Final Report Summary - SEISMIC (Slip and Earthquake Nucleation in Experimental and Numerical Simulations: a Multi-scale, Integrated and Coupled Approach)
We developed a system for monitoring and locating laboratory earthquakes in a rotary shear apparatus, which has the main advantage that very large displacements can be reached. Using this system, we were able to collect information on a large number of earthquakes and perform statistical analysis on them. We found that our laboratory earthquakes follow distributions that are similar to natural earthquakes.
Experiments, an analytical model and DEM simulations demonstrate the operation of pressure solution (stress-enhanced dissolution) and surface energy driven grain boundary healing can control the recurrence interval and stress drops of stick-slips (laboratory earthquakes) and additionally can cause the development of substantial cohesion during periods of no slip. Because cohesion needs to be destroyed when slip is re-initiated and has a short weakening distance associated with it, re-sliding of a cohesive gouge under our laboratory conditions results in unstable slip (“stick-slip”). The existence and development of cohesion should not be ignored, particularly in cases where effective normal stress is low such as in induced seismicity.
We have extended and modified a pre-existing microphysical model for fault gouge friction which specifically accounts for the operation of fluid-assisted deformation mechanisms such as pressure solution. The model is centered around the premise that time-dependent compaction is in competition with slip-dependent dilation which at steady state results in an increasing porosity with increasing sliding velocity. Such a competition resulting in a velocity-dependent porosity has been verified experimentally under in-situ hydrothermal conditions in a number of simulated fault gouges derived from natural faults such as the Alpine Fault in New Zealand. The model and results also show that this competition gives a high potential for the nucleation of unstable slip, i.e. earthquakes. We have implemented the microphysical model into a large scale seismic cycle simulator and used this simulate the occurrence of earthquakes on a time-scale of millennia. The results for a homogeneous fault are comparable to existing "classical" models that use an empirical friction law. When we apply our seismic cycle simulator to a heterogenous fault, i.e. one that consists of a distribution of asperities that are frictionally unstable and a matrix that is inherently stable, we find that the behavior of the fault changes dramatically when we change the fractal dimension of the size distribution of the asperities. Faults that relatively more larger asperities (low fractal number) display regular seismicity with earthquakes that rupture one or several asperities. Earthquakes that rupture the entire fault are sparse or entirely absent. In contrast, faults with relatively few large asperities (high fractal number) are seismically quiescent for long periods of time, until they rupture in one large earthquake which spans the entire fault. The earthquakes on these types of faults are periodic and should therefore be predictable.