ERROR ASSESSMENT FOR STRAIN MAPPING BY DIGITAL IMAGE CORRELATION

Abstract

The accuracy and sensitivity of strain mapping by DIC have been experimentally evaluated for a sheet metal tensile specimen and the error level of local strains in terms of their standard deviations from the means is found to be around 100 μstrain for the imaging system and data reduction processes used. In general, as the error in DIC measurements are affected both by digitization and optics for a given DIC processing routine, it is suggested that a system error calibration procedure similar to the one described here should be carried out for each batch of specimens prior to testing. A series of tests on a stationary specimen with a typical contrast pattern should be first conducted to identify the minimum number of frames needed for averaging during the image acquisition to achieve the desired accuracy in displacements and strains for the given image formation system. Then translation tests along the three axes and in-plane rotation tests that simulate the motion of a specimen during testing should be carried out and errors in displacements and strains should be computed immediately. Depending on the nature of the error and its magnitude, optics realignment, optimization of surface contrast features, and increase of digital image pixel resolution may further be explored to reduce the system errors to an acceptable level for each specific application. The reliability and accuracy of the strain maps obtained in the subsequent testing can thus be ensured.