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

1604

5

and temporal (constant) continuity of the result. Smoothing spline exhibited smooth data with smoothness factor between -0.5 and 0, with change of the result caused by smoothing in this range up to 1*10 -5 (2 MPa). Stress concentration factor varied between 3.13 to 3.15 respectively. Local regression with kernel sizes between 21x21 up to 31x31 has been used to ensure continuity of result. Absolute error caused by the filter has been between 1*10 -5 and 2*10 -5 (2 and 4 MPa). Stress concentration varied between 3.167 up to 3.19. In this case, smoothing spline resulted better results. The fact, that the smoothed displacement field contains only one peak of the gradient covered by large amount of grid points, predisposes the smoothing spline to better result. 4. Side notched specimen Authors consider the measurement on the cylindrical surface of the notch as the generally most suitable. Therefore, another specimen geometry has been utilized. Side notched specimen (see Fig. 4) allows to use different camera positions with respect to the specimen. Multiple specimens with different notch sizes have been used to perform the sensitivity analysis of the notch size on the result accuracy. As mentioned before, large amount of grid points leads to better spatial resolution and larger subset size leads to better displacement results. More pixels in the subset means more information about the position of the grid point. In other words, more grid points to be filtered with less noise. Therefore, number of pixels in the notch or number of pixels occupied by the notch corresponds to the amount of information about the displacement field and strain field in the notch. Considering a general notch with general strain field, the radius in the root of notch, indicates the sharpness of the strain peak and directly defines the curvature of the surface. Authors therefore consider the number of pixels pre notch radius as a good measure of strain peak proportions and amount of information about the displacement field. Purpose of the sensitivity analysis lies in definition of a minimal number of pixels per notch radius for satisfactory strain results. Two camera configurations have been tested (see Fig. 5). Horizontal configuration is similar to the symmetric configuration by open hole tensile test and the vertical could be used if only one-sided view is possible. Specimens has been loaded by MTS Bionix 370.02. Results has been exported as a mean of values on 2,5 mm line in the notch root. Stress concentration factor has been calculated by using 2D plane stress finite element model (using PLANE183 element). Small notch sizes led to several effects. Smoother parameter dependency, subset size and grid size dependency increase its influence by decreasing notch size. All these effects correlate with each other. Surprisingly, variance of the result causes largest negative influence. Lack of grid points by small notches led to the larger variance by multiple measurements in same load level. This causes deteriorating of shape of the hysteresis loop (see Fig. ).

Fig. 4. Side notched specimen

Fig. 5. (a) horizontal configuration (b) vertical configuration

Grid size and subset size increase its influence by smaller notches in comparable extent. Change of evaluation parameters affects the result by small notches. On the other hand, result by larger notch size remain constant by change of evaluation parameters. Displacement field contains gradient only in one direction and the whole field consists only of one strain peak without unnecessary