An optimal stochastic decision-making model incorporating pavement maintenance and rehabilitation works

An optimal stochastic decision-making model is proposed to generate a long-term maintenance and rehabilitation (M&R) schedule. It applies the discrete-time Markov model which incorporates pavement deterioration and improvement rates via the transition matrix. A transition matrix with 10 condition states includes 17 M&R variables, each representing a distinct M&R treatment. The model aims to optimise the pavement condition rating at the network level while enforcing variable and budget constraints. A simplified efficient procedure is proposed to optimise the model using a cost-effectiveness ratio, CE(Xi), defined as the ratio of a specific M&R treatment cost to the expected improvement outcome. M&R variables with lower CE(Xi) values are given priority for inclusion in the optimal solutions. The model data requirements include the stochastic parameters represented by the initial state probabilities and transition matrix, and cost units associated with specified M&R treatments. Three case studies are presented to demonstrate the efficacy of the proposed model. The first one employed all 17 M&R variables, the second applied 14 M&R variables with added vehicle operating cost (VOC), and the third only used 8 M&R variables with added VOC. The sample results indicated that more funding is spent on rehabilitation than maintenance as the annual budget increases. The solutions with added VOC provided more rehabilitation works for states with poor/bad pavements. The case with 8 M&R variables resulted in more rehabilitation work at a lower overall M&R cost.