Modeling trihalomethane formation in drinking water after alum coagulation or activated carbon adsorption

Abstract

Eight natural waters from throughout the United States were subjected to different degrees of alum coagulation and activated carbon adsorption treatment (not in series). The concentration of trihalomethanes (THMs) formed by the reaction between the natural aquatic organics and doses of free chlorine were determined by gas chromatography at ten discrete time intervals over a total of 168 hours of reaction time. Since bromide ion concentration and temperature were held constant, and the chlorine dose was based on the final nonvolatile total organic carbon (NVTOC) concentration, THM formation was modeled with the independent variables: reaction time, pH, and a surrogate parameter for organic precursor. Several surrogate parameters were investigated to quantify the THM formation potential. A multiplicative surrogate (UV*TOC) representing the product of the NVTOC concentration and the UV adsorbance at 254 nm was found to be the best surrogate for activated carbon treated waters and was found to be approximately equivalent to NVTOC for alum treated waters. The THM formation was found to fit a two stage reaction with respect to reaction time; relatively rapid formation in the first eight hours followed by a slower formation from 24 to 168 hours. The data base was analyzed with a statistical software package that performs multiple linear regression analysis. Three types of models were developed: a linear model, a nonlinear model and a logarithm transform model. The models were checked for predictive accuracy by a number of methods including the examination of statistics from the regression analyses, scatterplots of predicted versus observed THM values, and the number of predicted values within 20% of the observed value. The logarithm transform model was found to be the best overall model, although other models were more accurate for specific applications as to reaction time or water type. Gel permeation chromatography (GPC) was employed to examine the molecular weight distribution of the aquatic organics in three of the eight waters and to determine the effects of alum coagulation and activated carbon adsorption on these distributions. In general, alum coagulation preferentially removed high molecular weight organics while activated carbon removed a broad spectrum of molecular weight organics.