Browse Projects > Detailed View



Since the combined influence of mixed traffic and climate fluctuations can be incorporated effectively, pavement mechanistic-empirical (ME) design procedures are becoming more and more common practice for predicting pavement performance in terms of various pavement distresses. However, the precision of predicted distresses is usually low or questionable due to the lack of accurate distress models (i.e., empirical analysis) and the presence of uncertainties in pavement material characterization and the process of mechanistic analysis. By considering a static vehicle load, Layered Elastic Analysis (LEA) or Finite Element Method (FEM) is commonly utilized for mechanistic analysis in ME design. Additionally, it is assumed that the material is homogeneous and isotropic, and that the pavement layers are perfectly elastic. These assumptions about material characteristics and static vehicle loading, however, are never true in practice. As a result, the pavement responses predicted from the mechanistic analysis could be inaccurate. As the distress models are based on the predicted pavement responses, it affects the pavement performance predictions. Therefore, there is a need for improving the mechanistic model by utilizing field measured sub-surface pavement responses from moving wheel loads.


The current flexible pavement ME design algorithm is based on the LEA for estimating “mechanistic” pavement response due to static vehicle loadings by considering mixed traffic scenarios. That is, the flexible pavement models usually consist of “elastic” asphalt and aggregate base/subbase layers over subgrade “half-space.” The assumed “elastic” behavior can be conceived exclusively within a very limited range of low strain values but cannot be applied outside such strain range or would produce unrealistic results. Furthermore, the consideration of static vehicle load instead of actual moving wheel load also increases the uncertainty of the mechanistic analysis.

Pavement instrumentation, either in test sections or in-service pavements, has become a common practice among transportation authorities and research institutions. Information gathered from such pavement instrumentation projects have a huge potential for improving the current pavement ME design procedure.

The primary objectives of this study are two-fold:

  • Measurement of stress and strain responses in various pavement layers due to moving traffic at different climatic settings, and comparison of such measured responses with the numerical responses
  • Calibration of mechanistic models by reducing the bias in the pavement responses and accurately predicting performance which is the direct indicator of the remaining service life of pavement.

Accurate prediction of pavement performance is very crucial for transportation authorities such as FHWA, FAA, State DOTs for identifying and prioritizing maintenance and rehabilitation activities, managing transportation asset, allocating budget, and more importantly determining the remaining service life. Therefore, most of these authorities have adopted or going to adopt pavement ME design procedure for predicting pavement performance. As per transportation pooled fund study TPF-5(305), 17 state DOTs have already implemented AASHTO’s ME design for flexible pavements, and 27 state are planning to implement as of November 2021. However, none of the pavement ME procedures used today are precise and need to be improved. Therefore, the research needed for such improvement is outlined in this research need statement.

Aside from that, the research also creates an opportunity to improve the ME models for other pavement types such as rigid, composite as well as different construction techniques such as mechanical stabilization (e.g., geogrid, geotextile, geocell), chemical stabilization (e.g., lime, cement) and Full Depth Reclamation. However, this is outside the scope of the project, and a new research need statement should be created.

Related Research:

In 1990, the National Cooperative Highway Research Program (NCHRP) launched NCHRP Project 1-26, which marked the beginning of the development of mechanistic pavement analysis and design procedure. Through NCHRP 01-37A, the mechanistic-empirical pavement design guide (also known as MEPGD) was developed in 2002. Similar ME-based methods were also developed by the state DOTs (e.g., Washington State Department of Transportation Pavement Guide; MnPAVE by Minnesota Department of Transportation; CALME by California Department of Transportation) and industry (e.g., SPDM by Shell; Asphalt Institute method). Among these methods, pavement responses caused by external factors (traffic load and environmental) are computed in terms of stress, strain, or deflection within the pavement structure which are influenced by material characteristics and climatic conditions. In the early pavement ME design and analysis method, the pavement responses were empirically related to the allowable number of traffic passes to failures (e.g., fatigue life of asphalt; subgrade rutting life) using mathematical models. These empirical mathematical models along with the mechanistic response computation engines were the backbone of pavement ME design and analysis. In AASHTOWare Pavement (formerly MEPDG), the empirical models for predicting pavement distresses had been enhanced with the inclusion of the Long-Term Pavement Performance (LTPP) database, developed mainly from surface measurements. Nevertheless, there were still issues related to the accuracy of the mechanistic models. For example, the roughness model in AASHTOWare Pavement for predicting pavement IRI (international roughness index) relies on the area of fatigue cracking (or horizontal strain at the bottom of HMA) and average rut depth (or vertical deflection of each pavement layer). Inaccurate prediction of these pavement responses (strain and deflection) will result in incorrect estimation of pavement IRI. Similarly, the surface rut or deformation prediction is dependent upon the deformation of each pavement layer. The assumption of elastic pavement layers usually results in the inaccurate prediction of pavement rut. Therefore, several studies were conducted to understand the time- and temperature-dependent behavior of viscoelastic material (e.g., HMA) and the stress-dependent behavior of elasto-plastic unbound materials (e.g., aggregate base, subbase) through numerical simulations and laboratory testing. Although the large amount of data available, model reliability and accuracy in the prediction of pavement performance are still questionable and demand additional refinements of such theoretical models (based on LEA or FEA).


In order to prevent duplication of effort, it is advised to interact with various research institutions (such as the US Army ERDC, NCAT, FAA, and academic institutions) and transportation authorities about their present research and testing program. The following tasks must be carried out to meet the objectives:

Task 1: A detail review of various pavement ME design and analysis procedure (e.g., AASHTOWare Pavement ME, FAA’s Panda-AP, FHWA’s FlexPave) focusing on the mechanistic analysis (e.g., LEA type); Identification of the important input parameters for the analysis; Selection of critical outputs (pavement responses) from the analysis which are necessary for pavement performance prediction; Horizonal strain at the bottom of the asphalt layer, vertical strain on the top of subgrade and vertical deformation of each pavement layers are the recommended pavement responses

Task2: Review of current and previous research studies on pavement instrumentation, preferably In-service Pavement Test Sections or Full-Scale APT, by different research institutions (e.g., US Army ERDC, NCAT, FAA, Academia), and transportation authorities; Documentation of various sensor types (including their manufacturers), survivability rate of each sensor and easability of installation; A comprehensive review of recorded pavement responses, their accuracy and variability; Development of database of pavement responses based on the pavement thickness, applied load, sensor location and other site condition (these information can be used as a reference for validating and verifying the pavement responses for future pavement instrumentation test program)

Task 3: Selection of a specific Pavement ME design procedure (preferably AASHTOWare Pavement ME as it is recommended by most state DOTs); Selection of most common distress model(s) for determining the required pavement response measurements; As asphalt bottom-up cracking and surface rutting are the major pavement performance measures, the required pavement response will be asphalt strain and pavement layer deformation (including moisture level and temperature measurement)

Task 4: Selection of sensors for measuring the required pavement responses; Seek expert advice from APT owners (e.g., MnROAD, ERDC, NCAT) and users of in-service pavement instrumentation projects

Task 5: Installation of sensors in in-service pavement sections; Documentation of a detail geotechnical properties of each pavement layers including subgrade; It is recommended to select multiple in-service pavement test sections having different climatic zones and different subgrade types. The total number of test sections should be decided based on statistical significance and/or the variability of subgrade conditions, the design parameters such as material types, pavement thickness, and the needs of the agency

Task 6: Selection of traffic data collection method; It is recommended to collect a precise data of traffic loading because it has a direct influence on the pavement responses

Task 7: Integration of sensors and data acquisition systems for effective data collection; It is recommended for an automated data collection system triggered by the traffic; Such an automated system will help in saving energy consumption of the system as well as reduce the amount of unusable data

Task 8: A detail review of collected data from sub-surface sensors and traffic information collection device; Processing of the collected data by determining the actual traffic load and the associated pavement responses; Conducting numerical analysis using the pavement ME software (i.e., performing LEA) and estimating theoretical pavement responses; Comparing the measured (experimental) and theoretical (numerical) pavement responses and finding the bias; Developing a methodology to reduce the bias, and hence improving the accuracy of the numerical model (note that this bias reduction process is not the same as the process of finding local calibration coefficients of various distress models)

Task 9: Summarizing the key findings; Outlining the process of improving the numerical model; Preparing the final report


Highway Pavements Design and Evaluation of Pavement Structural Pavement Material Pavement Mechanistic Responses Subsurface Instrumentation Pavement Performance

Sponsoring Committee:AKG60, Geotechnical Instrumentation and Modeling
Research Period:24 - 36 months
Research Priority:High
RNS Developer:Prajwol Tamrakar
Source Info:AKG60 (formerly AFS20) survey results showed that there is a high need for measuring stress/strain response in pavement layers
Date Posted:08/19/2022
Date Modified:08/30/2022
Index Terms:Mechanistic-empirical pavement design, Pavement distress, Flexible pavements, Field tests, Instrumentation,
Cosponsoring Committees: 

Please click here if you wish to share information or are aware of any research underway that addresses issues in this research needs statement. The information may be helpful to the sponsoring committee in keeping the statement up-to-date.