[S12P-03] High-velocity friction experiments for samples of various size: Scale effect in the critical slip rate for dynamic weakening
There is a large difference in length scale between laboratory and natural large events. Then it is crucial to study the scale effect in friction and evaluate applicability of lab-experiments. It is known that friction decreases dramatically at coseismic slip rate due to frictional heating and various associated physico-chemical effects (e.g., Tsutsumi and Shimamoto, 1997 (TS97); Di Toro et al., 2011). High-velocity friction experiments for centimeter-scale samples showed the dynamic weakening at slip rates V of the order of 0.1 m/s. On the other hand, recent experiments by Yamashita et al. (2015) (Y+15) demonstrated that a large sample with 1.5 m-long and 0.1 m-wide frictional surface showed the dynamic weakening at V of the order of 0.01 m/s, arguing existence of the scale effect on the weakening slip rate Vw. Noda (2023) pointed out that it can be explained by concentration of frictional power due to thermoelastic instability (e.g., Dow and Burton, 1972). Heterogeneous thermal expansion of wall rocks of a frictional surface due to heterogeneous normal stress causes unlimited growth of the heterogeneity at V>Vcr, where Vcr is the critical slip rate proportional to the wavenumber of the heterogeneity. A larger sample hosts heterogeneity of a smaller wavenumber, and thus should weaken at a lower slip rate. In this case, Vw is expected to be inversely proportional to the sample size. Here, we present our preliminary results of high-velocity friction experiments for samples of various size conducted in order to confirm the scale effect and the hypothesis of the thermoelastic instability. We used gabbro of sub-millimeter grain size (“Zimbabwe black”) which may be comparable to experiments by TS97 and by Y+15. Because thermal properties of the apparatus matters (Yao et al., 2016), we used the same apparatus that TS97 used, which is now at Yamaguchi University. Width of the frictional surface W is 4.5 mm in TS97, 0.1 m and 11.5 mm in Y+15. So far, we conducted experiments with W of 3 mm and 13.5 mm. The experimental conditions are similar to TS97; the normal stress was kept at 1.5 MPa and successive positive velocity steps of a factor of about 1.5 were applied every about 50 m of slip. The sample with W=3 mm weakened at V=0.4 m/s, while that with W=13.5 mm did at V=0.25 m/s. When plotted together with TS97 and Y+15, Vw ranges for about 2 orders of magnitude. Normalization of the slip rate by Vcr, which is inversely proportional to W, reduces this variation to about 1 order of magnitude (Figure). The comparison indicates the existence of the scale effect, reconfirming the discovery by Y+15 and inferring significance of thermoelastic instability although more data is needed.