The morpho-mechanical behaviour of one artificial granite joint with hammer
ed surfaces, one artificial regularly undulated joint and one natural schis
t joint was studied. The hammered granite joints underwent 5 cycles of dire
ct shear under 3 normal stress levels ranging between 0.3 and 4 MPa. The re
gularly undulated joint underwent 10 cycles of shear under 6 normal stress
levels ranging between 0.5 and 5 MPa and the natural schist replicas underw
ent a monotonics shear under 5 normal stress levels ranging between 0.4 and
2.4 MPa. These direct shear tests were performed using a new computer-cont
rolled 3D-shear apparatus. To characterize the morphology evolution of the
sheared joints, a laser sensor profilometer was used to perform surface dat
a measurements prior to and after each shear test. Based on a new character
ization of joint surface roughness viewed as a combination of primary and s
econdary roughness and termed by the joint surface roughness, SRs one param
eter termed 'joint surface degradation', D-w, has been defined to quantify
the degradation of the sheared joints. Examinations of SRs and D-w prior to
and after shearing indicate that the hammered surfaces are more damaged th
an the two other surfaces. The peak strength of hammered joint with zero-di
latancy, therefore, significantly differs from the classical formulation of
dilatant joint strength. An attempt has been made to model the peak streng
th of hammered joint surfaces and dilatant joints with regard to their surf
ace degradation in the course of shearing and two peak strength criteria ar
e proposed. Input parameters are initial morphology and initial surface rou
ghness. For the hammered surfaces, the degradation mechanism is dominant ov
er the phenomenon of dilatancy, whereas for a dilatant joint both mechanism
s are present, A comparison between the proposed models and the experimenta
l results indicates a relatively good agreement. In particular, compared to
the well-known shear strength criteria of Ladanyi and Archambault or Saeb,
these classical criteria significantly underestimate and overestimate the
observed peak strength, respectively, under low and high normal stress leve
ls. In addition and based on our experimental investigations, we put forwar
d a model to predict the evolution of joint morphology and the degree of de
gradation during the course of shearing.
Degradations of the artificial undulated joint and the natural schist joint
enable us to verify the proposed model with a relatively good agreement. F
inally, the model of Ladanyi and Archambault dealing with the proportion of
total joint area sheared through asperities, a(s), once again, tends to un
derestimate the observed degradation. Copyright (C) 2001 John Wiley & Sons,
Ltd.