A parametric optimization framework for fin-and-tube heat exchangers based on response surface methodology and artificial intelligence

This study presents a numerical investigation and provides a novel framework for the rapid parametric optimization of attack angles (α) and aspect ratios (e) in plate fin-and-tube heat exchangers (PFTHEs) featuring elliptical tubes. The methodology incorporates three-dimensional numerical simulations, response surface methodology (RSM), backpropagation neural network (BPNN), and Entropy-VIKOR. The study focuses on a PFTHE with a single elliptical tube, varying the attack angles (α) from −30° to 0° in 5° increments and the aspect ratios (e) from 0.4 to 1.0 in a step of 0.1. Air velocities (u) range from 0.4 to 3.2 m/s with an interval of 0.2 m/s, corresponding Reynolds number from 200 to 2200. The results indicate that using elliptical tubes reduces Δp and enhances heat transfer capacity, with the most significant improvements at an e of 0.4. Exponential relationships between dimensionless parameters Nu or f and Re are established for different range of α at specific e values. The synergistic effect of e, α, and Re or u on PFTHE performance is confirmed through RSM and BPNN analysis. Their R² values are no less than 0.896 and reach a maximum of 0.998, indicating excellent predictive accuracy. Utilizing Entropy-VIKOR method, the framework identifies a suitable design solution with e = 0.6, α = -20°, and u = 3.2 m/s. This solution exhibits a 5 % increase in h and a 58 % reduction in Δp relative to the corresponding circular tube. The study provides specific expressions relating design variables to thermal–hydraulic behavior and indicates carefully selecting the α and e can improve the overall performance. In summary, these findings have significant implications for reducing optimization design period for PFTHEs with small tube diameters.