Fluid Force Optimization on Multi-Way Valve Blades via CFD Simulation and Surrogate Modeling
Abstract
To optimize fluid forces on the multi-way valve blades within thermal management systems of new energy vehicles, this study employs a systematic design methodology integrating parametric modeling, surrogate modeling, and multi-objective optimization. Using the Tesla Model Y 8-way valve as a case study, a parametric model is established. A high-fidelity sample dataset is generated through computational fluid dynamics (CFD) simulations utilizing optimal Latin hypercube sampling (OLHS). A radial basis function-thin plate spline (RBF-TPS) surrogate model is subsequently developed to replace computationally expensive CFD analyses. Global sensitivity analysis is performed using an improved Sobol’s method. Structural optimization of the valve core blades is then conducted via the NSGA-II genetic algorithm. Results indicate that valve core structural parameters significantly influence the fluid force on individual blades, with inner diameter, outer diameter, and blade thickness exhibiting the greatest impact. Multi-objective optimization achieves a substantial reduction in the fluid force acting on each blade. Simulation verification confirms the optimization outcomes with minor discrepancies.