Optimization-based decision support for inspection and maintenance of infrastructure networks
Infrastructure networks that provide basic services such as transportation, telecommunications, electricity distribution, and water supply and drainage are critical for the smooth functioning of a nation’s economy and its society. To provide efficient and uninterrupted services, these infrastructure networks need to be periodically inspected, upgraded, and maintained. However, infrastructure networks are expensive to operate and maintain; many infrastructure service providers allocate more than half of their total capital investments to network maintenance and improvement. With increasing customer expectations, intensifying global competition, and challenging financial environments, the infrastructure service providers need to develop models that can optimize all of the different factors that must be taken into consideration when making important decisions related to infrastructure network inspection and maintenance. This dissertation, which consists of three essays, focuses on some of the key decision issues associated with inspection and maintenance of these large infrastructure networks. Specifically, the first two essays, respectively, address a project management problem to maintain and expand a large-scale network and a periodic network inspection problem. The third essay, motivated by the computational challenges of the first two problems, addresses the network reduction and approximation problem within the same context. These problems are deterministic optimization problems over large-scale networks, which are very difficult to solve, and have not been extensively studied in the literature. In this dissertation, we introduce new optimization models for each problem, develop theoretical and algorithmic strategies that exploit problem structures to effectively solve the problems, and implement and test these methods on actual problems using data provided by an infrastructure service provider.