Artificial Neural Network- Genetic ‎Algorithm based Optimization of Baffle ‎Assisted Jet Array Impingement Cooling ‎with Cross-Flow

• Published in: Journal of applied fluid mechanics Vol. 15; no. 2; pp. 441 - 451
• Main Authors: Kurian, S, Johnson, J, Tide, P S, Biju, N
• Format: Journal Article
• Language: English
• Published: Isfahan Isfahan University of Technology 01.03.2022
• Subjects: Algorithms; Arrays; artificial neural network; Artificial neural networks; baffles; blow ratio; Clearances; Computational fluid dynamics; Computer simulation; Configurations; Cross flow; genetic algorithm; Genetic algorithms; Heat flux; Heat transfer; Jet impingement; K-omega turbulence model; Mathematical models; Neural networks; Optimization; Orifices; Parameters; Pressure drop; Pressure effects; Turbulence models; Two dimensional jets
• ISSN: 1735-3572; 1735-3645; 1735-3645
• DOI: 10.47176/jafm.15.02.33034

The objective of this research is to numerically investigate heat transfer and pressure drop characteristic ‎of a baffle assisted multi-jet impingement of air on a heated plate subjected to constant heat flux and ‎cross flow. Two baffle configurations were considered for the present study. An array of jets with 3 x 3 ‎configurations discharging from round orifices of diameter D=5 mm and with jet-to-heated plate distance ‎ranging from 2D to 3.5D were studied. SST k-ω turbulence model was used for numerical simulation to ‎examine the effect of blow ratio and baffle clearance on heat transfer and pressure drop characteristics. ‎Blow ratios of 0.25, 0.5, 0.75 and 1.0 and baffle clearances of 1 mm, 2 mm, and 3mm were considered ‎for CFD simulations. The split baffle configuration with baffle clearance of 3 mm is found to be more ‎advantageous when both heat transfer and pressure drop are considered. However, the segmented baffle ‎configuration with a baffle clearance of 1 mm gave better results for heat transfer alone. The present ‎study also deals with determination of optimal operating parameters with the help of Genetic Algorithm ‎and Artificial Neural Network. A pareto front was obtained for selecting the desired value of heat transfer ‎or pressure drop. It was found that Artificial Neural Network based predictions strongly agree with CFD ‎simulation results, and hence seems to be very useful in arriving at the optimum values of operating ‎parameters‎.