Evaluation of suitable chlorine bulk-decay models for water distribution systems

Maintaining the chlorine residual is a major disinfection goal for many water distribution systems. A suitable general chlorine bulk-decay model is required for simulation of chlorine profiles in networks to assist disinfection planning/management efficiently. The first-order model is unsuitable due to inaccuracy and inability to represent rechlorination. Three potentially suitable, simple, reactant models were compared. The single-reactant model was found to be unsuitable, as it was inaccurate when restricted to using a single set of invariant parameters. The two-reactant model was more suitable than the variable-rate-coefficient model, although both models were accurate under the same restriction. The two-reactant model was then calibrated against datasets consisting of multiple decay tests for five distinctly different waters. It accurately predicted data reserved for validation over the chlorine concentration range of 0–6 mg/L, using a single set of invariant parameters, and is therefore the simplest, generally suitable model for simulating chlorine profiles in distribution system networks. [Display omitted] ► Three second-order chlorine decay models were compared for accuracy, over the range 0–3 mg/L. ► The single-reactant model had unacceptable accuracy at high initial chlorine concentrations. ► The variable reaction coefficient model and two-reactant model had similar, acceptable accuracy. ► Only the two-reactant model parameters showed consistent temperature dependency. ► The two-reactant model was rigorously validated against data from five different drinking waters.