Predicative Models and Online Monitoring to Predict THM Formation at a Full-Scale WTP
- Allison Reinert, Justin Irving, Ben Stanford - Hazen and Sawyer
- Tiffanie Hawley - Cape Fear Public Utility Authority
As many utilities strive for compliance with the Stage 2 disinfectants/disinfection byproducts (D/DBP) rule, utilities are evaluating different techniques and tools to monitor DBPs and predict DBP formation in their distribution systems. Several predictive models have been developed for both trihalomethanes (THMs) and haloacetic acids (HAAs), however, most of these models are limited in their wide-spread applicability. Additionally, in recent years online THM monitoring systems have gained traction with water utilities looking to better understand the formation of THMs in their plants and distribution systems. Monitoring systems have greatly enhanced the understanding of the formation and distribution of THMs, but do not provide any predictive capability on its own. As such, the principal objective of this paper will be to present results from a recently-completed study where we compared various predictive THM models and then used on-line, continuous THM data with 3-D fluorescence excitation emission matrices (EEMs) and principal component analysis (PCA) to develop a new predictive model for THMs at a full-scale run of river drinking water treatment plant (WTP), even with periodic high-concentration bromide influence.
In this study, online THM monitoring (hourly samples) at the point of entry was conducted for four months with concurrent water quality sampling. Fluorescence sampling of the combined filter effluent was also conducted for one month. The online THM monitor was able to capture both total THMs (TTHMs) and speciated THMs. This large data set allowed for the development of baseline THM conditions as well as baseline organic matter conditions. With water quality and online THM measurements in hand, three different predictive models previously developed for both raw water and treated water applications were compared. Additionally, a predictive model based off of the data from the WTP was developed using PCA. Through the use of the statistical software R, the components for the predictive model were discerned from the excitation/emission wavelength pairs that most influence the formation of THMs. As a result, a three component model was developed with an R2 value of 0.96 for chloroform and 0.86 for bromoform prediction. Additional models were also created for each of the other speciated THMs and TTHM prediction which illustrated the influence of the bromide concentration in the source water.
The evaluation of the predictive models and the development of a new predicative model is beneficial to individual utilities or treatment plants looking to further evaluate ways of understanding DBP formation. By providing utilities tools to anticipate how their water quality will impact subsequent DBP formation, informed decisions can be made on how to operate the plant, and distribution system. By developing effective predictive models that utilize advances in technology, such as the EEMs, utilities will be better equipped to respond to quickly changing water quality in order to maintain DBP compliance as well as continue to produce high quality effluent water.
For more information, please contact the author at firstname.lastname@example.org.
Hear about new publications with our email newsletter
Horizons showcases significant water, wastewater, reuse, and stormwater projects and innovations that help our clients to achieve their goals, and can help you achieve yours. Articles are written by top engineers and process group leaders, demonstrating and explaining the beneficial application of a variety of technologies and tools.