Kinetics of Free Chlorine Decay in Water Distribution Networks

Oxidizers such as chlorine are commonly added to water before it leaves the utility plant to provide residual disinfection in distribution networks. However, their concentration and thus their effectiveness decreases as water flows through distribution networks due to a combination of processes in bulk water (natural decay and reactions with chemical and biological contaminants) and at the pipe wall (adsorption and chemical reactions such as corrosion). We have conducted batch reactor tests and closed-loop tests in a PVC pipe loop with and without a biofilm at the EPA's T&E Facility in Cincinnati, OH, to measure the rates of free chlorine decay in bulk water and at the pipe wall. Our tests have shown that the free chlorine decay rate remains almost constant with time (1 to 2 days) indicating zeroth order kinetics. Accordingly, we fitted these data to a zeroth order reaction model and determined the rate constants for bulk and wall decay. In most of the studies published in the literature, free chlorine decay rates measured in batch reactors, closed pipe loops and water distribution networks have been fitted to first order kinetics models. It is difficult to verify the accuracy of first order kinetics fits since the raw test data are usually not included in most of these papers. When raw data were published, we found that the zeroth order kinetics model fits the data similar to or better than the first order model. Since it was not possible to directly compare our free chlorine decay kinetics with those from the literature, we carried out parametric calculations using our H₂OFate software to examine the effect of kinetics models on TFC concentration predictions in a sample water distribution network. These calculations have shown that (1) the extent of chlorine decay at the pipe walls is comparable to that in bulk water suggesting that both processes are equally important in water quality analysis; (2) the presence of biofilm grown on PVC pipe walls does not affect the TFC concentrations in the network; and (3) the TFC concentration predictions are sensitive to the kinetics model selected.