Sampling of Residential Water Use for Leak Control via Water Budgets

The paper provides insights into the sampling techniques that may be employed to assess domestic water use for developing water budgets at district scale aimed at losses evaluation. The basic idea is to use only few meters to provide inference on the total use of a district metered area by standard statistical sampling procedures. Random and stratified sampling are compared with reference to the sampling fraction necessary to obtain estimates of the daily use totals (or means) with given precision and confidence level. An application with real data from a small, homogeneous group of connections in Palermo (Italy) shows how stratified sampling allows a considerable reduction of the sampling fraction. A sensitivity analysis on model parameters (group variance, precision, confidence levels) shows how stratified sampling is always a better option than random sampling and that results are mainly driven by precision, with a considerable increase of the required sample size when the desired precision shifts from 5% to 1%.