Bibliographic details and abstracts are available to all. Downloads of full-text dissertations are restricted to University of Portsmouth members who must login. MPhils may be accessed by all.

Abstract

This dissertation will investigate the various factors that influence the frequency of bursts and the amount of leakage on potable water mains and creates predictive models to determine the likelihood of pipe failure occurring in water distribution networks. The study involves a thorough analysis of the existing literature on pipe failure processes, pipe deterioration factors, and pipe failure prediction models.
The research is based on a historical burst dataset that is divided into three groups: 1, 2, and 3. Each group contains information on pipe failure incidents, pipe features, and environmental conditions. The study uses logistic regression modelling to determine connections between pipe failure incidents and other predictor variables such as seasonal variation, Day average pressure (AZNP) , sea proximity, pipe material, and pipe diameter. The area under the receiver operating characteristic (ROC) curve (AUC) is used as one of the outputs to assess the predictive ability of the models.
Seasonal variation, pipe material, and pipe diameter are all important predictors of water pipe failure incidents, according to the findings. The prediction models perform well, with AUC values ranging from 0.74 to 0.78. The findings are consistent with earlier studies, emphasising the significance of taking both static and dynamic elements into account when estimating pipe failure risks.

Course: Civil Engineering - BEng (Hons) - C0177

Date Deposited: 2023-07-13

URI/permalink: https://library.port.ac.uk/dissert/dis14154.html