Finite-element model validation and its execution algorithm

The paper presents the finite element (FE) model validation. The definition of the term is given as a sequence of operations on a construction and its FE model to get the model reflecting the stress-strain behavior, and/or the dynamic properties of the construction as accurate as only possible. In this study, the validation was illustrated by a FE model of an antenna which has a complex divergent structure. Eigenmodes were selected as responses for FE model validation. Special matrices which are calculated using the FE method program are analyzed to define whether a sensor placement is optimal. Sensors are utilized to get the structural responses during structure tests. This analysis shows dominant eigenmodes that correspond with the deformation of the considerable amount of construction mass, location of the sensors which are placed in the nodes having the maximum values of kinetic energy fractions and the excitation points of all the required structure eigenmodes. According to the previous analysis, the location of the sensors (e.g. accelerometers) and excitation points are chosen. Mathematically, the selection of the sensor positions is accompanied with a reduction of the global FE model matrices to the nodes (degrees of freedom) where the sensors will potentially be placed. The next step is to confirm that the location of the sensors is optimal by means of special correlation matrices. They are employed to compare eigenvectors in the selected nodes of the base and reduced FE models. The certain values of the elements of the correlation matrices are the confirmation that the sensors are located in the optimal way. Further, structural tests are conducted by utilizing the results of the previous analysis. Afterwards, the correlation matrices are calculated again to compare the base finite element model of eigenvectors in the previously chosen nodes (points) with the reduced ones. If the correlation degree of the eigenvectors is high, the base FE model is considered as validated. If the correlation degree of the eigenvectors is low, the base FE model must be updated.