Structural dynamic modification using experimentally improved finite element model

Abstract

It is often desirable to predict the effect that a physical modification will have on a structure's dynamic characteristics, without the time and expense required to build a prototype. Experimental modal models may be constructed to be more accurate than finite element models because damping and boundary conditions are difficult to model analytically. However, they do not usually contain rotational degrees of freedom which are required for beam and shell type modifications. A method for improving a finite element model's mass and stiffness matrices, using experimental results, was applied to a simple H-Frame structure. A unique mapping method was applied to expand the experimental mode shapes to the dimension of the finite element model. These mode shapes and the experimental natural frequencies were then used with the orthogonality conditions to produce an improved finite element model. The improved model reproduces the experimental natural frequencies used in the process. The original and improved finite element models were used to predict the effect of a beam addition to the original structure. The improved model produced the better prediction when compared to experiment.