The accuracy of distribution system water quality models for reactive substances such as
chlorine is dependant on how the reaction processes are modeled, the hydraulic flow
model and on the quality and quantity of field data available for model calibration. The
current trend in modeling reaction processes is to separate bulk and wall reactions.
However, current water quality models still treat bulk reaction coefficients as pipe
properties. The limitation of this approach in multi-source networks is that the bulk
reaction coefficient associated with a given pipe remains constant with time irrespective
of changes in the blend of water passing through the pipe.
A more appropriate approach in multi-source networks may be to treat the bulk reaction
coefficient as a source water property and to treat the bulk reaction coefficient in a given
pipe as a variable dependent on the blend of water passing through the pipe. This
approach requires an appropriate method for computing bulk reaction coefficients for
different blends of source waters given the bulk reaction coefficients of the sources.
This paper presents and investigates an approach to modeling and calibrating water
quality models that accounts for the variation of bulk reaction coefficients within pipes
due to in-network source water blending. This approach is then applied to model water
quality in one of the service areas of South Central Connecticut Regional Water
Authority receiving water from groundwater and surface water sources. Includes 2 references, figures.