Experimental identification of structural changes and cracks in beams using a single accelerometer

Abstract

Several advances in the structural health monitoring field and in crack identification techniques were achieved in recent years. Nonetheless, the use of those techniques for crack identification in beams by the industry is still modest. A few reasons can be pointed to explain this fact: some proposed methods are unfeasible from the economic or logistic point of view, or the cracks are detected only when they already present an advanced depth, or the structures intended to be monitored are subjected to random loads, causing methods using deterministic excitations to be unrepresentative of the actual situation. Considering this, the objective of this study is to propose a method that could make it possible to identify and monitor cracks in beams aiming at operational conditions, i.e., a method to identify small cracks remotely and in almost real time, in beams subjected to unknown random loading, minimizing the measurement equipment used to a single accelerometer and a remote computer. To achieve so, the proposed method combines an operational modal analysis (OMA) based experimental procedure, a numerical-computational model of the damaged beam using the finite element method and an optimization problem, solved by using the genetic algorithm (GA). The method was preliminary tested on a steel beam, into which structural changes simulating cracks with different depths were inserted. The method was also tested on numerically generated data with noise. The found results are encouraging, since they have shown that crack position and depth can be determined with appropriate accuracy for many engineering applications. The limitations on the proposed method were also discussed.