Using stochastic experimental modal data for identifying stochastic finite element parameters of the AIRMOD benchmark structure

Abstract

Stochastic simulations are more and more used in industrial scale applications to determine the probabilistic description of desired responses. Consequently the need to improve the credibility of the randomised input parameters has come up. To identify parameter stiffness values of e.g. a bolted joint connection in a serial production process in terms of means and covariances an inverse procedure must be applied to a set of uncertain data. In the past five years several stochastic model updating (SMU) procedures in the field of structural dynamics have been published. Nevertheless an application to a set of uncertain experimental modal data that characterises variability from production tolerances is still missing. In an extensive modal test campaign a generic AIRcraft MODel (AIRMOD), which is a replica of the GARTEUR SM-AG19 benchmark, frequency response functions (FRF) from random excitation have been measured 130 times at the German Aerospace Center (DLR) in Göttingen. Between succeeding tests the bolted joint aluminium structure has been disand reassembled to introduce locally well-defined stiffness variability. The FRFs have been evaluated with an automated modal parameter estimation technique. Up to 30 different modes have been identified in each sample. The uncertain eigenfrequencies and mode shapes from the experimental modal analysis are utilised to identify model parameter means and covariance matrix using a method described in [6]. Therefore 18 random variables have been chosen in combination with 14 modes in the residual, the rest is taken to check the prediction quality of the adjusted stochastic finite element model.