Monday  |  Salon 13  |  01:50 PM–02:10 PM

#15863, Image Based Quantification of Complex Fracture

Traditional fracture mechanics experiments such as photoelasticity, caustics, and DIC based methods can extract properties like fracture toughness but are limited to single non-interacting cracks under carefully controlled loading conditions. This work presents a supplementary analysis technique aimed at extracting quantitative fracture information from in-situ and post-mortem imagery of complex crack networks that are typically of use only for qualitative comparisons. Computer vision techniques are used to extract cracks from images ranging from high speed photography during loading to simulation outputs. A statistical analysis of the geometry of the digitized cracks is then used to quantify the both the individual properties of each crack, represented by averages and distributions of lengths, orientations, and similar properties, as well as geometrical relationship between cracks, including spacing, angular distance between neighbors, and number of branches. This information has value in two separate areas. Data for each individual experiment can be used to gain insight through the crack behavior into the effect of loading conditions and structural influences through measurements such as crack tortuosity and average crack velocity, as measured by the change in crack length from frame to frame of an image time series. Additionally, the statistical profiles of each experiment or simulation can be used for quantitative comparisons, rather than typical qualitative comparisons. Multiple case studies are presented as demonstration, including an analysis of in-situ X-ray imagery of ballistically loaded boron carbide ceramic and simulation outputs of fracture in skull caps.

Logan Shannahan   DEVCOM Army Research Laboratory
Phillip Jannotti   DEVCOM Army Research Laboratory