Calibration and cross-validation of a car component using frequency response functions and a damping equalization technique
Paper in proceedings, 2014
The calibration of an FE model of a car subframe is reported. In this calibration the minimum deviation between finite element model data and experimental data is searched for. The outcome of the calibrated model and cross-validation results are compared with results of testing being made on an ensemble of seemingly identical subframes. The subframe model has 250,440 degrees-of-freedom and the calibration is made for 16 uncertain model parameters. The efficiency of the calibration procedure under these conditions is reported. With model calibration being a cornerstone of the finite element validation procedure, the calibration problem is normally characterized as being a large scale optimization problem with many model parameters to solve for and with deviation metric that are nonlinear in these parameters. The calibrated parameters need to be found by iterative procedures, starting from initial estimates of parameter values. Sometimes these procedures get trapped in local deviation function minima and do not converge to the globally optimal calibration solution that is searched for. Here, a calibration formulation which gives a smooth deviation metric with a large radius of convergence to the global minimum is used. A method that utilizes damping equalization is used to avoid the mode correlation and mode pairing problem that need to be solved in various other model updating procedures. The method is combined with model reduction for increased speed and employs the Levenberg-Marquardt minimizer with randomized starts to achieve the calibration solution. The MATLAB-based open-domain software tool FEMcali has been used for calibration, validation and cross-validation.