Surface modification for enhanced condensation heat transfer of saturated humid air: A numerical investigation and evaluation model

Enhancing condensation heat transfer is of great significance for recovering low-grade steam energy. Therefore, constructing a heat transfer performance prediction model to guide equipment optimization is necessary. In this process, the effective application of surface modification technology plays a crucial role. However, existing models exhibit deficiencies in the controllable adjustment of surface properties. This study focuses on analyzing the influence of surface characteristics (water contact angle and thermal conductivity) on the condensation heat transfer of saturated humid air and proposes a Computational Fluid Dynamics data-driven machine learning model. During the pilot test, the reliability for evaluating the heat transfer performance of coating structures was validated. The simulation results indicate that superhydrophobic surfaces enhance heat transfer through droplet coalescence and jumping, increasing the heat transfer coefficient to 150 % at 90 ℃. The key range for thermal conductivity adjustment lies between 0.1 and 1 W/(m·℃), within which the heat transfer coefficient can increase to 191 %. Among the seven tested models, Particle Swarm Optimization-Back Propagation Neural Network and Physics-informed Neural Network demonstrated higher generalization ability, with mean absolute percentage errors for heat transfer coefficient prediction in actual environments being 8.5 % and 9.5 %, respectively. Nevertheless, their prediction accuracy in the high heat transfer coefficient range still requires improvement. Overall, this study verifies the feasibility of transferring this model from the pilot test stage to practical application and provides reliable data for the future development of condensation heat transfer technology.