Abstract -
Outcrop studies of normal faults indicate that their internal structure often depends on the nature of the sequence. Fault traces, for example, often exhibit steps and/or orientation changes at lithological contacts, resulting in fault plane irregularities that frequently control the formation of splays and lenses. The mechanical properties of faulted sequences and the effective pressure during the time of faulting are however typically poorly constrained from the study of natural faults. In the present study a numerical modelling approach is used in which both the nature of the sequence and the confining pressure during faulting can be controlled to elucidate how these intrinsic and extrinsic properties affect the internal structure of normal faults.
The commercially available Particle Flow Code in three dimensions (PFC-3D), which implements the Distinct Element Method (DEM), is used. Brittle rock is represented by randomly packed spheres that are bonded at their contacts. The breakage of bonds corresponds to fracture. This method allows modelling the large and discontinuous strains typically associated with fault zones. Modelling of normal fault growth is achieved by movement of a pre-defined fault at the base of the bonded particle model while a constant confining pressure is applied to the top of the model. The model comprises layers of particles with different properties that mimic mechanically layered rock sequences (e.g., sandstone-shale). The systematically varied parameters in the present study include sequence strength, confining pressure, net-to-gross and fault obliquity. The latter is achieved by changing the orientation of the pre-defined fault at the base of the sequence.
Analysis of the model faults is conducted on horizon maps extracted from the layer centres. The newly developed DEM model fault analysis tool allows automatic generation of fault polygons, the generation of displacement profiles and the determination of both external and internal displacement accommodating rotations, i.e. due to drag and fault-bound lenses. The analysis therefore permits quantification of displacement partitioning and associated layer juxtapositions as a function of the nature of the sequence and confining pressure.
The DEM models reveal that faults become better localised with increasing confining pressure and that the average fault strand orientation depends on fault slip obliquity. The average fault strand orientations are consistent with infinitesimal strain theory for transtensional shear zones, i.e. fault segments localise normal to the greatest incremental extension direction. Examination of displacement partitioning at branch points however reveals that there is no tendency towards hanging wall or footwall preferred displacement.
The DEM models also reveal that the degree fault localisation depends on the net-to-gross (N:G) of the sequence. For the range of N:G and for the mechanical properties considered in the present study the modelling suggests that the best localised faults develop in N:G = 2/3 sequences. At lower N:G a significant amount of the total displacement is accommodated by rotations, i.e. drag and lenses. In contrast, at high N:G bedding parallel slip is suppressed and
the more brittle layers impinge on each other early during fault zone growth, which suppresses localisation.
The DEM modelling results illustrate that in the future this numerical tool may be used to predict the likelihood of certain fault zone structure (e.g. drag, paired slip surfaces, relays etc.) within certain sequences deformed at a given depth. DEM modelling may hence be incorporated into existing workflows to complement other data (seismic, flow, well) for better assessing the sealing potential of faults.
Abstract of talk given to:
Geometry and Growth of Normal Faults, Geological Society of London, June 2014.