Questions and answers: anisotropy and inhomogeneity 169(a) establishing the purpose behind identifymg such zones;
(b) have the appropriate rock mechanics properties been selected in
conjunction with, say, structural geology information to identify
these domains? and
(c) given the criteria for zoning the rock mass, how do we distinguish
sampling artifacts from real differences in different domains?
To answer the question, it would be necessary to have further in-
formation on the exact purpose of the site investigation and the other
properties measured, but dividing the rock mass into different struc-
tural domains can be useful, e.g. different domains could have different
susceptibilities to slope failure.
Based just on the data given, the question is whether the values indic-
ate a real rock structure difference or whether the variation in the values
could have arisen by the sampling process. If we thought it justified, we
could perform statistical tests to examine whether statistical deficiencies
exist. However, this would not indicate the existence of structural do-
mains; it could be highlighting some attribute of the sampling process.
Moreover, we have not been told the purpose of the project and which
rock mechanics properties will be required in the modelling. Thus, there
is insufficient evidence to conclude from these data alone that structural
domains are required.410.9 The following list of fracture locations, quoted in metres, is
taken from the fracture log of a borehole core which transects a
well known stratigraphic boundary between two units of limestone.
Evidently, this boundary is clear to sedimentologists, but not to
geotechnical engineers.
5.780,6.391,6.761,7.105,7.180,7.401,7.478,8.142,8.455,9.139,
10.072, 10.264, 10.470, 10.539, 10.678, 11.421, 11.541, 12.178,
12.596,12.620,12.736,12.936,13.134,13.325,13.430
Use the concept of moving averages to help locate this boundary.A10.9 To attempt to find this boundary using moving averages, we
firstly compute the spacing values and then, taking these spacing data
in groups, compute the mean spacing and mid-point distance of the
group. The mean spacing is plotted against mid-point distance, and the
was based on the records of 8698 people who had their car stolen in the area from April
1998 to March 1999. The number of cars stolen versus the star sign data were
May 21-Jun 21 Apr 21-May 20 Jun 22-Jul23 Jul24-Aug 23 Mar 21-Apr 20 Feb 20-Mar 20
Gemini 811 Taurus 794 Cancer 785 Leo 756 Aries^754 Pisces^730
Aug 24-Sep 23 Dec %Jan 20 Sep 24-0ct 22 Jan 21-Feb 19 Oct 25Nov 22 Nov %Dec 22
Virgo 719 Capricorn 696 Libra 686 Aquarius 671 Scorpio 657 Sagittarius 639
Interpretations of these data are, for example, that Geminis are inattentive and leave their
keys in the car, whereas Sagittarians buy their dream car and then ensure that it is secure.
What do you think? Is the adoption of such a predictive model valid, based on the data
alone?