PSI - Issue 42

T. Koščo et al. / Procedia Structural Integrity 42 (2022) 1600 – 1606 Koščo, T., Chmelko, V. / Structural Integrity Procedia 00 (2019) 000 – 000

1602

3

problem to a slightly different position. This effect causes noise in displacement field. By computing a numerical gradient of displacement field, superimposed noise is magnified. Therefore, several filtering or smoothing approaches are used to minimize the effect of noise. Polynomial functions are common for this application, because of the possibility of analytic differentiation. Main problem of notched specimens and specimens with sharp strain peaks is the smoothing parameter sensitivity of the strain result. Overcoming this disadvantage while retaining usability in real operation is the main aim of the work. 2.3. Measurement equipment Authors used a Dantec dynamics Q400 DIC measurement system with two Baumer VCXU-50 USB 3.0 5 MPx monochromatic cameras in master-slave configuration. Q400 system evaluates the result in Istra4D 4.4.7 software, that two smoothing methods: local regression and smoothing spline for contour and displacement field. Contour does not significantly affect the strain data. Therefore, the smoothing of displacement field by using both methods will be mentioned. Loading of specimens were caried out by using MTS Bionix 370.02 and Inova ZUZ 200. 3. Open-hole tensile test Initial measurements were caried out on the open hole tensile loaded specimen for its simplicity and presence of analytical solution. Stress concentration has been computed analytically and numerically by using 2D plane stress (PLANE183) and 3D hexahedral elements (SOLID186), All FEM model were created based on methodology from Chmelko (2022). Specimen (see Fig. 1a) has been tensile loaded by INOVA ZUZ 200 testing machine. Specimen is made from structural steel with Young modulus of 206 GPa, loaded in linear elastic region only. Stress concentration factor has been computed by analytical formula proposed by Howland (1930) and calculated for this geometry as 3.154 and by 2D plane strass finite element model determined as 3.15. Two camera configurations with respect to the specimen has been tasted. First is configuration with partial view into the notch (Fig. 1b), drawback of this configuration is a need of relatively large hole diameter. Second configuration with symmetric view from both sides (Fig. 3). For displacement field filtering, both smoothing types were utilized and compared.

Fig. 1. (a) open-hole specimen (b) partial view of open hole tensile test

Local regression smoothing is based on polynomial regression of close neighbourhood, only one parameter, Kernel size, that defines the size of fitting region, can be set. Therefore, more localized effects can be better visualized, especially for small kernel sizes. On the other hand, smoothing spline creates polynomial function between the points and the smoothing is based on restriction of the curvature of the function. Zero smoothing parameter means minimum curvature.