NDMA formation during drinking water treatment: A multivariate analysis of factors influencing formation

The formation of the carcinogen N-nitrosodimethylamine (NDMA) during drinking water treatment has raised concerns in the drinking water industry. Many bench-scale laboratory tests and pilot plant studies have been completed to try to determine which factors during water treatment increase or decrease the amount of NDMA formed in drinking water. This study used data from over 20 drinking water treatment plants in the United States and Canada to determine which factors are most highly correlated with the NDMA concentration in delivered water using a mixed effects model with a random intercept. This type of analysis has not been used previously with trihalomethane (THM) models due to the fact that those studies did not sample such a large number and range of plants as was done in this NDMA study. Ultraviolet absorbance at 254 nm (UV254) in the plant influent and pre-chlorination time used at the plant were highly correlated in all models with NDMA concentration in finished water as well as the percentage change between NDMA formation potential in the plant influent and actual formation in the finished water. Specifically, an increase in UV254 absorbance in a model was associated with an increase in NDMA and an increase in pre-chlorination time in a model was associated with a decrease in NDMA. Other water quality parameters including sucralose concentration in the plant influent, polyDADMAC polymer dose, pH, and chlorine-to-ammonia weight ratio used in the plant were also correlated with NDMA concentration in the distribution system. Lastly, NDMA precursor loading was correlated with the use of polyDADMAC (where precursors were added) and the use of ozone and granular activated carbon (GAC) treatment (where precursors were removed).