Validation of Aeronautical Analytical Models at DLR

Kurzfassung

Today's development and construction processes for aeronautical structures impose increasing demands on the reliability and accuracy of mathematical models. These mathematical models have to satisfy high standards with regard to e.g. flutter calculations, fan-blade-off and windmilling certification calculation within the aeroelastic stability certification process. To validate these analytical models (FE-models), experimental data, e.g. from normal mode tuning or experimental modal analysis, may be utilized. The classical way to reduce the discrepancies between analytical and test data is to modify by trial and error the assumptions made for the mechanical idealizations and parameters of the FE-model until the correlation of analytical predictions and results of the test satisfy practical requirements. If only single parameters are considered a so-called manual updating, which is still state of the art in industry, of the analytical model based on engineering skills may be successful. However, if the number of parameters becomes too large it is often impossible to assess easily the effects of parameter changes on the FE-analysis results. As a result considerable effort has been made in the development of computational procedures for updating the parameters of the analytical models. These Computational Model Updating (CMU) procedures can be used instead of the manual correction approach, which allow for a simultaneous update of several correction parameters such that deviations of identified and predicted natural frequencies and mode shapes are minimized. This lecture shows some extracts of the experiences the author made in the last two years within the topic of model validation. After a short presentation of the mathematical background of the used CMU procedure, some model validation results of a benchmark test structure which exhibits a typical aircraft structural dynamic behaviour are shown. For this example the updated analytical model is validated using experimental data from modified structure. After this aeronautical research example an efficient method for the validation of very large industrial aeronautical FE-models is described. The method is applied to a large civil four-engine aircraft model.