Mitigation of Disinfection By-Product Formation through Development of a Multiple Regression Equation and a Bayesian Network
Issues of Disinfection By-Product (DBP) formation in response to chlorination in drinking water treatment systems is a common issue encountered by WTP operators. Efforts to minimize DBP formation are complicated by the presence of zebra mussels, which may inhabit the raw water intake of WTPs. While chlorination at the intake to control zebra mussel populations is effective, the formation of DBPs is exacerbated. Methods for reducing DBPs are explored, including adjusting the location for chlorine additions in the treatment sequence. Multivariate models for Total Trihalomethane (TTHM) and Haloacetic Acid (HAA) subspecies are employed to show that in some instances pre-chlorination can be reduced to lower DBP formation, while post-chlorination can be increased. A Regression model (R2 of 0.75) predicts that DBP levels can be lowered by post-chlorination rather than pre-chlorinating raw water for portions of the year except during the combatable life stage to assist in zebra mussel control. A second multivariate regression model for TTHM (R2 = 0.91) which includes bromide, a variable which, due to lack of data, was previously unused, is described and demonstrates that DBP levels can be reduced by lowering pre-chlorination levels. Finally, a Bayesian network is developed using the Webweavr-IV Toolkit, utilizing causal relationships between raw water quality parameters in the form of conditional probabilities. The results show that the average cancer risk can be decreased by while still maintaining zebra mussel control and simultaneously decreasing the incremental cancer risk, which currently fluctuates between 1 in 50,000 to 100,000 in Ontario.