Shear heating in granular layers

Heat-flow measurements imply that the San Andreas Fault operates at lower shear stresses than generally predicted from laboratory friction data. This suggests that a dramatic weakening effect or reduced heat production occur during dynamic slip. Numerical studies intimate that grain rolling or localization may cause weakening or reduced heating, however laboratory evidence for these effects are sparse. We directly measure frictional resistance (μ), shear heating and microstructural evolution with accumulated strain in layers of quartz powder sheared at a range of effective stresses (G_n = 5-70 MPa) and sliding velocities (V = 0.01 - 10 mm/s). Tests conducted at G_n ≥ 25 MPa show strong evidence for shear localization due to intense grain fracture. In contrast, tests conducted at low effective stress (G_n = 5 MPa) show no preferential fabric development and minimal grain fracture hence we conclude that non-destructive processes such as grain rolling/sliding, distributed throughout the layer, dominate deformation. Temperature measured close to the fault increases systematically with G_n and V, consistent with a one-dimensional heat-flow solution for frictional heating in a finite width layer. Mechanical results indicate stable sliding (μ ∼ 0.6) for all tests, irrespective of deformation regime, and show no evidence for reduced frictional resistance at rapid slip or high effective stresses. Our measurements verify that the heat production equation (q = μσ_nV) holds regardless of localization state or fracture regime. Thus, for quasistatic velocities(V ≤10 mm/s) and effective stresses relevant to earthquake rupture, neither grain rolling/sliding or shear localization appear to be a viable mechanism for the dramatic weakening or reduced heating required to explain the heat flow paradox.