Injection induced earthquakes and fluid-fault interactions

During my post-doc at Tufts I began collaborations with Robert Viesca on understanding the mechanics of aseismic slip activated by fluid injections. As part of these interests, we have combined theoretical work with modeling and mathematical inversion to analyze in-situ observations of fluid activated fault deformation recorded in the experiments of Yves Guglielmi and Fred Cappa on shallow carbonate faults (Guglielmi et al. (2015), Seismicity triggered by fluid injection-induced aseismic slip. Science 348, 1224-1226).

Fits to in-situ observations of fault-slip (blue-squares) in response to shallow fluid injection from the experiments of Guglielmi et al. (2015). Figure taken from Bhattacharya and Viesca (2019). The color-coded regions in the insets show rupture extent at different time instances for the pink curve fit, the blue dashed lines show extent of the fluid-pressurized region. For details, see Bhattacharya and Viesca (2019).

We found that the in-situ pore-pressure and fault-slip data simultaneously recorded during these experiments require that fluid-activated aseismic slip outpace pore-fluid migration. This indicates that aseismic slip could trigger ‘induced’ earthquakes within regions yet to be pressurized by pore-fluid migration! This work has recently been published in Science (Pathikrit Bhattacharya and Robert C. Viesca (2019), Fluid-induced aseismic fault slip outpaces pore-fluid migration, Science, 364, 6439, 464–468, https://science.sciencemag.org/content/364/6439/464).

I continue to work on the implications of these results for the post shut-in behavior of the fault and how it might explain the observed continued occurrence of large magnitude induced earthquakes after the cessation of injection.

Physics of rate-state friction

From my PhD days, I have been involved in working on the physics underlying the popular rate-state friction constitutive equations widely used in modeling earthquake deformation. Using laboratory data derived with novel experimental techniques in combination with a self-tuning Bayesian parameter inference code that I developed, I have worked on demonstrating that the decades old laboratory data that had been traditionally inferred to support the time-dependent strengthening of stationary frictional contacts can be re-interpreted to show more support for the theory that frictional interfaces heal only with slip. This approach of dealing with friction data can motivate the introduction of new physics into the description of fault friction; this is my ultimate goal.

Comparisons of fits to laboratory observed friction evolution during periods of near zero-sliding called holds. Conventional frictional wisdom points towards the superiority of the time-dependent Aging law under these circumstances, my work has demonstrated that the slip-dependent Slip law is unequivocally superior instead. Taken from Bhattacharya, Rubin and Beeler (2017).

I continue to closely collaborate with the rock mechanics laboratories at Brown and Penn State and the theory group at Princeton to devise and examine data from increasingly ambitious experiments to explore the full parameter space of fault friction.

Recently, we are probing the physics of fault frictional response to normal stress perturbations during sliding. Stay tuned for some exciting stuff on this front soon!

Modeling of earthquake nucleation and aseismic slip response to stress perturbations

I am also interested in understanding the implications of variable fault friction in deciding the fate and origin of earthquakes. In the past, I have worked on understanding the implications of a hitherto unknown stressing-rate effect on the evolution of earthquake nucleation in numerical models. I continue to be interested in these problems, in particular tying these models to modern, large scale friction experiments.

Evolution of slip-pulses during earthquake nucleation with a stressing-rate dependent friction law. Taken from Bhattacharya and Rubin (2014).

I continue to be interested in understanding how stress perturbations on otherwise potentially unstable and stable faults lead to modulation of nucleation and aseismic slip. My particular interest is in using these models to understand aseismic fault response to shear and normal load perturbations of different time signatures, e.g., the seasonal oscillations of the hydrologic cycle or extreme rainfall events.

Ultrasonic monitoring of frictional interfaces

There has been recent interest in ultrasonic monitoring as a means to probe the mechanics of contact interfaces during rock friction interfaces. Ultrasonic waves transmitted across sliding surfaces in rocks have been shown to be sensitive to the ‘state’ of the frictional interface, and these hold promise as meaningful probes of the deformation of microcontacts during frictional sliding. Such ultrasonic data have revealed a hitherto unknown shear stress dependence in fault friction. Our analysis of 1-D nucleation with this new rate-state formulation reveals that the introduction of the stressing rate dependence significantly alters the response to large velocity steps, and hence the style of nucleation.

Left to right: Shear stress and ultrasonic responses to 3 times, 30 times and 300 times velocity step increases plotted versus time (in units of 1000s). We simultaneously monitor P- and S-signals to better probe the rheology of the interface. The transmitted amplitudes and travel times both show sensitivity to sliding conditions.

In continuing collaboration with the Penn State Rock Mechanics laboratory, we have been trying to improve our understanding of the contact scale mechanics by simultaneously using P- and S-transmitted amplitudes and travel times. Based on even the simplest models of contact interfaces, one can argue that S- and P-transmissivities would exhibit differences in their sensitivities to the deformation of elasto-plastic contacts. When one considers additional complexities arising out of shear loading and slip across contacts, it is not unreasonable to argue that, when compared to earlier experiments using P-wave data alone, the combined use of P- and S-transmissivities might provide new information on the contact-rheology of the frictional interfaces in these experiments.