Effective rehabilitation strategies for water pipes play a very important role in both sustaining the reliability of water distribution systems and reducing costs. Pipe breakage prediction models provide a platform for effective rehabilitation strategies. The strength of the rehabilitation strategies is just an extension to those predictive models. There are different techniques and methods for modeling pipe breakage based on identifying breakage patterns using statistical or data-driven (mining) techniques. This review addresses those techniques from the perspective of rehabilitation strategy applications. Therefore, the rehabilitation strategies presented in the literature were reviewed according to three criteria: the level of pipe breakage prediction (pipe-group level or individual-pipe level), the phase, according to the bathtub curve, in which the predictive model is applicable and the performance of the system after rehabilitation. The use of artificial neural networks (ANNs) was found superior over statistical techniques for predicting pipe failure rates and consequently in rehabilitation strategies. However, ANNs are relatively less concerned with identifying specific relations between the variables involved. A proposal for the future research of environmentally integrated, optimal, dynamic and proactive rehabilitation and operation strategies is highlighted at the end of the article.

You do not currently have access to this content.