Streamlining Pavement ME for ALDOT

The AASHTOWare Pavement ME software has been slowly being adopted nationwide. While many agencies still rely on the tried-and-true 1993 AASTHO Guide for Design of Pavement Structures, there is also a benefit to capturing the complexities of asphalt pavement performance prediction – “black magic” to some – that is covered in the Pavement-ME software. But Pavement-ME has a lot of material-specific inputs, so how specific do we need to be? Does an agency need to test every single asphalt mix to get an accurate prediction of pavement performance?

These are some of the questions that the research team investigated for the Alabama DOT. But first, let’s remember some of the basics: The AASHTOWare Pavement ME software is built off the Mechanistic-Empirical Pavement Design Guide (MEPDG). While the trusty 1993 AASHTO pavement design methodology was purely empirical, the MEPDG seeks to combine the mechanistic (engineering) properties of pavement materials from the subgrade to the surface with a series of empirical equations to predict pavement performance.

Breaking down the levels of inputs

There are three levels of mechanistic inputs:

• Level 1 inputs provide for the highest level of accuracy and would have the lowest level of uncertainty or error, since input parameters are measured directly and are considered site or project-specific. This level requires laboratory or field testing (e.g., FWD of existing base materials at paving site).

• Level 2 inputs provide an intermediate level of accuracy. Level 2 inputs are typically
user-selected, possibly from an agency database, could be derived from a limited testing program, or could be estimated through correlations.

• Level 3 inputs provide the lowest level of accuracy and are provided as defaults within the software. This level might be used for design where there are minimal consequences of early failure, such as lower volume roads, as it may not represent the actual materials that will be used.

Over the years, a lot of work has gone into understanding Pavement-ME predictions, whether it’s looking at the different material inputs or the empirical transfer functions that compute predicted performance. This study sought to statistically analyze the difference between mixes. The team worked with ALDOT to collect 13 different dense-graded surface mixes, with at least two coming from each of ALDOT’s five regions. The mixes were collected from asphalt plants and compacted in the NCAT laboratory, then tested for dynamic modulus, repeated load permanent deformation, creep compliance, and tensile strength, in addition to the performance grading of the recovered asphalt binder. The inputs of the 13 unique surface mixes were then used on 41 Long Term Pavement Performance (LTPP) pavement sections across Alabama to predict performance. The average properties from all 13 mixtures were also calculated and applied to all 41 sections to see if “representative” ALDOT-specific Level 3 inputs would yield similar results.

Ultimately, the research team found that
ALDOT-specific Level 3 inputs for mechanical properties yielded similar predicted performance compared to mixture-specific Level 1 and Level 2 inputs over the 41 LTPP pavement sections. This supports the use of the ALDOT-specific Level 3 inputs for Pavement ME design in Alabama.

This finding could simplify the implementation of Pavement ME within ALDOT, but more work needs to be completed to fully understand the implications of adoption. It is also important that other states consider this type of approach, as changes in material types (aggregates, binders, etc.) could yield different results.

This study shows that well-developed Level 3 inputs can provide reliable pavement performance predictions, making Pavement ME more practical for agencies like ALDOT. It offers a path to streamline design without sacrificing accuracy. Other agencies may benefit from similar validation efforts to create cost-effective, region-specific approaches.

Continued research will help maximize Pavement ME’s value nationwide.

Contact Ben Bowers for more information on this research.