Optimization and ANN analysis of convective cooling of double triangular heat sources in expanded channel with elastic fin and magnetic field

Abstract

Efficient cooling of complex-shaped heated components is important for many engineering applications. This study numerically investigates forced convection cooling of two triangular heated objects in an expanded channel, incorporating the combined effects of an elastic fin, slot jet impingement, and an inclined magnetic field via the finite element method. The numerical analysis is conducted for a range of channel flow Reynolds numbers (Rec = 100-300), jet flow Reynolds numbers (Rej = 50-300), magnetic field strengths (Hartmann number; Ha = 0-30) and inclinations (gamma = 0 degrees-90 degrees), as well as elastic fin parameters, including length (Lf = 0-0.3H) and inclination (eta = 0 degrees-45 degrees). The elastic fin generates additional downstream vortices, along with those behind the step and near the heated objects, whose size varies with channel flow Re, jet flow Re, magnetic field, and fin parameters, potentially affecting convective heat transfer. At the highest channel flow Re, cooling performance for the first block rises from 22% (A, front wall) and 91% (B, rear wall) without fins to 107% and 86% with a fin, while for the second block it increases from 42.7% (C, front wall) and 67% (D, rear wall) to 54.5% and 27%. The use of slot jets near the step complicates the flow field but enhances the cooling performance. When a fin is employed, the cooling improvement reaches approximately 553.5% and 112.3% for the C and D walls of the second object, respectively, when comparing the lowest and highest Rej values. The trend of average Nu with changing Ha is similar to that observed with varying gamma. The optimal parameters (Ha, gamma, Lf, eta) = (0.2, 66.7, 0.89H, 18.5) maximize the cooling performance, resulting in improvements of 129.3% and 80.9% for the first and second triangular objects, respectively, compared to the reference case without a magnetic field and fin. The ANN model predicts near-optimal cooling, with average Nu deviations of 5.3% and 1.3% for the first and second objects, respectively.