System reliability-based design optimization of flexible pavements using adaptive meta-modelling techniques

Recognizing the need for sustainable and long-lasting structures, reliability-based designs have been gaining acceptance in the pavement engineering community. The Reliability Based Design Optimization (RBDO) is formulated to design cost-effective flexible pavements in the presence of uncertainties. Two meta-modelling approaches are considered in this study, the second-order adaptive Response Surface Model (RSM) and the adaptive Polynomial-Chaos based Kriging (PC-Kriging) meta-model. The study highlights the need for adaptive meta-modelling techniques in the reliability-based design of pavement structures, through the quantification of the epistemic uncertainty. For the numerical examples considered in this study, the model uncertainty was found to be around 1% for three-layer sections and below 2.5% for four-layer pavements. To address the correlation between the pavement failure modes, a System Reliability Based Design Optimization (SRBDO) approach is proposed in this paper. The SRBDO methodology is further adapted to design the optimum combination of pavement layer thicknesses and moduli, to meet the target levels of reliability, given the expected traffic demand and subgrade strength at a pavement site. Thus, the SRBDO formulation using adaptive meta-models, presented in this study, efficiently integrates the economic analysis of various pavement design alternatives with the Mechanistic-Empirical procedure in a system reliability framework.

Automated summary

Recognizing the importance of sustainable and durable structures, reliability-based designs are increasingly being adopted in pavement engineering. This study proposes the Reliability Based Design Optimization (RBDO) to create cost-effective flexible pavements considering uncertainties. Two meta-modelling techniques, second-order adaptive Response Surface Model (RSM) and adaptive Polynomial-Chaos based Kriging (PC-Kriging), are considered. The study highlights the need for adaptive meta-modelling approaches to address epistemic uncertainties. The proposed System Reliability Based Design Optimization (SRBDO) method integrates economic analysis and the Mechanistic-Empirical procedure to optimize pavement layer thicknesses and moduli.