Abstract:
This study investigates the development of a predictive model for failure rates in Polyvinyl
Chloride (PVC) water distribution pipes using existing data on pipe locations and diameters.
Understanding and predicting pipe failures are essential for proper infrastructure management,
avoiding water loss, and optimizing maintenance schedules. This study strives to identify
statistically significant correlations between pipe locations and diameters, and number of failures
encountered. Historical records of failures of PVC pipes, combined with geographical information
system (GIS) data on pipe networks and their surrounding environment, will be statistically
modeled using modeling techniques such as regression analysis or machine learning algorithms.
The predictive model will provide a valuable tool for water utilities to predict high risk pipe
sections, there by implementing specific inspections and replacement policies consequently saving
on operating costs and improving service quality. The findings of this study conducted in the
context of Uganda's water infrastructure will shed light on the specific variables influencing
failures in PVC pipes in this country and help advance general knowledge regarding water
distribution network asset management
