Optimizing inspection routes in pipeline networks

Abstract

Maintaining an aging network is a challenge for many water utilities due to limited budgets and uncertainty surrounding the physical condition of buried pipeline assets. The deployment of robotic inspections provides high quality data, but these platforms have limited use due to cost and operational constraints. To facilitate cost-efficient inspections, operators need to identify high-risk assets while accounting for the effectiveness of the tools at hand. This paper addresses inspection planning with the goal of finding an optimal route while considering tool limitations. An exact integer programming formulation is presented where only three factors are used to characterize tool constraints. Two classes of solution methods are explored: 1) tree based searches, and 2) integer programming. This paper demonstrates how each method can be used to identify optimal paths within a real water distribution system. Empirical trials suggest that tree-based search methods are the most efficient when the path limit is short, but do not scale well when the path length increases. In contrast, integer-programming methods are more effective for longer path lengths but have scalability issues for large network sizes. Data preprocessing, where the input network size is reduced, can provide large computation time reductions while returning near-optimal solutions.