March 2001
POSIVA OY
Tlnkatu 4, FIN-001 00 HELSINKI, FINLAND
Tel. +358-9-2280 30
Fax +358-9-2280 3719
-tTr ENERGY
Research organisation and address Customer
VTT Energy, Nuclear Energy Posiva Oy
P.O. Box 1604 Tlnkatu 4
FIN-02044 VTT, FINLAND 00100 Helsinki
Fracture orientation distributions have been fitted for the Hstholmen, Kivetty, Olkiluoto and Romuvaara site models.
The fitted orientation distributions are based on the combined surface and borehole data. Surface and borehole data
have been combined by the Geological Survey of Finland (GTK) for Kivetty, Olkiluoto and Romuvaara. The
combined data sets were also crosschecked by GTK to ensure that they agreed with the general geological
characteristics of the sites. Combined surface and borehole data was not available for Hstholmen. As a part of the
present study a combined data set of the surface and borehole fractures was constructed for Hstholmen. The
combined data set of Hstholmen was not crosschecked against the general geological data in the same way as at the
other sites.
Fracture orientations are represented by three fracture sets for all sites. Each set is characterised by Fisher distribution.
Resulting fitting of the orientation distributions show reasonable agreement between the corrected measured
(corrected for the observation biases) and fitted distributions.
Fracture sizes were studied for by utilising the surface fractures. Based on the fracture trace lengths and orientations it
is concluded that there is no statistical implication for correlation between fracture orientation and size, except for the
Hstholmen site. At Hstholmen the horizontal fractures seem to be shorter than vertical ones. However, it should be
noted that the horizontal fractures were not well represented at any of the sites in the fracture size analysis. Especially
at the Romuvaara site the number of horizontal fractures observed on outcrops is too small for reliable statistical
analysis.
Distribution
A. Hautojrvi (Posiva Oy), ENE Arkisto
R~Q,~-L
Principal author or Project manager
~'v~
~oteri, Senior Research Scientist Heikki Raiko, Group Manager
Availability statement
Confidential
The use of the name of the T echnical Research Centre of Finland (VTT) in advertising or publication in part of
this report is only permissible by written authorisation from the T echnical Research Centre of Finland
Working Report 2001-10
Antti Poteri
March 2001
Working Report 2001-07
VTT Energy
March 2001
ABSTRACT
Fracture orientation distributions have been fitted for the Hstholmen, Kivetty,
Olkiluoto and Romuvaara site models. The fitted orientation distributions are based on
the combined surface and borehole data. Surface and borehole data have been combined
by the Geological Survey of Finland (GTK) for Kivetty, Olkiluoto and Romuvaara. The
combined data sets were also crosschecked by GTK to ensure that they agreed with the
general geological characteristics of the sites. Combined surface and borehole data was
not available for Hstholmen. As a part of the present study a combined data set of the
surface and borehole fractures was constructed for Hstholmen. The combined data set
of Hstholmen was not crosschecked against the general geological data in the same
way as at the other sites.
Fracture orientations are represented by three fracture sets for all sites. Each set is
characterised by Fisher distribution. Resulting fitting of the orientation distributions
show reasonable agreement between the corrected measured (corrected for the
observation biases) and fitted distributions.
Fracture sizes were studied for by utilising the surface fractures. Based on the fracture
trace lengths and orientations it is concluded that there is no statistical implication for
correlation between fracture orientation and size, except for the Hstholmen site. At
Hstholmen the horizontal fractures seem to be shorter than vertical ones. However, it
should be noted that the horizontal fractures were not well represented at any of the sites
in the fracture size analysis. Especially at the Romuvaara site the number of horizontal
fractures observed on outcrops is too small for reliable statistical analysis.
TIIVISTELM
Tiivistelm
Abstract
1. Introduction ................................................................................................................. 2
2. Data ............................................................................................................................. 3
2.1. Fracture mappings ................................................................................................ 3
2.2. Orientation distributions ...................................................................................... 4
2.3. Fitted orientation distributions ............................................................................. 6
4. Conclusions ............................................................................................................... 13
5. References ................................................................................................................. 14
1. INTRODUCTION
Extensive fracture mapping programs have been carried out in the site investigation
program. Fractures have been measured both on the surface and in the boreholes. At the
Kivetty, Olkiluoto and Remuvaara sites more than ten boreholes have been extended to
the depth of 500 m or deeper. At Hstholmen five deep boreholes exists. Surface data on
the fracturing have been collected on the outcrops and exposed rock surfaces.
In the present study fracture orientation distributions have been estimated for the
Hstholmen, Kivetty, Olkiluoto and Remuvaara sites. In addition to the orientation
distributions, also the correlation between the fracture size and orientation was
examined. Also, as a part of this work the maps of the fracture traces on the low-level-
waste silo of the VU -repository at Olkiluoto and on the walls of the waste gallery of the
VU -repository at Hstholmen were digitised. Fracture traces were calculated from the
digitised images and converted into FracMan tracemap files for later use by the Golder
Associates Jne. in the earthquake simulations.
3
2. DATA
The borehole measurements include mapping of the core of the borehole, borehole-TV
and dipmeter measurements. At least few thousands of orientated fractures have been
mapped at each site. Surface mappings were first made on the outcrops but then at
Kivetty, Olkiluoto and Romuvaara also therock surface was exposed by making several
hundreds of meters Iong research trenches.
In the present study the fracture orientation distribution are based on combined data set
of the surface and borehole observations. Both outcrop and borehole measurements
include bias that prevents direct application of the measurements. The bias follows from
the geometrical limitations to observe planar fractures by one or two-dimensional
observations. From purely geometrical point of view it is possible to correct the bias in
the measured orientation distribution by assuming that the fracturing is statistically
homogeneous and that the sample is representative. In reality, a complete and reliable
description of the fracture orientations cannot be based on the observation in one
direction only. Several observations that scan the fracture population from different
orientations are needed.
In the Posiva's site selection program the Geological Survey of Finland have worked out
representative fracture orientation distributions for the Kivetty, Olkiluoto and
Romuvaara sites by using both the surface and borehole information (Anttila et. al.
1999b, 1999c, 1999d). In that study the orientation distributions were based on the
measured fracture orientations from the exposed research trenches on the surface and
from carefully ehosen boreholes. The bases of the combined data sets are presented in
the Table 1. The combined data sets were also crosschecked to ensure that they agreed
with the general geological characteristics of the sites. The representative data sets
received from the Geological Survey of Finland for the Kivetty, Olkiluoto and
Romuvaara were applied in the present study.
For the Hstholmen site this kind of interpreted and representative orientation data set is
not available. Orientation data set for the Hstholmen was composed as a part of the
present study. There are no exposed trenches at Hstholmen. Therefore, the data set of
the fracture orientations at Hstholmen is based on the data from the outcrops and five
deep boreholes (KR1 to KR5) (see Table 1). The data set used in the boreholes was a
combination of the core, dipmeter and borehole-TV measurements that was extracted
from the Posiva' s TUTKA-archive. The procedure applied in the combination of the
surface and borehole data is explained in Section 2.2.
4
Table 1. Combined data sets of the fracture orientations from the surface and boreholes
for different sites (for Kivetty, Olkiluoto and Romuvaara data is collected from Anttila
et. al. 1999b, 1999c, 1999d).
Site Surface data Borehole data
Hstholmen Outcrops ~1,~2,~3,~4,~5
Kivetty Trenches TK1, TK2 ~5,~13
Olkiluoto Trench TK2 ~1
Romuvaara Trench TK2 ~1
The fitting of the fracture orientation distributions is based on the combined data sets of
measured fracture orientations from the surface and boreholes. For Kivetty, Olkiluoto
and Romuvaara the combined data sets were composed by the Geological Survey of
Finland. The method they applied was following (e.g. Front et. al., 1998):
Select a compact subset of the fracture orientations. For each site this was based on
fractures from one or two boreholes and exposed research trenches on the surface. If
all the fractures would have been taken into account then there had been
considerable disproportion between the surface and borehole fractures. That would
have caused great difficulties in the scaling of the surface and borehole fracture data
sets.
Biases in the fracture orientation distributions caused by the geometric factors was
corrected by multiplying the number fractures by a factor 1/cos( a), where a is the
angle between the pole of the fracture and the borehole or the trench.
Before combination, the corrected trench and borehole data sets were scaled to the
same number of fractures by increasing the number of fractures in the smaller set.
For the Hstholmen site the surface data from the outcrops and from the boreholes
~1-~5 were combined by applying similar procedure as for the other sites was
applied. The major difference is in the combination of the data sets. They were not
scaled to the same number of fractures but to the same scanline length. The applied
procedure was following:
The bias caused by the geometrical factors was corrected. Plotting the fracture
orientations with a suitable projection on the stereoplot did this. The selected
projection adjusts the fracture orientation so that the number of fracture observations
for different angles between the fracture pole and the borehole or outcrop follows
the appropriate geometrical probability (appendix).
5
The fracture frequency in the boreholes seems to be much higher, almost three times
higher, in the boreholes than on the outcrops. This may indicate that the outcrops are not
representing the same rock as the boreholes. The boreholes sample the rock mass more
randomly than the outcrops. It is possible that the existence of an outcrop correlate with
some properties of the rock, like intensity of the fracturing. However, this cannot be
taken into account in the present study and it is simply assumed that the outcrop data
and borehole data are equally applicable. It should also be noted here that the combined
data set of the Hstholmen site is not crosschecked against the general geological data in
the same way as the other sites.
The combined fracture orientation distributions for different sites are presented 1n
Figures 1 and 2.
N 4.1
N
5.2
3.2
4.4
2.3 3.6
2.9
1.4 2.1
1.3
0.5
Figure 1. Combined orientation distributions for the Hstholmen (left) and Kivetty.
6
N %
5.2
4.4
3.6
2.9
2.1
1.3
0.5
Figure 2. Combined orientation distributions for the Olkiluoto (left) and Romuvaara.
Three fracture orientation sets were fitted to the corrected and combined orientation
distributions for each site. The fitting was performed by minimising the squared
difference between the measured orientation distribution and the summed total
distribution of the three orientation sets. All fitted orientation sets were Fisher
distributions. Fisher distribution is defined by Equation ( 1)
where (} is the angle from the axis of the mean orientation that is also the axis of the
symmetry of the distribution and yis the measure of the dispersion.
Three sets appeared to be suitable for all the sites. At Hstholmen, for example, the
cubic fracturing of the rapakivi-granite is clearly visible in the combined orientation
distribution (c.f. Figure 1). Similar features can also be seen at Kivetty, Olkiluoto and
Romuvaara. Although, at Olkiluoto the fractures dipping gently to the southeast are very
dominant. At Romuvaara the groups of the horizontal fractures and vertical from
northeast to southwest striking fractures are very concentrated. Vertical fractures that are
striking from northwest to southeast are dispersed to a couple of different subsets and
orientations of these fractures are clearly more dispersed than for the two other sets.
Parameters of the fitted orientation distributions are presented in Table 2 and fitted sets
are presented in Figure 3 and 4. The figures of the fitted sets are produced by generating
5000 fractures using data in Table 2 and then plotting the generated orientation
distribution. In some cases, like at Romuvaara, two of the orientation sets are very
concentrated around the mean orientation. It is possible that the rather high factor (i.e.
maximum of 10) applied to increase the number fractures in the less sampled areas of
7
the orientation distribution could have produced sharp peaks to the region of the "blind
zone". This in tum has result fracture sets with perhaps unrealistic small dispersion in
the orientation. For Romuvaara this problem becomes even more evident later when the
outcrop fractures are divided into the fracture sets for the trace length analysis
(Figure 8).
N 4.1
N
3.2
2.3
1.4
N 3.3
N
2.6
1.9
1.2
0.5
Fracture size may correlate with the orientation. The possible correlation was studied by
using outcrop data. Outcrop data included both fracture orientation and size information.
There are limitations in this analysis resulting from the nature of the surface data. First,
horizontal fracture sets observed in the boreholes are not well represented in the outcrop
data. Secondly, the size of the observation window on the surface is limited and it
truncates the long fractures making the sizes of the fracture traces more similar.
There seems to be some differences in the fracture sizes for fracture sets in different
orientations. The fracture trace lengths for different fracture orientations are presented in
Figures 5 and 6. The figures represent the mean fracture trace lengths in the 5% search
area of the stereonet. Based on the Figures 5 and 6 the differences in the fracture size for
different orientations are significant. However, it should be noted, that the number of
fractures and therefore also the confidence of the mean value varies depending on the
fracture orientation. For example, a couple of large horizontal fractures at Romuvaara
cause an impression that the horizontal fractures in general are large at Romuvaara.
To have a more quantitative estimate on the possible correlation between the fracture
orientation and fracture size the outcrop fractures were divided into the fitted fracture
sets. This was performed by simply assigning the individual fractures into fracture sets
where they, according to the fitted orientation distributions, most probably belong. In
Figures 7 and 8 there are presented how the outcrop fractures were divided into different
fracture sets. The fracture sets are given in Table 2.
The mean values of the samples from the same population are normally distributed if the
samples are non-biased. The variance of sample mean value is the population variance
divided by the sample size. This holds although the distribution of the whole population
is not known. The variance of the population was estimated by the variance of the
sample. This is a good estimate if the sample is large enough and for small samples it is
the best that we can get. Using this principle the mean trace lengths are compared for the
different fracture sets by calculating the 95% confidence ranges of the mean fracture
trace lengths for each set. This was done by using Equation (2)
r1 = flr -1.960" r
(2)
ru = flr + 1.960"r
where r1 is the lower and ru is the upper limit of the range, f.lr is the mean trace length of
the set and O"r is the standard deviation of the mean trace length of the set. The standard
deviation O"r is calculated from the standard deviation of the set by Equation (3)
(3)
The results (Table 3) show that only at Hstholmen there is difference in fracture size at
this confidence level. At Hstholmen the shortest fractures are the horizontal ones. The
longest fractures are vertical and striking from northeast to southwest. The difference in
the mean fracture size is about 3 m between these two sets. The mean size of the
fractures in the third set, vertical fractures striking from northwest to southeast, is in
between the two other ones. The very small number of fractures in the first set of the
Remuvaara can be seen in the wide confidence range. Taking the confidence range into
account it is not possible to separate the fracture size of the first from the others.
However, it should be noted that the basis of the calculating the confidence limits of the
mean fracture size follows from the assumed normal distribution of the sampled mean
values. Strictly, this is true only when the sample size is large enough and also the
sample variance gives better estimate of the population variance if the sample size is
large enough. Quantitatively, the given confidence limits might be in some cases just
allusive but they can be used in the judgement how important the differences in the
mean fracture size are.
Figure 5. Mean fracture size for fractures of different orientation at Hstholmen (left)
and Kivetty. The scale is in meters.
11
Figure 6. Mean fracture size for fractures of different orientation at Olkiluoto (left) and
Romuvaara. The scale isin meters.
N N
X ." ::,f
~ ~
X~
:>o 0
oO
:
% .. , .,:
-r..;:
X
{JO 0 @0 0 . 0
0 0
0 <68) ~ 0
" ~.. o o o o0
0 ... X
0
... X
.x
..
Figure 7. Distribution of the outcrop fractures into the fitted orientation sets for
Hstholmen (left) and Kivetty.
12
N N
:.
Figure 8. Distribution of the outcrops fractures into the fitted orientation sets for
Olkiluoto (left) and Romuvaara.
Table 3. Comparison of the fracture trace length data of different fracture sets. N is the
number of traces in set, a is the variance of the measured trace lengths in set, r1 is the
lower limit and ru upper limit of the 95% confidence range of the mean fracture trace
length rm.
Site N a[m] n [m] rm[m] ru[m]
set
Hstholmen 1 183 6.2 5.2 6.1 7
2 1806 7.6 7.4 7.8 8.1
3 1998 8.3 8.2 8.6 9
Kivetty 1 735 1.4 1.8 1.9 2
2 26 1.1 1.8 2.2 2.6
3 812 0.86 1.6 1.7 1.8
Olkiluoto 1 126 2.5 2.1 2.5 3
2 520 1.5 2.1 2.2 2.4
3 1439 1.7 2.2 2.3 2.4
Remuvaara 1 4 1.1 1.5 2.6 3.7
2 84 0.71 1.4 1.6 1.7
3 457 0.93 1.6 1.7 1.8
13
4. CONCLUSIONS
Fracture orientation distributions have been fitted for the Hstholmen, Kivetty,
Olkiluoto and Remuvaara site models. The fitted orientation distributions are based on
the combined surface and borehole data. Surface and borehole data have been combined
by the Geological Survey of Finland (GTK) for Kivetty, Olkiluoto and Romuvaara. The
combined data sets were also crosschecked by GTK to ensure that they agreed with the
general geological characteristics of the sites. Combined surface and borehole data was
not available for Hstholmen. As a part of the present study a combined data set of the
surface and borehole fractures was constructed for Hstholmen.
Fracture orientations are represented by three fracture sets for all sites. Each set is
characterised by Fisher distribution. Resulting fitting of the orientation distributions
show reasonable agreement between the corrected measured (corrected for the
observation biases) and fitted distributions.
Fracture sizes were. studied for by utilising the surface fractures. Based on the fracture
trace lengths and orientations it is concluded that there is no statistical implication for
correlation between fracture orientation and size, except for the Hstholmen site. At
Hstholmen the horizontal fractures seem to be shorter than vertical ones. However, it
should be noted that the horizontal fractures were not well represented at any of the sites
in the fracture size analysis. Especially at the Remuvaara site the number of horizontal
fractures observed on outcrops is too small for reliable statistical analysis.
14
5. REFERENCES
Anttila, P., Ahokas, H., Front, K., Hinkkanen, H., Johansson, E., Paulamki, S.,
Riekkola, R., Saari, J., Saksa, P., Snellman, M., Wikstrm, L. & hberg, A.,
1999a. Final disposal of spent nuclear fuel in Finnish bedrock - Hstholmen site report.
Posiva Oy, Helsinki, 1999. Report Posiva 99-08.
Anttila, P., Ahokas, H., Front, K., Hinkkanen, H., Johansson, E., Paulamki, S.,
Riekkola, R., Saari, J., Saksa, P., Snellman, M., Wikstrm, L. & hberg, A.,
1999b. Final disposal of spent nuclear fuel in Finnish bedrock - Kivetty site report.
Posiva Oy, Helsinki, 1999. Report Posiva 99-09.
Anttila, P., Ahokas, H., Front, K., Hinkkanen, H., Johansson, E., Paulamki, S.,
Riekkola, R., Saari, J., Saksa, P., Snellman, M., Wikstrm, L. & hberg, A., 1999c.
Final disposal of spent nuclear fuel in Finnish bedrock -Romuvaara site report. Posiva
Oy, Helsinki, 1999. Report Posiva 99-10.
Anttila, P., Ahokas, H., Front, K., Hinkkanen, H., Johansson, E., Paulamki, S.,
Riekkola, R., Saari, J., Saksa, P., Snellman, M., Wikstrm, L. & hberg, A.,
1999d. Final disposal of spent nuclear fuel in Finnish bedrock - Remuvaara site report.
Posiva Oy, Helsinki, 1999. Report Posiva 99-11.
Front, K., Paulamki, S. & Paananen, M., 1998. Updated lithological bedrock model
of the Remuvaara study site, Kuhmo, northeastem Finland. Posiva Oy,
Work Report 98-55. (ln Finnish with abstract in English).
A-1
This Appendix explains how the fracture orientations measured at Hstholmen were
corrected for the observation bias. Fracture orientations in the boreholes are biased
because the probability to see fractures perpendicular to the borehole is higher than for
the other fractures (Terzaghi, 1965). The same phenomenon affects also the orientation
distribution from the outcrops and tunnels, although the bias in the case of tunnel is not
as strong as in the case of boreholes.
The orientation distribution of the fractures is based on the density of the points
representing individual fracture orientations on the equal-area diagram. The observation
bias of the orientation distribution is corrected by modifying the standard equal-area
projection to take the observation probability into account. The probability to observe a
fracture of the orientation in question is obtained as the ratio between the projected
differential unit areas in the modified and equal-area projections.
Fracture orientation can be presented by giving the orientation of the normal vector to
the fracture plane. This orientation can be giving by two angles, trend and dip-angle.
The orientation can also be given as an intersection point on unit sphere and the normal
vector of the fracture plane. Bach normal vector have two intersection points on the
sphere which means that it is enough consider only half of the unit sphere. For a uniform
orientation distribution the intersection points on the unit sphere are also distributed
uniformly. Equal-area projection is a way to represent the points on the unit sphere in a
two-dimensional figure so that the uniform distribution points is preserved for uniform
orientation distribution.
Equal-area projection can be calculated in a following way, for example. Let us define
plunge angle B that is the angle from the horizontal plane. If examine the surface area of
the sphere that is defined all the points that have plunge angle larger than B. For the
sphere with a radius of p the surface are depends on the plunge angle 8 according to
Equation (A-1)
A( (} )= 2n p 2 ( 1- sin( (} )) . (A-1)
2
dA=- 2n p cos( B )dB . (A-2)
W e have to now calculate the projection that preserves the ratios of surface areas
between different plunge angles on two-dimensional plane. Line representing fixed
plunge angle is a circle on the surface of the sphere. Similarly they are projected to
concentric circles on two-dimensional plane. Only the function that relates the plunge
angle and radius in two-dimensional plot should be determined. This is done requiring
A-2
that the ratios of surface areas be preserved in projection. The area that is corresponding
to the orientations that have plunge angle larger than 0 is for the two-dimensional plot
simply the area of the corresponding circle (Equation (A-3))
(A-3)
The rate of change in the projected area of the two-dimensional plot can be expressed as
(A-4)
By making the rates of changes dA (Equation A-2) and dAp (Equation A-4) equal we
obtain
(A-5)
from which the radius r(O) can be solved. Integrating both sides of Equation (A-5) gives
For simplicity the radius of pole diagram is normalised so that r(O)=l. We must also
require that the area goes to zero when the plunge angle 0 goes to n/2, i.e., r(rc/2)=0.
This latter requirement merely means that the area of the lower hemisphere is also
normalised to the same area as the pole diagram area. Applying the normalisation to
Equation (A-6), we finally obtain an equation for the equal-area projection
r( B ) = ~1 - sin( B ) . (A-7)
The horizontal orientation, trend angle et>, does not influence the observation probability
and therefore it is not involved in the equations. This definition (Equation A-7) of the
equal-area projection is the same that has been given by e.g. Terzaghi (1965).
Applying the same line of reasoning as above it is easy to modify the projection in
Equation (A-7) so that also the fracture observation probability is taken into account.
Probability to observe a fracture is proportional to the sin(theta), where theta is the angle
between the fracture plane normal and borehole. It is now assumed that the borehole is
vertical. This is no restriction, because measured orientations can always be rotated so
that the borehole is vertical.
points in the equal-area plot is proportional to the probability to observe the fractures
(assuming that the underlying orientation distribution is uniform). The uniform
orientation distribution is reconstructed if the observation probability is taken into
account in the plotting projection. This can achieved if the projected area changes by the
observation probability. The observation probability is proportional to the cosine of the
angle between the borehole and the normal of the fracture plane. Therefore, the
observation probability changes as the sine of the angle 8 (plunge angle 8 is complement
to the angle between vertical borehole and fracture normal vector). This can be written
in the following form
dA
_P =sin(B) (A-8)
dA
(A-9)
(A-10)
Again, normalising the pole diagram to unit circle, it can be seen that the constant in
Equation (A-1 0) must be zero and the corresponding reference sphere must also be a
unit sphere, i.e., p=l. Substituting these changes in Equation (A-10), we finally obtain
the projection
r( () ) = cos( () ) . (A-11)
Projection (A-11) is now constructed so that plotting uniform fracture orientations that
are sampled from borehole give uniform distribution also on this modified equal-area
plot. Benefits of this method are that there is no need for generation of new fractures and
an immediate view is got how the unbiased orientation distribution looks like.
A similar correction as for the borehole fractures can also be applied for the outcrop
fractures. An outcrop is now assumed to be a horizontal plane. The probability of
observing a fracture is proportional to the complement of the angle used in the case of
borehole. Therefore, the observation probability changes to cos(8) instead of sin( 8) for
the borehole fractures. Applying this to Equation (A-8), we obtain
A-4
dA
_P=cos(8) . (A-12)
dA
(A-13)
Again, integrating both sides of the Equation (A-13) following equation is obtained
(A-14)
Normalising the projection to the unit circle (r(0)=1) and requiring that r(Jr/2)=0, the
equation of the projection finally becomes:
It should be noted that the angle () in Equation (A-15) should be given in radians.
For the fractures observed on the tunne! wall the projection can be constructed from the
projections (A-11) and (A-15). On the tunne! wall the probability to observe fractures
parallel to the tunne! axis is significantly higher than for the borehole. If the mean radius
of the fractures is R and the radius of the tunne! is L, then the probability to observe
fractures parallel to the tunne! can be approximated roughly by the quotient RI(R+L).
This approximation is exact in two-dimension and if the spatial density of the fractures
uniform.
The probability Rl(R +L) can be used to form an interpolation formula for the modified
equal-area projection. It is required that the projection reduces to the ordinary equal-area
projection if the tunne! is very large compared to the fracture size, and on the other
hand, it should give the projection of the borehole fractures in Equation (A-11) if the
fractures are much larger than the tunne!. These requirements can be fulfilled with a
projection in Equation (A-16)
r( 8) = cos( 8)
2
(-R-) +
R+L
(1- ~)( 1- sin( 8))
R+L
(A-16)
Terzaghi, R. D., 1965. Sources of Error in Joint Surveys. Geotechnique 15, 287-304
(1965).