Damage evaluation in fibre reinforced composites (WP4)

As an anisotropic material, fibre reinforced composites (FRC) have a wide range of design and application possibilities due to the variation of fibre direction, layer structure as well as fibre and matrix selection. This anisotropy and heterogeneity of FRCs results in different types and mechanisms of damage (fibre fracture, inter-fibre fracture, delamination etc.). Furthermore, the high directional stiffness generates a brittle behaviour with abrupt mechanical failure due to critical macroscopic crack formations. In the literature, some publications already exist on impact - detection in small plates of a quasi-isotropic sandwich plate. Nevertheless, the fibre and matrix material, the fibre-volume content, the fibre-direction and stacking order of the lamina have an additional influence on the measured vibrational signal. Therefore an isolated investigation of these parameters is necessary for a reliable damage detection in composite structures with VAM.
For first experiments, specimen of different layer structures were tested to evaluate the damage-dependent behaviour of the MI. A cross-ply lay-up [0,903]s was used to investigate whether severe matrix crack initiation and the propagation under tensile stress can be assigned to changes in the VAM signal. In addition, a [0;454]s lay-up was compared because of the occurrence of many small interface cracks and delaminations. GFRP1 was used. Two different test procedures were defined to evaluate the influence of the composite lay-up and the damage type on the measurement results. First, a sequentially increased load led to a damage formation in the cross-ply samples. After the loading, a constant measurement procedure was conducted. These results were compared to fatigue tests which were combined with VAM measurements of the same design as the sequential test before. 
From the sequential test in the cross-ply laminate, we can deduct a constant MI value before the damage is occurring and afterwards a linear relationship of the MI and the number of cracks in the sample as shown in below Figure. Additionally, these results are much less influenced by the excited frequency of the piezoceramics compared to the experiments on aluminium samples. For the fatigue test on both lay-ups, we observed a logarithmic increase of the MI. This is in strong contrast to the aluminium samples, where the MI-increase is starting at around 80% of the lifetime. However, the MI curve reveals a strong correlation with the damage index and, therefore, the change of modulus of the sample. When fitting a logarithm to the MI curve, a laminate and amplitude specific set of parameters could be obtained to describe the curve. With these parameters a differentiation between the laminates and the amplitudes of the fatigue load is possible. An abstract was submitted. Based on these results, an analytic approach to estimate the damage state of composite parts deducted from the MI-curve is part of the ongoing research.
Therefore, coupon specimen will be tested to create a general heuristic before analysing complex composite parts.
Varying the lay-up in cross-ply laminates also results in different fundamental characteristics of VAM. We obtained this result by comparing glass-fibre cross-ply laminates with [0,903]s and [0,90]s lay-up. The bigger the volume of critical connected 90° layers in composites, the more high-order sidebands are pronounced. Due to the more complex stiffness variation within the composite and the more complex propagation velocity in GFRP [0/90]s there is no linear gradation of sidebands as in [0/906]s. Thicker 90° layers result in larger critical cracks, while thinner ones increase the number of smaller cracks, which typically describes the thin-ply effect. Accordingly, the VAM-signal, expressed by the high-order sidebands, is very sensitive to crack size and crack density. To support this assumption, the layer structure has to be systematically varied and it has to be clarified whether the intra-laminar or inter-laminar interaction of the propagation velocity dominates the VAM signal. Furthermore, fatigue tests with CFRP2 (higher propagation velocity) will also be carried out with acoustic emission measurements to detect crack hits and to correlate the number of cracks with the increasing high-order sidebands.