Data processing automation for bulk water supply monitoring

Abstract

Water as a resource is becoming more scarce with South Africa having several provinces being struck with droughts. Up to 30% of water is lost through leaks in water distribution networks. It is common practice to monitor water usage in large water distribution networks. These monitoring systems unfortunately lack the ability to alert on high flow rates and detect water leaks unless the data is reviewed manually. The paper will explore statistical and Artificial Intelligence approaches to test the viability to detect leaks. This will can then be used as an alerting team to improve operational efficiencies of small teams and reduce repair time of leaks and thus reduces water lost through leaks.