SCEC Award Number 18043 View PDF
Proposal Category Individual Proposal (Integration and Theory)
Proposal Title Inversion of fault hydro-mechanical properties from borehole observations of fluid flow and fault slip
Name Organization
Robert Viesca Tufts University
Other Participants Pathikrit Bhattacharya, Postdoctoral Scholar
SCEC Priorities 3f, 2c, 2a SCEC Groups FARM, Seismology, CS
Report Due Date 03/15/2019 Date Report Submitted 12/14/2021
Project Abstract
Recent controlled fluid-injection experiments on well instrumented faults bridge a knowledge gap by simultaneously monitoring fault slip and pore pressure evolution on a fluid-activated fault [Guglielmi et al., 2015]. Such datasets can reveal interesting fault behavior: e.g., Guglielmi et al. find that fluid-injection first activates aseismic slip, which is followed only subsequently by the onset of micro-seismic activity. In recent work, we used these observations to estimate the hydro-mechanical properties of a fluid-activated fault using numerical models of pore-fluid flow and aseismic fault slip within an expanding shear rupture. While the estimated hydrological properties of the fault are in agreement with relevant laboratory derived values, a simple model for fault slip assuming a constant friction coefficient fails to fit the time history of slip for realistic mechanical fault parameters. Therefore, besides producing meaningful and internally consistent bounds on in-situ fault properties like fluid permeability and storage, as well as the background stress state and the shear modulus, our results also suggest that the complete observed time history of slip probably requires the mechanical properties of the fault to evolve with slip. This work explored replace the constant friction coefficient constraint by incorporating slip-weakening in order to improve the fit to the observed slip data and obtain in situ constraints on the relevant new model parameters. Therefore, besides illuminating our mechanistic understanding of fluid-induced aseismic slip, this work established quantitative constraints on key fault parameters from direct field-scale observations, a crucial step towards building more realistic models of earthquake hazard.
Intellectual Merit Actively deforming faults are rarely accessible. While remote geodetic and seismological measurements of active faults and geologic observations of exhumed faults have provided an abundance of useful information, there remains a noticeable dearth of observational constraints. For example, it is difficult to infer the extent and distribution of slip in a well constrained manner, as well as to measure the mechanical and hydrologic properties of deforming faults. A field experiment, in which fault deformation was provoked by borehole injection and measured in situ, has provided novel data [Guglielmi et al., 2015]. In initial work, we developed a coupled hydro-mechanical model with the purpose of determining whether characteristic representations of fault-zone fluid flow and deformation could reproduce the observations and to what extent model parameters could be inferred, including the hydraulic permeability and storativity of the fault, host rock stiffness, fault friction coefficient, and the stress state. This work looked to examine model assumptions, particularly as a model assuming a constant Mohr-Coulomb coefficient of friction was found to be insufficient to reproduce the observed history of slip.
Broader Impacts This project provided support for one postdoctoral scholar (Pathikrit Bhattacharya) and for work relevant to injection-induced seismicity.
Exemplary Figure Figure 4: Fits to the observed slip (blue squares) at the borehole in response to fluid injection [Guglielmi et al., 2015]. (A) The model time history (purple and orange curves) reflects the slip at the center of nearly elliptical ruptures whose strength is determined by the local effective normal stress and a piecewise-linear slip-weakening friction coefficient with peak strength f_p, residual-to-peak strength ratio f_r/f_p and slip-weakening distance D_c. The shaded regions around the colored lines represent 10% of the accepted samples with shading density representing frequency of occurrence. The purple curve fits the entire time history of slip, the orange curve the first 1180 seconds only. The three contour insets show the spatial distributions of slip corresponding to the purple curve at the indicated times. The dashed blue contours represent increases in pore-pressure by 2, 1 and 0.5 MPa. (B) The linear slip-weakening model derived from the fits in A. (C)–(G) The Bayesian posteriors corresponding to the fits in A. The reported values of G_o and \tau are derived assuming f_p = 0.6. The vertical dashed lines represent the least misfit models shown in A.