MODSIM 3DXCONFERENCE FEA variation simulation Probabilistic design
Modeling_and_Simulation_from_an_Industry_Perspective
My presentation:
The performance and safety of a mechanical design is crucial to most engineering companies. However, it is difficult to achieve a design that is highly optimized while maintaining an adequate safety margin. There are two main methods aimed at providing a reliable and safe design while maintaining performance; the commonly used Safety Factor (SF) method and the Probabilistic Risk Assessment (PRA) approach. Since its roots in the nuclear section, PRA has been aimed at estimating the likelihood of failures instead of covering uncertainties with over-engineering. The criticism has been that applying such a methodology relies so heavily on accurate models to reduce ambiguity that the result is considered to be untrustworthy. In the field of mechanical engineering, developments in simulation techniques question if this claim still holds true. In a time where we are trying to colonize space, the aim should be to reduce over-engineering and turn complexity into risk assessment. This presentation presents a case study in which FEA combined with Design of Experiments (DOE) results in a probabilistic approach for evaluating the output of a design subject to geometric variation of twelve parameters, with the goal of optimizing three conflicting functional requirements. Output consists of defining and estimating important parameters, creating meta-models based on FEA, conducting optimization and Monte-Carlo simulations to estimate the failure rate.
Presentation Replay
I'm available here to answer your questions.
@TN
Ph.D. Student, Technical University of Denmark
You can find all event presentations on the wiki page:
