Detection of notches and cracks based on the monitoring of local strain and the solution of inverse problems

Abstract Engineering structures are in general exposed to cyclic or stochastic mechanical loading. Exhibiting incipient cracks, particularly light‐weight shell and plate structures, suffer from fatigue crack growth, limiting the life time of the structure and supplying the risk of a fatal failure. Due to the uncertainty of loading boundary conditions and the geometrical complexity of many engineering structures, numerical predictions of fatigue crack growth rates and residual strength are not reliable. Most experimental monitoring techniques, nowadays, are based on the principle of wave scattering at the free surfaces of cracks. Many of them are working well, supplying information about the position of cracks. One disadvantage is that those methods do not provide any information on the loading of the crack tip. In this work, the development of a concept for the detection of straight and simply kinked notches or cracks in finite plate structures under mixed mode loading conditions is presented. In this approach, the distributed dislocation technique is applied to model the direct problem, and a genetic algorithm is used to solve the inverse problem. Solving the inverse problem, eg, with a genetic algorithm, this allows the identification of external loading, crack or notch position parameters, such as length, location or angles, and the calculation of stress intensity factors, as long as the shapes and the number of the cracks are a priori known. Experiments are performed using plates with notches under tensile loading.