This page describes fracture toughness tests we did on a CT specimen in our lab. These experiments are interesting because they involve discontinuous displacement fields, so this adds a layer of difficulty for DIC analysis. Anyway, the specimen was patterned using filtered acrylic spray paint, and was then set above a bath of acetone for a few seconds to improve adhesion of the pattern to the sample. A diagram and sample image(s) are shown below:

This specimen was precracked, and then a picture of the unloaded precracked specimen was taken and stored as the reference image. The current image was taken after a applying a load of 920 lbf. Using standard DIC analysis will result in poor results in the region near the crack because the displacement field is discontinuous there. To account for this, I've utilized the Eulerian to Lagragian conversion algorithm with subset truncation in Ncorr. To more easily determine the location of the crack, I utilized a polarizing filter in conjunction with a QM100 long distance microscope. A comparison of the regular image and polarized image are shown below:

Another added benefit of the polarizer is that it shows the formation of shear bands around the crack tip. These are visible because the pattern we used was simply a white speckle pattern over a polished surface, so the sample surface is visible between the speckles. The polarized and current image can then be superimposed and the crack can be removed from the ROI. This process is shown below:

The idea of converting the Eulerian displacements to the Lagrangian perspective is demonstrated below (although it is important to note that the Eulerian displacements have been multiplied by -1, so they are technically not true Eulerian displacements):

The overall crack tip displacement is not very large, so the comparisons are somewhat subtle; for softer and more ductile materials, the results would be more dramatic. Anyway, a comparison is shown (for E_yy Green-Lagrangian Strain) between this method and the standard DIC method:

By visual inspection, it's apparent the standard DIC method will give incorrect strain information near and across the crack. Furthermore, points located on the crack (and very near it) are omitted from the analysis, because strain in this region essentially doesn't make sense, since the material splits in this region. As an additional step of analysis, I've shown the region near the crack tip for the polarized image and the E_yy strains which demonstrates a nice correlation:

Even though the E_yy strains are with respect to the reference image and the polarization image is with respect to the current image, the overall strain is low enough such that the morphologies are similar, which instills some confidence in the results. 

• The files I used can be downloaded here: fracturetoughness.zip