Error Localization Methods With Static Modes and Thermographic Data

The goal of updating methods in dynamic is to improve the finite element models by means of experimental tests. In fact, we try to estimate a better value of the mass (M) and stiffness (K) matrices. Several ways are possible to complete the updating of M and K : - Global techniques which consist to correct the entire matrices without keeping a physical meaning of the M and K modifications. This approach transforms the finite element “knowledge” models into “representation” models which are more accurate in the measured frequency range but don’t represent the actual structure. This kind of method has been developed first by Baruch (1978) and Berman (1979) and improved by lots of researchers (Kabe, 1985 and Wei, 1990). - Local techniques try to compute the best values of the physical parameters defined on each element of the modelling. This method is a very stiff one since the optimization of the numerical modes is required over a very large number of parameters. To be able to use this approach we must select only the elements which must be modified to reduce the size of the optimization problem. To do so, we will apply the localization methods. In this paper we will present two ways to locate the modelling errors as well as a new approach based on thermographic, data. Each localization methods will be tested and we will show the influence of the expansion technique and the improvement due to the use of the static modes. The thermographic measurements give us a large number of data but partial because the information gathered by the camera concernes only the sum of the principal strains (Olivier et al, 1988 and Ryall et al, 1992). But the advantages of this method are the number of points which is important and the nature of the data which is a strain information (very sensitive to the local defaults of modelling).