Assessment of variation of natural frequencies of FE model based on the application of alpha-cut strategy and genetic algorithms

Engineering tools which are currently available within the realm of virtual prototyping allow for the assessment of variation of mechanical structure properties. This goal can be achieved through the application of uncertainty analysis. Both probabilistic and possibilistic approaches are employed for studying uncertainty propagation in FE models. Numerical simulations enable shortening time-to-market for designed mechanical systems. However, great number of iterations required in non-deterministic analyses may dramatically reduce the scope of possible applications, especially in respect of time-consuming dynamic analyses. In the dissertation, the effective CFGA method is proposed to assess the variations of natural frequencies of FE model. Briefly, the advantages of the method include reduced number of eigenproblems to solve in dynamic analyses and relatively high quality of results. Hence the significant reduction of computational time. It is of a great importance, especially in the case when a large number of degrees of freedom are considered in an FE model and, as a consequence, a solution of dynamic problem is very time-consuming. The paper includes selected application of described method.