High-resolution domestic water consumption data – Scope for leakage management and demand prediction

Abstract

Challenges such as water scarcity and ever-increasing demand put an additional strain onto water distribution networks. Better asset management through leakage mitigation and demand forecasting can offset the current and future implications of these challenges. This paper shows how new high-resolution logging (e.g. 1 litre ticks) is able to enhance traditional methods of investigating leakages (e.g. minimum night flows) and instantiate novel methods for demand prediction (through micro-component analysis). Machine learning or other statistical analytical techniques coupled with the high-resolution data can be used in an adaptive way for leakage detection and demand forecasting. As a proof of concept, this paper investigates example datasets obtained from a UK based water company. The analyses suggest that it is possible to: extrapolate leakage from night flow time series data; predict water consumption patterns for different types of households and create consumption profiles based upon water user characteristics/behaviour.