Civil Engineering ETDs

Publication Date



Traffic is one of the key inputs in pavement design. The pavement Mechanistic-Empirical (ME) design allows three different types of input level of traffic data based on the availability of the data. They are: site specific data (Level 1), regional data (Level 2), and the national data (Level 3). Level 1 inputs (e.g., load magnitude, configuration, and frequency) are generated from Weigh-in-Motion (WIM) station installed in each site. However, it is not always practically possible to install WIM station due to high cost of WIMs. Therefore, often time the designers have to rely on the Level 2 or Level 3 traffic data. But it is not known yet how good the national data or the regional data compared to New Mexico's site specific data in predicting interstate pavement performances. To this end, this study examines the effects of different levels of traffic inputs on predicted pavement distresses in New Mexico. Two major interstate highways were considered in this study: Interstate-40 (I-40) and Interstate-25 (I-25). Site-specific inputs were developed using installed WIM stations at the pavement sites. WIM data was analyzed using an advanced and updated software developed by the UNM researchers. Traffic data were simulated through the ME design software for predicting pavement performances. Results show that axle load spectra (ALS) and lane distribution have a great influence on predicted interstate pavement performance. Vehicle class distribution (VCD), directional distribution, and standard deviation of lateral wander have a moderate impact on pavement performance. Monthly adjustment factor, axles per vehicle, axle spacing, and operational speed have very little effect on the predicted pavement performance. On the other hand, predicted pavement performance is insensitive to hourly distribution and wheelbase distribution. Hence, regional traffic data were developed from ten site specific data using both arithmetic average and clustering methods. Since, ALS and VCD are two inputs which affect the predicted distresses significantly, these two values were considered for this case. Finally, using the regional inputs, the national inputs, and the site-specific inputs of VCD and ALS, pavement ME predicted performances were determined. Results show that predicted performance by the cluster data are much closer to those by the site-specific data. Performance generated by the ME default values are significantly different from those generated by the site-specific or cluster values. When comparing performance by the ME design default to those by the statewide average data, the ME design default VCD produces less error than the ALS. Therefore, this study recommends using clustered data or site-specific WIM data instead of ME default or statewide average value. In addition, a guideline was successfully established to select appropriate axle load spectra inputs based on vehicle class data.


Asphalt Pavement, Input Level, Pavement Performance, Weigh-in-Motion, Traffic Load, Cluster Analysis


New Mexico Department of Transportation (NMDOT)

Document Type




Degree Name

Civil Engineering

Level of Degree


Department Name

Civil Engineering

First Advisor

Tarefder, Rafiqul

First Committee Member (Chair)

Stormont, John

Second Committee Member

Ng, Tang-Tat