Optimization of airflow over an evaporator using blade arrangements: A combined CFD, machine learning, and experimental approach

This study presents a comprehensive investigation into the optimization of airflow over a household refrigerator's evaporator through the strategic use of guide blades. Unsuitable airflow in conventional designs often leads to inefficient heat exchange, non-uniform temperature distribution, and accelerated frost accumulation. To address these issues, guide blades were introduced near the outlet of return duct. A total of 52 blade configurations were analyzed using two-dimensional computational fluid dynamics (CFD). By using a neural network model, optimal blade geometry can be predicted based on thermal performance of trained CFD results. The optimal configuration, identified through simulation and confirmed experimentally, yielded a 0.62 °C reduction in average surface temperature and a 7 % decrease in total frost mass over time. The results demonstrate that even minor geometric adjustments can significantly enhance heat transfer performance and suppress frost formation, offering a practical, low-cost approach for improving the energy efficiency of domestic refrigeration systems.