Bi-objective pavement maintenance and rehabilitation optimization decision-making model incorporating the construction length of preventive maintenance projects

Abstract

Existing optimization decision-making approaches for pavement maintenance and rehabilitation (M&R) ignore the construction length of preventive maintenance (PM) projects, and its negative effects are difficult to be transformed into cost. To address this issue, this study proposes a bi-objective decision-making model that incorporates the problem as the second objective into the two-stage bottom-up approach. The proposed model contains selection of performance indicators, Bayesian neural network-based probabilistic deterioration model, evaluation of initial M&R actions on a segment level, and bi-objective decision-making. It is solved by the enumeration method and the non-dominated sorting genetic algorithm II. Finally, the Pareto solutions are obtained. A solution is an M&R plan, where an initial action (treatment type) is selected for a pavement segment. Among the Pareto solutions, the one with the second objective greater than or equal to and closest to the shortest construction length, is the optimal M&R plan. In addition, compared with the model that converts the problem into a constraint, the proposed model recommends a better plan that can achieve higher performance at lower cost, which validates the strength of the proposed model. Decision-makers can adopt the proposed model to optimize pavement M&R plans that consider the construction length of PM projects.