Crack detection in plated T-joints through vibration techniques

The study presents investigations on different aspects of vibration techniques for crack detection in welded T-joints; it incorporates experimental testing, finite element analysis, system identification and model updating procedures in developing a methodology for detecting and quantifying fatigue cracks in plated T-joints. The experimental study measured modal parameter changes that occur in a structure as a fatigue crack grows in the critical region of plated T-joints. The finite element model was generated and refined by comparing the frequency response functions (FRFs) obtained from the finite element (FE) model and the intact experimental model. Finally, combining the experimental and finite element results and utilizing a model updating procedure a methodology was developed to find out the location, length and depth of the crack at different intervals of fatigue crack growth; the methodology used the correlation between the finite element and experimental frequency response functions. Local strain frequency response functions, which are very sensitive to cracking, were used as the objective functions in the model updating procedure. Perturbation based updating procedure, which incrementally modifies the design parameters (i.e. crack location, length and depth) at every iteration, was used to minimize the error in FRFs between the experimental and finite element models. Changes in strain mode shapes were used to determine the approximate location of the crack initially; subsequent estimation of crack profile was carried out utilizing model updating procedure and using line spring (crack) elements to model the crack. -- The crack location, length and depth measured from five experimental T-joint specimens have been correctly identified and quantified by this methodology. To the author's best knowledge, the application of this method to a plate-type structure and the accuracy that could be achieved have never been investigated in the published literature so far. At the conclusion of this study, a promising method for crack detection is provided. It is believed that this will help in the further development of new vibration-based NDE techniques for industrial applications