Linking Residential Smart Water Data with Household Activities in the Context of Supportive Housing

Abstract Submission: Residential water use is an important component of engineered water systems. However, understanding specific residential water end uses requires fine-resolution data or intrusive sub-metering. Advances in smart water metering systems and machine learning algorithms can reveal additional information about residential water-use behaviors. Understanding water-use behaviors and household activities can be valuable in supportive housing contexts, such as living facilities for elderly adults or individuals struggling with homelessness that combine housing with social support services. Supportive services can include daily well-being checks, which can be invasive and compromise privacy. As an alternative, we present a proof-of-concept approach using smart water meter data to non-intrusively monitor water end-use behaviors. We demonstrate this approach using synthetic yet representative 1-second resolution flow data for residential water end uses, categorizing household activities into typical and atypical conditions. These results can be useful for educating residents about potential benefits from smart water metering systems and creating feedback alerts to inform service providers when well-being interventions are necessary while maintaining resident privacy.

Learning Objectives/Expected Outcome (Optional) :