Multi-objective optimization for designing of high-speed train cabin ventilation system using particle swarm optimization and multi-fidelity Kriging

Maintaining a high level of thermal comfort and indoor air quality for occupants while minimizing the system energy consumption is crucial for the long-haul High-speed train cabins. The traditional way of handling the multi-objective problem relies on the “trial and error” design which involves lengthy manual design parameter adjustment and performance evaluation based on on-site measurements or analytical and empirical models. To shorten design optimization process, a multi-objective optimization platform has been developed using the nondominated sorting-based particle swarm optimization (NSPSO) algorithm for searching the trade-off optimal design of the ventilation system in a fully occupied high-speed train (HST) cabin. A computational model of the HST cabin occupied by four full rows of passengers was constructed using ANSYS Fluent. To ensure the accuracy of the CFD model, high resolution computational thermal manikins were adopted to simulate the thermal and pollutant dispersion under influence of the passengers. Different combinations of ventilation operation parameters were evaluated against its performance in terms of thermal comfort, air quality and energy consumption. Furthermore, to reduce the computational cost of constructing the training sample, a Multi-fidelity Kriging technique is also proposed a surrogate method in replacing the time-consuming CFD simulations while maintaining acceptable accuracy. The result demonstrates that the presented approach is capable to perform a multi-objective optimization for indoor ventilation system design and yield accurate Pareto-front result with up to 35.61% saving of computational time.