Multi-Objective Optimization of Imperfect Pavement Repair Policy: A Reliability-Based Approach

Abstract

This study proposes a reliability-based framework for scheduling pavement repairs in a way that considers the imperfect effect of repair. Highway agencies can use this framework in the absence of comprehensive data sets for pavement performance prediction. To model imperfect pavement repairs, this study uses a hybrid hazard rate model. Further, a multi-objective optimization technique called NSGA II (non-dominated sorting genetic algorithm II) was used; it was integrated with a pavement reliability model. Rather than consider a fixed planning horizon, the number of repair cycles required (N) is kept as a decision variable considering reconstruction at the end of N cycles. Maximizing system reliability under repair at the end of the replacement cycle and minimizing the expected cost function were considered objective functions. Along with N, the algorithm also found the optimum target reliability for repair intervention. In this study, the uncertainty in construction quality (CQ) and traffic characteristics were considered in the reliability analysis. They are represented with appropriate probability distributions. Demonstrating the use of the framework, a case study was undertaken, and the reliability–cost trade-off solutions for different uncertainty levels of traffic and CQ were identified. The framework captured the effect of increased uncertainty resulting in frequent repair needing replacement much earlier. Notably, the impact of uncertainty in CQ on the Pareto fronts was more prominent than that of traffic attributes, highlighting the importance of CQ. A sensitivity analysis of the adjustment factors was performed, which found that the age reduction factor affected the optimum solutions substantially.