Abstract

Water distribution systems are the most expensive part of the water supply infrastructure system. In Canada and the United States, there are 700 water main breakae every day, and there have been more than 2 million breaks since the beginning of this century, which have cost more than 6 billion Canadian dollars in repairs costs for the two countries. Municipalities and other authorities that manage potable water infrastructure often must prioritize the rehabilitation needs of their water main. This is a serious challenge because the current potable water networks are old (i.e. deteriorated) and require certain modifications to bring them up to acceptable reliability and safety levels within a limited budget. In other words, municipalities need to develop a balanced rehabilitation plan to increase the reliability of their water networks by rehabilitating (first) only those pipelines at high risk of failure. The objective of this research is to develop a risk model for water main failure, which evaluates the risk associated with each pipeline in the network. This model considers four main factors: environmental, physical, operational, and post-failure factors (consequences of failure) and sixteen sub-factors which represent the main factors. Data are collected to serve two purposes: to build the model and to show its implementation to case studies. The required data are collected from literature review and through a questionnaire sent to the experts in the field of water distribution network management. From the collected data, pipe age is found to have the most significant indication of water main failure risk, followed by pipe material and breakage rate. In order to develop the risk of failure model, hierarchical fuzzy expert system (HFES) technique is used to process the input data, which is the effect of risk factors, and generate the risk of failure index of each water main. In order to verify the developed model, a validated AHP deterioration model and two real water distribution network data sets are used to check the results of the developed model. The results of the verification show that the Average Validity Percent is 74.8 %, which is reasonable considering the uncertainty involved in the collected data. Based on the developed model, an application is built that uses Excel ® 2007 software to predict the risk of failure index. At last, three case studies are evaluated using the developed application to estimate the risk of failure associated with the distribution water mains.