Structural lightweight design of bogie bolster based on surrogate models

The bogie bolster is one of the key components of railway freight cars, and its lightweight design plays an important role in improving the transportation capacity of freight vehicles. However, the bolster features complex geometry, leading to high costs and long design cycles in structural development. In this paper, finite element analysis (FEA) is performed to evaluate the static strength and stiffness of the bolster. An optimization model is established taking the geometric dimensions of process holes as design variables, the minimum mass of the bolster as objective function, and allowable stress and deformation as constraint conditions. The Kriging surrogate model is constructed for mass, stress and deformation responses. Then expected improvement (EI) criterion is adopted for sample infilling to improve model prediction performance, and genetic algorithm (GA) is employed to solve the optimization problem. Global Sobol sensitivity analysis is further implemented to quantify the contribution of each design variable to structural responses. After obtaining the lightweight scheme, finite-element-based fatigue simulation and modal characteristic analysis are carried out for post-verification to assess the service reliability of the optimized bolster. Numerical results show that the optimized structure satisfies the static-strength and stiffness constraints in the optimization formulation. Post-verification demonstrates that the optimized structure also meets fatigue and modal-performance specifications. The total mass of the bogie bolster is reduced by 16.20 kg, achieving the lightweight target for heavy-duty freight bogie bolsters.