<scp>CFD</scp> and <scp>ML</scp> ‐based optimization of a modified trapezoidal wavy channel heat exchanger

Abstract Engine exhaust heat recovery systems (EHRS) are increasingly used to improve thermal efficiency, reduce fuel consumption, and lower emissions during cold‐start operation. Their performance depends heavily on compact heat exchangers, where fin geometry governs turbulence generation, heat transfer enhancement, and hydraulic losses. Yet conventional fin designs often struggle to achieve a favorable balance between heat transfer and pressure drop. This study addresses this limitation by developing and optimizing a modified trapezoidal wavy corrugation for EHRS heat exchangers. A combined CFD–Gaussian Process Regression (GPR) surrogate‐based framework is employed to identify geometries that maximize convective heat transfer while minimizing the friction factor. CFD simulations were performed for 150 geometries generated through Latin Hypercube Sampling, covering five key fin parameters: profile height, profile width, rise angle, fall angle, and fillet radius. The resulting dataset trained GPR surrogate models to predict Nusselt number and friction factor across the design space, enabling efficient extraction of Pareto‐optimal designs. A full‐scale heat exchanger using the optimized corrugated profile increased heat transfer by 54.42% over a flat‐channel configuration, accompanied by a 978.23% rise in pressure drop. Despite this trade‐off, the enhanced thermal performance highlights the potential of the proposed design for compact, high‐effectiveness waste heat recovery systems.