Publication Date


Date of Final Oral Examination (Defense)


Type of Culminating Activity


Degree Title

Master of Science in Materials Science and Engineering


Materials Science and Engineering

Major Advisor

Yang Lu, Ph.D.


Janet Callahan, Ph.D.


Peter Mullner, Ph.D.


The Anm model used for creating virtual concrete consisting of irregular shapes has been improved by integrating two existing algorithms: the extent overlap box (EOB) method for detecting contact between two irregular shapes and the uniform thickness shell algorithm. The EOB method has been compared with the previously used Newton-Raphson method and shown to be able to detect inter-particle contact with better accuracy and with less computational cost. Two parameters that define the balance between accuracy and performance of the EOB method have been identified and studied. The uniform thickness shell has been used to specify the minimum inter-particle distance in the 3D model of irregular shaped particles. A clear relation between shell thickness and packing density has been established through a series of simulations. To further improve the performance of the Anm model, the performance bottlenecks in the code have been identified and data parallelism has been introduced with minimal amount of code change. Another variation of the Anm model has been explored where the uniform thickness shell overlaps with other uniform thickness shells and other particles. The overlapping uniform thickness shell model is representative of microstructures such as the interfacial transition zone (ITZ) present in concrete. Studying such processes that depend on the Euclidian distance from the particle surface in three dimensions can be challenging. A new method for obtaining two dimensional slices of this model has been developed and resultant images showing the spatial distribution of the different phases are analyzed. It has been observed that the apparent thickness of the shell in the 2D slices can be larger than the prescribed normal distance from the particle surface and this is dependent on the angle between the slice and the particle surface normal. The 2D analysis has been shown to be useful to explain surface features observed in actual slices of concrete samples. The “wall effect” observed in the Anm model (and real concrete) is characterized with a radial distribution function utilizing the 2D slicing feature and the methods for performing this characterization is developed.