Heuristic optimization of a continuous flow point-of-use UV-LED disinfection reactor using computational fluid dynamics | www.esteco.com

Heuristic optimization of a continuous flow point-of-use UV-LED disinfection reactor using computational fluid dynamics

Author(s): 
Richard M. Jenny, Micah N. Jasper, Otto D. Simmons, Max Shatalov, Joel J. Ducoste (North Carolina State University)

CHALLENGE - The increased flexibility in design options for alternative disinfection sources such as UV-LED reactors opens up to new methodologies to inactivate pathogenic microorganisms and reduce the exposure to carcinogenic disinfection by-products (DBPs) in drinking water. However this flexibility also increases the level of complexity when developing such reactors, particularly with regards to shape, size, spatial orientation of light, and germicidal emission wavelength. In this study, a numerical optimization routine was coupled with the Computational Fluid Dynamics analisys to optimize a point-of-use continuous flow UV-LED disinfecting reactor. Coupling CFD models with optimization software will offer an opportunity to heuristically determine the most efficient UV-LED reactor design under specified design constraints. 

SOLUTION - modeFRONTIER was coupled with COMSOL Multi-physics, and the two multi-objective functions are: 1) Maximize the log10 inactivation and minimize the system supply power; 2) Minimize the log10 inactivation and maximize the system supply power. The third function was a single-objective analysis, to minimize the supply power by maintaining at least 2.0 log10 inactivation of the microorganism. The multi-objective analysis provides a set of optimal designs, that is, the Pareto front. The single objective function presents one optimal solution. In modeFRONTIER, the MOGA-II algorithm was chosen as the evolutionary algorithm to determine the optimal three-dimensional design

BENEFITS - In this project, batch scripts and Matlab code were used to link modeFRONTIER with COMSOL. Coupling the optimization routine with CFD was very effective at decreasing engineers' design decision space and allowed to identify a potentially near-optimal UV-LED reactor solution. This configuration consists of seven UV-LEDS (all emitting 275 nm) with five of the seven lights located on the effluent side of the reactor.