Artificial boundary conditions for finite element model update and damage detection

Abstract

In structural engineering, a finite element model is often used to estimate its dynamic or static behavior, and in conjunction with measured data, to monitor the health of a structure. This research is focusing on the use of experimental data for the improvement of a finite element model such that it accurately represents the structure, and in the localization of artificial damage in the model. The method of Artificial Boundary Conditions in model update and damage identification is used. To update the finite element model, identified stiffness changes are typically applied at each element of the model. This research introduces and demonstrates a new method using the mean stiffness correction at every element, to produce improved prediction of natural frequencies. The use of this method in conjunction with a composite sensitivity matrix created from the application of artificial boundary conditions, and utilizing the higher modes, is shown to more accurately update the finite element model than the usual method. Furthermore, this research demonstrates a new method for damage identification that uses subtraction of residual stiffness values, which can identify the damage regardless of location.