Modeling Residential Water Demand to Uncover High Consumption and Potential Drivers

Water distribution systems operations rely on the demand exerted by their customers. Measuring water demand at monthly time steps is mainly used for billing purposes. From a management perspective, the link between water demand and exogenous variables may shed light on consumption patterns and pinpoint high consumption areas in the hydraulic system. In this project, we collect residential water demand data from January 2016 to December 2020 reported at monthly time steps from accounts located in a suburban Chicago area. We then model water demand by including public data from the Township Assessor’s Office. These public data include the property area, number of bedrooms, number of bathrooms, and age of the building. Our analysis aims to benchmark residential water demand and its relationship with publicly available exogenous data. Preliminary findings show that water demand from June to October is consistently higher than the rest of the year, with its peak occurring in August. The lowest consumption is reported in March for these five years. When seasons are analyzed, Summer and Fall are the seasons with the highest consumption. The predictive model reports that the property characteristics explain a low percentage of residential monthly water demand. The developed model can assist utility planners in targeting areas with high consumption and provide a demand management strategy to foster water conservation policies.