Maximizing FCEV Stack Cooling Performance: Developing a Performance Prediction Model Based on Machine Learning for Evaporative Cooling Radiator 2024-01-2586

Recently, regulations on automobile emission have been significantly strengthened to address climate change. The automobile industry is responding to these regulations by developing electric vehicles that use batteries and fuel-cells. Automobile emissions are environmentally harmful, especially in the case of vehicles equipped with high-temperature and high-pressure diesel engines using compression-ignition, the proportion of nitrogen oxides (NO_x) emissions reaches as high as 85%. Additionally, air pollution caused by particulate matter (PM) is six to ten times higher compared to gasoline engines. Therefore, the electrification of commercial vehicles using diesel engines could potentially yield even greater environmental benefits. For commercial vehicles battery electric vehicles (BEVs) require a large number of batteries to secure a long driving range, which reduces their maximum payload capacity. However, fuel-cell electric vehicles (FCEVs) use hydrogen as a fuel to generate electricity, allowing them to achieve a long driving range with relatively fewer batteries. Therefore, FCEVs are more suitable for heavy-duty trucks. However, FCEVs require a significant increase in the number of cooling components to ensure the performance of key parts, including fuel-cell. As a result, the development of a new cooling system is essential in FCEVs to achieve high cooling performance within the constraints of the vehicle package. In this study, we addressed the insufficient fuel-cell cooling performance by harnessing the evaporative cooling effect of exhaust water, a byproduct of fuel-cell power generation, which is injected into the stack cooling radiator using the nozzles. Through test on the ‘Hyundai XCIENT Fuel-Cell’ and conducting 108 times of system evaluations, we confirmed that injecting water into the stack cooling radiator resulted in an additional cooling performance of 4~5°C due to evaporation. We also analyzed the key factors for improving cooling performance through data analysis. Furthermore, we implemented a predictive model, using machine learning techniques such as Python’s PyCaret, to optimize and maximize cooling performance based on ‘cooling perfomance improvement’ and ‘evaporation contrifbution’ when applying the evaporative cooling effect in acutal vehicles.