Mechanistic-Empirical Design and MnROAD

MnROAD Lessons Learned – December 2006

MnDOT Contact: Shongtao Dai

1 Abstract
In its first ten years of operation, MnROAD’s data and road research contributed to many
issues in pavement engineering. In particular, MnROAD made its greatest contribution
in the field of mechanistic-empirical (ME) design. MnROAD’s data has used to calibrate
and verify a number of pavement design guides, including MnPave, an ME design
program created by MnROAD engineers and adopted by MnDOT. Furthermore, the use
of this data as inputs into existing design methods has exposed some of the inadequacies
of commonly used design methods. This brief details MnROAD’s involvement in ME
design and describes the capacity of MnROAD as a lasting influence on mechanistic-
pavement design for years to come.

2 Background
In its fourteen original objectives for research, MnROAD had a significant focus upon
both verifying and developing mechanistic-empirical (ME) models for pavements using
MnROAD data. At the time of the creation of these objectives, the empirical data for
existing design methods came from the AASHO Road Test of the late 1950s and early
1960s. Seeing as how a number of important variables had changed since the time of the
AASHO Road Test (e.g. both frequency and maximum weight had increased), MnROAD
recognized the opportunity to re-evaluate the standard design procedures, all of which
were built upon the same AASHO Road Test data. This re-evaluation would of course
begin with the material, environmental, traffic, and response data from MnROAD and
then involve the analysis of MnROAD/MnDOT engineers or other out-of-state
researchers.
As reflected by MnROAD’s research objectives, pavement engineers and
pavement design methods of the early 1990s were concerned almost exclusively with
design thickness. Soon after opening to traffic, MnROAD began to observe that its test
sections were experiencing more distress and damage from environmental effects than
from traffic. In line with this observation, many MnROAD engineers were quick to point
out during the Lessons Learned project that “MnROAD was a thickness experiment that
turned into an environmental experiment.” Hence, MnROAD engineers were quick to
shift their focus on thickness to a new emphasis on environmental effects in the
performance of a given pavement system. This shift succeeded, of course, excellent work
dealing with the “thickness” question that will be detailed later in this brief, including an
evaluation of concrete design methods and the development of a local mechanistic-
empirical design method known as ROADENT.
After observing the effects of environment on MnROAD’s test sections, cold
regions studies became a priority for MnROAD. Much of this work centered around
seasonal variations in pavements and accounting for these effects in the MnPave design
method under development by MnROAD engineers. Other work involved closely
observing and recording low-temperature cracking and working toward a model of this

ME Design Brief 1
phenomenon. Many times, the work in cold regions topics coincided with pavement
design, and this brief will summarize the work in a few of these topics.
Overall, MnROAD continued to be a source for a number of more general topics
in pavement design and modeling. Its main focus as a source has been the contribution of
pavement data and expertise in one or more of the following:
1. climatic data;
2. traffic information;
3. construction and material information; and
4. pavement performance and response data
This brief will also detail MnROAD’s contribution of the above critical data sets to
general topics in mechanistic-empirical research, including NCHRP Project 1-37A, the
Mechanistic-Empirical Pavement Design Guide (MEPDG), and pavement modeling.

3 MnROAD Research and Data in Mechanistic-Empirical Design

MnROAD experience in and contributions to research in pavement design and modeling
were some of the foremost concerns of its first ten years. The following topics present
some experiences that point to the value of MnROAD and MnROAD data in advancing
the state-of-the-art in pavement design methods and models.

3.1 Cold-regions Design and Modeling

An early partnership between MnROAD and the U.S. Army Cold Regions Research and
Engineering Labs (CRREL) represented some of the first work to use MnROAD data to
refine a mechanistic-empirical design method. In the case of the CRREL work,
researchers at CRREL used MnROAD test section data to further develop their
mechanistic pavement design procedure for seasonal frost areas. A main reason that the
MnROAD facility was attractive to CRREL for its research can be found in sources such
as Figure 1.

Figure 1. Seasonal freezing index with time at Buffalo, Minnesota (Bigl and Berg 1996).

The relatively high freezing indices repeated over the course of thirty years is evidence of
the fact that MnROAD experiences the “cold” in cold-regions research, and for this
reason CRREL partnered with MnROAD for a number of research topics in the first

ME Design Brief 2
years of MnROAD’s first decade. This work was detailed in MnDOT Report 1996-22 by
Bigl and Berg and MnDOT Report 1996-23 by Bigl and Berg.
The CRREL research began with extensive laboratory tests on materials from
MnROAD test sections. (These tests are described in the Lessons Learned technical brief
on Climate Studies at MnROAD.) The material studies conducted by CRREL
contributed the necessary input parameters for CRREL to model test sections at
MnROAD and predict performance. The first of CRREL’s two MnDOT reports gives
details about the nature of CRREL’s mechanistic design procedure and its four
components:
1. FROST, to predict the amount of frost heave, settlement, temperature, ware
content, pore water pressure, ice content, and density of the pavement
structure and its components;
2. TRANSFORM, to divide the structure into layers and assign each layer a
corresponding resilient modulus, Poisson’s ratio, and density value;
3. NELAPAV, to calculate the stresses, strains, and deflections at a given
location in the pavement using a nonlinear layered elastic model
4. CUMDAM, to calculate the damage experienced by the pavement system.
Fine points on these components are provided by CRREL in MnDOT Report 1996-22.
The modeling done by CRREL using MnROAD data had three phases that are
closely detailed in the CRREL reports. CRREL’s first simulation was to model
temperatures from a year close to the mean freezing index and apply these conditions to
eleven test sections at MnROAD. The second simulations were for adjustments in
parameters and special environmental conditions (temperature extremes, etc.) over a
season on a total of nine test sections. The third and final simulation was to model two
test sections over 21 seasons. Overall, the models suggested that mechanistic design was
very uncertain in predicting the performance of a pavement system. However, the
research provided CRREL with an opportunity to refine their mechanistic procedure and
prepare for MnROAD environmental and performance data to refine their models.
Unfortunately, the comparison and use of this data was never put into practice by CRREL
due to time constraints.
Later work by Berg in MnDOT Report 1997-21 calculated maximum frost
penetration depths, using the modified Berggren equation, for the test sections at
MnROAD over three winters. These calculated depths were then compared against the
measured frost depths for the same period of time and discussed. Berg found that the
calculations overestimated frost depths substantially. As the author suspected that some
of this error might have been due to measurement error, Berg recommends that
MnROAD reevaluate its measurement data from the test sections that were particularly
divergent from the expected frost depths. Berg also found, however, that by adjusting
certain inputs (thermal conductivity of materials and mean annual soil temperature), he
was able to obtain calculations that better agreed with MnROAD measurements. Berg’s
study, as an appendix to the CRREL reports, concludes by providing MnROAD with
more directions for research and environmental measurements.

3.2 Evaluating Concrete Design

Another early project at MnROAD was desribed in MnROAD Report 1997-14 by
Thomas Burnham and William Pirkl. This research project involved the application of

ME Design Brief 3
data characterizing MnROAD concrete test sections to the MnDOT rigid pavement
design guidelines, the 1993 AASHTO Guide for Design of Pavement Structures
(AASHTO-93), and the 1984 Portland Cement Association Thickness Design for
Concrete Highway and Street Pavements (PCA-84). This report sought to examine the
parameter values used as inputs for the respective designs and determine the predicted
service lives of these test sections using each of the three design methods and the
parameter values.
The first step in this process was to determine the needed parameter values for
each design method based upon MnROAD data at hand. These parameter values were
then compared to the assumed values used to design the sections at the time of their
construction. Burnham and Pirkl discovered a number of discrepancy between the
assumed and as-built parameter values for each of the three design methods. While some
of the discrepancies could be explained by construction or material difficulties, the
authors felt that others spoke to the limitations and assumptions involved in certain
influential parameters. The authors made recommendations for future research on other
parameters that could not be addressed in the scope of this report (load transfer and
drainage coefficients).
Burnham and Pirkl found that the predicted serviceable life of each test section
was highly variable as the researcher moved between design methods and levels of
reliability. The inaccuracies of these models and the discrepancies between their
predictions as exposed by Burnham and Pirkl’s study was the first major use of full-scale
test track data to evaluate existing pavement design methods, and as noted by survey
subjects in the Lessons Learned project, this particular study illustrated that the concrete
design methods of the early 1990s were inadequate.

3.3 ROADENT
In the planning stages of MnROAD, one of the primary research concerns, as mentioned
above, was the creation of a Minnesota-specific mechanistic-empirical design method for
pavements that was both built upon MnROAD data and developed locally by MnDOT
and UM researchers. This objective was realized through an FHWA pooled fund study
involving MnDOT whose product was the flexible design procedure known as the
“Minnesota Method,” which was incorporated into the flexible thickness design program
called ROADENT. ROADENT was developed by David Timm, David Newcomb, and
Bjorn Birgisson of UM and called upon the expertise of a number of MnROAD
engineers.
The so-called heart of ROADENT—that is, its mechanistic pavement model—is
the Waterways Experiment Station Layered Elastic Analysis (WESLEA) to for design
and analysis in a flexible pavement system. WESLEA uses information for loading,
layer thickness, layer modulus of elasticity, and layer Poisson’s ratio for up to five layers
to calculate critical strains in the system. Based upon an evaluation of existing layered
elastic analysis programs using MnROAD response data, UM researchers felt that
WESLEA would best serve the Minnesota method of flexible pavement design.
ROADENT handles inputs in a number of innovative ways. First is ability to use
load spectra, in addition to ESALs, to describe traffic. Furthermore, in concert with work
in seasonal variation in pavements that would later be published in MnDOT Report 2000-
35 by Ovik et al., ROADENT accounts for the seasonal variability of the pavement

ME Design Brief 4
system by requiring the seasonal modulus value and season duration for that modulus for
each layer. A seasonal temperature for that season is also incorporated to account for
variability in the asphalt surface. To incorporate reliability analysis into ROADENT,
layer moduli and thickness variability were established by UM researchers using data
from MnROAD.
Researchers also calibrated the performance transfer functions used by the fatigue
and rutting equations in predicting pavement distress and failure. These transfer
functions were calibrated using MnROAD data. Finally, ROADENT calculates damage
on a given pavement system using Miner’s hypothesis. The flowchart for the
mechanistic-empirical design of ROADENT is illustrated by Figure .

Figure 2. Minnesota flexible design flowchart (Timm et al. 1999)

While this design procedure was used locally, due to the large number of calculations
involved in specifying ROADENT to the user’s needs, MnDOT pursued an expansion

ME Design Brief 5
upon ROADENT in terms of the front-end experience. This expansion became what is
now known as MnPave.

3.4 MnPave
The MnDOT flexible pavement design method, built upon ROADENT, was integrated
into a software package called MnPave in 1999. MnPave is designed to make pavement
design methods more accessible to practitioners in pavements. The MnPave software
utilizes a graphical user interface that allows users to point and click to adjust various
inputs in the design. As MnPave is built on ROADENT, many of the features of
ROADENT are reproduced in MnPave.

Figure 3. The start-up screen of MnDOT’s MnPave Design software (Chadbourn et al. 2002).

The MnPave package also places an emphasis on seasonal affects and the work
done in MnDOT Report 2000-35 by Ovik et al. This work includes dividing the year into
five seasons, two of which are “Early Spring” and “Late Spring.” These periods are
defined by the level of moisture in the aggregate base and the degree of freezing in the
subgrade: these factors contribute largely to layer stiffness. Furthermore, MnPave allows
users to characterize the duration of their seasons. By placing the so-called fifth season
front and center in the MnPave software, MnDOT is introducing to state and local
pavement engineers the idea that thickness concerns must be considered alongside and
understanding of climate and environment.
The MnPave software has been adopted across the state quite readily. Low-
volume road construction practices for the state of Minnesota were revised in 2002 (in
MnDOT Report 2002-17) to account for the existence of MnPave. Furthermore, MnPave
is used in a number of pavement engineering classes at UM to educate students on
pavements and introduce them to tools for pavement design. These and other steps taken
to promote MnPave ensure that regional mechanistic-empirical design methods,
especially those such as MnPave that account for seasonal variation, will be closely
studied and familiar to the next generation of pavement engineers.

3.5 Mechanistic-Empirical Design Guide

ME Design Brief 6
As the creators of the MEPDG recommend that local agencies should verify and calibrate
the MEPDG prior to implementation, MnDOT and UM used MnROAD data substantially
in adapting the MEPDG for Minnesota environmental conditions. The data generally
required in this process is that of: climate, traffic, construction/materials, pavement
performance, and pavement response. It goes without saying that MnDOT called upon
the resources of MnROAD to provide this data in considerable detail.
MnROAD climate data was useful in verification of the EICM temperature
predictions for concrete pavements. Weigh-in-motion (WIM) data collected on the
mainline test sections provided critical traffic data and allowed for a comparison of
MnROAD’s traffic versus the predicted traffic by the MEDPG’s assumed traffic. As the
MnROAD database has distinguished itself from other databases with the spectrum of
information it provides on the construction and constituents of its test sections, there was
no lack of material or construction data for the verification/calibration efforts. Finally, it
was in pavement response/performance data where MnROAD was particularly useful.
Forensic trenching work done at MnROAD provided very distinctive information
to the verification/calibration of MEPDG, so much so that the information recovered
from the forensic studies of rutted test sections was incorporated into the general rutting
models of the MEPDG. This information was discussed in reports by Isackson et al. and
Mulvaney and Worel. These reports contributed to MEDPG’s ability to predict rutting by
accounting for the structural failure of pavement system in rutting and examining the
rutting in various lifts of the pavement. This information was invaluable during
calibration of the MEPDG and will be detailed in later documents to come out of the
MEPDG project.

4 MnROAD Contributions to Mechanistic-Empirical Design

MnROAD’s most important contribution to the pavement community is arguably its
contributions to mechanistic-empirical design. MnROAD data has provided valuable
input to a variety of mechanistic-empirical design methods and models. More
importantly, MnROAD strain gauge data has been used to verify that layered elastic
theory, an assumption upon which most flexible design methods and models are based, is
a valid approximation of pavement response. This verification is taken for granted by so
many researchers that it is virtually ignored. However, in confirming that the
assumptions of layered elastic theory can be used as an approximation of pavement
response and thus are a valid basis for mechanistic-empirical design, MnROAD indirectly
bolstered all existing flexible design methods, including MnPave and MEPDG.
Also in line with this verification of layered elastic theory is MnROAD’s work in
evaluating existing pavement design methods for concrete (rigid) pavements. The work
in MnDOT Report 1997-14 was non-anecdotal evidence of the fact that the mechanistic-
empirical design methods of the early 1990s were not adequately predicting performance
of pavements, which led to many costly under- and over-designed roadways. Many
professionals in pavement engineering feel that this contribution alone justifies
MnROAD’s entire first ten years of existence—such is the need for non-anecdotal
evidence on a full-scale basis about pavement design methods.
MnROAD has also been a valuable source of quality data to professionals
working on pavement design. The most notable example of this is MnROAD’s

ME Design Brief 7
relationship with MEPDG. Not only has MnROAD data been used by the MEPDG, a
great deal of MnROAD expertise and observations have gone into the MEPDG as well.
For instance, the MEPDG software currently assumes that traffic volume is uniform
throughout the year. Analysis of MnROAD traffic data has shown that this assumption is
not valid for the WIM on the mainline test sections. Furthermore, as noted above,
MnROAD’s forensic trench studies provided the MEDPG with new insights on rutting.
In providing its data for studies in pavement design, MnROAD has most notably been of
great assistance to those looking to verify the predictions of a given method or model.
All of the above highlights feature some kind of verification, most notably that of
MnPave, which is built upon MnROAD data and expertise.

5 Recommendations
MnROAD’s contributions to pavement design methods and pavement modeling are one
of MnROAD’s strongest virtues in its first ten years. By making its data public and
openly discussing MnROAD experiences with its test sections, MnROAD engineers
made its data and pavement expertise a desired commodity for pavement engineers
seeking to verify or calibrate their method or model with empirical data. For this reason,
very few recommendations that have not been stated in other briefs are possible in
regards of additional contributions of MnROAD to pavement design methods or
pavement models. In following other Lessons Learned recommendations, such as those
addressing MnROAD’s database or its release of information, MnROAD will in turn
make its contributions to pavement design and modeling even more influential. So long
as MnROAD’s focus remains, in part, upon providing detailed climatic data; traffic
information; construction and material information; and pavement performance and
response data, MnROAD data will influence pavement design methods and pavement
models for decades to follow.

6 References
1. Alvarez, C. and M. R. Thompson. Mechanistic-Empirical Evaluation of the
MnROAD Mainline Flexible Pavement Sections. Report FHWA-IL-UI-263.
Illinois Department of Transportation, 1998.
2. Ariza, P. and B. Birgisson. Evaluation of Water Flow through Pavement Systems.
Report MN-RC-2002-30. Minnesota Department of Transportation, 2002.
3. Bao, W. Calibration of Flexible Pavement Structural Model Using MnROAD
Field Data. M.S. Thesis. University of Minnesota, Minneapolis, 2000.
4. Berg, R. L. Calculating Maximum Frost Depths at MnROAD: Winters 1993-94,
1994-95, and 1995-96. Report MN-RC-1997-21. Minnesota Department of
Transportation, 1997.
5. Bigl, S. R and R. L. Berg. Material Testing and Initial Pavement Design
Modeling: Minnesota Road Research Project. Report MN-RC-1996-23.
Minnesota Department of Transportation, 1996.
6. Bigl, S. R and R. L. Berg. Modeling of MnROAD Test Sections with the CRREL
Mechanistic Pavement Design Procedure. Report MN-RC-1996-22. Minnesota
Department of Transportation, 1996.

ME Design Brief 8
 7. Burnham, T. R. and W. M. Pirkl. Application of Empirical and Mechanistic-
Empirical Pavement Design Procedures to MnROAD Concrete Pavement Test
Sections. Report MN-RC-1997-14. Minnesota Department of Transportation,
1997.
8. Chadbourn, B., S. Dai, P. Davich, J. Siekmeier, and D. Van Deusen. MnDOT
Flexible Pavement Design - MnPave Beta Version 5.1. Commercially
Unpublished Report for the Office of Materials and Road Research. Minnesota
Department of Transportation, 2002.
9. Forst, J. Calibration of Rigid Pavement Structural Model Using MnROAD Field
Data. M.S. Thesis. University of Minnesota, Minneapolis, 1998.
10. Isackson, C., S. Dai, and D. Boerner (1999). “Minnesota Road Research Project
Forensic Investigation 199701.” Report MN-RC-1999-33, Minnesota Department
of Transportation.
11. Mateos, A. and M. B. Snyder. Validation of Flexible Pavement Structural
Response Models with Data from the Minnesota Road Research Project. In
Transportation Research Record: Journal of the Transportation Research Board,
No. 1806, TRB, National Research Council, Washington, D.C., 2002, pp. 19–29.
12. Mateos, A. M. Load Equivalency Factors from the Structural Response of
Flexible Pavements. M.S. Thesis. University of Minnesota, Minneapolis, 2000.
13. Mulvaney, R. and B. Worel (2001). “MnROAD Cell 26 Forensic Investigation.”
Internal Report to Office of Materials and Road Research, Minnesota Department
of Transportation,.
14. Mulvaney, R. and B. Worel (2002). “MnROAD Mainline Rutting Forensic
Investigation.” Internal Report to the Office of Materials and Road Research,
Minnesota Department of Transportation.
15. Newcomb, D. E., D. M. Johnson, H. B. Baker, and S. M. Lund. Minnesota Road
Research Project: Work Plan for Research Objectives. Report MN-RC-1990-03.
Minnesota Department of Transportation, 1990.
16. Ovik, J. M., B. Birgisson, and D. E. Newcomb (2000). “Characterizing Seasonal
Variations in Pavement Material Properties for Use in a Mechanistic-Empirical
Design Procedure.” Report MN-RC-2000-35, Minnesota Department of
Transportation.
17. Skok, E. L., D. H. Timm, M. L. Brown, and T. R. Clyne (2002). “Best Practices
for the Design and Construction of Low-Volume Roads.” Report MN-RC-2002-
17REV, Minnesota Department of Transportation.
18. Timm, D. H., B. Birgisson, and D. E. Newcomb. Mechanistic-Empirical Flexible
Pavement Thickness Design: The Minnesota Method. Report MN-RC-1999-10.
Minnesota Department of Transportation, 1999.
19. Timm, D. H., B. Birgisson, D. E. Newcomb, and T. V. Galambos. Incorporation
of Reliability into the Minnesota Mechanistic-Empirical Pavement Design
Method. Report MN-RC-1999-35. Minnesota Department of Transportation,
1999.
20. Zhang, Z., H. K. Stolarski, and D. E. Newcomb. Development and Simulation
Software for Modelling Pavement Response at MnROAD. Report MN-RC-1994-
31. Minnesota Department of Transportation, 1994.

ME Design Brief 9