Field Measurements in Mechanical Testing Using Close-Range Photogrammetry and Digital Image Analysis

Abstract

This thesis contains a study on using numerical post-processing of recorded digital images from mechanical experiments to obtain continuous measurements of both displacement and strain fields of a particular specimen surface. Both two-dimensional and three-dimensional field data are measured. Two different measurement techniques have been applied in this study, i.e. structured light (projected fringes) and Digital Image Correlation (DIC). By using the technique of structured light combined with a calibrated camera model, continuous three-dimensional full-field out-of-plane deformations of plates subjected to impact have been measured. However, this technique does not allow for measurements of strain fields in the test specimen. By applying DIC on plane specimens, two-dimensional inplane displacement and strain fields may be measured. A DIC code for analysis of mechanically recorded image series has been developed in this Ph.D.-work, and the DIC functionality has been further improved to be suitable for plane specimens experiencing crack growth. Effort has been made to increase the accuracy and robustness of the DIC algorithm in measuring such cracked specimens, and the discontinuous displacement fields of propagating cracks have been successfully captured. The developed DIC functionalities have mainly been tested for two-dimensional problems. However, the presented functionalities are extendable to three-dimensional problems. In this study, three-dimensional DIC analyses have been carried out to investigate the necking of mechanical specimens during loading.

A main part of the Ph.D.-work has concerned the development and testing of the imageanalysis code. Particularly, increasing the computational speed of the code has been a challenging and time-consuming task.

The thesis starts with a synopsis, giving an introduction and the organization of the research work, together with a brief summary of the main findings and conclusions. The synopsis also includes a brief introduction to the image-analysis techniques of interest and an overview of the development of a 2D-DIC code for analysis of images recorded in mechanical experiments. Following the synopsis, four independent papers are presented.

Paper I presents a study where the technique of structured light has been applied to measure the continuous out-of-plane deformations of aluminium plates subjected to low velocity impact. The principles of the measurement technique are presented in detail as well as the results from an experimental test series. The experimental measurements have further been used to validate numerical and analytical models.

Paper II presents an experimental and numerical investigation of fracture in a cast aluminium alloy using a modified Arcan test setup. Two-dimensional DIC has been applied to measure displacement and strain fields in the specimens and further used to validate finite element simulations where the fracture parameters were assumed to follow a modified weakest-link Weibull distribution. Comparison of strain fields and crack paths between the experimental and numerical studies was emphasized.

Paper III presents a study where two-dimensional DIC, incorporating a mesh of elements suitable for node-splitting, has been applied to investigate fracture in small-scale SENT tests of a pipeline steel. The node-splitting approach proved valuable to measure fracturemechanics parameters such as crack path, crack length, CTOD and CMOD as well as capturing the discontinuous displacement and strain fields of the cracked SENT specimen. The results from the DIC analyses were compared with traditional clip-gauge measurements. In addition, an assessment of the measurement uncertainties in two-dimensional versus threedimensional DIC in the presence of specimen necking is included in the paper.

Paper IV contains an evaluation of mesh adaption techniques, as well as a crack-path optimization procedure in DIC, by analyzing synthetic image series of a specimen with a single propagating crack. A node-splitting approach is applied, together with an approach based on overlapping meshes. Using this latter approach, a crack-path mask is applied, which lets a random crack path to be defined at pixel level. The synthetic images are generated from finite element simulations, and the known displacement fields are used as a validation tool for the DIC analyses.