摘要: |
The mechanistic-empirical pavement design guide (MEPDG), developed under the recently completed NCHRP Project 1-37A, is a complex tool. Sensitivity analyses typically vary a single input parameter while holding other parameters constant. A pilot project is outlined to identify the feasibility of a global sensitivity analysis of the design process by using random sampling techniques over the entire input parameter space. This study sampled the following flexible pavement input variables: hot-mix asphalt (HMA) base nominal aggregate size, climate location, HMA thickness, annual average daily truck traffic (AADTT), subgrade strength, truck traffic category, construction season, and binder grade "bump." A total of 100 design sections were randomly sampled from these input parameters. The resulting predicted performance of longitudinal and alligator cracking, HMA and total rutting, and international roughness index were analyzed by using the Pearson‘s and Spearman‘s correlation coefficients. These coefficients indicate that AADTT, HMA thickness, and subgrade strength have a significant impact on performance, whereas the remaining parameters have lesser impacts. The performance predicted with trial designs with default values of the least significant variables was compared with the original performance predictions using the original input factors. This comparison indicated that the default parameters produced similar performance to designs using all the input parameters. These results demonstrated that this type of sensitivity analysis may be used to identify important input parameters across the entire parameter space. The utilization of a "bumped" binder grade did not appear to have a significant influence on the performance predicted by the MEPDG models. These results indicate that further research to refine the MEPDG performance models may be necessary. |