Empirical bromate formation models were developed from observed batch
ozonation data to simulate the effects of important water quality char
acteristics and treatment processes on bromate formation. Data generat
ed from true batch ozonation studies with an orthogonal matrix design,
served as the base for development of empirical models for bromate fo
rmation. The variables examined include pH, initial bromide concentrat
ion, alkalinity, ozone dose, ammonia and dissolved organic carbon leve
l. This array of variables was selected to evaluate the effects bf imp
ortant water quality characteristics and treatment processes on the fo
rmation of bromate ion during water ozonation. Limited temperature var
iation data were also generated. Isolation and fractionation of natura
l organic matter (NOM) by ultrafiltration and reverse osmosis membrane
techniques allowed assessment of source water variations and the role
of molecular size on bromate formation. Multiple linear regression of
logarithmic transformations was used to generate the models. Results
obtained from the empirical models indicate that bromate formation is
favored at high pH, initial bromide concentration, alkalinity and high
ozone dose. On the other hand, increasing DOC and ammonia concentrati
on decreased bromate formation. The models also indicated that NOM iso
lated from different water sources have a strong influence on bromate
formation upon ozonation of model solutions composed of different memb
rane isolates. Both internal and external validation of the models dem
onstrated that the bromate models predicted bromate formation well. Fi
nally, the models indicated that acid addition (pH depression) strateg
ies can be used to control bromate formation. Copyright (C) 1996 Elsev
ier Science Ltd.