Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Aeronautics and Astronautics

Committee Chair

Weinong Chen

Committee Member 1

Charles Bouman

Committee Member 2

James Doyle

Committee Member 3

Michael Sangid

Committee Member 4

Tyler Tallman


Kolsky bar is commonly used to characterize materials at high strain rates from 102 to 104/s. Insitu visualization of specimen fracture process is vital to understand its dynamic mechanical behaviors. In this dissertation, compression Kolsky bar is combined with a high speed multichannel flash X-ray tomography methodology to capture snapshots of the dynamic in-situ 3D specimen volume information to visualize the dynamic damage and fracture process. This method is capable to produce sub-millimeter-resolution tomography reconstructions of dynamic fracture process from very limited number (4 in this research) of projections. This method enables a precise and repeatable control over the loading history of the specimen and the flash X-ray projection time, thus a correlation between the stress-strain/force-displacement response and the reconstructed volume can be clearly defined. The 4-channel flash X-ray setup is built for low geometric unsharpness (0.15 mm) to improve reconstruction resolution by placing intraoral size phosphor storage plate detectors close to the specimen without interference from unnecessary exposures. For each 2D projection, raw image scanned from detector was aligned according to the positions of the Kolsky bars in the projection via edge detection. Dark current was measured for each experiment and imposed to the reconstruction log input. The log inputs with and without the normalization to the un-deformed state were used. To reduce number of artifacts, Algebraic Reconstruction Technique was used to iteratively update the reconstruction. Three different static phantoms were imaged and reconstructed, where the result shows the image processing inputs are correct and the relative locations of X-ray beams, specimen and phosphor storage plate (PSP) detectors are within the tolerance. Three sets of preliminary Kolsky bar experiments were conducted, where the results indicate that a good tomography reconstruction requires a sufficient feature-signal-to-noise ratio in the 2D projection and a small number of cracks inside the specimen for reduced number of artifacts.

Applying this technique, dynamic spheroconical indentation experiments on two types of machinable ceramics (Macor and Mykroy/Mycalex 550) and uniaxial compression on 3D printed sandstone around 100/s were conducted. For indentation experiments, the post-peak-force reconstructions show that the Macor specimen fractures due to two major cracks - one parallel and the other one oblique to the loading direction, while the M/M specimen breaks due to only one major crack parallel to the loading. Such finding matches with the cross-sectional microstructure images, where the particle is randomized distributed in Macor, while the particle is highly directional in M/M. For 3D printed sandstone, it is found that the adhesive infiltrant coated layer apparently affects its mechanical behaviors: the specimen with higher percentage of coated region has higher strength. Under ~100/s loading, the 3D printed sandstone exhibits a brittle behavior, and the reconstructions show that the cylindrical specimen fractures due to cracks separating the side coated layer. Such cracks are initialed near one of specimen-bar interfaces. Under 0.001/s loading, however, the sandstone exhibits a ductile behavior, and the crack is initialed from the center of the specimen. Based on the reconstruction results, the limitations and several potential improvements of the flash X-ray tomography are discussed.