Demand Management Policies for Improving Urban Water Infrastructure Performance

Abstract

Residential outdoor water use constitutes a significant portion of urban water demand, and in arid and semi-arid regions outdoor irrigation can exceed 50\% of single-family use during the warm season. Outdoor use is a primary driver of seasonal and sub-daily demand peaks, which determine the capacity and the operating cost of water distribution systems (WDSs). Advanced Metering Infrastructure (AMI) has made high-resolution consumption data broadly available, but the hourly records that utilities typically collect report only undifferentiated volume at each meter and do not reveal the composition or the timing of demand. This dissertation develops methods to estimate residential outdoor and end-use water demand from hourly AMI data, and these estimates are applied to evaluate and design demand-management policies. Household demand is coupled with a hydraulic model of the distribution network throughout. The methods are developed using a labeled end-use dataset collected at 31 single-family homes (SFH) in Utah and are applied to two years of hourly AMI records collected at approximately 18,000 SFH in Lakewood, California.

Three studies are presented. In the first study, a set of minimum-consumption methods is developed to disaggregate hourly demand into indoor and outdoor use without supplementary data. The methods are validated using the labeled Utah dataset, where the account-level Minimum Week Method (MWM-A) estimates the outdoor share within approximately 3\% of the measured value and reproduces the early-morning irrigation peak. The MWM-A is applied to the Lakewood dataset, and the average outdoor share is estimated as approximately 54\%. In the second study, the account-level outdoor estimates are used as input to a hydraulic model of the Lakewood system to compare outdoor water use restrictions (OWR) with dynamic pricing (DP). Restrictions reduce total demand and energy cost and raise the minimum nodal pressure, but they increase background leakage and water age and reduce utility revenue, while dynamic pricing produces the largest reductions in peak pipe flow with smaller impacts. Fixed-rate restrictions distribute the reduction uniformly across income groups, and affordability remains low under all policies. In the third study, the disaggregation is extended to six end uses using a stacked-ensemble model, and the End-Use Dynamic Pricing Policy (EUDPP) is formulated as a bi-level simulation-optimization problem, in which each household schedules its end-use consumption in response to price and the utility selects an end-use, time-of-use, and seasonal tariff subject to hydraulic, affordability, and revenue constraints. The optimized tariff reduces total weekly demand by 28.0\% in summer and 20.8\% in winter and reduces peak pump power by 11.1\% and 9.5\%. The reduction is concentrated in discretionary irrigation, essential indoor uses are protected, and all constraints are satisfied.

The methods and findings of this research provide a framework for using smart water meter data to estimate residential outdoor and end-use demand and to design demand-management policies that are equitable and account for infrastructure performance. The framework supports data-driven decisions for the management of urban water distribution systems.