Joint Optimization of Resource Allocation and Maintenance Planning for Multi-Facility Infrastructure Systems

This research considers an integrated resource allocation and maintenance planning with complex maintenance effects for infrastructure systems consisting of multiple facilities. We formulate the problem as the sum of multiple finite-horizon Markov decision processes subject to budget constraints and solve it using Lagrangian relaxation method. A two-stage method is presented to solve the Lagrangian dual problem via subgradient optimization. In Stage 1, the Lagrangian relaxation problem is decomposed into individual Markov decision process problems, which generally leads to maintenance policies violating budget constraints. A Lagrangian repair heuristic is proposed to repair the infeasible policies to high-quality feasible ones in Stage 2. A computational study is conducted to demonstrate the proposed model using a pavement system with two sections.