Estimation
AC Temperature Correction

Load Transfer Efficiency

Non-Linear Materials

Raw Data QC Tools

Segmentation Tools
Void Detection
Program
BAKFAA
Bousdef
DAPS ●
ELMOD ● ● ● ● ● ● ● ●
Evercalc ● ● ● ●
MICHBACK ● ●
MODCOMP ● ●
MODULUS ● ●
MODTAG ● ● ● ● ● ● ● ●
DARWin ● ●

4.5.1 Non-Linear Materials

ELMOD, EVERCALC, MODCOMP and MODTAG are capable of analysis of the non-
linear elastic properties of pavement layers. MODCOMP includes nine stress-
dependency models, the most of any program investigated. Every stress-dependency
model included in the other programs is also included in MODCOMP.

LTPP used MODCOMP for backcalculating FWD data and all nine stress-dependency
models were utilized. It was found that none of the models consistently converged on an
acceptable solution for all of the pavements in the study, and that the root-mean-squared
errors were more often unacceptably high for nonlinear models than for the linear
model.[4] These results indicate that non-linear analysis requires detailed knowledge and
experience to apply correctly, and that rigid guidelines for their application are
inappropriate.

21
4.5.1.1 MODCOMP

MODCOMP allows the user to choose one of nine different stress-dependency models
for backcalculated moduli. Depending on which model is chosen, the user must also
enter the unit-weight of each pavement layer, the lateral earth pressure ratio, or the
percentage passing the #200 sieve. Data for at least four different load levels at each test
point is required.

4.5.1.2 MODTAG

MODTAG uses MODCOMP as its backcalculation engine, but currently limits the user
to one stress-dependency model. The model used is the Uzan Model of the form shown
in Equation 2.

E = k1θ k 2 τ oct
k3

Figure 2. Equation 2.

where: E = elastic modulus, psi

θ = bulk stress, psi
τoct = octahedral shear stress, psi
k1,k2,k3 = regression constants

The developers of MODTAG plan on including all of the non-linear models currently in
MODCOMP in the future.

4.5.1.3 ELMOD

ELMOD accounts for non-linearity in the subgrade material based on the shape of the
outer portion of the deflection basin. This is not a formal stress-dependency model, and
it does not yield coefficients that can be entered into any stress-dependency model.
Rather, it is an empirical method that is used to estimate the subgrade modulus directly
underneath the load plate at the test load. Typical linear backcalculation yields the
subgrade modulus at some distance from the load plate, at a point where the stress is
generally lower than that directly underneath the load plate. ELMOD uses this method
by default, but it can be disabled. An advantage of this method is that it does not require
drops at more than one drop height.

Alternatively, through the use of the FEM/MET/LET add-in module (which must be
purchased separately) ELMOD can use a formal stress-dependency model. The model
used is the principal stress model of the form shown in Equation 3.

22
 E = k1σ 1
k2

Figure 3. Equation 3.

where: E = elastic modulus, psi

σ1 = major principal stress, psi
k1,k2 = regression constants

ELMOD does not include the overburden stress in the major principal stress calculation,
so unit weight of the pavement material is not required. As the overburden stress in the
subgrade is typically larger than the load-related stress, the neglection of overburden
stress magnifies the perceived non-linearity of the subgrade response.

4.5.1.4 Evercalc

Evercalc uses two stress-dependency models, one for coarse-grained materials and one
for fine-grained materials. For coarse-grained materials a bulk stress model is used, as
shown in Equation 4.

M r = k1θ k 2
Figure 4. Equation 4.

where: Mr = resilient modulus, psi

θ = bulk stress (the sum of the principal stresses), psi
k1,k2 = regression parameters.

EVERCALC uses a deviator stress model for fine-grained materials, as shown in

Equation 5.

M r = k1σ d
k2

Figure 5. Equation 5.

where: Mr = resilient modulus, psi

σd = deviator stress, psi
k1,k2 = regression parameters.

As with ELMOD, overburden stress is not included and material unit weights are not
required inputs. At least two different load levels are required to perform non-linear
analysis with EVERCALC.

23
4.5.2 Depth to Bedrock

Six of the nine backcalculation programs investigated will estimate the depth to bedrock.
ELMOD, Evercalc, MODULUS and MODTAG use a technique originally developed by
Per Ullidtz of Dynatest. This technique first plots measured deflection versus the inverse
of deflection sensor offset (1/r). This plot typically has a linear section followed by a
non-linear concave-down section at high values of 1/r. Sometimes there is also a non-
linear concave-up section to the plot at low values of 1/r. The linear portion of the plot is
extrapolated down to the X-axis, and the inverse of this intercept is taken as an estimate
of the depth to bedrock. MODULUS and Evercalc further refine this estimate using
empirical regression equations developed by Rhode and Scullion.[5]

DAPS and MICHBACK will also estimate the depth to bedrock, however the
methodology is not explained in the accompanying documentation.

4.5.3 Asphalt Concrete Temperature Correction

Three of the nine backcalculation programs investigated will correct the modulus of
asphalt concrete layers for temperature effects.

4.5.3.1 ELMOD

ELMOD includes a “Temperature Table”, which includes ratios of Et / Eref for each
whole degree Celsius between -45 at 59. This table is populated by default with a
reference temperature of 25º C, but it is user-modifiable.

4.5.3.2 Evercalc

Evercalc performs temperature correction using a regression equation developed by

Washington DOT, as show in Equation 5.

(
E ref = E t × 100.000147362 T 2 − Tref2 )
Figure 6. Equation 6.
where: Eref = elastic modulus at reference temperature, psi
Et = elastic modulus at test temperature, psi
Tref = reference temperature, degrees Fahrenheit
T = test temperature, degrees Fahrenheit

This equation is not modifiable, however the reference temperature is. By default, the
reference temperature is set to 77º F.

24
4.5.3.3 MODTAG

MODTAG performs temperature correction using Equation 6.

(
k Tref −Tcalc )
E ref = E calc × 10
Figure 7. Equation 7.
where: Eref = Modulus at the reference temperature
Ecalc = Backcalculated modulus
k = -0.0195 for testing in the wheel path, -0.021 for mid-lane
Tref = Reference temperature, ºC
Tcalc = Mid-depth temperature at the time of testing, ºC

This equation is not modifiable. The reference temperature is set by the user, and there is
no default value.

4.5.4 Asphalt Concrete Mid-depth Temperature Prediction

The asphalt concrete modulus correction methodologies described above require the mid-
depth pavement temperature at the time of testing as an input. Measuring the mid-depth
pavement temperature is tedious, and typically takes longer then the actual FWD test.
For this reason several equations have been developed to estimate the mid-depth
pavement temperature based on more readily available data such as the pavement surface
temperature at the time of testing, and the previous day’s average air temperature.
ELMOD, Evercalc and MODTAG all include such predictive equations.

4.5.4.1 ELMOD and MODTAG

ELMOD and MODTAG both use the BELLS3 predictive equation. This equation
requires the pavement surface temperature at the time of testing and the previous day’s
average air temperature as inputs.

4.5.4.2 Evercalc

Evercalc uses the Southgate predictive equation. This equation requires the pavement
surface temperature at the time of testing and the average air temperature over the
previous five days as inputs.

4.5.5 Raw Data QC Tools

Raw data quality control (QC) tools assist the user in evaluating the quality of the raw
data. They may also enable the user to eliminate erroneous or suspicious data points.
Three of the nine backcalculation programs include raw data QC tools.

25
4.5.5.1 MICHBACK

MICHBACK provides some basic raw data QC tools. It will provide an automatic
warning when the load data varies from the target load by more then 10%. It also plots
the raw deflection data and allows the user to eliminate entire test points or individual
sensor data at a test point which appear anomalous.

4.5.5.2 ELMOD

ELMOD also provides some raw data QC tools. It provides the ability to plot the raw
deflections and composite modulus. Anomalous test points can be removed, but
individual sensor data can not be removed. During backcalculation, the user can choose
to review the composite modulus plot for each deflection basin and remove individual
sensor data from further analysis.

4.5.5.3 MODTAG

MODTAG provides a wide array of raw data QC tools. It includes a series of checks that
it calls “Pre-analysis”. These checks include: non-decreasing deflections, zeros in data,
overflow, non-linearity, and erroneous sensor position.

• “Non-decreasing deflections” is a situation where the reported deflections do not

decrease with increasing distance from the load plate.
• “Zeros in Data” is triggered when one or more deflection value is reported as
zero.

• “Overflow” triggers when a deflection sensor reports more than 2032 microns (80
mils) of deflection, which is the maximum deflection that most Dynatest FWDs
can measure.

• “Linearity test failed” triggers when a linear regression line drawn through the
maximum deflection vs. load level data at a test point does not intercept the y axis
close to zero. This is not necessarily an indication of problems with the raw data,
however it does indicate that non-linear backcalculation of the data is warranted.

• “SLIC warning” triggers when the reported deflection data does not agree with
the reported deflection sensor position.

Deflection basins that that fail these checks may be deleted.

MODTAG also allows the user to review the composite modulus plots for each deflection
basin, and either delete the entire basin or remove specific sensors from the analysis.

26
4.5.6 Load Transfer Efficiency

Among the nine backcalculation programs reviewed, only ELMOD and MODTAG will
compute load transfer efficiency (LTE). Both programs compute LTE according to
Equation 7 for joint approach testing, and Equation 8 for joint leave testing.

d12
LTE a = × 100
d0
Figure 8. Equation 8.

d −12
LTE l = × 100
d0
Figure 9. Equation 9.

where: LTEa = load transfer efficiency from joint approach testing, %

LTEl = load transfer efficiency from joint leave testing, %
d0 = deflection at center of load plate, mils
d12 = deflection at 12 inches in front of center of load plate, mils
d-12 = deflection at 12 inches behind center of load plate, mils

The major difference between these two programs is how joints are identified in the FWD
data file.

4.5.6.1 ELMOD

ELMOD identifies joints based on specific codes imbedded in the Dynatest F25 and
FWDWin file formats. During testing the FWD operator is presented with a graphical
representation of a PCC slab and can select his position on the slab. These positions are
saved in the data file as codes. ELMOD reads these codes out of the data file, and
automatically computes LTE for codes that indicate load transfer testing. These codes
can be manually edited in ELMOD prior to analysis. In addition to joint approach and
joint leave testing, ELMOD will also compute LTE across transverse joints for
appropriate test locations.

4.5.6.2 MODTAG

MODTAG requires that load transfer tests be labeled “JA” (i.e. joint approach) or “JL”
(i.e. joint leave). It can automatically determine the test type from the FWD data file if it
is properly commented in the field, or the user can manually input the test type in
MODTAG. MODTAG does not have the capability of calculating load transfer across
transverse joints or cracks.

27
4.5.7 Void Detection

Among the nine backcalculation programs reviewed, only ELMOD and MODTAG will
perform void detection. The methodology used in these programs is similar. Both
programs a linear regression line through the center deflection vs. load data collected at a
given location. The y-intercept of this line is taken as an indication of the presence of a
void at that location.

ELMOD will perform void detection at all test locations, regardless of the position of the
FWD or number of drop heights.

MODTAG will perform void detection only at test locations that are labeled “C” (i.e.
corner testing). Void detection will only be performed if data at three or more load levels
is collected.

4.5.8 Segmentation Tools

Segmentation tools assist a user in breaking up FWD data into logical segments for
further analysis. Segmentation may be performed in order to allow for different
structural properties such as layer thickness or as a way of presenting analysis results.
Three of the nine backcalculation programs investigated allow for segmentation of FWD
data.

4.5.8.1 ELMOD

ELMOD allows an analysis project to consist of one or more FWD data file. Prior to
backcalculation, plots of raw deflection data or composite modulus vs. station may be
reviewed. Based on these plots the user may choose to break the project up into two or
more segments with different structural properties. No automated assistance is provided
to the user in making that determination. Subsequent to backcalculation the user may
further segment the project for the purpose of reporting statistics.

4.5.8.2 MODULUS

MODULUS allows an analysis project to consist of only one FWD data file. Prior to
backcalculation raw deflection data vs. station can be plotted for deflection sensors 1, 2
and 7. MODULUS does not directly allow segmentation based on structural properties
prior to backcalculation; however it does allow test locations to be excluded from
analysis. So, MODULUS could be run twice on the same FWD data file with two
different sets of layer thicknesses, and then all but the appropriate test locations could be
excluded from analysis.

After backcalculation, MODULUS can automatically segment the project based on the
analysis results. The methodology used in automatic segmentation is not explained in the
documentation which accompanies the software. Alternatively, the user may manually

28
create segments. The mean and standard deviation of the analysis results are reported for
each segment.

4.5.8.3 MODTAG

MODTAG allows an analysis project to consist of one or more FWD data file. Prior to
analysis, several plots may be reviewed to assist the user in making segmentation
determinations. The cumulative differences of deflections plot is provided expressly for
segmentation purposes. Depth to stiff layer and composite modulus vs. station may also
be used for segmentation. Based on these plots the user may choose to break the project
up into two or more segments with different structural properties. No automated
assistance is provided to the user in making that determination. There is no way to
perform segmentation based on backcalculation results, except to re-segment the project
and perform backcalculation again.

4.6 Recommendations

We recommend that the following programs be excluded from further consideration

because of the listed drawbacks and lack of redeeming features.

• BAKFAA – Maximum of seven sensors

• Bousdef – No FWD file import capability
• DAPS – Cost
• MICHBACK – DOS program, limited to KUAB files
• MODCOMP – DOS program, no FWD file import capability

The remaining four programs are Evercalc, ELMOD, MODULUS and MODTAG.

4.6.1.1 Evercalc

Evercalc was developed by Washington DOT for their own usage, although it is freely
available to others. The last major update was in 2001.

Pros

• Depth to bedrock estimation

• Temperature correction of AC moduli
• Estimation of mid-depth AC temperatures

Cons

• File import is limited to Dynatest F20 and F25 formats

• Lack of segmentation capability
• Lack of data quality control checks

29
4.6.1.2 ELMOD

ELMOD was developed by Dynatest as a revenue-generating product. The latest version

was issued in December of 2005.

Pros

• User interface – the most polished of the programs reviewed

• Customer support
• Full range of additional features

Cons

• Cost ($5000 for a site license)

• Data import is limited to Dynatest formats.

4.6.1.3 MODULUS

Modulus was developed by the Texas Transportation Institute (TTI) under contract to
Texas DOT. The last major update was in February of 2003.

Pros

• Simplicity of operation – the most straight-forward and easy to understand of the

programs reviewed
• Depth to bedrock estimation

Cons

• Relative lack of features

• Data import is limited to Dynatest F20 format

4.6.1.4 MODTAG

MODTAG was developed by Virginia DOT and Cornell University. It is currently in use
by Maryland DOT and Virginia DOT, although they do not use its backcalculation
features. Development is ongoing, and the latest public release was in October 2005.

Pros

• Ability to import data in non-manufacturer specific PDDX format

• Full range of features
• Most comprehensive data QC tools among programs reviewed
• Ability to perform data analysis using the methodology included in the AASHTO
1993 Pavement Design Guide

30
Cons

• Program is currently in a usable, but not finished state

• Users may find the number of options overwhelming

31
32
 REFERENCES

1 Burmister, D.M., “The General Theory of Stresses and Displacements in

Layered Systems”, Journal of Applied Physics, February 1945.

2 Odemark, N., “Investigations as to the elastic properties of soils and design of

pavements according to the theory of elasticity”, Meddelande No. 77 Stateus
vaginstut, 1949.

3 Westergaard, H.M. “Stresses in Concrete Pavements Computed by

Theoretical Analysis”, Public Roads Vol. 7, 1926.

4 Von Quintus, H.L., Simpson, A. “Back-Calculation of Layer Parameters for

LTPP Test Sections, Volume II: Layered Elastic Analysis for Flexible and
Rigid Pavements”, FHWA-RD-01-113, October 2002.

5 Rhode, G.T. and Scullion, T., “MODULUS 4.0: Expansion and Validation of
the Modulus Backcalculation System.” Texas Transportation Institute
Research Report 1123-3, November 1990.

33