Vibration-based Non-destructive Detection Of Delaminations In FRP Curved Laminates

Fiber-reinforced Polymer(FRP), as advanced engineering materials, has been widely used in the civil engineering, aerospace industry, vehicle, marine engineering and chemical sectors due to their high specific strength and specific stiffness, high temperature corrosion resistance, fatigue resistance and the good design performance. FRP materials are commonly used to construct the engineering structure in the form of layered laminates, of which the binding force between layers is mainly determined by the resin matrix and therefore are prone to peel and form delaminations. Delamination is a main destructive form of FRP materials in engineering applications and may lead to a serious decline in performance and service life of the engineering structure, affecting the structural safety and even cause severe disasters. Therefore, it is imperative to detect delaminations in FRP materials at an early stage. The stiffness of structure will be reduced after the damage occurs, which will cause the changes in the structural dynamic parameters, such as frequency, mode and damping ratio.In current project, the delamination damage in FRP materials are tried to be assessed through the changes in frequencies. Frequency measurement has the advantage of simply testing process, data stability, and good reproducibility, thus changes in frequencies has been widely reported in the field of damage detection. However, the existing literatures reported to detect damage in FRP laminated through frequency shifts is mostly based on FRP laminated beams with through-the-width delamination damage. Very rare detection work was reported for FRP laminated plate with embedded delamination damage, and even less, if not none, has reported the delamination detection in the FRP curved plates which are actually the more real application form of FRP materials. Therefore, current project mainly focus on delamination detection for the FRP curved plates through changes in frequencies.There are three main part of work in the project. Firstly, a relatively accurate finite element model of FRP curved plates with delamination damage was constructed in ANSYS and was used to study the influence in frequencies from the existence of delaminations with varied sizes and locations. Then the inverse algorithms, the artificial neural network and surrogate-based optimization, were developed in order to predict delaminations in FRP curved plates through frequency shifts. Finally, the two algorithms were validated by numerical test cases to evaluate their performance in delamination assessment. The results show that both algorithms can predict the delaminations with enough accuracy, but the surrogate-assisted optimization can provide more accurate delamination location and size. Besides, compared to the direct optimization without using surrogate to substitute the finite element model used during the iteration in optimization, surrogate-assisted optimization can save much computation time and make the whole inverse detection much faster. Therefore, surrogate-based optimization algorithm can be implemented quickly and accurately identify the delamination damage in FRP curved laminated plates.