C. Marone et Sjd. Cox, SCALING OF ROCK FRICTION CONSTITUTIVE PARAMETERS - THE EFFECTS OF SURFACE-ROUGHNESS AND CUMULATIVE OFFSET ON FRICTION OF GABBRO, Pure and Applied Geophysics, 143(1-3), 1994, pp. 359-385
We describe experiments in which large (14 x 40 cm nominal contact are
a) blocks of gabbro were sheared in a direct shear apparatus at room t
emperature, 5 MPa normal stress, and slip velocities from 0.1 to 10 mu
m/s. The apparatus was servocontrolled using a displacement feedback
measurement made directly between the gabbro blocks. Two surface rough
nesses were studied (rough, produced by sandblasting, and smooth, prod
uced by lapping with #60 grit) and accumulated displacements reached 6
0 mm. Measurements of surface topography were used to characterize rou
ghness and asperity dimensions. Step changes in loading velocity were
used to interrogate friction constitutive properties. Both rough and s
mooth surfaces showed appreciable displacement hardening. The coeffici
ent of friction mu for rough surfaces was about 0.45 for initial slip
and 0.7 after sliding 50 mm. Smooth surfaces exhibited higher mu and a
greater tendency for unstable slip. The velocity dependence of fricti
on a-b and the characteristic friction distance D-c show systematic va
riations with accumulated displacement. For rough surfaces a-b started
oat positive and became negative after about 50 mm displacement and D
-c increased from 1 to 4 mu m over the same interval. For smooth surfa
ces, a-b began negative and decreased slightly with displacement and D
-c was about 2 mu m, independent of displacement. For displacements <3
0 mm, rough surfaces exhibit a second state variable with characterist
ic distance about. 20 pm. The decrease in a-b with displacement is ass
ociated with disappearance of the second state variable. Our data indi
cate that D-c is controlled by surface roughness in a complex way, inc
luding but not limited to the effect of roughness on contact junction
dimensions for bare rock surfaces. The data show that simple descripti
ons of roughness, such as rms and peak-to-trough, are not sufficient t
o infer D-c. Our observations are consistent with a model in which D-c
scales with gouge thickness.