| dc.description.abstract | Gouge-bearing fault rocks often develop characteristic (micro)structural features, such as R1-, P- and Y- micro-shears. In spite of the quite different PT-conditions under which fault rocks may develop, these shear band microstructures are often very similar and show characteristic orientations with respect to the macro-scale sense of shear. This has been widely observed and well described in both experimentally- and naturally produced fault rocks. However, no microphysical model has been developed which explains both the macroscopic rheology and why the observed shear bands are produced. In this thesis, a model is proposed for fault rock friction and the role played by micro-scale shear bands in determining macro-scale behaviour. The approach adopted is a follows. To simulate the initial stress-strain response and shear localising behaviour of a model shear zone, prior to macroscopic yield, it is assumed shear stress increases linearly with shear strain. After a certain amount of strain has been accommodated, the stress state reaches a point at which the system fails in a Mohr-Coulomb fashion producing the first shear band set. The next shear band set is then assumed to form at the orientation that is most favourable for on-going deformation, i.e. the orientation that leads to the lowest macroscopic shear strength. In order to assess what orientation is preferred, the entropy production function is used to identify the lowest entropy production rate. This gives a set of equations where quantities such as the normal stress, the system' s cohesive strength, and various frictional parameters can be varied to match experimental observations. In this way, different hypotheses about what causes particular shear band orientations to form, and in what sequence, were systematically tested. The model predicts a range of shear band orientations that falls within the range of orientations observed in both experimentally simulated and in natural fault zones, although comparison with specific experiments reveals that different frictional parameters for the R- versus P-shears are required to reproduce specific orientations seen in such experiments. Moreover, if it is assumed that the R- and P-shears can have different frictional parameters, the macroscopic rheological properties such as the peak stress and steady-state stress observed in specific experiments can be reproduced. The model also predicts that the R-shears activate before the P-shears, which is the order of shear band activation that is typically obtained in experiments. After activation, both these shear bands will then follow an antithetic rotation with respect to the imposed vorticity. This implies that Y-shears can form as a result of R-shear rotation, whereas boundary shears can form by local dilatation at the shear zone boundaries as a result of shear band activity. Even in the current stage of this presented model, valuable insights can be gained regarding the underlying mechanisms for shear band formation and strain accommodation: 1) After the R-shears have formed by Mohr-Coulomb failure, the P-shears form in the orientation that results in the lowest macroscopic shear strength; 2) After shear band formation, isovolumetric strain accommodation results in rotation of both shear bands; 3) Y-shears and boundary shears can form as a result of strain accommodation on the R- and P-shears; | |
