Global Sensitivity Analysis of Mechanistic–Empirical Performance Predictions for Flexible Pavements

Schwartz, Charles W.; Li, Rui; Ceylan, Halil; Kim, Sunghwan; Gopalakrishnan, Kasthurirangan · 2013 · OpenAlex-citations

DOI: 10.3141/2368-02

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study addresses the limitations of previous sensitivity analyses for the AASHTO Mechanistic–Empirical Pavement Design Guide (MEPDG), which often relied on one-at-a-time methods, small input subsets, and outdated software versions. The authors conducted a comprehensive global sensitivity analysis (GSA) to quantify how variations in design inputs affect performance predictions for flexible pavements. The goal was to identify which input parameters most significantly influence distress outcomes, thereby guiding designers on where to prioritize data collection and model refinement. The research evaluated 15 base cases representing five climatic zones (hot-dry, hot-wet, temperate, cold-dry, cold-wet) and three traffic levels (low, medium, high). Over 40,000 MEPDG simulations were performed using Latin Hypercube Sampling to vary all inputs simultaneously across their plausible ranges. Inputs included traffic volume, layer thicknesses, material properties (such as dynamic modulus and resilient modulus), and groundwater depth. To analyze the complex relationships between inputs and outputs, the authors developed response surface models using both multivariate linear regression (MVLR) and artificial neural networks (ANNs). The ANNs proved superior, providing robust approximations of the nonlinear relationships with high goodness-of-fit statistics. Sensitivity was quantified using a design limit normalized sensitivity index, which relates percentage changes in inputs to percentage changes in predicted distress relative to design limits. The results identified specific inputs as consistently hypersensitive or very sensitive across all distress types. The hot-mix asphalt (HMA) dynamic modulus master curve parameters (alpha and delta), HMA thickness, surface shortwave absorptivity, and HMA Poisson’s ratio were the most influential factors for longitudinal cracking, alligator cracking, rutting, and roughness. Fatigue cracking was also highly sensitive to granular base thickness and the resilient moduli of both the base and subgrade layers. Thermal cracking sensitivity was primarily driven by the HMA aggregate coefficient of contraction and binder properties. The study noted that standard binder grades rarely produced significant thermal cracking, requiring specialized simulations with stiffer low-temperature grades to assess this distress accurately. The significance of this work lies in its provision of practical guidance for pavement designers and model developers. By identifying the most critical input parameters, the study highlights where additional effort is needed to obtain high-quality, certain input values to improve prediction accuracy. Furthermore, the findings help developers identify specific model components that may require reexamination or enhancement. The successful application of ANNs as response surface models suggests they could serve as efficient substitutes for computationally intensive geomechanics computations in future pavement design applications.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• road surface friction