Understanding the post-peak deformation behavior of rock is important for underground rock engineering. Laboratory property testing is commonly employed to investigate the post-peak deformation behavior. However, the test conditions of laboratory testing, especially the Loading System Stiffness (LSS) of stiff test machines, are usually varied and the influence of this variation on the test results has not been fully elucidated. In addition, studying the influence of test conditions on the post-peak deformation behavior of rock is crucial for interpreting test results and subsequently applying the results to rock engineering design.
The goal of this dissertation is to identify how the post-peak deformation behavior of a rock specimen is affected by three major aspects of test conditions––the specimen geometry, the contact conditions, and the LSS. To achieve this goal, an FEM/Explicit tool was employed to carry out numerical experiments, in which the same material property was assumed for the rock specimens and each wanted test condition was isolated for analysis.
The well-observed slenderness effect and the recently-observed cross-sectional shape effect on the Uniaxial Compressive Strength (UCS) of rocks were studied. The modeling results suggest that the numerical tool and models are suitable for investigating the problem and the hoop tension theory could be flawed. Next, the cross-sectional shape effect on the post-peak deformation behavior was investigated. The modeling results reveal that although the influence of the cross-sectional shape on the UCS of rocks is small, the cross-sectional shape affects the post-peak deformation behavior considerably.
The actual contact condition and the end effect in true triaxial compression tests were simulated while the legitimate intermediate principal stress (σ₂) effect was excluded from the rock material contacts are frictional and the specimen in the σ₂ loading direction is squat. Thus, existing 3D empirical failure criteria based on previous true triaxial compression test results may overestimate the rock strength.
The influence of LSS on the post-peak stress–strain relations of stable rock failure was examined. Key loading components of stiff test machines were considered in the numerical model. The modeling results clarify that LSS affects the post-peak stress–strain curves of rocks even when the failure process is stable. Unless LSS is either perfectly rigid or equivalent to the critical LSS (λ), the post-peak stress–strain curves obtained under various LSS (with LSS > λ ) are varied and all steeper than the one under an ideal loading condition.
This dissertation demonstrates the cross-sectional shape effect in the post-peak deformation stage, the long overlooked σ₂ effect caused by the end effect, and the variation of post-peak stress– strain curves due to the LSS. This dissertation also makes a contribution to examining the hoop tension theory and recognizing the correct choice of cross-sectional shape for test specimens, offers insights into improving 3D empirical failure criteria and true triaxial test settings, and suggests new requirements for developing stiff test machines in the future.