| Summary: | 博士 === 國立交通大學 === 環境工程系所 === 96 === Water distribution networks (WDNs) should supply safe, stable, and high quality water to consumers. During the distribution, pipeline breaks in a WDN could deteriorate the water quality, cause insufficient water pressure at demand nodes, and subsequently reduce its ability to deliver water. Therefore, vulnerability analysis should be implemented to determine proper maintenance strategies, such as pipeline replacement, remedial actions for a pipe break, and installation of a water quality monitoring network, to improve the performance of a WDN.
This study used the valve-enclosed segment as the basic unit to simplify a complex WDN and proposed two depth-first search based algorithms to efficiently identify segments and the critical segments that can shut off water supply to downstream segments when closed. Two failure analysis procedures based on the connectivity among segments and the hydraulic simulation procedure with consideration of insufficient pressure at demand nodes were implemented to evaluate the impact of a failure event for determining proper maintenance strategies. The failure analysis procedures were then applied to simulate the shortage impact caused by shutting off one or more segments, and the results were used to determine effective pipe replacement decisions by a previously developed optimization model.
The valve-enclosed segment was also treated as a basic unit for developing the analysis procedure to evaluate the valve criticality, with consideration of the potential shortage impact expansion caused by a failed valve. A search algorithm was developed to identify all possible failure paths which might reach the failed valve. The criticality of a valve is determined based on the summation of potential shortage impacts multiplied by their failure probabilities for all possible failure paths. The valve criticality information can be used to develop an efficient valve maintenance plan.
A monitoring network is essential to ensure the quality of water in a WDN. To avoid the tedious trial-and-error procedure proposed in a previous study, a mixed-integer programming model was developed in this study to determine an optimal monitoring network directly. Two additional alternative models were also proposed to provide extra information for making a good decision.
A web-based decision support system was developed based on the proposed WDN vulnerability and maintenance analysis procedure and models. Vulnerability analysis related modules, a shut-off analysis module, and a valve failure impact expansion tree analysis module were developed and integrated into the system to facilitate related decision analyses. A connectivity based method was also proposed to define the criticality of a value according to the potential impact expansion of a valve failure.
The proposed vulnerability analysis procedure and maintenance models were applied to several WDNs for evaluating and demonstrating their applicability. The results show that the proposed procedures and models could effectively facilitate related decision analyses.
|
|---|
