Development of Reliable Pavement Models
José Pablo Aguiar-Moya, Jorge Prozzi, University of Texas at Austin, May 2011, 150 pp. (161025-1)
The current report proposes a framework for estimating the reliability of a given pavement structure as analyzed by the Mechanistic-Empirical Pavement Design Guide (MEPDG). The methodology proposes using a previously fit response surface, in place of the time-demanding implicit limit state functions used within the MEPDG, in combination with an analytical approach to estimating reliability using First-Order and Second-Order Reliability Methods (FORM and SORM). Additionally, in order to assess the accuracy of the FORM and SORM reliability estimates, Monte Carlo simulations are also performed.
A case study based on a three-layered pavement structure is used to demonstrate the methodology. Several pavement design variables are treated as random; these include HMA and base layer thicknesses, base and subgrade modulus, and HMA layer binder and air void content. Information on the variability and correlation between these variables are obtained from the Long-Term Pavement Performance (LTPP) program.
Response surfaces for limit states dealing with HMA rutting failure are fit using several runs of the MEPDG based on a factorial design of combinations among the aforementioned random variables, as well as traffic, structural, and climatic considerations. These response surfaces are then used to analyze the reliability of the given pavement structure.
Using the second moment and simulation techniques, it was found that on average the reliability estimate by the MEPDG is very conservative. Additionally, the validity of the methodology is verified by means of direct simulation using the MEPDG. Finally, recommendations on fitting the response surface are provided to ensure the applicability of the methodology.
Keywords: Mechanistic-Empirical Pavement Design Guide (MEPDG); First-Order and Second-Order Reliability Methods (FORM and SORM); Pavement Structure; Pavement Response
ENTIRE REPORT (Adobe Acrobat File – 4.6 MB)