Local Crack Closure Measurements: Development of a Measurement System Using Computer Vision and a Far-Field Microscope

ABSTRACT: An accurate and relatively simple methodology for estimating crack closure loads has been developed. Using this method, measurements may be taken at a user-specified position behind the crack tip during the entire fatigue crack growth process. The methodology has three distinct components: (a) an imaging system having adequate magnification with minimal distortion. (b) a simple, Windows- based procedure for image acquisition and image analysis, and (c) techniques for applying a random, high contrast pattern on the specimen's surface. To meet the imaging requirements, a far-field micmscope objective capable of high magnifications was employed to image small regions on the order of 0.5 mm by 0.5 mm. The regions were near the crack tip. To meet the requirements of a user-friendly system, a Windows-based data-acquisition interface was developed to run the system on a common PC. Using the interface, images are acquired automatically during a loading/unloading cycle and stored digitally. Image analysis is performed on the saved images to rapidly obtain the crack opening displacement as a function of load; these data are used to estimate the crack closure load. Finally, two methodologies for applying a random, high-contrast pattern with average sizes of 4 to 20 nanometers were developed. The first method uses 11 nanometer filter paper and a low-pressure compressed air supply to apply small particles of photocopier toner powder to the surface of the specimen. The second method uses contact lithography to achieve a random pattern with smaller feature sizes, on the order of 2 to 8 nanometers. Baseline tests of the overall system have demonstrated that it is both easy to use and accurate. Specifically (a)the PC interface has demonstrated that images can be acquired automatically while the loading frame is cycling at 0.01 Hz, and (b) the crack tip opening displacement data have been shown to have errors on the order of 0.05 pixels for the toner powder patterns, corresponding to 27 nanometers for the magnification used.

Comparison of strains from strain gages and from DIDS. Schematic of measurement locations on a flat tensile specimen.
Full Paper