Investigating bromide incorporation factor (BIF) and model development for predicting THMs in drinking water using...

Area of Science:

• Environmental Chemistry
• Water Quality Analysis
• Toxicology

Background:

• Disinfection byproducts (DBPs) in drinking water, particularly trihalomethanes (THMs), pose cancer risks.
• Brominated THMs are more toxic than chlorinated ones, yet models for individual THM prediction are scarce.
• THM formation is influenced by natural organic matter (NOM), bromide ions, disinfectants, pH, temperature, and reaction time.

Purpose of the Study:

• To develop and evaluate machine learning models for predicting individual THMs (chloroform, bromodichloromethane, dibromochloromethane, bromoform) and total THMs.
• To investigate the contribution of different molecular weight fractions of NOM to THM formation and bromide incorporation.
• To assess the impact of pH, bromide-to-chlorine ratio, and NOM characteristics on bromide incorporation factors (BIFs).

Main Methods:

• Fractionation of NOM by molecular weight.
• Investigation of DOC, THMs, and BIFs across NOM fractions.
• Development and validation of predictive models using Support Vector Regressor (SVR), Random Forest Regressor (RFR), and Artificial Neural Networks (ANN).

Main Results:

• BIFs ranged from 0.08-0.16 and 0.07-0.15 per mg/L DOC at pH 6.0 and 8.5, respectively.
• Higher BIFs were observed at lower pH and lower NOM molecular weights, and with increased bromide-to-chlorine ratios.
• Models demonstrated excellent predictive performance (R² = 0.870–0.988) in testing datasets, with SVR and RFR showing superior results.

Conclusions:

• The developed machine learning models accurately predict individual THMs in drinking water.
• These models can aid in controlling specific THMs to ensure regulatory compliance and minimize human health risks associated with DBPs.
• Understanding the role of NOM fractions and bromide is crucial for effective THM management.