A granular‐physics‐based view of fault friction experiments

Slide‐Hold‐Slide Protocols and Frictional Healing in Discrete Element Method (DEM) Simulations of Granular Fault Gouge

“…These plots indicate that the magnitude of gouge compaction in the granular model is in general agreement with laboratory observations, after both are normalized by their appropriate value of D c . For the granular simulations this is the sensible normalization; Ferdowsi and Rubin (2020) found that the ratio of gouge thickness changes to D c was of the nominal gouge thickness over the range they explored. For the lab data, normalization by D c is intended to account for the fact that deformation is typically localized over a layer of unknown thickness; inherent in this approach is the assumption that both slip and compaction are concentrated within this layer.…”

“…The stress decay response for the three velocities differ more significantly at the lower (lab-like) stiffnesses of Figures 2a-2c also include the predictions of the Aging and Slip laws for the stiffnesses used in the granular model. These predictions are obtained using the RSF parameter values determined independently from Slip law fits to simulated velocity steps performed on the identical granular system (Ferdowsi & Rubin, 2020). We do not use parameter values determined using the Aging law because that model is clearly inappropriate for modeling velocity steps, in both laboratory experiments and our DEM, as the Aging-law estimate of D c depends entirely upon the magnitudes and signs of the velocity steps one chooses to fit (Bhattacharya et al, 2015) (for stiff systems, such as that used by Ferdowsi and Rubin (2020), only the value of D c , and not a and b, depend upon the adopted state evolution law).…”

“…In addition, for comparison to those data we explore a wider range of system stiffnesses. All the SHS simulations in Ferdowsi and Rubin (2020) were conducted at the highest stiffness we could achieve, that limit being set by the elastic stiffness of the gouge layer itself. For velocity-step tests this is desirable; a high stiffness ensures that the inelastic sliding velocity is always nearly the load point velocity, which allows one to infer the RSF parameters directly from the transient frictional response without having to account for a varying velocity.…”

Earthquake dynamics constrained from laboratory experiments: new insights from granular materials

“…Although the fit seems to be of great quality merely from a numerical point of view (R 2 is always greater than 0.85) the fit produces values that are sometime controversial (see Tables 2 and 4 and section 3) and often suppresses the dependence on the slip velocity, thus giving only a state-dependent friction. et al, 2018] and granular models [Ferdowsi and Rubin, 2020], who also reported that none of the existing rate and state laws reproduce all the robust features emerging from laboratory data.…”

“…They are not proper of the constituent materials, but are due to the granular nature of the medium, as also shown from numerical simulation with materials whose friction coefficient is 0.5 [Mair, 2007]. Comparable values are also found with microscopic models [Van den Ende et al 2018; Ferdowsi and Rubin, 2020]. It can also be observed that there is no stress rise at the beginning of the slip.…”