Date of Graduation

8-2012

Document Type

Dissertation

Degree Name

Doctor of Philosophy in Engineering (PhD)

Degree Level

Graduate

Department

Civil Engineering

Advisor/Mentor

Kevin D. Hall

Committee Member

Edward A. Pohl

Second Committee Member

Kelvin C. Wang

Third Committee Member

Stacy G. Williams

Fourth Committee Member

Robert P. Elliott

Keywords

Applied sciences, Monte Carlo simulation, Pavement design, Probabilistic design, Reliability analysis, Risk analysis

Abstract

Reliability used in the Mechanistic Empirical Pavement Design Guide (MEPDG) is a congregated indicator defined as the probability that each of the key distress types and smoothness will be less than a selected critical level over the design period. For such a complex system as the MEPDG which does not have closed-form design equations, classic reliability methods are not applicable. A robust reliability analysis can rely on Monte Carlo Simulation (MCS). The ultimate goal of this study was to improve the reliability model of the MEPDG using surrogate modeling techniques and Monte Carlo simulation.

To achieve this goal, four tasks were accomplished in this research. First, local calibration using 38 pavement sections was completed to reduce the system bias and dispersion of the nationally calibrated MEPDG. Second, uncertainty and risk in the MEPDG were identified using Hierarchical Holographic Modeling (HHM). To determine the critical factors affecting pavement performance, this study applied not only the traditional sensitivity analysis method but also the risk assessment method using the Analytic Hierarchy Process (AHP). Third, response surface models were built to provide a rapid solution of distress prediction for alligator cracking, rutting and smoothness. Fourth, a new reliability model based on Monte Carlo Simulation was proposed. Using surrogate models, 10,000 Monte Carlo simulations were calculated in minutes to develop the output ensemble, on which the predicted distresses at any reliability level were readily available. The method including all data and algorithms was packed in a user friendly software tool named ReliME.

Comparison between the AASHTO 1993 Guide, the MEPDG and ReliME was presented in three case studies. It was found that the smoothness model in MEPDG had an extremely high level of variation. The product from this study was a consistent reliability model specific to local conditions, construction practices and specifications. This framework also presented the feasibility of adopting Monte Carlo Simulation for reliability analysis in future mechanistic empirical pavement design software.

Share

COinS