PSI - Issue 37

Dániel Antók et al. / Procedia Structural Integrity 37 (2022) 796–803 Dániel Antók, Tamás Fekete et. al. : Evaluation Framework … / Structural Integrity Procedia 00 (2019) 000 – 000

801

6

Fig. 2. Sample image and the evaluation of graphical information

3.2. The Measurement System The test system is based on a Gleeble 3800 thermomechanical simulator. Tensile tests are based on standard test setups, using flat and square cross-section specimens. Force and crosshead displacement are measured by in-system sensors; standards-compliant elongation is measured by clip-type extensometers. The testing machine is equipped with an optical system that provides spatial geometric images of the test specimens during measurements, allowing beyond standard measurements to be made. For flat specimens, one camera is used, while two cameras are used for rectangular specimens to view both sides of the specimen. The active part of the test specimens has initially a regular mesh along the gauge length. During measurements, the optical system takes images of the specimen at times synchronised with the sampling, controlled by the machine's trigger signal. When processing the recorded raw data, the displacement vector field and shape of the specimen are generated from the images in a fixed coordinate system. A sample image is presented in Figure 2. The first stage of image processing is to create blurred images from the raw pictures (using a low-pass filter or Gaussian blur) to reduce the noise-like signals resulting from the uneven surface of the sample. After blurring, segmentation is made by thresholding. Subsequently, the centroids of white spots and their movements are determined. Centroid tracking is made by the Lucas-Kanade algorithm – Lucas and Kanade (1981) – . Given that a simple, single-constant thresholding, which initially yields good results, does not work well for large deformations, an adaptive thresholding is needed, which is being optimized. 3.3. The Digital Twin based Evaluation/Simulation System At the current stage of research, the Evaluation/Simulation System can be described in a couple of sentences as follows. Digital Twins of the tensile tests are implemented using Finite Element ( FE ) modelling technology. The DTs of the specimens are modelled in the MSC Marc-Mentat FE system, and corresponding simulations are performed using that system respectively. In the following, the modelling strategy and the models developed for the square cross-section 8 x 8 x 50 mm tensile specimens are reviewed. The modelling strategy is guided by the principles of targeting accuracy of evaluation and rational and economical use of resources. Therefore, the specimen models were designed as 3D models, one being a simple model, exploiting the symmetry properties of the specimen geometry, the other being a full geometric model of the specimen. Only 1/8 th of the sample is modelled in the simple model, considering all three planes of symmetry, so reducing both number of nodes and computation time. The full model is more resource intensive but can also be used in situations where the symmetries of the simple model cannot be exploited. Gauge sections are meshed using hexahedron elements, with increased mesh density around the contraction zone and tetrahedron elements around the heads. Von Mises plasticity theory is used with strain-hardening material response. The material is considered initially homogeneous and isotropic. Boundary conditions are provided by the corresponding kinematic constraints. The calculations use large strain – large displacement kinematics, in updated Lagrangian setting.