Multi-objective co-optimization of operation and geometry for liquid-cooled heat sinks using an integrated CFD–RSM–NSGA-II framework

This study proposes a multi-objective optimization framework integrating computational fluid dynamics, response surface methodology, and non-dominated sorting genetic algorithm-II to enhance the overall performance of straight-channel liquid-cooled heat sinks. Both operational parameters (inlet temperature and Reynolds number) and geometric dimensions are considered to minimize thermal resistance, pressure drop, temperature non-uniformity, and deviation from the target temperature, while maximizing the Nusselt number. Response surface analysis reveals significant trade-offs among these conflicting objectives. A set of Pareto-optimal solutions provides a range of optimal design alternatives. The best compromise solution (point E), selected using the technique for order preference by similarity to ideal solution (TOPSIS), reduces the pressure drop by 85.2% compared to the high-performance extreme (point A) and decreases the thermal resistance by 66.8% relative to the low-flow-resistance extreme (point B), significantly improving overall system performance. The proposed framework effectively balances thermal performance and flow resistance, offering a reliable tool for multi-objective thermal design and providing valuable insights into thermal management in high-performance electronic devices.