Civil Engineering ETDs

Publication Date



In the Mechanistic-Empirical Pavement Design Guide (MEPDG), traffic loading is more complex and precise than in the old AASHTO 1993 Guide. Traffic load in MEPDG is characterized by a larger number of traffic inputs such as Annual Average Daily Truck Traffic (AADTT), traffic growth, traffic directional distribution, lane distribution, vehicle class distribution, hourly distribution, monthly distribution, axle load spectra, number of axles per truck class, traffic wander, speed, and tire pressure. Many of these inputs can only be obtained from Weigh-in-Motion (WIM) data. There are about 15 WIM stations in New Mexico. Data from these WIM sites are evaluated for their use for local calibration of the MEPDG. An algorithm is developed in Visual Basic Application (VBA) to check the quality of WIM data. In essence, frequency distributions of the gross vehicle weight and the front steering axle weight are calculated and compared to the criteria recommended by the Traffic Monitoring Guide (TMG). It is observed that only three WIM sites have consistent and reliable weight data. The reasons for the inconsistency in collected WIM data are found to be lack of calibration and bad condition of the pavement surface near WIM sites. Also, axle load spectra are developed for all WIM data using an algorithm implemented in VBA. The influence of axle load spectra on pavement performance is predicted using MEPDG. For the sections analyzed, it is shown that the effect of axle load spectra is very high on fatigue cracking, moderate on permanent deformation, and non-existent on thermal cracking and roughness. In this study, local data related to traffic, climate, pavement structure, materials, and distress are collected from different NMDOT sources and stored in MEPDG Oracle database. A total of 29 New Mexico pavement sections are found to have all MEPDG inputs, however data lack quantitative distress values required for MEPDG calibration. This is because New Mexico has collected qualitative distress data rather than actual measurements of rut depth and crack length over the past years. Instead of using these 29 sections, only 11 sections from the Long Term Pavement Performance (LTPP) database located in New Mexico are used for local calibration. The permanent deformation, alligator cracking, and longitudinal cracking models are calibrated. In calibration methodology, the target is fixed to reduce the residual sum of squared errors, defined by the difference between predicted and measured distress, so that any bias is eliminated and precision is increased. The optimized calibration coefficients are: βr1 = 1.0, βr2 = 0.9, βr3 = 1.2, βGB = 0.8, βSG = 0.8 for the rutting model; C1 = 0.73, C2 = 0.09, C3 = 7200 for alligator cracking; and C1 = 5.5, C2 = 2.56, C3 = 1000 for longitudinal cracking. It is concluded that the newly developed calibration coefficients reduce the error in the MEPDG prediction and are beneficial for designing pavements using MEPDG in New Mexico.


Pavements--Performance--New Mexico--Measurement, Pavements--Live loads--New Mexico--Measurement, Pavements--New Mexico--Design and construction.

Document Type




Degree Name

Civil Engineering

Level of Degree


Department Name

Civil Engineering

First Advisor

Tarefder, Rafiqul

First Committee Member (Chair)

Brogan, James

Second Committee Member

Ng, Tang-Tat