Office of Operations Freight Management and Operations

Comprehensive Truck Size and Weight Limits Study: Pavement Comparative Analysis Technical Report

Appendix A: Revised Pavement Desk Scan Report

Table of Contents


Chapter 1: Introduction

This report presents a revised version of the Desk Scan (Subtask V.B.2) developed to support the Pavement Comparative Analysis (Task V.B.) of the 2014 Comprehensive Truck Size and Weight Limits Study (2014 CTSW Study). This revised Desk Scan addresses the recommendations made by the National Academy of Science (NAS) Peer Review Panel concerning the originally submitted version of this scan.

1.1 Purpose

The purpose of the revised Desk Scan is to:

  • Reorganize and enhance the original Desk Scan; and
  • Add any additional, relevant content that may have been identified since the submission of the original Desk Scan.

Specifically, the NAS Peer Review Panel recommended that the original Desk Scan be reorganized to address four issues:

  • Survey of analysis methods and a synthesis of the state of the art in modeling impacts
  • Identification of data needs and a critique of available data sources
  • Assessment of the current state of understanding of the impacts and needs for future research, data collection and evaluation
  • Synthesis of quantitative results of past studies including reasonable ranges of values for impact estimates.

The team reviewed the most relevant previous studies comparing pavement costs of vehicle use, including state, national, and international cost allocation and truck size and weight studies, as well as any other studies that include estimates of vehicle-induced pavement costs on either an absolute or relative basis. They also include pavement analysis or design studies that will help in the application of AASHTOWare Pavement ME Design® or in the compilation of data required for that model. The principal objective of the search was to gain a thorough understanding of the current state of research and practice concerning pavement cost analysis related to heavy vehicle use. The literature search included a variety of information sources: (1) engineering and scientific periodicals and journals; (2) conference proceedings; (3) federal, state, international, and university reports that show up in library search engines, such as Compendex, based on key words; and (4) studies identified during the May 29, 2013 public hearing for the 2014 CTSW Study or by USDOT officials.

Cost allocation studies develop detailed estimates of costs related to vehicle weights and other characteristics in somewhat more detail than a typical Truck Size and Weight (TSW) study, so this desk scan included both highway cost allocation (HCA) and TSW studies at the federal and state levels and in other countries. No studies with new methodologies were found, but several will help in the pavement cost analysis for this 2014 CTSW Study, and several more lend support or perspective to the proposed approach. Section 1.6 includes a list and brief synopsis of all publicly-available reports reviewed as part of this desk scan.

1.2 Overview of Alternative Approaches to Analyzing Pavement Costs

The pavement team’s review of previous studies and techniques for analyzing pavement costs associated with changes in traffic loads reveals approaches that fit into three broad categories: (1) using traditional "equivalent single axle loads" (ESALs) derived from the half-century-old AASHO Road Test as a measure of pavement damage, and therefore pavement damage costs, (2) applying pavement deterioration models to a representative group of pavement sections with a large number of traffic loading conditions to derive a new set of load equivalence factors (LEFs) and deterioration curves that vary by distress type, or (3) directly applying current pavement design models to a small number of sample pavement sections under scenario traffic loadings to derive estimates of changes in pavement life and therefore pavement cost changes. Each of these three alternative approaches is discussed below.

1.2.1 Using ESALs as a Measure of Pavement Damage

As used in this report, the term "ESAL" refers exclusively to the AASHO-Road-Test-based factors as calculated by formulas in the in the 1993 AASHTO Pavement Design Guide. All other factors are referred to by the generic term "LEF".  FHWA’s HCA and TSW studies stopped using unmodified AASHO-Road-Test-based ESALs in 1979, after the Congressional Budget Office (CBO) strongly criticized their continued use, based on the outdated assumptions used to derive the formula for ESALs, which was based on a short term test of a small set of pavement cross sections in a single environmental zone. It should be noted that only a limited range of axle types were included in the study, and the calculation of ESALs for tridem axles is based on extrapolating a dummy variable. Most, but not all, states followed the federal lead and discontinued use of ESALs for HCA studies, but typically continued to use them when they commissioned TSW studies.

By far the largest number of truck-size-related pavement studies in the past fit into the first category: using ESALs as an assumed determinant of pavement damage and deriving cost estimates in various derivative approaches based on that initial assumption. The most prevalent approach involves calculating the number of ESALs before and after the proposed or implemented changes in vehicle traffic loads, calculating either an average or marginal cost per ESAL through either a micro or macro approach, and simply multiplying the two factors.

The 2009 Wisconsin Truck Size and Weight Study provides an example of a typical ESAL-based approach. As stated in the study, the analysis used a four-step approach to estimate pavement (and, in this case, bridge deck) impacts of each size-and-weight scenario:

  • Step 1 – Estimate cost to highway agencies and other road users associated with an additional ESAL mile of travel for various types of highways and highway conditions;
  • Step 2 – Estimate ESALs as a function of operating weight for Base Case and Scenario trucks;
  • Step 3 – Calculate the change in ESAL miles due to freight shifting from Base Case to Scenario trucks; and
  • Step 4 – Calculate the change in pavement and bridge deck costs as the product of 1) the change in ESAL miles and 2) cost per ESAL mile.

One could argue that ESALs provide a reasonably credible job at describing the average effects of single and tandem axle loads under typical conditions, despite their general limitations cited above, since the Road Test did use a range of weights of each of those axle types, and since its measured variable, roughness, is a function of most of the common flexible and rigid pavement distresses. Since some of the Wisconsin scenarios, however, used tridem axle trucks, using ESALs that are based upon extrapolating a dummy vehicle for those vehicles is considerably less credible.

Recent studies in Virginia (Allen et al., 2010) and Kansas (Bai et al., 2010) derived statewide and corridor-specific, respectively, per-ESAL-mile estimates of variable pavement maintenance costs to use as basis for estimating the pavement cost impacts of heavier truck traffic. The Virginia study is notable in that it provides an excellent example of using current expenditures and imputed foregone maintenance costs to derive a statewide estimate of damage per truck-mile of travel. Although marred by the use of ESALs as the loading metric, rather than a more up-to-date set of LEFs, the overall approach of calibrating to actual costs has merit.

Since using ESALs as a basis for differentiating among trucks for national policy considerations is neither technically defensible nor politically feasible, the second and third types of approach have more potential for use in the CTSW.

1.2.2 Deriving Pavement Damage Relationships from Pavement Performance Models

Most of the studies that did not use ESALs used another form of axle load factors (ALFs) or load equivalence factors (LEFs), that were typically derived for a single particular distress to describe the relative damage by one axle weight and type compared to the damage of a standard axle. As used in this report,  the term "LEF" refers to a factor that describes the relative damage caused by one axle to a standard reference axle. ESALs, as used in this report, are a specific type of LEF. Some of the LEFs derived in previous studies were based on mechanistic primary-models, and some of these were calibrated to a small amount of observed empirical data. Some approaches based their LEFs on the MEPDG model in its various versions. Some of the studies used reduction in time-to-failure as the variable that determined LEFs, and some used target distress levels.

The 1997 Federal Highway Cost Allocation Study (HCAS) Final Report and the 2000 U.S. Department of Transportation Comprehensive Truck Size and Weight Study both used the same version of the National Pavement Cost Model (NAPCOM) for estimating the relative shares of pavement damage caused by each vehicle of a given type and operating weight.

NAPCOM was originally developed to enhance the 1982 HCAS approach of assigning costs to vehicles based on their estimated contribution to each pavement distress weighted by the importance of each distress to the need to repair or replace a pavement. Its earliest version used newly-developed mechanistic-empirical pavement models to derive a set of pavement damage equations for six of the most important distresses observed on each type of pavement (flexible and rigid). Each distress equation used axle types and weights as primary variables with the weight exponent independent of ESALs, so each the equation for each distress produced an LEF that varied not only by weight and type of axle, but could also vary by pavement, base, and climate characteristics as well. FHWA had updated NAPCOM regularly as new and better mechanistic-empirical pavement models became available.

The NAPCOM version used in the 1997 and 2000 studies included the third set of major updates to the damage equations. Unlike in earlier versions, however, FHWA used a simplified version of estimating axle weights for each operating group and vehicle configuration, rather than using an array of all observed axle weights in an effort to speed up run times. For each vehicle class and operating gross weight (OGW) group, axle weights for each position on a vehicle were expressed as an average of all observed weights at that position on the vehicle (steering axle, drive axle, first load axle, etc.). Given the non-linear aspect of most LEFs, this produced compromised results of unknown magnitude.

FHWA’s 2012 Vermont Pilot Program Report used a new set of LEFs that were developed in 2011 and 2012 as part of a major restructuring of NAPCOM. The LEFs had been fully incorporated into a spreadsheet version of NAPCOM designed for use at a state level. New damage models were based on running MEPDG thousands of times, systematically varying base traffic by one axle weight and type at a time to determine the relative effect that each has on pavement deterioration in a range of pavement types and climatic conditions. The LEFs were applied to the pavement sections in Vermont that were affected by the pilot project and weighted by the prevalence of each distress on these sections. The approach used all available WIM data and classification counts to determine before-and-after axle load spectra.

Oregon’s 2013 Highway Cost Allocation Study used the same LEFs in an adapted spreadsheet version of NAPCOM to assess appropriate levels of weight-distance tax rates and other user fees. Oregon uses smaller OGW increments than have been used previously in national studies--2,000-pound increments rather than 5,000 pounds-- a feature that greatly improves the precision of the results. The approach uses imputed axle weights for OGW groups on each side of a WIM observation to compensate for the smaller number of observations in each smaller OGW increment, an approach that seems to have merit. As in Vermont, LEFs were weighted based on available estimates of pavement distress prevalence, with the weighted LEF-miles used to allocate load-related pavement costs. The results allow assessment of the pavement damage costs associated with any particular type of vehicle at each OGW, so could be readily applied to a truck size and weight study.

A University of Texas paper (Bannerjee et al., 2013) applied a later version of MEPDG, DARWin-ME, in a similar manner to the approach used by FHWA for PaveDAT and the latest version of NAPCOM to derive what the authors termed "Equivalent Damage Factors" (EDFs). Focusing on flexible pavements and using a smaller number of distresses than the FHWA approach (combining AC rutting and other rutting and combining all cracking components), the study found that the EDFs for rutting varied with pavement thickness somewhat more than was observed in the FHWA study. A probable explanation is that the ratio of surface-to-total rutting changes as pavements get thicker; even though the AC layer and subsurface rutting LEFs stay fairly constant, the increased prevalence of AC layer rutting with thicker pavements changes the relative importance of the two LEF components.

A Michigan DOT study (Chatti, 2009) used laboratory studies and mechanistic models to determine "axle factors" (AF) for single, tandem, tridem, quadrem, and larger axle groups. AFs were defined as the ratio of damage of a full axle grouping to the average weight of each single axle in that grouping. AF values were then used to correct AASHTO-based LEF (ESAL) values for the average single-axle weight, in effect producing new LEFs that were not dependent upon the original ESAL values for multi-axle groupings. The study reinforced MEPDG-derived LEF results from other recent studies cited above: tridems and quadrems have less relative effect than tandems or singles (with the same weight per individual axle) on cracking, but a greater relative effect on rutting. The results could allow extension of mechanistic model findings to quadrems—currently outside the scope of AASHTOWare Pavement ME Design®.

All the previous approaches in this group of studies have relied upon damage relationships built from earlier mechanistic-empirical models or from earlier versions of AASHTOWare Pavement ME Design®. Although sound in concept, their use of now-superseded models may affect their credibility, which leads us to the third group of approaches that can be used for truck size and weight analysis.

1.2.3 Estimating Pavement Performance Directly from Models

Previous truck size and weight studies have not used this approach, partly because until recently there has not been a mechanistic-empirical model that has achieved such broad acceptance among pavement engineers as has AASHTOWare Pavement ME Design®. A 2007 FHWA-sponsored study (Timm et al., 2007) used this approach by applying MEPDG to a small number of hypothetical pavement sections, all having the same 15-inch crushed stone base, the same A-6 subgrade, and using the same MEPDG default Alabama climate file. For each of four traffic levels, flexible surface layer thicknesses were selected that resulted in reaching MEPDG-predicted terminal pavement conditions in about 20 years (24 years for the lowest–traffic section). Similarly, 15-foot rigid slab thicknesses were selected that reached terminal conditions in 20 to 28 years for each traffic level. All traffic levels used the same base case truck class and axle load distributions, varying only in annual average daily truck traffic (AADTT).

Base case predicted service lives were compared to predicted lives for three different loading scenarios: (1) shifting entire weight distributions toward heavier axles, (2) adding specific heavier axles, and (3) changing the GVW from 80,000 to 97,000 lbs. while adding an axle to the rear tandem group and using an idealized weight distribution for each vehicle type. The first scenario showed very large decreases in pavement life (and increases in cost), the second showed significant cost increases when the number of added heavy axles exceeds 10% of the number of legally loaded axles, and the third showed no practical difference between the 80,000-pound 5-axle vehicle and the 97,000-pound 6-axle vehicle. Mechanistic analysis outside of MEPDG showed only slight difference in pavement response, confirming the finding of insignificant changes in pavement costs.

The authors noted that their findings represent a limited set of conditions and that results were pavement-specific-- in several of the scenarios, one or two of the pavement sections showed much higher rates of change in service life than the other sections.  The authors did not report which terminal pavement condition (IRI, alligator cracking, AC rutting, or total rutting for flexible pavements; IRI, transverse cracking, or faulting for rigid) was reached first for each of the eight pavement sections included in the study. Subsequent work using the next MEPDG version to support FHWA’s NAPCOM and PaveDAT models showed large variation in the relative effects of axle weights among the various pavement condition metrics, so it is perhaps not surprising that different sections could show vary different results if two different failure mechanisms were involved.

A 2010 TRB paper (Tirado et al., 2010) coupled the use of a primary response model with damage predictions from mechanistic-empirical analysis to quickly estimate relative levels of distress by particular combinations of axle loads, including groups of more than three axles. Together with the Chatti 2009 paper cited earlier, this approach could perhaps extend damage analysis to axle groups with more than three axles in the group (tridems), a current limitation of the AASHTOWare Pavement ME Design® model. The team does not have the time or budget to apply the approach in this study, but may find it informative to tabulate the prevalence of multi-axles in our WIM analysis and apply a rudimentary approach for considering the likely effect of considering quadrem and larger axle groupings, rather than arbitrarily dividing them into tridem and tandem groupings as is current common practice.

Using models to directly estimate changes in service lives, building especially upon the lessons learned in the work by Timm et al. for FHWA, could potentially achieve the objectives of this study. Assuming accurate estimates of actual axle loading spectra under the base case and each scenario for this study, as well as a small number of tightly-defined scenarios, the approach would eliminate the need to estimate LEFs for the impact assessment phase of the study thereby potentially improving the accuracy of the results calculated in this study.

1.3 Data Requirements for Pavement Comparative Analysis

Each of the types of approaches outlined above requires a variety of data inputs, with some variations. The traffic data requirements are similar to the data needs of the bridge, safety, and modal shift analyses, except that the pavement and bridge phases need more detailed information on axle load distributions.

1.3.1 HPMS Section Data

Using any of three approaches to develop a valid national estimate of changes in pavement cost would rely upon detailed knowledge of the national highway system network characteristics and traffic levels. FHWA’s compilation of HPMS section data provides the best available collection of traffic estimates, single-unit truck traffic estimates, combination truck traffic estimates, pavement condition, pavement design, and age data currently available, although the level of detail about many of these parameters has to be supplemented with other data sources. Average daily truck traffic, for example, does not supply nearly enough information to properly apply any of the three approaches, so must be supplemented with vehicle classification and weigh-in-motion (WIM) data. All three approaches require approximately the same level of detail regarding traffic data.

An ESAL-based approach, although not recommended for this 2014 CTSW Study, requires the least supplemental detail for most of the non-traffic factors supplied by HPMS. ESALs vary by pavement type and thickness, which are supplied quite reliably by the section data, as well as by terminal PSI value, which can be assumed to be relatively uniform for all sections on a given highway class.

Applying a derived model such as NAPCOM requires not only the pavement type and thickness information required by an ESAL-based approach, but also detailed information about pavement condition, since modern LEFs vary by distress type. If a rigid pavement section fails by faulting, for example, LEFs increase much less with axle weight than if the section fails by cracking. Unfortunately, HPMS has only recently added information about the states of pavement distress, and states have been somewhat slow in supplying the information. Also, HPMS fails to distinguish between top-down and bottom-up cracking, or between AC surface rutting and total rutting, and LEFs for each of these distresses have significantly different exponents and significantly different offsets among the axle types (single, tandem, and tridem). Thus, the HPMS data has to be supplied with ad hoc data from other sources—either special studies or information from state pavement engineers. Some of this supplemental data has been gathered from a few states, but more information is needed.

Applying AASHTOWare Pavement ME Design® requires much more detail about a pavement structure and its material properties than is available on HPMS, as discussed below in section 1.3.4.

1.3.2 Vehicle Classification Data

All approaches require obtaining as much vehicle classification data as possible—whatever FHWA can provide and deems appropriate for initial estimates of truck travel for broad classes of trucks in each state on functional class. FHWA no longer publishes or compiles formerly available HPMS area wide travel counts reported by the states for the 13 HPMS vehicle classes on each highway system, but will provide VMT estimates for regional groups of states for six broad summary vehicle classes (two truck and four passenger vehicle classes). State-reported raw vehicle classification station counts are available to support further break down of FHWA’s summary VMT reports, as has been done in previous cost allocation and size and weight studies.

Raw classification data contains many errors and inconsistencies, as well as a strong tendency in most states to use class 13 as a "catchall" category. The data needs to be not only reviewed and edited, but also systematically corrected using the additional information available from WIM data. Further, FHWA is able to supply slightly over 1400 stations that have a continuous year’s worth of 24-hour data (a necessary criterion to avoid temporal bias) to provide a high degree of accuracy in estimating travel by detailed vehicle class. Triple this number of stations would significantly increase the accuracy of detailed truck travel estimates. The inadequate number of classification stations and the lack of adequate quality control in the reported data are perhaps the greatest data limitation in the 2014 CTSW. This data inadequacy affects all three approaches equally, since all three depend upon accurate estimates of travel by detailed vehicle class upon which to project axle weight distributions provided by WIM data.

1.3.3 Weigh-in-Motion (WIM) Data

All three approaches also require the same level of detail on operating weight and axle weight distributions, so will make use of all available WIM data compiled by FHWA for multiple purposes in this 2014 CTSW Study, as well as the most recent years of WIM data collected for LTPP. Previous compilations of national travel estimates and truck travel characteristics have frequently used the most recent consecutive 12 months of WIM data for each state in order to lessen potential seasonal bias. There may be value in using 12, 24, or 36 months of consecutive data from the WIM sites, since that many years are generally available and easily obtained.

In general, there is much more WIM data available than ever before, increasing the accuracy of estimating the distributions of operating weights and axle loads at each station. WIM data limitations stem mostly from the insufficient number of stations reported to FHWA, as well as the lane bias of the stations. Nearly all the 19 LTPP WIM sites and the 451 FHWA-compiled, state-reported WIM sites, for example, systematically erase data collected from light vehicles (which negates the opportunity to estimate the percentage of trucks in the traffic stream) and collect WIM data from only the right lane on four-or-more-lane highways (which may bias truck type and weight estimates). Further, the potentially large differences in north-south vs. east-west truck traffic cannot be accounted for in the absence of a comprehensive truck-travel network and a sufficient number of WIM stations to populate that network.

1.3.3.1 Detailed Vehicle Class Travel Estimates

Since raw WIM data reported to FHWA or under the LTPP program includes axle weights and distances between axles for each observed vehicle, vehicle classifications provided by the standard axle-spacing algorithms used by the states can be corrected based on the additional information. Also, the data provides enough detail to sub-classify the 13 standard classification vehicle classes into the more detailed classes required by the 2014 CTSW Study. In previous FHWA studies, individual WIM observations have been evaluated for validity based on the reported axle weights and spacings, and either reclassified or rejected according to explicit edit criteria. The team will work with FHWA and the pavement team to update, refine, and adjust these edit criteria for this 2014 CTSW Study based on the collective expertise.

1.3.3.2 Operating Gross Weight (OGW) Distributions for Each Vehicle Class

Based on the refined WIM-record edit criteria, the team will compile the operating weight distributions for each detailed truck class in each state and on each available highway class. Ideally, each state would report enough WIM data to FHWA to allow independent operating weight distributions for each vehicle class on each type of highway. In most cases, however, states collect WIM data on Interstate and arterial highways, especially rural arterial highways. Also, many states do not have enough use by some of the vehicle classes, since some are allowed only by special permit or not at all. Therefore, the team has to group highway types and sometimes states to develop valid OGW distributions for many vehicle classes. The team will take care to distinguish among states with varying weight regulation on Interstate and non-Interstate highways in developing the estimates of OGW distributions.

1.3.3.3 Axle Weight and Type Distributions

Axle weights and types have large effects on pavement deterioration and service life. WIM data provide an excellent source of knowledge about the actual distribution of axle weights for the weight groups in each vehicle class, so that the pavement team does not have to use unrealistic idealized axle weights to typify a weight class. For example, an 80,000-pound 3-S2 is often characterized as having a 12,000-pound steering axle and two 34,000-pound tandem load axles. If the actual distribution of axle weights is 10,000 / 37,000 / 33,000-pound, however, the vehicle will cause significantly more pavement damage than would be estimated by the standard weight distribution.

For consistency with Pavement ME Design® traffic input requirements, the team will tabulate axle weight frequencies in 1,000-pound weight groups for steering axles and single load axles, 2,000-pound increments for tandem axles, and 3,000-pound increments for tridem axles, and will develop separate frequency distributions for each weight group and each vehicle class.

1.3.4 Pavement Design and Materials Data

An ESAL-based approach requires only rudimentary pavement design information (pavement type, thickness, and terminal PSI), since those are the only design variables in the ESAL equations. HPMS supplies type and thickness information, and reasonable assumptions can be made about terminal PSI values as a function of highway class.

A NAPCOM-based approach requires primarily observed distress data, since LEFs used in NAPCOM vary mostly by pavement type and distress type, and much less by climate, base type, and other design details. HPMS supplies much of the information needed for these subtle variations, but is only just beginning to supply rudimentary pavement distress data. States are including distress information for an increasing number of pavement sections, thereby improving the accuracy of NAPCOM’s cost-share estimates. A fundamental source of uncertainty, however, stems from the HPMS data reporting structure that fails to distinguish between (a) bottom-up and top-down cracking (for both rigid and flexible pavements), and (b) surface and total rutting (for flexible pavements). In each case, LEFs for the grouped distresses vary widely, so not knowing the relative importance of each component compromises the accuracy of the estimates.

Directly applying AASHTOWare Pavement ME Design® to a set of pavement sections requires a large number of pavement design details, soil data, and other materials data. The software package includes the climatic data needed for proper program operation, and includes a large quantity of nationally-derived default data for nearly everything else. To properly analyze the sample pavement sections, the team needs to carefully match materials, design, and calibration parameters to a representative set of pavement cross-sections in a representative range of climates. Fortunately, sample LTPP sections have developed all the required information, so will provide the details for the selected sample sections.

1.4 Needs for Future Research and Data Collection

The accuracy of future TSW (and HCA) studies could be improved if the most significant data deficiencies and analytic uncertainties were lessened. The following two sections describe what the study team believes are the greatest sources of uncertainty—both of them current data deficiencies.

1.4.1 Vehicle Classification Data

All three approaches depend heavily upon detailed knowledge about the types of vehicles using the national highway system—knowledge that currently depends on periodic ad hoc analysis of large quantities of WIM and classification data. Both WIM data and classification data have their deficiencies, as does an analysis system that does not continuously compile and evaluate the state-reported data so that it can be compared from year to year and better evaluated as it is submitted.

WIM data collection alone cannot provide vehicle class travel estimates by itself unless the 400 or so stations reported annually: (a) increase in number by at least 10-fold, (b) collect data in all lanes of a multi-lane roadway instead of just the right lane, (c) report data for all vehicles instead of screening out light vehicles, and (d) are located more rationally—either randomly or as part of a truck transportation network. Since most of these improvements are unlikely, vehicle classification data is likely to be a necessity far into the future.

Current classification data falls far short of what is needed for accurate estimation of travel by vehicle class. States report far too few stations, do not adequately review or edit the data, and do not report the weighted importance of each station. FHWA does not attempt to annually compile the information and report detailed truck class travel as part of the Highway Statistics series, which would go a long way to improving the quality of data that now seems to only be compiled on an ad hoc basis every five years or so.

1.4.2 Pavement Condition Data

Applying NAPCOM or a similar model requires information about pavement distresses on the national highway system, since LEFs vary substantially depending upon distress type. Current HPMS section data needs to be more complete in order to apply an approach of this type, and distinctions need to be made between rutting components (surface and total rutting) and among cracking components (bottom-up, top-down, transverse), either through statewide reporting or special studies. The other two approaches do not necessarily require detailed information about pavement distresses observed on the highway system.

1.5 Linkage with Project Plan

Based on evaluating previous studies and available current models and approaches, as described in the previous sections, the 2014 CTSW pavement comparative analysis will focus on a small number representative pavement sections covering a range of locations with varying climates, pavement types, pavement types, and surface thicknesses. The AASHTOWare Pavement ME Design® model will be used in this analysis and run for each of these sections to determine a base case of the expected pavement performance under traffic conditions appropriate for each thickness (mix of vehicle types and operating weights as well as truck traffic levels). Locations will be selected that avoid climate extremes and thus represent typical weather effects several groups of states. To the extent possible, Long Term Pavement Performance Program (LTPP) sections will be used as a basis for each sample section and will adjust base case parameters as required to make sure that each sample section represents the pavement performance history that would typically be expected.

For each sample section, the first step will be to perform a base traffic performance analysis. Next, traffic inputs will be varied in ways that represent traffic shifts that occur as a result of the various truck scenarios. This will require a series of runs of Pavement ME Design® during which all factors except traffic are held constant.

The multiple runs for each sample section will enable an evaluation of changes in pavement service life as a result of changes in truck travel associated with each modal shift scenario. These changes in pavement service life will be translated into pavement cost changes associated with size and weight scenarios using rudimentary life cycle cost analysis. The approach used in the project plan coincides with the third approach outlined in this report, "Estimating Pavement Performance Directly from Models."

The first approach, "Using ESALs as a Measure of Pavement Damage" is ruled out because it relies on ESALs—widely discredited because (a) calculating ESALs for tridems has no empirical or theoretical validity since it requires extrapolating a dummy variable, (b) ESALs apply only to roughness, which has many components that vary in their sensitivity to magnitude of axle load, and (c) ESALs derive from the AASHO Road Test, performed long ago as a short term performance test in a single location.

The second approach, "Deriving Pavement Damage Relationships from Pavement Performance Models" is ruled out because it (a) relies upon LEFs derived from an earlier version of AASHTOWare Pavement ME Design® that need to be verified using the latest version, and (b) requires an inventory of distress observations that is currently incomplete.

1.6 Comparison of Results with Previous Studies

Unlike most other recent truck size and weight studies, the 2014 CTSW Study contained some scenarios that result in anticipated increases in average axle loads and some that resulted in decreases. In the 2000 CTSW Study, all scenarios resulted in significant reductions in average axle loads, as did the 2004 Western Uniformity Scenario Study and state studies in Minnesota and Wisconsin. Only the Vermont pilot study resulted in increases in average axle loads.

Table 1 contains summary results from each of these recent state, regional, and national studies. Note that, as might be expected, scenarios with lower average axle loads tended to see reduced pavement costs, while cases with higher average axle loads tended to show increased costs. Note, however, that some scenarios resulted in somewhat more subtle interactions between reduced VMT and increased average loads per axle. Average axle loads, after all, are not as important as the distribution of axle loads at the higher ends of the axle load range, given the non-linearity of pavement damage as a function of axle load.

Table 1: Summary Pavement-Related Analysis Results
Study Vehicles and Weights Analyzed
k = thousands of pounds
Change in Truck VMT Change in Pavement Costs
Nationwide Studies
USDOT, Comprehensive Truck Size and Weight Limits Study (2014) 3S2-88k
3S3-91k
3S3-97k
DS5 33s-80k
TS7-105.5k
TS9-129k
-0.6%
-1.0%
-2.0%
-2.2%
-1.4%
-1.4%
+0.4%
-2.4%
-2.6%
+1.8%
+0.1%
+0.1%
USDOT, Comprehensive Truck Size and Weight Study (2000) 3S3-90k; DS9 33s-124k
3S3-97k; DS9 33s-131k
RMD-120k; TPD-148k; Triple-132k
Triple-132k
-10.6%
-10.6%
-23.2%
-20.2%
-1.6%
-1.2%
-0.2%
0.0%
Regional Studies
USDOT,  Western Uniformity Scenario Analysis (2004) RMD-129k; TPD-129K; Triple-110k -25% -4.2%
WsDOT, Wisconsin Truck Size and Weight Study (2009) 3S3-90k
3S4-97k
SU7-80k
DS8-108k
3S3-98k
SU6-98k
-0.4%
-1.2%
-0.5%
-0.02%
-0.4%
-0.04%
-$14.6 M
-$19.9 M
-$1.5 M
-$16.8 M
-$10.2 M
-$0.3 M
FHWA, Vermont Pilot Program Report (2011) SU3-55k; SU4-69k; CS5-90k; 3S3-99k expanded to Interstate for one year +1.7%, Int
-1.5% Non-I
+12%, Int
-0.5%, Non-I
MnDOT, Minnesota Truck Size and Weight Project (2006) 3S3-90k
3S4-97k
3S3-2-108k
SU6/7-80k
Not
Reported
-$1.3 M
-$2.2 M
-$1.3 M
-$0.6 M

1.7 Summary of Publicly-Available Reports Reviewed

Below are listed all studies reviewed as part of this process including comments about the utility of each study for this project. The first four groups of reports include readily-available reports that were identified through web search or prior knowledge, while group 5 includes studies suggested at the May 29, 2013, 2014 CTSW Study’s Public Hearing webinar that in some cases were less easily located.

(1) Using ESALs as a Measure of Pavement Damage

Allen, Gary, Audrey Moruza, and Brian Diefenderfer, Oversize and Overweight Vehicle Studies. Virginia DOT Presentation to the Transportation Accountability Commission, August 4, 2010.

http://dls.virginia.gov/GROUPS/transaccount/meetings/080410/oversize.pdf

Researchers used the array of all axle weights from Virginia WIM data, as well as historical expenditure data, to determine an average cost per ESAL-mile of travel. If overweight vehicles are charged only for the extra costs (beyond the legal axle load limit), they would be assessed 3.56 cents per ESAL mile, but that rate needs to be reviewed and updated over time as truck characteristics change. This is one of the few studies that attempted to calibrate relative pavement damage to actual expenditures and imputed costs, but the ESAL assumption requires an updated form of the analysis to reflect better current knowledge. Replacing the ESALs with updated, distress-specific load equivalence factors could overcome this limitation and make the report findings useful for truck size and weight studies.

Bai, Yong, Steven D. Schrock, and Thomas E. Mulinazzi, Estimating Highway Pavement Damage Costs Attributed to Truck Traffic. Mid-America Transportation Center, Report # MATC-KU: 262. 2010.

http://matc.unl.edu/assets/documents/matcfinal/Bai_EstimatingHighwayPavementDamageCostsAttributedtoTruckTraffic.pdf

Sponsored by the USDOT University Transportation Centers Program, this University of Kansas study collected highway data on 41.13 miles of U.S. Highway 50/400 in Kansas, and applied HERS and AASHTO methods to derive average maintenance expenditures per ESAL mile. This became the basis for estimating the additional costs that would be associated with an increase in meat packing truck traffic. As with the study by Allen et al., the ESAL assumption makes the findings of only general interest to the current 2014 CTSW Study, since we now know that ESALs do not adequately measure the relative effects of tridems, particularly.

Fortowsky, J. Keith, and Jennifer Humphreys, "Estimating Traffic Changes and Pavement Impacts from Freight Truck Diversion Following Changes in Interstate Truck Limits," Transportation Research Record: Journal of the Transportation Research Board, No. 1966, TRB. National Research Council. Washington, D.C. 2006, p. 71.

This TRB paper assumes all pavement damage is directly related to ESALs. We will not be using the assumptions necessary to rely upon ESALs, as cited above, so the study does not help us in our current 2014 CTSW Study.

Hajek, Jerry J, Susan L. Tighe, and Bruce G. Hutchinson. "Allocation of Pavement Damage Due to Trucks Using a Marginal Cost Method." Transportation Research Record: Journal of the Transportation Research Board, No. 1613, Paper # 98-1283. TRB. National Research Council. Washington, D.C., 2008.

https://journals.sagepub.com/doi/10.3141/1613-07

Ontario Ministry of Transportation determined the marginal cost of providing pavement structure for one additional passage of an ESAL on various roads, and found that a typical additional truck mile resulted in marginal costs that varied significantly across the highway system, ranging from a low of C$0.004 per km ($0.006 / mile) on a southern Ontario freeway to C$0.46 per km ($0.72 / mile) on a local road. Ontario used standard Road-Test-derived ESALs for single and tridem axles, and used elastic layer theory to extend the Road Test results to derive ESALs for other axle groupings. Unlike FHWA’s cost allocation procedures, however, which used average ESAL costs, Ontario’s method uses marginal ESAL costs for the particular heavy vehicles of interest. Thus, the overweight vehicles receive the full benefit of the existence of other heavy vehicles, which is much more significant on major highways than on lightly-traveled local roads-- hence the much higher difference in costs than usually appears in U.S. analysis. We do not suggest reviving the incremental design approach (abandoned for pavement cost analysis in this country in the 1970s), and cannot use the ESAL assumption, so the findings are of only general interest to the current 2014 CTSW Study. The wide scatter in the results, however, by type of roadway provides a cautionary tale to using only a small number of pavement sections without considering the context of a national sample of pavement sections.

Roberts, Freddy L., Aziz Saber, Abhijeet Ranadhir, and Xiang Zhou. Effects of Hauling Timber, Lignite Coal and Coke Fuel on Louisiana Highways and Bridges, LTRC Report No. 398. USDOT. March 2005. http://www.ltrc.lsu.edu/pdf/2005/fr_398.pdf

Using the 1986 AASHTO Design Guide and standard ESALs shows that heavier tandem axles (up to 48 kip) require additional overlay thickness and reduce pavement life. The current $10 annual overweight fee for an 86 kip 3S2 timber truck in Louisiana should be raised to many times higher per year if the axles are evenly loaded, and much higher, still, if the 48 kip axle is permitted. Allowing 100 kip trucks should not be permitted because pavement overlay costs double compared with an 86 kip truck. The ESAL assumption makes the findings of only general interest to the 2014 CTSW Study, since we will not be assuming that ESALs adequately measure relative effects of axle loads, for the reasons cited above.

Saber, Aziz, Mark Morvant, and Zhongjie Zhang. "Effects of Heavy Truck Operations on Repair Costs of Low Volume Highways". Presented at TRB 200 Annual Meeting, on CD-ROM of 2009 Meeting Proceedings. January 2009.

https://trid.trb.org/view/880573

Using standard ESALs, the study analyzed two vehicle types and three gross weights and concluded that 100 kip sugarcane trucks should be paying an annual fee of many times higher than their current annual fee if they are use the standard 3S2 configuration, but would not need an increase in that fee if they use a 3S3 configuration. The ESAL assumption makes the findings of only general interest to this 2014 CTSW Study, for the reasons cited above.

Study of Impacts Caused by Exempting Currently Non-exempt Maine Interstate Highways from Federal Truck Weight Limits, Appendix C: Pavement Cost Impacts, Development Process for the Study Network, Wilbur Smith Associates Study Team, June 2004.

This report assumed that all pavement damage is related to ESALs, so has limited information useful to this 2014 CTSW Study, for the reasons cited above.

Wisconsin Truck Size and Weight Study: Final Report. Prepared for Wisconsin Department of Transportation by Cambridge Systematics with National Center for Freight and Infrastructure, University of Wisconsin- Madison and Others. June 15, 2009.

https://rosap.ntl.bts.gov/view/dot/17499

Pavement analysis considered differential effects of traffic under various temperature and moisture conditions, and effects of load and non-load factors, but assumed that all vehicle-related damage is measured by and related to traditional ESALs and that the Road Test ESALs can be extended to tridems by extrapolating a dummy variable from a regression equation. The ESAL assumption makes the findings of only general interest to this 2014 CTSW Study, for the reasons cited above.

(2) Deriving Pavement Damage Relationships from Pavement Performance Models

Bannerjee, Ambarish, Jorge A. Prozzi, and Prasad Buddhavarapu, A Framework for Determination of load Equivalences Using DARWin-ME, Paper Number 13-1770, TRB 2013 Annual Meeting, on CD-ROM of 2013 Meeting Proceedings. January 2013.

The study used DARWin-ME to compute Equivalent Damage Factors (EDF) consisting of two partial factors: Axle Load Factor (ALF) and Group Equivalency Factor (GEF), based on pavement responses that result in the same distress level, following a procedure used earlier by an FHWA research project. The overall load equivalency for a truck is equal to the sum of the EDFs for each constituent axle group. Three AC distresses were analyzed: rutting, fatigue cracking, and roughness. After analyzing EDFs for a wide range of AC pavement designs, the authors concluded that there is little evidence that EDFs are affected by structural capacity for the latter two distress types. For rutting, however, EDFs had in inverse relationship with thickness for single axles, while EDFs of multi-axle groupings peaked for structural numbers between 3.5 and 4.0. The findings verify findings of the LEF derivations for the updated NAPCOM and PaveDAT models, and variation of thickness adds a nuance that will be useful in this base pavement section design.

Chatti, Karim. "Effect of Michigan Multi-Axle Trucks on Pavement Distress." Michigan DOT and Michigan State University, Final Report, Executive Summary, Project RC-1504. February 2009.

https://www.michigan.gov/documents/mdot/MDOT_Research_Report_RC-1504__ExecSum_272183_7.pdf

Laboratory studies were used to determine axle factors (AF) for each tridem (and larger) grouping at each weight. AFs were defined as the ratio of damage of a tridem, for example, to a single axle weighing one-third as much. The AFs were multiplied by the ESALs for each axle grouping on a truck and subbed to derive a truck factor (TF). When combined with empirical data on selected Michigan highways with flexible pavements, the study concluded that tridems (and n-groups) had less relative effect on cracking but more relative effect on rutting than single or tandem axles of an equivalent weight per axle. The results of this study could be useful in extending study findings to quadrem and larger axle groupings.

Ioannides, Anastasios M., and Lev Khazanovich, "Load Equivalency Concepts: A Mechanistic Reappraisal." Transportation Research Record: Journal of the Transportation Research Board, No. 1388, pp. 42-51. TRB. National Research Council. Washington, D.C., 1993.

The paper reviews the evolution of load equivalency concepts, both prior to and after the 1958 -1960 AASHO Road Test. The Road Test’s mechanistic-empirical ESAL concept varies considerably from the purely mechanistic equivalent single-wheel load (ESWL) and equivalent single-axle radius (ESAR) approaches. The latter mechanistic approach, however, appears to offer advantages over either of the other two approaches. When the paper was written, load equivalency factors (LEFs) were vital for designing pavement for mixed traffic, since they allowed the relative effects of each vehicle to be incorporated into design. To the extent that mechanistic-empirical models become prevalent for design, however, a truck size and weight study can avoid the use of LEFs if there is no need to report the relative effects of various vehicles on pavement life.

Nicholas, John, Roger Mingo, Mark Berndt, and Eulois Cleckley, Pavement Damage Analysis Tool (PaveDAT) for Overweight Truck Permit Calculation, Talking Freight Seminar Series, June 12, 2012.

https://www.fhwa.dot.gov/planning/freight_planning/talking_freight/june202012.cfm

PaveDAT builds upon the National Pavement Cost Model (NAPCOM) and the improvements made to it in recent work by FHWA. New damage models were based on running MEPDG thousands of times, systematically varying traffic to determine the relative effect that each type and weight of axle has on pavement deterioration in a full range of pavement types in a full range of climatic conditions. PaveDAT is a simplified version of the complicated, nationally representative NAPCOM model, but uses the same relative damage factors. These new load equivalence factors (LEFs) are similar in concept to the traditional ESAL concept, but vary widely across the important distresses for each type of pavement. PaveDAT was applied in the District of Columbia in a recent assessment of the costs associated with overweight vehicles.

1997 Federal Highway Cost Allocation Study (HCAS) Final Report, FHWA. https://www.fhwa.dot.gov/policy/hcas/final/index.htm

New pavement costs were allocated to vehicles based on the same minimum pavement approach used in the 1982 HCAS, wherein costs of pavement thickness above a sidewalk or bikeway standard are assigned to vehicles based on traditional ESALs. Costs for pavement reconstruction, rehabilitation, and resurfacing (about 25% of all federal obligations) were allocated using the latest version of NAPCOM, following the 1982 approach of assigning costs to vehicles based on their estimated contribution to each pavement distress weighted by the importance of each distress to the need to repair or replace a pavement. For both types of cost, FHWA developed estimates of travel by vehicle class and operating weight group. Unlike in 1982, however, FHWA used a simplified version of estimating axle weights for each operating group and vehicle configuration, rather than using an array of all observed axle weights. The team intends to use an array of axle weights for each weight group and configuration, rather than a regression equation describing the average weight of each axle.

Highway Cost Allocation Study: 2013 - 2015 Biennium, Final Draft. Prepared for Oregon Department of Administrative Services, Office of Economic Analysis by ECONorthwest, with R.D. Mingo and Associates, Jack Faucett Associates, HDR Engineering, and Mark Ford. January 2013.

https://www.nrc.gov/docs/ML1430/ML14308A129.pdf

Every two years, Oregon evaluates its anticipated highway program and its current highway usage patterns to determine how to adjust user fees to match highway user cost responsibilities. As in 2011, a new version of NAPCOM / PaveDAT was adapted to vehicle classes weight categories, and simplified highway classes, was updated to include the most recent Oregon WIM and pavement condition data, and was used for pavement cost allocation.

Trucks and Infrastructure Maintenance Costs. State Smart Transport Initiative. Undated

https://ssti.us/wp/wp-content/uploads/2011/11/Trucks%20and%20Infrastructure%20Maintenance%20Costs.pdf

Compiles truck estimated per-mile pavement costs from a variety of cited sources, including CBO and FHWA reports. May be of general interest to this 2014 CTSW Study as a point of comparison for baseline per-mile pavement cost estimates.

U.S. Department of Transportation Comprehensive Truck Size and Weight Study, FHWA. August 31, 2000. https://www.fhwa.dot.gov/reports/tswstudy/

The study found that pavement wear is an important area of interest in conducting truck size and weight studies because rough pavement affects the cost of travel via vehicle operating costs, delay costs, and crash costs. Pavement wear increases with axle weights and the number of axle loadings applied to a pavement. To analyze the magnitude of changes in pavement wear given alternative mixes of weights and axle configurations, the study used the same version of NAPCOM that was used in the 1997 HCAS, using the same baseline estimates of travel by vehicle class and operating weight group and the same simplified version axle weight distributions. The team recommends using an array of axle weights for each weight group and configuration for this 2014 CTSW Study, but will attempt to modify the older study’s approach of using axle weights and types as the primary units of analysis in favor of considering all the axle weights and types in each operating weight group as a single unit.

Vermont Pilot Program Report, FHWA Report to Congress Required by P.L. 111-117, 2012.

https://ops.fhwa.dot.gov/freight/sw/reports/vt_pilot_2012/vt_pilot.pdf

Vermont raised size and weight limits on its Interstate highways for one year beginning in December 2009. This study estimated traffic and infrastructure impacts and energy consumption and compared them to the pre-pilot (control) case. For pavements, the study team used an expanded version of the PaveDAT model, with its newly derived, distress-specific LEFs. Since traffic shifted mostly to 4-axle single units and 6-axle combination trucks as a result of the temporary allowance of 51 kip tridems on the Interstate system, pavement damage attributable to these vehicle classes increased considerably. Pavement damage on the Vermont Interstate system increased by 12 percent, which translates to significant increases in pavement maintenance and repair costs and more frequent work zones. There was a negligible decrease (less than 0.5%) in pavement damage off the Interstate system.

(3) Estimating Pavement Performance Directly from Models

Timm, David H., Rod E. Turochy, and Kendra D. Peters. Correlation between Truck Weight, Highway Infrastructure Damage and Cost. Auburn College of Engineering for FHWA, DTFH61-05-Q-00317, Subject No 70-71-5048. October 2007. http://www.eng.auburn.edu/files/centers/hrc/DTFH61-05-P-00301.pdf

Using MEPDG for a small sample of pavement sections, the study team determined the time until terminal pavement distress for a base case of traffic, then under three different loading scenarios: shifting entire weight distributions toward heavier axles, adding specific heavier axles, and changing the GVW from 80,000 to 97,000 lbs. while adding an axle to the rear tandem group. The first scenario showed very large decreases in pavement life (and increases in cost), the second showed significant cost increases when the number of added heavy axles exceeds 10% of the number of legally loaded axles, and the third showed no practical difference. Mechanistic analysis outside of MEPDG showed only slight difference in pavement response, confirming the finding of insignificant changes in pavement costs. The authors noted that their findings represent a limited set of conditions and that results were pavement-specific. They recommended that future work identify other loading scenarios for MEPDG simulation and establish a methodology to more accurately predict changes in loading spectra. FHWA followed up on these recommendations and initiated a project that systematically varied axle loadings for a larger number of pavement sections, and derived a general set of findings that could apply to any set of traffic shift scenarios (see Nichols et al., above). If the team cannot successfully use the primary approach proposed in this work plan, based on the pilot pavement section, the findings of this study and especially the follow-up study are directly applicable to the proposed back-up approach.

Timm, David, and Kendra Peters. Effects of Increasing Truck Weight Limit on Highway Infrastructure Damage. ICWIM 5, Proceedings of the International Conference on Heavy Vehicles: 5th International Conference on WIM of Heavy Vehicles, March 2013.

http://road-transport-technology.org/HVTT10/Proceeding/Papers/Papers_WIM/paper_123.pdf

Using MEPDG, no change in pavement life was found under idealized vehicle loading conditions when the same weight of freight was carried on a 97,000-lb vehicle or an 80,000-lb. vehicle. The idealized loading assumption makes the study unusable for this analysis, since we will be using actual observed axle weights as the basis for analysis, and the results are likely to be very different.

Tirado, Cesar, Cesar Carrasco, Jose M. Mares, Nasir Gharaibeh, Soheil Nazarian, and Julian Bendaña. "Process to Estimate Permit Costs for Movement of Heavy Trucks on Flexible Pavements." Transportation Research Record: Journal of the Transportation Research Board, 2154, pp. 187-196. TRB. National Research Council. Washington, D.C., 2010.

http://pustaka.pu.go.id/files/pdf/BALITBANG-03-C000066-610032011103843-process_to_estimate_permit_cost.pdf

The paper describes use of a primary-response model, coupled with damage predictions from a mechanistic-empirical analysis, to quickly estimate relative levels of distress caused by particular combinations of axle loads. It is interesting for the current study, since it does not vary traffic within the M-E model, but external to the model, thus allowing much more rapid estimation of the relative effects of axles based solely on their primary responses. The approach certainly has merit, but expanding it to this 2014 CTSW Study would require a fairly major research effort that is probably beyond the scope, since we do not have enough calendar time or staff-hour budget to substantially extend the findings of AASHTOWare Pavement ME Design

Zapata, C., and C. Cary. Integrating the National Database of Subgrade Soil-Water Characteristic Curves and Soil Index Properties with the MEPDG. National Cooperative Highway Research Program Project 9-23B, Preliminary Draft Final Report, National Research Council, Washington, D.C., 2012.

http://onlinepubs.trb.org/onlinepubs/nchrp/docs/NCHRP09-23B_FR.pdf

Findings of this report, and, more importantly, the associated ASU Soil Maps software tool can be used to establish what the substructure properties will be for any of the sites analyzed in any (and within any) of the four LTPP climatic regions evaluated. The team will use this report in compiling the data necessary for each pilot section.

(4) Description of Modeling Techniques, Potential Improvements, and Inputs

Cenek, P., R. Henderson, I. McIver, and J. Patrick, Modelling of Extreme Traffic Loading Effects. Opus Central Laboratories for New Zealand Transport Agency, Research Report 499. October 2012.

http://www.nzta.govt.nz/resources/research/reports/499/docs/499.pdf

The study investigated the premature failure of low-volume, low-strength roads that were sometimes associated with significant increases in heavy truck traffic on New Zealand highways, as might occur with road detours or with new mining or forestry operations. A key finding of the study was that extreme traffic loading does not immediately show increased distress or added maintenance costs. Thus, traffic deterioration models are more useful than examining historical pavement management data in assessing vehicle-related pavement costs. Although the findings are not directly applicable to this 2014 CTSW Study, the amount of scatter in the data as well as the length of time needed to observe accelerated wear serve as cautionary tales in analyzing the effects of heavy trucks on pavements via solely empirical data.

Chatti, Karim, Hassan Salama, and Chadi El Mohtar. "Effect of Heavy Trucks with Large Axle Groups on Asphalt Pavement Damage." Presented at 8th International Symposium on Heavy Vehicle Weights and Dimensions, Johannesburg, South Africa, March 2004.

http://road-transport-technology.org/Proceedings/8%20-%20ISHVWD/EFFECT%20OF%20HEAVY%20TRUCKS%20WITH%20LARGE%20AXLE%20GROUPS%20ON%20ASPHALT%20PAVEMENT%20DAMAGE%20-%20Chatti.pdf

Laboratory studies of a particular asphalt mix subject to pulse loadings representing various axle groupings (1 to 8 axles per group, 3.5-foot spacing) indicates that the normalized distress per ton goes down as the number of axles in a group goes up. Linear regression of LTPP distress and WIM data confirms this observation. The results of this study could be useful in extending study findings to quadrem and larger axle groupings. It is likely that considering actual axle groups, rather than arbitrarily dividing large groups of axles into tridems and tandems, would increase the accuracy of pavement damage analysis, and we will attempt to partially incorporate this approach.

Mallick, R., S. O’Brien, D. Humphrey, and L. Swett, Analysis of Pavement Response Data and Use of Nondestructive Testing for Improving Pavement Design, First Year Report 04-1A, Maine Department of Transportation, August 2006.

This report presents a description of instrumentation at the first fully instrumented flexible test pavement test section in Maine. Strain gauges were installed at the bottom of the HMA layer as well as in the subbase and subgrade, while pressure cells were installed in the subbase and the subgrade. Other instruments consist of thermocouples, moisture and thermal resistivity probes. Models relating temperature at two depths of the HMA layer with ambient temperature and solar radiation were developed. Stress/strain data were collected using a loaded truck running at different speeds at different temperatures. The response pulses at different layers were modeled with the Haversine equation and its slight variations. The effect of speed on the time of loading at the different layers was examined, to develop equations for predicting time of loading for laboratory testing, for example, for different traffic speeds for similar structures in Maine. The effect of time of loading on HMA strains, especially at higher temperatures, was well manifested in the measured data. Comparisons of predicted versus measured responses showed that the tensile strains in the HMA layer match with the predicted ones at lower temperature and lower time of loading. For subbase, the stresses were under predicted, whereas predicted strains matched quite well with the measured strains. In the case of subgrade, both the stresses and the strains were consistently higher than the predicted values - the difference increased with an increase in time of loading and temperature. The results from this ongoing study provide much needed information on response of typical reconstructed pavement in Maine, which can be used for laboratory testing and theoretical modeling, as well as in structural design using mechanistic procedures. This section could be used as one of the sites for analysis for the Northeast zone since the Maine DOT has been collecting real-time data on that site since 2006. Specifically, they have the following information available on this section (on Rt. 15) that could be used with complete data for one of the pavement analysis sites: test section cross-sectional layer features (layer thickness and material types); material properties for the subgrade, base, and HMA courses; temperature data for the mid-depth of the asphalt base and at the bottom of the asphalt base; pavement mechanical response data on the speed versus time of loading in the different pavement layers.

Oh, Jeongho, E.G. Fernando, and R.L. Lytton. "Evaluation of Damage Potential for Pavements Due to Overweight Truck Traffic, Journal of Transportation Engineering," 133(5), 308-317. DOI:10.1061/(ASCE)0733-947X(2007)133:5(308). 2007

http://www.academia.edu/937877/Evaluation_of_Damage_Potential_for_Pavements_due_to_Overweight_Truck_Traffic

Researchers installed multi-depth deflectometers (MDDs) along a section of highway in Brownville, where overweight trucks were routinely allowed starting in 1998, in order to establish a correlation between field measurements of pavement response to overweight trucks and the observed critical strains of rutting and fatigue cracking. The analysis was done in the overall framework of cross-anisotropic modeling of pavement response. The researchers found excellent correlation between damage and primary response, meaning that primary response is a good proxy for expected pavement damage. The study could be used to check consistency of their findings with the AASHTOWare Pavement ME Design® model, but we do not have enough calendar time and did not propose enough effort to second-guess the models incorporated in AASHTOWare Pavement ME Design®.

Sadeghi, J. M., and M. Fathali. Deterioration Analysis of Flexible Pavements under Overweight Vehicles. Journal of Transportation Engineering, 133(11), 625-633. DOI:10.1061/(ASCE)0733-947X(2007)133:11(625). 2007.

http://www.nlcpr.com/Deterioration%20Analysis%20of%20Flexible%20Pavements.pdf

The authors used layer theory, following the Burmeister approach, to derive operational life reduction factors for two-axle and three-axle single unit trucks and for 3S2s. Not really credible for our purposes given alternative available models. We have opted to use the AASHTOWare Pavement ME Design® model, and do not have enough calendar time and did not propose enough effort to second-guess the incorporated damage models in AASHTOWare Pavement ME Design®.

Schwartz, Charles W., Rui Li, Sung Hwan Kim, Halil Ceylan, and Kasthurirangan Gopalakrishnan. Sensitivity Evaluation of MEPDG Performance Prediction. NCHRP Project 1-47, Final Report. TRB. December 2011.

http://onlinepubs.trb.org/onlinepubs/nchrp/docs/NCHRP01-47_FR.pdf

The study systematically varied all the user inputs for the MEPDG model to determine the sensitivity of the pavement performance predicted by the model to the variability of the input factors for five types of pavements-- new HMA, HMA over a stiff foundation, new JPCP, JPCP over a stiff foundation, and new CRCP-- and five climate types (the usual four, plus temperate). Although design inputs were varied, traffic composition was not-- only AADTT and operating speed were varied. The study derived normalized sensitivity indices (NSIs) for each distress for each input variable, expressing the percentage change in the normalized distress divided by the percentage change in the design input. Key findings were that design inputs for the surface layers were the most important; longitudinal cracking, alligator cracking, and AC rutting were substantially more sensitive to inputs than were IRI and thermal cracking; design input sensitivities for thermal cracking had little overlap with the design input sensitivities for the other distresses; and little thermal cracking occurred when binder grades were properly matched to the climate. The study will be helpful in designing base case pavement sections, but the lack of traffic variations make it less useful for the overall analysis for the current 2014 CTSW Study.

(5) Not Directly Usable but Supplying Background Information

Acimovic, Benjamin, Leela Rejaseker, and Reza Akhavan. Forensic Investigation of Pavement Failure on Vasquez Boulevard. Colorado DOT Research Branch, Report No. CDOT-2007-7. May 2007.

https://www.codot.gov/programs/research/pdfs/2007/vasquez.pdf

Vasquez Boulevard in Commerce City, Colorado, as part of U.S. 6, provides a main trucking route in the I-25 corridor for overweight and over-height trucks. After reconstruction in 2001, parts of the pavement showed severe rutting in less than one year. Pavement failure was found to be related to repeated heavy loads, exposure of a layer constructed in the 1940s that did not contain an anti-stripping agent, inexperience with the stone-matrix asphalt technique used in the rehabilitation, and variable mix gradation and AC content. Although the study confirms that heavier trucks do, indeed, contribute to accelerated pavement wear, especially with faulty pavement designs, the findings are too general to contribute to the analysis methods we rely upon in the current 2014 CTSW Study.

Barnes, Gary, and Peter Langworthy. "Per Mile Costs of Operating Automobiles and Trucks."Transportation Research Record: Journal of the Transportation Research Board, No. 1864. TRB. National Research Council. Washington, D.C., 71-77. 2004. Available online in pre-published form at

http://www.hhh.umn.edu/centers/slp/pdf/reports_papers/per_mile_costs.pdf

Citing other studies, the report concludes that IRIs below 80 (PSIs greater than 3.5) add nothing to vehicle operating costs, but IRIs of 170 (PSI 2.0) result in 2.5 cents per mile in additional operating cost. The additional cost derives from reduced vehicle life and in increased repair and maintenance costs. User costs are not being modeled in the current 2014 CTSW Study, so the findings are of only general interest.

Dodoo, Nii Amoo, and Neil Thorpe. "Road User Charging for Heavy Goods Vehicles." Presented at 7th International Symposium on Heavy Vehicle Weights and Dimensions, Delft, The Netherlands, June 2002.

http://road-transport-technology.org/Proceedings/7%20-%20ISHVWD//Road%20User%20Charging%20For%20Heavy%20Goods%20Vehicles%20-%20%20Dodoo.pdf

Although many countries in Europe and North America have explored charging vehicles based on operating axle weights and the associated pavement damage, charging for actual damage at the point of use through use of WIM or other scales becomes problematic because of the high cost of installing weight station and the poor correlation between static and dynamic axle load. The authors instead propose an on-board system consisting of dynamic axle-load measurement combined with vehicle location measuring devices (now widely known as GPS systems). Interesting approach, but well beyond the scope of this analysis in this 2014 CTSW Study, since we have not been asked to consider alternative user-fee charging mechanisms.

Fernando, Emmanuel G. "Investigation of the Effects of Routine Overweight Truck Traffic on SH4/48." Texas Transportation Institute, Project 0-4184, Summary Report. April 2006.

http://d2dtl5nnlpfr0r.cloudfront.net/tti.tamu.edu/documents/0-4184-S.pdf

After Texas authorized 125,000-lb trucks to routinely use a state highway in Brownsville, TTI collected data to assess the impact of overweight trucks on that route. They first used ground-penetrating radar to estimate layer thicknesses and to subdivide the route into uniform subsections, where they used falling weight deflectometer tests to monitor load response over time. They also took cores at selected locations to both verify the penetrating radar thickness estimates and to characterize asphalt concrete properties. The research found good correlations between AC moduli back calculated from static and dynamic analysis and that the additional ESALs from overweight truck traffic will likely result in accelerated pavement deterioration. The study could be used to check consistency of their findings with the revised AASHTOWare Pavement ME Design® model, but we do not have enough calendar time and did not propose enough effort to second-guess the models incorporated in AASHTOWare Pavement ME Design®.

Gibby, A. R., Ryuichi Kitamura, and Huichun Zhao. "Evaluation of Truck Impacts on Pavement Maintenance Costs," Transportation Research Record: Journal of the Transportation Research Board, No. 1262, pp. 48 - 56. TRB. National Research Council. Washington, D.C., 1990.

https://itspubs.ucdavis.edu/wp-content/themes/ucdavis/pubs/download_pdf.php?id=1008

The study randomly selected 1,100 one-mile sections of state highways, collected data on traffic, weather, geometric conditions, and pavement maintenance costs on those sections, and used that data to develop a model of pavement maintenance costs. Incremental maintenance costs were expressed in terms of average annual maintenance cost per vehicle. An interesting analysis that is not directly applicable to the current 2014 CTSW Study.

Hernandez, Sarah, Andre Tok, and Stephen G. Ritchie, "Integration of Weigh-in-Motion and Inductive Signature Technology for Advanced Truck Monitoring." Institute of Transportation Studies, University of California, Irvine, Report # UCI-ITS-WP-13-3, August 2013.

http://www.its.uci.edu/its/publications/papers/ITS/UCI-ITS-WP-13-3.pdf

The study points out the high rates of error when inductive loop technology alone is used to classify trucks and demonstrates how the error rates can be reduced by including axle weight data from WIM. Further, the study explores using inductive loop devices with high sampling rate detector cards to identify characteristic body type signature. This allows users to identify truck body types and dramatically reduce classification vehicle classification errors. While the technology is not currently used by states in reporting WIM data to FHWA, this study provides analysis of error rates for several vehicle classes that the team can compare to our error rates, and perhaps use to refine our WIM data classification algorithm.

Luskin, David, and C. Michael Walton. Effects of Truck Size and Weights on Highway Infrastructure and Operations: A Synthesis Report. Center for Transportation Research: The University of Texas at Austin. Report No. FHWA/TX-0-2122-1. March 2001.

http://www.utexas.edu/research/ctr/pdf_reports/2122_1.pdf

In reviewing a number of truck size and weight studies, including FHWA’s 2000 study, the authors found that shifting away from the dominant 3S2 and increasing gross vehicle weights would not necessarily increase pavement costs, and might make them lower, but it would likely increase bridge costs. Safety effects were inconclusive. The ESAL assumption, as well as the wide range of diversion assumptions, makes the findings of only general interest to the current 2014 CTSW Study, although it does illustrate that heavier vehicle weights do not automatically result in higher pavement costs.

Papagiannakis, Athanassios, Nasir Gharaibeh, Jose Weissmann, and Andrew Wimsatt. Pavement Score Synthesis. Texas Transportation Institute, Report No. FHWA/TX-09/0-6386-1. January 2009.

https://d2dtl5nnlpfr0r.cloudfront.net/tti.tamu.edu/documents/0-6386-1.pdf

The synthesis summarizes the use of pavement scores by states, including rating methods and how the scores are used for recommending pavement maintenance and rehabilitation actions. Some states considered only the dominant distress in rehab strategies, while others considered all the distresses present. Most states considered both range and severity of distress. Differences in rating systems make comparison of overall pavement conditions among states invalid. Good overview of rating systems, but not of direct relevance to this 2014 CTSW Study.

Regehr, Jonathon David, Exposure Modelling of Productivity-Permitted General Freight Trucking on Uncongested Highways. Doctoral Dissertation for University of Manitoba Civil Engineering Department. October 2009.

The paper describes a methodology for improving estimates of LCV exposure data for the Canadian Prairie Region. The dataset for the study integrated output from a classification algorithm, field observations, and industry intelligence. The classification algorithm is of particular interest to this 2014 CTSW Study, since it broke the LCV classes into a larger number of vehicle types than FHWA commonly uses, thereby allowing a higher degree of certainty in some of the most important LCV classes. The team will use the algorithm to refine the WIM analysis and to help us evaluate how many of vehicle classes to use in the analysis.

Regehr, J. D, J. Montufar, and D. Middleton. Applying a Vehicle Classification Algorithm to Model Long Multiple Trailer Truck Exposure. Published in IET Intelligent Transport Systems, February 2009. Abstract available at: https://digital-library.theiet.org/content/journals/10.1049/iet-its.2008.0066

The paper describes an algorithm also described in the Regehr dissertation. The team will use the algorithm to refine the WIM analysis and to help us evaluate how many of vehicle classes to use in this analysis.

Rouen, Chhooeuy, and Mom Mony. "Damage Effects of Road Pavements Due to Overloading in Cambodia", Academia.edu, Undated.

https://www.academia.edu/1375429/Damage_Effects_of_Road_Pavements_due_to_Overloading_in_Cambodia

Synthesis of previous studies in many other countries shows that truck overloading is a serious problem that can greatly increase pavement costs. Not directly usable for this 2014 CTSW Study, since there is insufficient information about the axle loads, the pavements, or the materials.

129,000 Pound Pilot Project: Report to the 62nd Idaho State Legislature. Idaho Transportation Department (IDT). January 2013.

https://rosap.ntl.bts.gov/view/dot/26522/Email

Idaho raised the operating GVW limit from 105.5 kips to 129 kips as a pilot project on selected routes in the state in 2003, 2005, and 2007. The 105.5 kip trucks typically operated with 8 axles, while the 129 kip trucks typically operated with 10 or 11 axles. The state legislature asked IDT to study the impact of the pilot on safety, bridges, and pavements and report to the legislature every three years. Participating trucking companies reported making 264,169 trips by 1,359 trucks between 2004 and 2012, and ITD did not observe any significant effects on safety, bridges or pavements, while participating trucking companies reported great savings in costs and number of trips. Normal maintenance and repair activities occurred during the pilot, but ITD did not tabulate their relative frequency on the pilot and non-pilot routes, so one cannot conclude that there was no effect on pavement or bridge deterioration, only that regular maintenance and repair activities were able to compensate for any change in deterioration rates. That lack of data, plus the small sample size of routes, trips, and pilot duration make any conclusions from the project somewhat tentative at this point.

Estimating Truck-Related Fuel Consumption and Emissions in Maine: A Comparative Analysis for a 6-axle, 100,000 Pound Vehicle Configuration, American Transportation Research Institute, September 2009.

The performance of a 6-axle vehicle configuration operating at a maximum GVW of 100,000 pounds was analyzed over two roughly parallel routes between Augusta and Brewer, Maine. The existing route (Route 9) reflects current conditions where trucks greater than 80,000 pounds GVW are not allowed on I-95 north of State Route 3 due to federal weight restrictions. The alternative route (I-95) assumes trucks up to 100,000 pounds GVW would be allowed to travel on I-95 north of State Route 3. This report relates only very limited information that relates to the impact of increased truck loads on pavement response. It deals instead with energy consumption and emissions.

"How Vehicle Loads Affect Pavement Performance." Wisconsin Transportation Bulletin No. 2, Undated.

http://epdfiles.engr.wisc.edu/pdf_web_files/tic/bulletins/Bltn_002_Vehicle_Load.pdf

Explains ESALs and the basics of pavement fatigue and pavement strength to a lay audience. The ESAL assumption makes the findings of only general interest to this 2014 CTSW Study, but the explanation of why pavement damage goes up faster than axle weight could be helpful in summary reports intended for a non-technical audience.

Research Projects – Multiple documents. Multimodal Transportation & Infrastructure Consortium. Available at http://www.mticutc.org/research/research-projects/

Several projects underway appear to have some possible relevance, but all of the final reports for these projects are listed as "Coming Soon" so will not be available soon enough for this 2014 CTSW Study.

"Section 5 - Truck Weight Monitoring", Traffic Monitoring Guide. Federal Highway Administration, May 1, 2001

Describes the truck weight monitoring program under which states collect and report WIM data. Excellent reference material for using and understanding the WIM data that we will use in the 2014 CTSW Study.

(6) Additions Suggested During May 29, 2013 Webinar

Di Cristoforo, R., Regehr, J.D., Germanchev, A., and Rempel, G. (2012). "Survival of the Fittest: Using Evolution Theory to Examine the Impact of Regulation on Innovation in Australian and Canadian Trucking," Heavy Vehicle Transport Technology 12, Stockholm, Sweden. This publication is not directly related to pavement issues. It examines the impact of regulation on trucking in Australia and Canada by applying evolution theory and deals more with regulatory issues.

Jablonski, B., Regehr, J.D., Kass, S., and Montufar, J. (2010). "Data Mining to Produce Truck Traffic Inputs for Mechanistic-Empirical Pavement Design," 8th International Transportation Specialty Conference, Canadian Society for Civil Engineering, Winnipeg, Manitoba. This presentation is related to the overall 2014 CTSW Study, but not overly useful for our task.

Jablonski, B., J.D. Regehr, G., Rempel, T. Baumgartner, A, Nuñez, K. Patmore, M. Moshiri, H. Hernandez, and J. Montufar, J. (2010). "Traffic Data Requirements for the Mechanistic-Empirical Design of New and Rehabilitated Pavement Structures in Manitoba," prepared for Manitoba Infrastructure and Transportation by UMTIG in association with Regehr Consulting. This paper is useful from its title but is propriety (a consulting report) and therefore not publicly-available.

Malbasa, A., Regehr, J.D., and Clayton, A. (2005). "A Performance-Based Approach to On-Road Regulatory Compliance of Commercial Vehicle Operations in Manitoba," UMTIG, prepared for the Compliance and Regulatory Services Branch, Manitoba Transportation and Government Services. This paper seems related more to compliance from its title, plus it is propriety (a consulting report) and therefore not publicly-available.

Montufar, J., J.D. Regehr, G. Rempel, T. Baumgartner, and B. Jablonski (2008). "The Impacts of Increased Truck Gross Vehicle Weights: Environmental Scan," Montufar & Associates and UMTIG, prepared for Alberta Infrastructure and Transportation. This paper may be useful from its title but is propriety (a consulting report) and therefore not publicly-available.

Montufar, J., J.D. Regehr, C. Milligan, and M. Alfaro (2011). "Roadbed Stability in Areas of Permafrost and Discontinuous Permafrost: A Synthesis of Best Practices," prepared for Transport Canada – Surface – Prairie and Northern Region by Montufar & Associates in association with Regehr Consulting and UMTIG. This paper is actually most pertinent to railroads in the northern Canadian context and a publicly-available paper is forthcoming in ASCE Journal of Cold Regions Engineering.

Radstrom, B., Regehr, J.D., Arango, J., Steindel, M., Rempel, G., Jablonski, B., Montufar, J., and Clayton, A. (2007). "Traffic on Manitoba Highways 2006," University of Manitoba Transport Information Group, prepared for the Traffic Engineering Branch, Manitoba Infrastructure and Transportation. This paper may be marginally-useful for its Level 1traffic data, specifically the percentages and load range for 3-S2 and B-train truck types.

Regehr, J.D. (2012). "Truck Exposure to Inform Size and Weight Policy Decisions," presentation prepared for the Transportation Research Board Annual Meeting, Washington, D.C. This presentation is related to compliance.

Regehr, J.D. (2011). "Understanding and Anticipating Truck Fleet Mix Characteristics for Mechanistic-Empirical Pavement Design," Transportation Research Board Annual Meeting CD-ROM, Washington, D.C. This paper analyzes vehicle classification data to support the implementation of the Mechanistic-Empirical Pavement Design Guide (MEPDG). A cluster analysis and expert judgment are applied to vehicle classification data from Manitoba to produce six jurisdiction-specific truck traffic classification groups (TTCGs). These groups are used to estimate truck volumes by class at locations where no site-specific classification data exist. The unique vehicle classification distributions evident from these groups, particularly the relative predominance of six-axle tractor semitrailers and multiple-trailer trucks within the fleet, demonstrate the importance of developing truck traffic data inputs based on local conditions and expertise. This publication is relevant to pavements, but specifically looks at vehicle class rather than weight.

Regehr, J.D. (2010). "Leveraging Truck Traffic Data from Mechanistic-Empirical Pavement Design to Support Other Transportation Engineering Decisions," presentation prepared for the North American Travel Monitoring Exposition and Conference, Seattle, Washington. This presentation is related by not overly useful for this 2014 CTSW Study as it is not detailed enough.

Regehr, J.D. (2009). "Truck Loading on Highway Infrastructure in the Canadian Prairie Region," presentation prepared for the Vehicle-Infrastructure Interaction Workshop, Winnipeg, MB. This presentation is related by not detailed enough to be useful for this 2014 CTSW Study.

Regehr, J.D. (2003). "Estimating Live Truck Loads for Roads and Bridges: A Sectoral Approach Applied to Grain Transport," presentation prepared for the Institute of Transportation Engineers (Manitoba Section), Winnipeg, Manitoba. This presentation is related by not detailed enough to be useful for this 2014 CTSW Study.

Regehr, J.D. (2002). "Aspects of Agriculture-Related Trucking in Manitoba," UMTIG. This presentation is related by not overly useful for this 2014 CTSW Study.

Regehr, J.D., Baumgartner, T., Nuñez, A., and Montufar, J. (2009). "Measuring and Estimating Recreational Traffic in Manitoba," presentation prepared for the Recreational Traffic Monitoring Workshop, Lakewood, CO. This presentation is related by not detailed enough to be useful for this 2014 CTSW Study.

Regehr, J.D., Jablonski, B., Rempel, G., Baumgartner, T., Nuñez, A., Patmore, K., Moshiri, M., Hernandez, H., and Montufar, J. (2010). "Traffic Data Requirements for Mechanistic-Empirical Design of New and Rehabilitated Pavement Structures in Manitoba," presentation prepared for the MEPDG User Group, Transportation Association of Canada Spring Technical Meetings, Ottawa, ON. This paper seems useful from its title, perhaps for the traffic classification, but it is propriety (a consulting report) and therefore not publicly-available.

Regehr, J.D. and Montufar, J. (2007). "Classification Algorithm for Characterizing Long Multiple Trailer Truck Movements," Transportation Research Board Annual Meeting CD-ROM, Washington, D.C. This presentation is related by not overly useful for this 2014 CTSW Study. It deals with development of an algorithm that provides the core dataset for modelling long-truck exposure in terms of the volume of trips, and their weight and cubic characteristics. It is embedded within a modelling approach in which exposure is an explanatory variable needed for predicting transportation system impacts related to long-truck operations. Table 2 may be useful in that it contains WIM data related to long trucks from highways between Winnepeg and Brandon, MB, and Figure 3 includes the load spectra.

Regehr, J.D. and Montufar J. (2012). "Traffic Data and the State of the Practice in Canada," presentation prepared for the North American Travel Monitoring Exposition and Conference, Dallas, Texas. This presentation is related by not overly useful for this 2014 CTSW Study.

Regehr, J.D., Montufar, J., and Clayton, A. (2009). "Lessons Learned about the Impacts of Size and Weight Regulations on the Articulated Truck Fleet in the Canadian Prairie Region," Canadian Journal of Civil Engineering, vol. 36, no. 4, pp. 607-616. This publication is not directly related to pavement issues. It deals more with the state-of-the-practice and policy issues, but includes some information on the shift in traffic percentages related to articulated trucks. This paper is not useful for this 2014 CTSW Study.

Regehr, J.D., Montufar, J., and Clayton, A. (2009). "Options for Exposure-Based Charging for Long Multiple Trailer Truck Permits," Transportation Research Record: Journal of the Transportation Research Board, no. 2097, pp. 35-42. This presentation is related by not overly useful for this 2014 CTSW Study as it appears to deal more with compliance issues.

Regehr, J.D., Radstrom, B., Arango, J., Isaacs, C., Han, K., Rempel, G., Montufar, J., and Clayton, A. (2006). "Traffic on Manitoba Highways 2005," UMTIG, prepared for the Traffic Engineering Branch, Manitoba Transportation and Government Services. This presentation is related by not overly useful for this 2014 CTSW Study.

Reimer, M. and Regehr, J.D. (2012). "Clustering of Vehicle Classification Data to Support Regional Implementation of the Mechanistic-Empirical Pavement Design Guide," presentation prepared for the North American Travel Monitoring Exposition and Conference, Dallas, Texas. This presentation will soon be published as a TRB TRR Journal article. It is relevant to pavements, but specifically looks at vehicle class rather than weight.

Transportation Research Board of the National Academies. All Motor Carrier Publications. http://www.trb.org/MotorCarriers/Publications1.aspx The relevant projects shown in this publications list appear to have already been included in this desk scan.

You may need the Adobe® Reader® to view the PDFs on this page.

previous | next
Office of Operations