Does fault strengthening in laboratory rock friction experiments really depend primarily upon time and not slip?

Abstract

The popular constitutive formulations of rate‐and‐state friction offer two end‐member views on whether friction evolves only with slip (Slip law) or with time even without slip (Aging law). While rate stepping experiments show support for the Slip law, laboratory‐observed frictional behavior near zero slip rates has traditionally been inferred as supporting Aging law style time‐dependent healing, in particular, from the slide‐hold‐slide experiments of Beeler et al. (1994). Using a combination of new analytical results and explicit numerical (Bayesian) inversion, we show instead that the slide‐hold‐slide data of Beeler et al. (1994) favor slip‐dependent state evolution during holds. We show that, while the stiffness‐independent rate of growth of peak stress (following reslides) with hold duration is a property shared by both the Aging and (under a more restricted set of parameter combinations) Slip laws, the observed stiffness dependence of the rate of stress relaxation during long holds is incompatible with the Aging law with constant rate‐state parameters. The Slip law consistently fits the evolution of the stress minima at the end of the holds well, whether fitting jointly with peak stresses or otherwise. But neither the Aging nor Slip laws fit all the data well when a − b is constrained to values derived from prior velocity steps. We also attempted to fit the evolution of stress peaks and minima with the Kato‐Tullis hybrid law and the shear stress‐dependent Nagata law, both of which, even with the freedom of an extra parameter, generally reproduced the best Slip law fits to the data.