Using artificial intelligence techniques to automate sewer system planning

This thesis explores the use of computing science algorithms in sewer system automation. Related research can be separated into two subproblems; design and layout. Design determines pipe properties such as size, depth and slope. Sewer system layout specifies the topology of the pipe network. Many layout techniques consider only high-level connectivity between key neighborhood points. This thesis improves the automated layout process by finding detailed pipe and manhole positions. A set of primitive algorithms for placing a pipeline between two fixed points is developed. These primitive algorithms are used to develop two algorithms to minimize the entire neighborhood cost. The first uses a local greedy optimization heuristic to quickly generate high quality solutions. A second algorithm implements a branch-and-bound search to generate the best layout based on a set of fixed points. These algorithms are validated within a complete sewer planning prototype using a third party design module.