A parametric study of DIC measurement uncertainties on cracked metals

Abstract

Displacements and strains in the presence of a crack in welded steels have been studied using digital image correlation (DIC) under zero load and selected loading conditions. A parametric study has been carried out to assess the effects of some of the key parameters, including subset size, step size, and the size of the measurement window, on the measurement uncertainties for systematic and random errors. The assessments were carried out a region around the crack tip, where high strain gradients and inhomogeneous deformation present significant challenges for DIC measurements.

Full‐field characterisation of crack tip deformation and fatigue crack growth using digital image correlation—a review

“…One way of estimation may be by the method of virtual strain gauge (VSG), where a VSG was placed along y ‐direction and strains were evaluated (Figure ) at the most sensitive position (crack tip) at selected subset and step sizes. A subset size of 43 pixels and a step size of 12 pixels were found to produce the most “converged” results amongst the results obtained by combinations of subset sizes (23‐75 pixel) and step sizes (2‐19 pixel) for this application …”

“…The full‐field displacement errors of the area correlated under K = 20 MPa√m: random errors (left column) and systematic errors (right column) for the selected subset sizes [Colour figure can be viewed at wileyonlinelibrary.com]…”

“…The normal strains were obtained under a stress intensity factor of (B) 20 MPa√m and (C) 30 MPa√m. Light‐weighted lines show the results from combinations of subset sizes (23, 35, 43, 49, 61, 75 pixels) and step sizes (4, 6, 9, 12, 19 pixels), with the most “converged” sets (subset size‐step size = 35‐12, 43‐9, 43‐12, and 49‐9) highlighted [Colour figure can be viewed at wileyonlinelibrary.com]…”

“…The correlation algorithm is applied to both images allowing an evaluation of measurement uncertainty. Most work to date have focused on the estimation of stress intensity factor (SIF) using the displacement data extracted from DIC images, 10,[20][21][22][23][24][25][26] although not all reported the strategies for parameter selection or measurement uncertainties. A range of field of view (FOV) was used (from 2 × 2 mm 2 to 18 × 24mm 2 ), with a portion of crack normally included.…”

“…Unlike a recommendation of excluding plastic zone area in the estimation of K, 21 they suggested that, if a sufficiently large FOV is used, plastic zone area does not seem to affect the estimation of SIF using the displacement data from DIC. Including about 25% of the crack in the area of interest is recommended, whilst including more terms in the Williams' expansion does not appear to further improve the accuracy of the estimated K. 25 A parametric analysis 26 of measurement uncertainties in displacements and strains of a cracked specimen was carried out under selected loading conditions based on Xu et al 24 Full-field errors in displacements were also estimated under loads for selected subset sizes, as shown in Figure 3. The random errors appear to decrease with the increase of subset size, whilst the systematic errors appear to increase with the increase of subset size; hence, a compromise is needed to minimise the overall errors.…”