Heuristic Methods for Pipeline Network Design

Abstract

We showcase geospatial heuristic methods for network design and optimization. We propose and adapt graph algorithms to achieve optimal (or close to optimal) fluid transportation networks meeting quantifiable criteria (such as minimizing cost for example). Typically, these are used on pipeline infrastructure design, for CO2 collection or H2 distribution for example. The pipeline cost functions involved in the optimization depend on both pipeline length and a concave function of pipeline capacity. As such, discrete optimization methods are required. We have extended the tool to integrate other known aspects of network design. A sink placement algorithm can identify the minimum-cost storage location (and in parallel construct the rest of the a priori unknown network structure). The tools have finally been adapted to allow the inclusion of pre-existing pipeline infrastructure at a lower cost. They can then propose networks that prioritize planning along pre-existing pipeline routes.