Robust and Resilient Water Distribution Systems

Abstract

The purpose of a water distribution system (WDS) is to deliver the required amount of water to the customer under the desired pressure and quality. However, demand change and component failure result in low pressures at customer taps and make it difficult to achieve the goal. To mitigate the impact of the disturbances, system performance measure such as robustness and resilience can be considered in the WDS design and operation. Robustness is generally defined as an ability of the systems to maintain its function under a defined set of disturbance. On the other hand, Resilience is a system's ability to prepare and recover from a failure. The goal of this dissertation is to develop methodologies to enhance WDS robustness and resilience. In robustness-based design, reliability has been considered. Reliability is generally defined as the system's ability to provide an adequate service to customers under uncertain system condition and measured by the probability that stochastic nodal pressures are greater than or equal to a prescribed minimum pressure. However, although improving reliability will improve system robustness, the question is how the reliability index will improve system robustness. Robustness incorporates the variation of system performance; an additional aspect of system performance that reliability does not encompass. Pipe bursts are the most common failure in WDS. Therefore, promptly detecting and locating bursts will decrease the failure duration and increase system resilience. While many burst detection methods are available, identifying the method with the highest detectability is important to system owners/operators. However, to date, no cross comparisons of these methods have been completed for burst detection using a common data set. In addition, most traditional burst detection methods do not have a mechanism to include system operational changes. This dissertation is composed of three journal manuscripts that address these three key issues on WDS robustness and resilience. For WDS robustness improvement, a new robustness index is developed and used for multi-objective robustness-based design. The robustness-based design is compared to conventional reliability-based design. For WDS resilience improvement, the best method among six Statistical Process Control (SPC) methods is identified in terms of detection effectiveness and efficiency. Finally, a burst detection method applicable under system operational condition change is posed.