Sampling- and SORM-based RBDO of a Knuckle Component by Using Optimal Regression Models

Abstract

In this work, a sequential linear programming method for reliability-based design optimization (RBDO) is applied to a knuckle component for a heavy truck. Optimal regression models are used as surrogate models for the objective and the stress constraint, where the latter is calculated by nite element simulations (contact analysis). Instead of using linear or quadratic regression models, optimal rational regression functions are found by using a genetic algorithm. First-order Taylor expansions of the objective and the stress constraint are created using intermediate variables dened by the iso-probabilistic transformation. For the constraint, this is done at the most probable point. The rst-order formulation is then augmented using second-order reliability methods (SORM), both Breitung’s and Hohenbichler, Rackwitz’ formulas are implemented. In addition, a sampling-based correction is also adopted, which in this work is performed by using crude Monte Carlo simulations. The proposed RBDO approach is shown to be ecient and robust for the investigated problem. Parametric studies with various levels of probabilities of safety and standard deviations are performed for dierent types of distributions, and the results are presented as trade-o curves.