PSI - Issue 28

F. Conrad et al. / Procedia Structural Integrity 28 (2020) 2195–2205 F. Conrad, A. Blug et al. / Structural Integrity Procedia 00 (2019) 000–000

2196

2

carried out using conservative approaches. Especially in industrial applications, the reason is that the primary goal is to generate a reliable and robust statement about the residual lifetime of the component. Another fact that arises is that experimental knowledge about crack behaviour under multiaxial conditions is often not available. Instead, the crack growth behaviour is investigated almost exclusively on standard fracture mechanic specimens under globally uniaxial loading cycles. For less conservative predictions of the remaining lifetime of the components, approaches that include multiaxial effects might be taken into account. Within a publicly funded research project a systematic experimental database will be established in order to extend and secure concepts for the evaluation of crack behaviour particularly under consideration of multiaxiality. One focus is an improved experimental detection and theoretical description of crack closure, the crack propagation rate and propagation direction. Therefore, a novel DIC system accelerated by a graphics processing unit (GPU) is under development that is suitable to both monitor and control uniaxial and biaxial crack-growth experiments. As cracks form discontinuities in local displacement, full-field DIC is a quantitative experimental method to measure crack growth in terms of crack path, crack tip position or strain distribution. Moreover, full-field DIC is an ideal method to measure quantities like local displacement �⃗��⃗� or local 2D strain tensor ̿��⃗� as calculated by the finite element method (FEM). Therefore, both, FEM and DIC, are frequently combined to verify theoretical description within the experiment by Alshammeri et. al. (2020), Breitbarth et. al. (2018), Du et. al. (2017), Roux Langlois et. al. (2015) and many others. When one would like to use the strain information as an experiment variable however, conventional DIC systems are often too slow to meet the recommendation of ASTM E 606 for strain-controlled measurement, i.e. about 400 global strain measurements per fatigue cycle. For that reason, video extensometers are often combined with other strain measurement devices like mechanical extensometers or strain gauges. However, these devices share the same specimen surface, which means that sensors have to be sequentially removed during experiment. To simplify the experimental procedure, Blug et. al. used the GPU to accelerate the cross-correlation calculation of 2D DIC measurements in order to achieve global strain measurement rates above 1 kHz for strain-control. For the measurements presented in this paper, the authors expanded the system to measure both, high-framerate global strain for strain-control and full-field local strain for crack growth investigations. In this paper, the authors report first crack growth results obtained in uniaxial and biaxial crack growth experiments with this double-purpose DIC system. In particular, the system needs to combine high frame rates for strain-control and high spatial resolution to measure crack opening and strain distribution near crack tip. Nomenclature � , � Coordinates of the biaxial testing machine axes ��, �� Coordinates of the camera of the DIC system � � , � � ����e� �e��een ��i�in �� c����ina�e� �� ca�e�a an� �e��in� �achine u�⃗��,�� Displacement vector in camera coordinates u�⃗�a,�� Displacement vector in machine coordinates � Base length of strain between ROIs 1-4 in Fig. 2 � Base length of local strain in full-field mode ε � , � , � , � Strain in directions a, b, x, or y �� , �� Shear strain in camera or machine coordinates ̿��, �� , ̿� , � Local strain tensor in camera or machine coordinates Crack length � , �� Crack opening displacement in y and � , � direction ∆ε Threshold for divergence � , � Sobel operators in direction of camera coordinates