You are on page 1of 1

N.

Barton / International Journal of Rock Mechanics & Mining Sciences 39 (2002) 185–216 189

A graphic example of this ‘scale effect’ is shown in correlations with an investigation of seismic velocity.
Fig. 2. While the logged values of RQD and Jn must Some recent examples of available seismic techniques
include all scales of the rock mass structure, every detail have been given in this Journal (vol. 38, No. 6, Sept
cannot be modelled discretely, and one is forced to 2001) and will be referred to shortly.
include in a distinct element model only the joint sets If shallow seismic refraction data, or deeper cross-
assumed to have most influence on, for example, the hole tomography, or VSP, can be extrapolated by means
stability of the modelled tunnels or cavern. A represen- of one or more correlations between Q and Vp ; then
tative value for the modulus of deformation of a limited some of the uncertainties in tunnelling or cavern design
rock mass volume is still required, onto which will be could be removed. However, there are several potential
superimposed the larger scale ‘REV’ (representative pitfalls, which have led some to assume that seismic data
elementary volume) response of the rock mass as a may be unreliable. The truth is probably that the physics
whole, in which a fault might also be modelled, if close have not been understood.
enough to the excavation under investigation. The Based on data from hard rock tunnelling projects in
stiffness of the fully consolidated major joint sets is several countries, including the 62 m span Gjvik cavern
likely to modify the assumed modulus of deformation, in Norway where NGI performed seismic tomography
and may also give anisotropic deformability, i.e. details and core logging, a preliminary hard rock correlation
not usually captured in continuum modelling. between Q and Vp was suggested [8]:
The less dominant jointing is reflected in a reduced Q-
Vp E3:5 þ log Q ð2Þ
value, and in a reduced P-wave velocity (Vp ) and static
modulus of deformation. We will define the latter as where Vp is in units of km/s. (Note: all logarithmic terms
Emass in this paper, to distinguish it from Young’s are log10 in this paper.)
modulus and from the physical joint aperture (E), where Important clues supporting this relation were subse-
E > e; the smooth parallel plate hydraulic aperture used quently discovered from extensive earlier work by
in the cubic law. This inequality is due to effects of joint Sjgren and co-workers [9], who had presented Vp ;
roughness JRC, and flow tortuosity around areas of RQD and joint frequency (l m1) data from 120 km of
rock-to-rock contact [7] which will be addressed when seismic refraction profiles and 2.8 km of adjacent core
discussing grouting. data. Fig. 3a shows the mean trends of this data (with
some slight extrapolations by the writer). Along the x-
axis of both figures is appended the simple Vp 2Q
3. Correlation of Q with V p relation given by Eq. (2).
The above relation between Vp and Q was subse-
In view of the importance of the site investigation quently generalised to include rock that could be weaker
phase that precedes preliminary design, and which gives or even stronger than the assumed ‘hard’ rock. Normal-
indications of the need for additional hydrogeologic isation of the Q-value was tested, using 100 MPa as
information, we will start this exploration of Q- the hard rock norm. For improving correlation to

Fig. 2. An illustrative ‘textbook example’ of dominant joints (strong continuity, low JRC, JCS and fr ) requiring distinct element modelling, and the
less dominant ‘back-ground’ jointing (higher JRC, JCS and fr ) which will nevertheless be represented in RQD and Jn and in a reduced deformation
modulus and seismic velocity. Kimmeridge Bay, South Coast, England [writer].

You might also like