| As part of an assessment of the feasibility and safety of deep geological disposal of nuclear fuel waste, a laboratory testing program was carried out at the University of Manitoba to investigate the course of fracture around a circular opening in granite specimens. Although some biaxial tests were conducted as well, the modeling in this thesis uses data from uniaxial compression tests. The block faces and occasionally the insides of the cavity were instrumented with strain gauges. These were used to detect fracture initiation at the three usual action areas: the primary fracture regions which occur at the tensile stress concentrations at the hole perimeter, the remote or secondary fracture regions which occur in quadrant positions a short distance from the perimeter of the hole, and the spalling fracture regions which occur in the compression zones near the hole perimeter. Fracture propagation along both the primary and the remote fracture directions were also tracked with the aid of strain gauges.; To interpret the results, two computer programs were constructed. One, InSight{dollar}rmsp{lcub}2D{rcub}{dollar}, is based on the finite element technique, the other, EJDBEM, on the boundary element method. What distinguishes these programs from others in the education or commercial spheres is that both represent a new approach to fracture through the implementation of a finite-width crack model. In these programs a crack is represented by a slot that, in contrast to classical fracture mechanics, has finite width. Fracturing events are detected through the use of a stress-based criterion that relies on contributions from both in-plane principal stresses. This allows fracturing to be controlled not only by the lateral tensile principal stress, as is commonly done in fracture mechanics, but also by the axial compressive stress. A stress averaging technique is used in conjunction with this stress-based fracture criterion to address the size effect commonly observed in laboratory testing.; The results from the physical model testing showed a pronounced size effect for the initiation and propagation histories of all three fracture types. In addition to the size effect it was found that the finite dimensions of the laboratory specimens significantly controlled fracturing events. The most intriguing result was that, through structural size changes, fracture propagation appeared to go from a very stable fracture propagation mode at small radius cavities to an (almost) unstable mode for the larger cavity sizes. Most of the existing analytical models ignore boundary effects (infinite medium assumed) and fail to explain the size effect on primary fracture propagation. Exceptions to this are Bazant's and Carpinteri's treatment of size-effect. The two numerical models of this thesis were successful in modeling the test results. (Abstract shortened by UMI.)... |
