Strength, fallouts and numerical modelling of hard rock masses

Sammanfattning: The prediction of compressive stress-induced failures is of concern for the design and construction of deep underground excavations in mining and civil engineering. An overstressed rock mass may result in fallouts of rock, which may cause occupational safety hazards, damage of equipment, and/or production disturbances. The purpose of this thesis was to improve on how to model compressive stress-induced failure and fallouts with appropriate material models and strength parameters. The thesis focuses on commonly used design methods for underground excavation with special application to hard rock mass strength assessment. The aim was to suggest the most appropriate material model for fallout prediction and to identify factors governing the strength of hard rock masses. A comprehensive literature review of existing classification/characterisation systems and rock mass failure criteria used to estimate the rock mass strength was conducted. Existing rock mass failure criteria and classification/characterisation systems were evaluated through three case studies. A Round Robin Test was conducted for two of these cases. The evaluation was performed in order to identify robust systems and criteria, as well as to identify the parameters having the strongest impact on the calculated rock mass strength. The case studies revealed that the N, Yudhbir - RMR76, RMi, Q-, and Hoek-Brown - GSI methods, appeared to yield reasonable agreement with the measured stresses at failure. The parameters reflecting joint shear strength have a major influence on the estimated rock mass strength. The RMi method has proven difficult to use. For quality determination of the rock mass, a stress reduction free Q-system (or N) is preferable, as the Q-system covers a wider range of geological situations and the parameters are better described than for the RMR system. For massive rock masses in areas with high stresses and tight interlocks the impact of jointing is less obvious and the GSI method can be used for determination of the rock mass quality. In the subsequent work, a total of six selected case histories of fallouts in hard rock masses were studied. These were collected based on a comprehensive investigation and survey of well described compressive stress- induced fallouts, in drifts, raises and/or tunnels. All six cases considered civil and mining engineering rock excavations where the rock mass properties, measured stresses, behaviour and fallout were well documented. The field observations were compared with predictions from numerical modelling using the finite element analysis program Phase2. The results of the applied brittle-plastic models were sensitive to changes of the peak strength parameters and less sensitive to variations in residual parameters. A cohesion-softening friction-hardening (CSFH) model, using peak cohesion equal to the intact rock strength, proved to be the most appropriate material model for capturing the observed rock behaviour. Yielded elements failed in shear and intersecting shear bands were found to be good indicators of compressive-stress induced fallouts. This is likely since shear is often the final mechanism in the failure process before fallout occurs. The potential compressive stress-induced fallouts can, using a CSFH model, be predicted using the following indicators: (i) intersecting shear bands with significantly elevated strains and which connect to the excavation boundary, and (ii) shear bands being located within the region of yielding. Both criteria must be fulfilled simultaneously. The results showed that the developed shear bands and the zone of yielded elements were sensitive to changes in mesh density. By using small elements (0.01 m) at and close to the boundary of the excavation and in the region of the predicted failure, the results showed no significant changes of the predicted failure zone, with a further decrease in zone size. The CSFH model was applied for prediction and follow-up of compressive stress-induced failure and fallouts of footwall drifts in the Kiirunavaara underground mine. A multi-stage analysis was carried out in order to gradually change the stresses to simulate mining progress. A parametric study was conducted in which strength properties, location, and shape of the footwall drift were varied. The modelling results were sensitive to the shape of the drift. The location of the predicted fallouts in the model was in good agreement with the location of observed fallouts for the case when the drift roof was simulated flatter than the theoretical cross-section. The results indicate that the true shape of the drift was different from the planned one. Simulating actual fallout by removing the indicated region of fallout in the model showed fewer occurrences of compressive-stress induced fallouts in later loading stages for footwall drifts in the Kiirunavaara mine. By scaling the damaged rock and creating a v-notch in the roof (similar to the predicted fall-out shape), in an access drift in the Kristineberg mine, the stability of the excavation was improved.