Identifying Time-of-use Demand Characteristics and Price Elasticity of Residential Water Users Using Smart Water Meter Data

Abstract Submission: Urban water decision-makers must manage peak demands to conserve energy consumption associated with delivering water and offset the need to repair and expand water infrastructure. Utilities can use digital water data to understand the characteristics of different water consumers, their consumption, and their response to price-based water management tools. This research explores different types of water users based on time-of-use profiles through an approach that integrates clustering analysis and econometric modeling. A machine learning clustering algorithm, dynamic time warping, is applied for hourly water demand data to establish clusters with unique time-of-use demand patterns. Census data and physical housing characteristics are used to assess differences among clusters based on property value, property size, income, household size, and head-of-household age. Weather and precipitation data are paired with demand and billing data, and econometric regression modeling is applied to assess price elasticities for each cluster. The results of the clustering analysis indicate that there are five distinct clusters. While the average demand pattern of the population shows two demand peaks, no single cluster of water users has the same pattern. The five clusters differ in the time of peak demand, magnitude of demand, and contribution to total demand peaks. Late night and early morning water users consume a higher average volume of water than other clusters, and mid-day water users consume the lowest volume of water. Mid-day water users have the lowest home values, and morning water users are more likely to have higher median incomes. Finally, results from an elasticity regression indicate mid-day water users are most responsive to water price, while early morning and late night water users are the least responsive to water price. This research quantifies distinct time-of-use patterns of water consumption and elasticity values for population segments to generate insight for developing tailored demand management programs.

Learning Objectives/Expected Outcome (Optional) :