SCEC Award Number 11055 View PDF
Proposal Category Individual Proposal (Integration and Theory)
Proposal Title Thermally Driven Shear Localization in Fault Zones
Investigator(s)
Name Organization
James Rice Harvard University
Other Participants Platt, John D.
SCEC Priorities A8, A7, A10 SCEC Groups FARM, Geology
Report Due Date 02/29/2012 Date Report Submitted N/A
Project Abstract
 We investigate strain localization in fluid-saturated gouge materials, driven by thermal processes during rapid shear. Our previous work based on thermal pressurization and rate-strengthening friction predicted a critical width above which uniform shearing was unstable. We have first shown that two additional effects, inertia and dilatancy of the gouge material, exert a negligible drag on strain localization at seismogenic depths. Using material parameters from the extensive studies of the Median Tectonic Line gouge, and frictional data unfortunately limited to low strain rate experiments, we predict a localized zone width of 5-40 microns, for a slip rate of 1 ms-1 and for ambient temperature and effective normal stress conditions intended to represent a centroidal depth (~ 7 km) of a typical crustal seismogenic zone. Recent observations have shown that thermal decomposition of fault materials may occur during rapid shear and modeling suggests that these endothermic reactions can cap the maximum temperature rise and generate large pore pressures. We thus investigated how these effects could affect strain localization. First performing a linear stability analysis we find a critical width as a function of ambient fault temperature. At low temperatures, when the reaction rate is negligible, we recover the prediction of Rice and Rudnicki. At high temperature where the reaction is dominant, we see significant additional strain localization and predict for calcite decarbonation a localized zone width of 2-6 microns, for a slip rate of 1 ms-1.
Intellectual Merit Well known studies for the Punchbowl and Chelungpu faults suggest a highly localized slip zone of primary earthquake shear, ~50-300 microns wide. More recently De Paolo et al. [2008] analyzed a series of faults in the Northern Apennines, Italy and found on all a nested structure, with a highly localized band of less than 100 microns, contained within a broad damage zone. Faults that had undergone more slip had broader damage zones, but the localized zone width was comparable for all faults. Further evidence for strain localization during rapid shear comes from high-velocity friction experiments; Brantut et al. [2008] indicated the presence of an “ultralocalized deformation zone”, interpreted as the main slipping zone of the experiment, 1-10 microns wide. The thinness of these zones is both a puzzle to explain, as we address in this work, and an essential input to fundamental understanding of the thermo-mechanical physics of fault zone processes in seismic shear (as we address in related studies). The amount of slip required to obtain a given strength drop via thermal pressurization is controlled by the slipping zone width, and that width influences the maximum temperature rise. Also, the study by Noda et al. [2009] showed how increasing the width of the slip zone, from 40 to 100 microns in a particular parameter range, causes the rupture to transition from a crack, to a growing slip pulse, and finally to an arresting slip pulse.
Broader Impacts The activity is a major part of the Ph.D. research of John D. Platt.

In addition to the educational contribution, this issue of highly localized shear, and when and why it occurs, is central to unraveling the hydro-thermo-mechanical physics of seismic fault zone processes.
Exemplary Figure Fig. 4. A plot showing the linear stability prediction for the critical width of shear localization in calcite as a function of ambient fault temperature. Note the two limits at extreme temperature values, one at low temperatures corresponding to classical thermal pressurization of in-situ pore fluid, with the other at high temperatures being dominated by the endothermic reaction processes. Each line corresponds to an order of magnitude increase of the reaction rate constant A (which multiplies the Arrhenius exponential factor), and we have shown that the limits are independent of A