Optimization of convective heat transfer using T-shaped object and elastic splitter in backward-facing step flow with ANN modeling

Abstract

Methods to enhance the effectiveness of cooling systems in complex-shaped channels, which are relevant in many engineering applications, are essential. In this study, forced convective cooling of nanofluid flow over a backward-facing step is analyzed under the combined effects of a T-shaped inclined object near the step and an elastic splitter in the upper channel. Cooling performance is studied with finite element method by varying Reynolds number (Re), splitter vertical position ( y _s ), and T-shaped object size ( d ₁ and d ₂ ), vertical location ( y _o ) and orientation ( γ ). The elastic splitter outperforms the rigid configuration, enhancing cooling performance by 14.2 % compared to the rigid case at the highest Re. At y _o  = 0.4 H , average Nu rises 86.5 %, and γ  = ±45 orientations shift the Nu peak and results in cooling improvement by 15–16 %. When optimization is invoked, ( d ₁ / a _o , d ₂ / a _o , y _o / H , γ ) value is (1.043, 2.96, 0.076, −30.85) for maximum Nu – objective case while it is (2.96, 2.51, 0.396, 42.174) for average Nu – objective case. For the optimized object configuration, the local Nu is predicted across varying nanoparticle loadings using an ANN with 20 hidden neurons. This model accurately captures the first and second peaks, as well as the variations in average Nu.