Skip to Main Content
 

Global Search Box

 
 
 
 

ETD Abstract Container

Abstract Header

Multi-Objective Optimization of Conventional Surface Water Treatment Processes

Abstract Details

2016, Doctor of Philosophy, University of Akron, Civil Engineering.
Optimization of the coagulation/oxidation process in drinking water treatment is essential to ensure that high quality drinking water is provided at minimal cost to the consumer. While there are many objectives in drinking water treatment, current research in process optimization typically focuses on only one objective. This research focused on the development of an approach for optimizing multiple treatment chemical doses to achieve multiple water quality objectives simultaneously that can be efficiently and effectively implemented for real time process optimization. Neural network process models for the removal of fluorescence components and turbidity were built using operational data collected from Akron Water Supply (AWS) in Akron, Ohio. Model results were generally good, with correlation coefficients for the final models ranging from 0.51 to 0.97. Using these process models, chemical doses were optimized using the Borg genetic algorithm. Results showed that optimization in three dimensions yielded different optimal dosing solutions than optimization in two dimensions, suggesting that optimizing the coagulation process for DOC and cost only would yield suboptimal turbidity removal. Results also showed that many different chemical dosing combinations are capable of meeting specific target water quality values, but their costs varied, suggesting that finding one chemical dosing combination to meet a required turbidity target could be suboptimal in terms of DOC removal and cost. Because genetic algorithms can be time and computationally intensive, an alternative search algorithm was also evaluated for potential online process optimization. Its performance compared with the brute force evaluation of all possible chemical dosing solutions. At dose increments of 1mg/L for four decision variables, the brute force approach required the evaluation of 96,959 potential solutions and required 3.5 minutes to complete the evaluation. In contrast, the search algorithm was able to identify the same optimal solution by evaluating only 376 nodes in 2.2 seconds. Over the 268 days of operational data provided by AWS, the search algorithm was able to identify the same optimal dosing solution for all days, requiring an average of 96% fewer solution evaluations. These results suggest that the search algorithm is ideally suited for automated process control.
Christopher Miller, PhD (Advisor)
William Schneider, PhD (Committee Member)
Chelsea Monty, PhD (Committee Member)
Stephen Duirk, PhD (Committee Member)
Richard Einsporn, PhD (Committee Member)
195 p.

Recommended Citations

Citations

  • Kennedy, M. J. (2016). Multi-Objective Optimization of Conventional Surface Water Treatment Processes [Doctoral dissertation, University of Akron]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=akron1477332989340079

    APA Style (7th edition)

  • Kennedy, Marla. Multi-Objective Optimization of Conventional Surface Water Treatment Processes . 2016. University of Akron, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=akron1477332989340079.

    MLA Style (8th edition)

  • Kennedy, Marla. "Multi-Objective Optimization of Conventional Surface Water Treatment Processes ." Doctoral dissertation, University of Akron, 2016. http://rave.ohiolink.edu/etdc/view?acc_num=akron1477332989340079

    Chicago Manual of Style (17th edition)