Characterizing stochastic water age in premise plumbing systems using conventional and advanced statistical tools

Abstract

Water age is often used as a proxy for water quality in drinking water systems, but the highly stochastic nature of water demands in premise plumbing systems, and nonlinear relationship between water age and quality, makes the average age a poor indicator of quality. In this research, we propose using statistical tools, including a formal extreme value analysis, to assess water age distributions. These tools were leveraged to provide qualitative and quantitative measures associated with the maximum water age across different time-windows. The water ages themselves were computed using a stochastic water demand generator (SIMDEUM) and a hydraulic pipe network model (EPANET). The approaches were compared by assessing water ages for varying water demands and varying number of occupants for a single-family home. For the shower, a 10× reduction in household water demands did not substantially change the maximum ages. However, changing the occupancy from two people to one person, with no change in the average water demands, had a greater impact than the 10× average water demand reduction. For example, for the shower's cold water supply in a house with one person, there was a 10% chance of water age exceeding 80 hours in a 31 day time period. However, for the same house and time period, but with two people together using the same amount of water as the single person, e.g., via low-flow fixtures, there was less than a 1% chance of the water age exceeding 80 hours. Overall, this research provides new approaches to quantify and analyze water age data and can provide insights into the effects of changing water demand patterns and occupancy on water age.