Logo image
Wastewater disinfection with peracetic acid: development of an artificial neural network-based metamodel
Thesis   Open access

Wastewater disinfection with peracetic acid: development of an artificial neural network-based metamodel

Ian Bakst
Master of Science (M.S.), Drexel University
Feb 2016
DOI:
https://doi.org/10.17918/etd-6703
pdf
Bakst_Ian_201617.16 MBDownloadView

Abstract

Anti-infective agents Disinfection and disinfectants Mechanical Engineering
Peracetic acid (PAA) is well-known as an antimicrobial agent for disinfection, and has particular applicability for treatment of wastewater for reuse. The performance of a per-acetic acid disinfection system is a function of site, specific water characteristics, intrinsic kinetics, and reactor hydraulics. Often on-site pilot testing is important to determine the site-specific kinetics. Prediction and optimization of disinfection performance is often difficult without modeling the full-scale disinfection system. Metamodels offer an elegant solution to this problem. In this thesis, we present the results of microorganism inactivation in a mobile serpentine pilot reactor using a three-dimensional computational fluid dynamics (CFD) model. A 3-D CFD model was used for the prediction of transport of momentum as well as the concentrations of the disinfectant PAA and its decay, and the microorganisms (dead and alive). The CFD model developed was calibrated with tracer data from the on-site pilot reactor and used to analyze the chemical and biological performance of such a system. Polynomial t and ANN-based metamodels were developed from the CFD model predictions of microorganism inactivation in lieu of experimental disinfection data. It was determined through cross-validation analysis that the 5-neuron ANN was the optimum metamodel for this application. The metamodel was applied to obtain algebraic relations of the disinfection rate as a function of the inlet conditions of the microbial concentration and PAA dose.

Metrics

29 File views/ downloads
12 Record Views

Details

Logo image