Optimization of Maintenance Planning for Water Distribution Networks under Random Failures

Abstract

Cost-effective and preventive maintenance for water distribution networks (WDN) is essential for the sustainable and reliable use of water resources. In this work, mathematical models were developed and optimization procedures were used for optimal WDN maintenance planning under random failures. A conditional probabilistic measure known as conditional expected supply impact (CESI) was derived, which took into account both WDN infrastructure conditions and dynamic customer demands. To seek an optimal WDN maintenance plan, an integer programming problem was formulated to minimize the CESI measure with binary decision variables and budget constraints. The mathematical models and optimization approach were demonstrated using a regional WDN in a large United States city. Genetics algorithms (GA) were applied to solve the optimization problem and find the optimum maintenance plan for the WDN. The optimum plan averaged a reduction of CESI by 60%.