A combined experimental and theoretical investigation of the damage process in hard rock with application to rockburst

Abstract

Modern day mining operations are being forced deeper due to the depletion of near surface deposits. As such, more rock stability problems are being encountered due to the higher stress and temperature conditions found at depth. Some of these problems can be dealt with using current knowledge and conventional rock mechanics, however, the rockburst phenomenon proves hard to predict and mitigate. As a result it is necessary to develop more focussed strategies to address rockburst and provide a means to test and model the mechanism. It was clear in the literature that the controlling factors in rockburst is the elastic strain energy stored in the rock prior to failure and the accumulated damage in the ‘skin’ layer of an excavation. Therefore, this study focussed on the formation of a strategy to better understand the damage processes in hard, burst prone rocks and hence provide a better testing and modelling platform for further rockburst research. Initial review of literature test data revealed a need for the implementation of a thorough testing methodology to measure the full stress-strain and damage evolution of conventional, compressive rock tests. As such, this dissertation proposes a coupled full circumferential control (FCC) and acoustic emission methodology to provide linked stress-strain-damage results. It is shown that the circumferential strain control method of testing provides a more comprehensive data set for the calibration of constitutive models. It also postulates that crack damage thresholds are very reliant on the load control method used during testing. Therefore, rocks that exhibit snap back behaviour should be controlled using circumferential strain rate. Using the proposed testing methodology this research was able to obtain a complete data set for granite under multiple confinement levels. After conventional testing, the gathered data should be incorporated into a theoretical model to enable the numerical analysis of rock failure. Most models in the literature calibrate using stress-strain and damage, however, they are dealt with separately or via the use of complex hardening and softening functions. Therefore, development of a flexible unified yield-failure criterion is given along with a new calibration procedure to allow more physicality to be maintained in a simple, phenomenological damage-plasticity model. The enhanced yieldfailure criterion enables the simple calibration of damage state with stress evolution and therefore, when subjected to unloading or to a load path change the model can correctly measure the level of accumulated damage in the material. Other phenomenological enhancements allowing pre-peak hardening and damage evolution law calibration are also presented. Generally, once a constitutive model has been calibrated using conventional experiments it is applied numerically to simulate engineering problems. However, as little is known about rockburst mechanisms, it is vital to provide the model with a relevant data set to validate the effectiveness of predicting the phenomenon. Existing research in rockburst testing is expensive and has inherent limitations. Therefore, to easily and consistently replicate the mechanisms in the laboratory an innovative method for small scale rockburst testing is proposed. The test was centred on a standard Hoek triaxial cell, utilising the three dimensional stress state of a thick walled hollow cylinder to replicate in-situ ground conditions. Internal pressure was then imparted, maintained and released using a new platen design to simulate the excavation of a tunnel during the loading phase of the test. The proposed method successfully predicted and replicated the conditions of rockburst in the internal bore of the specimen. Characteristic acoustic emission response and stress states were also identified to provide some indication of the propensity of burst under varying insitu pressure. Due to the success of the experimental apparatus, the data at bursting coupled with the acoustic emission response from inventive platen sensor housings gave a good indication that the apparatus has the potential for patent.