Reliability design optimisation for practical applications based on modeling processes

摘要

There has been extensive development of computer-aided design/computer-aided engineering software in support of product design. These computer-aided tools usually focus on the problem of deterministic design. Recently, probabilistic design, namely reliability design, is receiving greater emphasis for handling the greater or lesser variations that exist in loadings and strengths. This study reports on a modelling process meant to perform reliability design optimisation (RDO) that integrates reliability evaluation with the probabilistic design system (PDS) of Advanced Analysis Guides (ANSYS). Geometric models of the designs are constructed and the design variables are formulated in probabilistic form based on the stress-strength interference reliability model for performing RDO. The optimal solutions for the design variables are obtained by goal-driven optimisation utilising the PDS of ANSYS. The sensitivities of the design variables to the stress, deformation and mass are investigated so that the priority for design modifications can be determined accordingly. A horizontal joint robot is used as an example to describe the model-based design processes. The modelling processes, which integrate engineering analysis software with reliability evaluations, are important features in the probabilistic design. Using this approach, optimal design solutions can be rapidly obtained when the constraints and the reliability targets are given. This proposed approach empowers the design engineers to be able to use finite element methods to fulfil RDO.