Abstract

This article summarizes the basic formulation of two well-established finite element model (FEM) updating techniques for improved dynamic analysis, namely the response function method (RFM) and the inverse eigensensitivity method (IESM). Emphasis is placed on the similarities in their mathematical formulation, numerical treatment, and on the uniqueness of the resulting updated models. Three case studies that include welded L-plate specimens, a car exhaust system, and a highway bridge were examined in some detail and measured vibration data were used throughout the investigation. It was experimentally observed that significant dynamic behavior discrepancies existed between some of the nominally identical structures, a feature that makes the task of model updating even more difficult because no unequivocal reference data exist in this particular case. Although significant improvements were obtained in all cases where the updating of the FE model was possible, it was found that the success of the updated models depended very heavily on the parameters used, such as the selection and number of the frequency points for RFM, and the selection of modes and the balancing of the sensitivity matrix for IESM. Finally, the performance of the two methods was compared from general applicability, numerical stability, and computational effort standpoints.