Calculation of the mean residence time in distribution systems from tracer studies and models

Abstract

The residence time of water within the distribution system (DS) is a key parameter to characterize the extent of disinfectant loss and disinfectant by-product formation. While hydraulic models include calculation of a parameter referred to commonly as ‘water age’, the expense of development and calibration has restricted their availability. Chemical tracer studies provide a less expensive alternative way to estimate water age even if these may need to be repeated to capture the seasonal range of water demand. This research presents a technique to calculate the ‘mean residence time’ (MRT) from tracer studies by applying well-known principles from chemical reactor engineering. A numerical experiment was first performed using a pipe network from a case study. A negative step input of a conservative chemical to a case study DS was simulated using the EPANET model. The response curves at a series of nodes were predicted by the water quality sub model of EPANET from which the MRTmodel values were calculated. MRTmodel values were close to the average water age as predicted by the hydraulic sub model of EPANET. Actual tracer studies were conducted in the distribution systems of Raleigh (serving 250,000) and Durham (serving 190,000), North Carolina. MRTfield values were calculated from tracer response data at a series of sampling stations. A highly skeletonized hydraulic model was available for the Raleigh DS to generate a predicted average water age. The MRTfield values at 12 stations were consistently higher than average water age, most likely because the hydraulic model was too highly skeletonized. The tracer study in the Durham DS showed the use of two or more tracers to calculate an MRTfield that was weighted by the