You are on page 1of 35

Engineering Geology, 17 (1981) 19--53 19

Elsevier Scientific Publishing Company, Amsterdam - - Printed in The Netherlands

CORRELATION OF ROCK MASS CLASSIFICATION PARAMETERS


OBTAINED FROM BORECORE AND IN-SITU OBSERVATIONS

I.S. CAMERON-CLARKE and S. BUDAVARI


Engineering Geologist, Johannesburg (R.S.A.)
Department of Rock Mechanics, University of the Witwatersrand, Johannesburg (R.S.A.)
(Received July 17, 1979; accepted August 13, 1980)

ABSTRACT

Cameron-Clarke, I.S. and Budavari, S., 1981. Correlation of rock mass classification
parameters obtained from borecore and in-situ observations. Eng. Geol., 17: 19--53.

The paper descTibes an investigation carried o u t to establish correlation trends between


rock mass classification parameters derived from borecore and in-situ observations. The
purpose of this study was to determine the reliability of classifying rock masses for
engineering purposes from borecore data alone.
Rock mass classification parameters were obtained from three tunnels excavated in
widely differing geological environments; two in Johannesburg and one in the Cape
Province of South Africa. These parameters from the borecore and in-situ measurements
were used separately to classify the rock masses encountered. Both the Geomechanics
and Q Classification Systems were employed in determining the quality of rock masses.
The paper describes the geological environments encountered, the methods of obser-
vation used and the results obtained. From the comparison of the in-situ and borecore
classification results, conclusions are reached and discussed. Some comments are also
offered regarding the applicability of the two classification systems under the conditions
specified.

INTRODUCTION

The concept o f determining the engineering qualities of rock masses by


means o f their classification has been used for a number of years. During
this time, several classification techniques have been developed, ranging from
earlier, less quantitative, to presently accepted, sophisticated methods. The
Geomechanics and Q Classification Systems, developed by the South
African Council of Scientific and Industrial Research and the Norwegian Geo-
technical Institute, respectively, represent two of the more recently introduced
techniques of engineering rock mass classification. Both of these techniques
were developed mainly from reported case history studies and the in-situ
examination of completed underground excavations. The purpose of both
techniques is to give a quantitative assessment o f the rock mass quality in
terms of several well-defined, engineering geological parameters. The ultimate
aim of these classification systems is to provide the engineer with basic data

0013-7952/81/0000--0000/$02.50 © 1981 Elsevier Scientific Publishir/g Company


20

to enable him to design structures in rock and to predict the support


measures required under the prevailing conditions. Both classification systems
have invoked considerable interest in the appropriate professional circles and
their effectiveness in geotechnical investigations has been widely discussed
(Houghton, 1975; Payne, 1976; Bieniawski, 1976a; Barton, 1976a). The
study presented in this paper is~a further contribution to the evaluation of
the applicability of these two rock mass classification systems under the
conditions specified.
In order to be useful in the design and preconstruction phases of an under-
ground excavation, the parameters used in an engineering rock mass classifi-
cation system must be obtainable during the preconstruction site investigation.
In areas where surface rock exposures are minimal or non-existent, or where
the engineering qualities of the rock masses at the depth of excavation are
expected to be significantly different from those of the near-surface rocks,
core drilling may be the only exploration technique suitable for providing
data for rock mass classification purposes. The required classification param-
eters in the preconstruction phase, therefore, may have to be obtained
from borecores alone. It is to be expected that the reliability of the classifi-
cation parameters obtained from borecore data may be somewhat lower than
those obtained from the in-situ study of exposed rock surfaces at the same
locality. Consequently, rock mass classifications carried out using borecore
data alone are also expected to be less reliable:
Only a few attempts appear to have been made to compare borecore pre-
dictions of rock mass quality ratings and support requirements with related
values obtained from in-situ measurements. It follows from the lack of such
investigations, that only a limited amount of information is available on the
degree of reliability of borecore data in engineering rock mass classifications.
The most noteworthy of the relevant publications are those of Bergman
(1974) and Barton (1976a). Bergman studied the reliability of five methods
of subsurface investigation. He found that integral sampling, as developed by
Rocha (1971), and television examination of boreholes were the most reliable
methods for rock mass quality evaluation. Conventional core drilling was
found to provide ambiguous information which could result in under-rating
the rock mass properties. Barton determined the rock mass qualities of a
massive biotite gneiss in which an underground power house was excavated.
His results showed that the average in-situ Q values were approximately half
the average borecore Q ratings.
The purpose of the present paper is to evaluate the reliability of using
borecore data only for classification purposes under the conditions specified.
The degree of reliability of the data is inferred from the comparison of the
appropriate classification parameters obtained from the same locations by
both borecore and in.situ observations. In the correlation employed, the in-
situ data are accepted as the basis of comparison. Rock mass classification
parameters used in the study were obtained from three tunnels excavated in
widely differing geological environments. Both the Geomechanics and the Q
Classification Systems were employed in quantifying the relevant parameters
and the quality of rock masses.
21

DESCRIPTION OF THE GEOLOGICAL ENVIRONMENTS

The three tunnels from which data were obtained for this investigation are:
the Bushkoppies Tunnel, situated approximately 15 km to the south of
Johannesburg; the Delvers Street Tunnel, located in the central city area of
Johannesburg; and the Du Toitskloof Pilot Tunnel, situated approximately
60 km northeast of Cape Town. The locations of the three tunnels are shown
on the appropriate regional geological maps in Figs. 1 and 2. The Bushkoppies
Tunnel is to be used for the conveyance of sewage, and the Delvers Street

+ + + .~.

!. ''" (

o lo 20 3o km

LEGEND

Karroo Supergroup
Transvaal Supergroup
V Ventersdorp Supergroup
Upper
~- Witwatersrand Supergroup
Lower_J
Ancient Granite 8, schist remnants

/f Fault
Regional dip direction
Delvers Tunnel

o" Buskoppies Tunnel


Fig.1. Regional geological map o f the Johannesburg area showing the locations of the
Bushkoppies and Delvers Street Tunnels.
22

LEGEND

* Tertiary to Quaternary

; Cape Supergroup
[~ KlipheuwelFormation
~i Cape Granite
Malmesbury Beds

f Fault

D0 Du Toitskloof Tunnel

Fig.2. Regional geological map of the Cape Peninsula area showing the location of the Du
Toitskloof Pilot Tunnel.

Tunnel was planned to house feeder and distribution cables for a new central
electricity substation. The Du Toitskloof Pilot Tunnel forms part o f the in-
vestigation for a major highway tunnel between Paarl and Worcester. The
ultimate aim of the scheme is to have two tunnels, 36 m apart, each carrying
unidirectional traffic. All three tunnels have inverted "U" shaped cross-
sections with equal horizontal and vertical excavated dimensions o f about 3 m.
23

The Bushkoppies Tunnel

The Bushkoppies Tunnel has a total length of a b o u t 6.5 km and the depth
of cover over the tunnel varies from 5 m to 65 m, Fig.3. It follows an approxi-
mate northwest--southeast route through the lavas o f the Ventersdorp Super-
group which display a general east--west strike direction and dip at a low
angle to the south. The rocks are variable from soft, completely weathered
and closely jointed material at the east portal, to very hard almost massive
rock in other sections along the tunnel. Most o f the tunnel was excavated in
hard, widely to closely jointed rock with three or more joint sets and occasional
clay-filled faults and fault zones. Tests on intact rock specimens, obtained
from these rocks yielded a mean uniaxial compressive strength of 367 MPa
(Brink and Associates, 1974). The twelve sample sites studied in this tunnel
are shown in Fig.3.
According to the nature o f the prevailing geological conditions, the tunnel
was divided into four broad sections: (a) from the east portal at 150 m to
624 m; (b) from 624 m to 3500 m and 4000 m to 5150 m; (c) from 3500 m
to 4000 m; and (d) from 5150 m to 6688 m. These sections can be described
qualitatively as follows.

Section (a)
The weathered rock exposed at the east portal was typical of the rock
encountered along the first 100--150 m length of the tunnel and is due to
normal weathering processes apparent to a b o u t 10 m below the ground surface.
Fresher rock was encountered as the excavation advanced, b u t the presence
of numerous clay-filled joints contributed to the generally unstable rock mass
condition. Partially weathered, closely jointed rock, apparently related to a
fairly extensive fault zone, was encountered near the end of this section.
Heavy support was o f t e n required throughout this part of the tunnel and
mesh-reinforced shotcrete, together with steel arches, were used.

Section (b )
This section o f the tunnel typified the most characteristic rock conditions
encountered. With the exception of localized unstable zones caused b y the
presence o f minor faults, the rock was hard and widely jointed with closed
or cemented joints. Minor water flows were noted from some of the faults.
No systematic support was needed, however, spot bolting was o f t e n carried
out.

Section (c)
The rock along this section was hard, unweathered and widely jointed, but
had a more blocky appearance than was noted in section (b). Several major
fault zones, containing quite badly sheared rock, from which moderate flows
of'water were noted, occurred within this section. Systematic rock-bolting at
1-m to 1.5-m intervalswas used for support, Along one short section, in
addition to rock bolts, wire mesh was introduced to prevent the fall-outof
small blocks from the roof.
t~

o i o
n~ t~ nn 0 •

• , M tl M .
; ! , =, . ,. ,. , . ~., ~ , , ,..~ ~ ,~n . , , ._-_ ". 0J
6000 SO00 4000 3000 ""'20'00 1000 150
il
LEGEND 0 , 1000rn

Scale
V e n t e r s d o r p Lav~.

Diabas.e

S h e a r Zones

BUSHKOPPIES TUNNEL

Fig.3. Prof'de along the Bushkoppies Tunnel.


25

Section (d)
This section was characterized b y numerous intersecting clay filled joints
and a number o f quite major faults. The rock was hard and widely to closely
jointed. Weathering, particularly adjacent to the larger joints and minor
faults, became more pronounced as the depth of cover decreased westwards.
Moderate water flows were n o t e d in places. Owing to the tendency for the
rock mass to loosen as a result of the clay joint fillings, it was often neces-
s a l t to install quite closely spaced rock bolts (0.5 m in places). Along
particularly bad sections shotcrete and steel arches were used.

The Delvers Street Tunnel

The Delvers Street Tunnel has a total length of a b o u t 1.2 km and the
cover depth to its crown varies from 2.5 m to 11 m (Fig.4). The tunnel
passed through alternating arenaceous and argillaceous strata of the Govern-
ment Reef and J e p p e s t o w n Formations of the Witwatersrand Supergrnup and,
at its northern end, through a sheared fault zone. The regional strike direction
of the strata is east--west and dips are vertical or very steep towards the
south. The tunnel was divided into three sections based on differences in the
general rock mass properties with respect to the tunnel, the excavation
method, and the distance behind the face at which the initial steel arches
were installed. These are: (a) from the north end of the tunnel to 350 m;
(b) from 350 m to 850 m; (c) from 850 m to 1180 m.

Section (a )
This section of the tunnel was orientated parallel to the regional strike
and dominant joint directions, and was excavated in the vicinity of the
sheared fault zone mentioned above. The steep dip, and chlorite and clay
coated joint surfaces resulted in unstable sidewall conditions. The excavation
method consisted of drilling and blasting, and the steel arches were installed
within a metre o f the face.

Section (b )
The strike of the dominant structures was perpendicular to the tunnel
axis along this section, resulting in more stable rock conditions. The excava-
tion m e t h o d was the same as in the previous section and the arches were
installed approximately 3--5 m behind the face.

Section (c)
This section was located in colluvium and residual soils derived from the
various geological horizons intersected. Hand-held pneumatic diggers were
used for Iexcavating this section and arches were installed within a metre of
the face. In places, unstable sandy material necessitated the instaUation o f
continuous Lagging in the roof.
t~
O~

e~ r,~ co ~t

Kruis Sma{
t
Yon Wetlig h "u~ Jeppe K~rk Pritchard I I
st. st. st. • ''1 I st. st. I st President L (
| I ~ ~ I ~'~" ~ Market
I / [11. I~ I , ~ Commissioner~ 20

LO 0-~

? loo 200 300 0 " " .///,

LEGEND

CoUuvium

Quartzite
Sandy Quartzite
,o lOO 2oo=
Argillaceous Quartzite with Shale & Sandstone
Scale
Arenaceous Shale Weathered
Shale
Diabase

It!~llll Fault Zone


DELVERS TUNNEL

Fig.4. Profile along the Delvers Street Tunnel.


27

The Du Toitskloof Pilot Tunnel

The Du Toitskloof Tunnel is almost 4 km long and its depth of cover


varies from 10 m to 20 m at the portals to approximately 700 m below the
crest of the Klein Drakenstein Mountains (Fig.5). The tunnel passes through
Cape Granites and Table Mountain Sandstones. The t w o lithological groups
are separated by the Du Toitskloof Fault which has a d o w n t h r o w o f u n k n o w n
magnitude on its eastern side, and is located a b o u t 600--700 m in from the
east portal. At the time when this study was undertaken, the tunnel excava-
tion had reached a b o u t 2200 m from the east portal and 1 0 0 - 1 5 0 m from
the west portal. The data presented were obtained from six sample sites
within the eastern tunnel. Consequently, the general comments made below
are based largely on observations made along that section of the eastern
tunnel excavated up to the time of completion of the study.
The borecores examined for this project were obtained from ten boreholes
drilled horizontally or vertically from within the tunnel. Each core was
evaluated separately and no effort was made to combine data when both
horizontal and vertical boreholes had been drilled from the same site.
The Table Mountain Sandstones are generally moderately hard, to hard
rock and jointed with iron staining and occasional clayey infillings on joint
surfaces. The rock is soft and brecciated in the vicinity of the Du Toitskloof
Fault. Shotcrete and rock bolts were used to stabilize some sections of the
tunnel in this rock type and steel arches with continuous lagging were
necessary along the worst sections. Tests carried o u t on specimens of these
rocks yielded uniaxial compressive strength values between 25 and 274 MPa
(Van Niekerk et al., 1978).
The Cape Granite provided good tunnelling conditions when n o t associated
with major faults. It was hard and both massive and jointed. Generally, no
support measures were required under these conditions. In the vicinity of
major faults the rock was soft and highly sheared with poor self-supporting
characteristics. Steel arches, continuous lagging and concrete backfill were
necessary to support this t y p e of material while in places shotcrete had to
be a p p l i e d a t the face immediately after the blast. Mean uniaxial compressive
strength values of 105 MPa were obtained for the unweathered and unsheared
granite (Van Niekerk et al., 1978).

METHODS USED TO OBTAIN THE REQUIRED CLASSIFICATION PARAMETERS

The methods used to obtain and evaluate the parameters required b y the
Geomechanics and Q Classification Systems were adapted from existing
techniques. The main consideration was to obtain the relevant information
as simply and accurately as possible w i t h o u t unnecessary duplication of
effort. Since a number of the parameters required for both classification
systems are related to the jointing in the rock, a m e t h o d o f joint survey had
to be adopted. Scan line surveys, which record information a b o u t joints
intersecting a specific line, were suitable for both borecore and in-situ
t~
(2O

..%

-12 c~ 12
•11 ~ _ o_.a 117-
-10 ................... 10-
.9 _ ~ ~E~E! E~E ~'~ ~E ;E .-:'~._ c~ -~ 9-
"8 "~ ~ , ============================== 8-
7 o .,,,.j~. ~. , + ~ o
7-
6-

, , • T T -, ~ t t + \

Dut6 Dut5 Dut4 Dut3 Dut2 Dutl

LEGEND

Talus
0 1000m
Table Mountain Sandstone I i I

Cape Granite Sca~e

Fautt breccia

Dut~ Site Location


DU TOITSKLOOF TUNNEL
Modified after'~OCKCROFT&LOUDON. 1976

Fig.5. Profile along the Du Toitskloof Tunnel.


29

measurements and were, therefore, employed throughout the present study.


The Rock Quality Designation (RQD) values were determined using the
methods proposed by PalmstrSm (cited in Barton et ah, 1974) and Priest
and Hudson (1976). The latter method could only be used in conjunction
with detail line joint surveys and, for reasons to be described later, was not
employed at the Delvers Street Tunnel nor at certain locations in the Du
Toitskloof Tunnel.

Borecore measurements

The method used to record the borecore data was based on those
suggested by Franklin et al. (1970), Rankilor (1974) and the South African
Section of the Association of Engineering Geologists (1976). The underlying
principle considered when selecting the method was, that the core log should
present, as closely as possible, a true picture of the rock core with a minimum
amount of description and interpretation being used. The object was to
enable the rock mass to be classified by anybody from the core log alone,
without their judgement being affected by the unintentional bias of the
original compiler of the log.
This method of logging was used only for selected sections of cores. In
most situations it would only be necessary to undertake a detailed assess-
ment of the jointing in the core for a limited distance above and below the
planned excavation. The remainder of the core could be recorded in more
general terms. The Point Load Strength Index, determined by using an
Engineering Laboratory Equipment test machine, was used to assess the
strength characteristics of the cores.
In most cases the available cores were in reasonably good condition. At
some of the sites in the Bushkoppies Tunnel, however, a suitable core length
could not be obtained from the roof strata owing to losses during storage. It
was therefore assumed that the available core was representative of the rock
mass at the particular site. This was considered to be a reasonable assumption
in view of the generally homogeneous nature of the rock mass at this
tunnel.

In-situ m e a s u r e m e n t s

The in-situ data were obtained using a line joint survey method adapted
from that proposed by Piteau (1970) and the information was recorded on
joint survey data sheets. The in-situ measurements were carried out as close
to the underground positions of the relevant boreholes as possible. At the
Bushkoppies and Delvers Street Tunnels, the boreholes were drilled from
surface along and adjacent to the centre lines of the planned excavations. At
the Du Toitskloof Tunnel, the holes were drilled from within the tunnel.
The detail line joint survey method was used only at the Bushkoppies and
Du Toitskloof Tunnels. In the former tunnel the survey lines were a standard
length of 10 m, centred as closely to the intersection points of the various
30

boreholes as possible. At the Du Toitskloof Tunnel, the joint measurements


were made along the walls of drilling recesses cut into the sides of the tunnel.
It was, therefore, possible to obtain measurements along two directions
orientated approximately perpendicular to each other. The survey line lengths
varied with the dimensions of each recess, but an effort was made to use
similar lengths for both survey directions at a particular locality. Owing to
access problems at two of the sites examined in this tunnel, joint survey data
previously obtained from adjacent localities by Van Niekerk et al. (1978),
were used for the in-situ rock mass classification.
The limited space and continual installation of steel supports, together
with the simultaneous construction of the concrete lining, precluded the use
of the line joint survey in the Delvers Street Tunnel. The assessment of classifi-
cation parsmeters was, therefore, based on visual appraisals and approximate
measurements.

Processing the borecore data

The procedure listed below was used to assign values to the various param-
eters specified by the Geomechanics and Q Classification Systems.
(1) The number of joints with similar inclination angles with respect to the
core axis were recorded and corrected for orientational bias.
(2) The results of step (1) were plotted as histograms and the numbers of
joint sets estimated for the rock mass at each borehole.
(3) The average spacing of the joints in each set was estimated.
(4) The number of joints in each set with similar surface roughness and
alteration characteristics were recorded.
(5) The most typical surface roughness and alteration characteristics of
each joint set was estimated.
(6) Data forms were compiled for each classification system, showing the
conditions considered to be representative of the rock mass at each borehole
location.
(7) Calculation sheets were compiled and "rock mass rating" (RMR) and
Q values were evaluated.

Processing the in-situ data

A similar approach to the one described above was used to evaluate the
in-situ joint survey data. The steps followed are listed below.
(1) The number of joint sets were estimated with the aid of a stereogzaphic
projection.
(2) The field survey data was then grouped into the respective sets as
estimated in step (1).
(3) The intersection angles between the survey line and the strike direction
of each joint set were determined and the true joint spacings calculated
after correcting for orientational bias (Terzaghi, 1965). The RQD values
were also determined.
31

(4) The influence of each joint set on the tunnel stability was estimated.
(5--8) Data forms and calculation sheets were compiled for each classifi-
cation system, as for the borecore data.

PRESENTATION OF RESULTS

The results derived from the borecore and in-situ observations and
presented in Tables I and II are the net output for each classification
system. For the Geomechanics System these are the RMR-value, rock class,
stand-up time and active unsupported span of each sample site. The latter
two values were estimated from the diagram presented by Bieniawski (1976a).
Owing to the scatter of data shown on this diagram, a median line was
constructed through the zone representing the limits of South African ex-
perience (Fig.6). The support recommendations for each site are also given.
The results for the Q System are the Q value, limiting span, stand-up time,
and support categories for each site. The limiting span was calculated from

TABLE I

Rock classification ratings for the Geomechanics Classification System

Sample IAthology Boxecore r e s u l t s In-situ r e s u l t s


site
number RMR rock stand-up unsup- RMR rock stand-up unsup-
class time ported class time ported
span (m) s p a n (r~

Bos 1 Lava 44 3 2--3 days 3 65 2 5--6 wks. 6.5


Bos 2 Lava 75 2 6--8 mths. 8 75 2 6 - - 8 mths. 8
Bos 3 Lava, diabase 88 4 1 day 2.5 74 2 6---8 m t h s . 8
Bos 4 Diabase 72 2 3--5 mths. 7.5 81 1 12--18 mths. 9
Bos 5 Lava 74 2 6--8 mths. 8 84 1 2--3 yrs. 9
Bos 6 Lava 77 2 9--11 mths. 8.5 82 1 1--2 ym. 9
Bos 7 Lava 58 3 2--4 wks. 5.5 75 2 6-8 mths. 8
Bos 8 Lava 68 2 2---4 m t h s . 7 71 2 4--5 mths. 7
Bos 9 Lava 80 2 12--15 mths. 8.5 79 2 12--15 mths. 8.5
Bos 10 Lava 45 3 3---4 d a y s 3.5 70 2 3--4 mths. 7
Bos 11 Lava 57 3 2--3 wks. 5 84 1 2--3 yrs. 9
Bos 12 Lava 68 2 2--4 mths. 7 87 1 4 yrs. 9
Del 1 A~gil]. q u a r t z i t e 28 4 5--10 h 2 41 3 1--2 days 3
Del 2 A~gill. q u a r t z i t e 51 3 6--7 days 4 32 4 8--10 h 2
Del 3 Quartzite with 40 4 1--2 days 3 43 3 2--3 days 3
ArgilL h o t .
Del 4 Quartzite 66 2 2--3 mths. 6.5 74 2 6--7 mths. 8
Del 5 Quartzite shale,
siltstone 33 4 10--12 mths. 2 51 3 6--7 days 4
Q u a r t z i t e shale,
Del 6 siltstone 36 4 15--20 h 2.5 44 3 2--3 days 3.5
Dut I-H Quartzitic sandst. 56 3 2--3 wks. 5
Dut I-V Quartziflc sandst. 47 3 4--5 days 4 61 2 3--4 wks. 6
Dut 2-H Quartzitic sandst. 50 3 5---7 days 4
Dut 2-V Quartzitic sandst. 54 3 1--2 wks. 4.5 70 2 3--4 mths. 7
Dut 3-V Quartzitic sandst.
weath, and brecc. 30 4 8--10 h 2 44 3 2--3 days 3.5
Dut 4-H Granite sheared 35 3 4--5 days 3.5
Dut 4-V Granite sheared 46 3 4--5 days 37 4 20--24 h 2.5
Dut 5-H Granite 63 2 5--6 wks. 6 46 3 3--4 days 3.5
Dut 6-H Granite 69 2 3--4 mths. 7
Dut 6-V Granite 74 2 6--7 mths. 8 81 1 1--2 yrs. 9
32

TABLE H

Rock classification ratings for the Q Classification System

Sample Lithology Borecore results In-situ results


site
number Q support stand-up limiting Q support stand-up limiti
category time span (m) category time span

Bos 1 Lava 0.07 34 1 day--1 wk. 1.1 0.14 29 1 wk.--I mth. 1.5
Bos 2 Lava 2.38 21 I0 yr. 4.5 6.06 17 I0 yrs. 6.6
Bos 3 Lava, 0.02 34 1 h 0.7 1.23 21 5---10 y r s . 3.5
dlabase
Bos 4 Diabase 26.67 13 10 yrs. 11.9 11.08 13 10 yrs. 8.4
Bos 5 Lava 6.26 17 10 yrs. 6.7 32.82 13 10 yrs. 12.9
Bos 6 Lava 0.33 29 1--6 mths. 2.1 6.66 17 I0 yrs. 6.8
Bos 7 Lava 0.13 29 Iwk.--Imth. 1.4 0.98 25 5--10yrs. 3.2
Bos 8 Lava 0.80 25 6 . 1 0 yrs. 2.9 2.91 21 I0 yrs. 4.9
Bos 9 Lava 200.00 5 10 yrs. 26.6 16.66 13 10 yrs. 9.9 ~
Bos 10 Lava 0.63 25 1--5 yrs. 2.7 0.83 25 5-10 yrs. 3.0
Bos 11 Lava 0.13 29 1 wk.--1 mth. 1.4 0.44 25 6.12 mths. 2.3
Bos 12 Lava 2.67 21 10 yrs. 4.7 2.0 21 10 yrs. 4.2
Del 1 Argill. 0.08 33 1 day--I wk. 1.2 0.58 25 1--2 yrs. 2.6
quartzite
De| 2 Argill. 0.16 29 1 wk.--I mth. 1.6 2.05 21 I0 yrs.
quartzite 4.3
Quartzite
Del 3 with 0.32 29 1--6 mths. 2.0 2.99 21 10 yrs. 5.0
argill, h o L
Del 4 Quartzite 1.76 21 I0 yrs. 4.0 4.75 17 I0 yrs. 6.0
Quartzite
Del 5 s h a l e and 0.012 33 1 h 0.5 1.02 21 5-10 yrs. 3.2
siltstone
Quartzite
Del 6 s h a l e and 0.27 29 1--6 mths. 1.9 0.22 29 1 wk.--1 mth. 1.7
siltstone
Dut 1-H ~aartzitic 1.20 21 5 - 1 0 yrs. 3.4
sandstone 1.20 21 5-10 yrs. S.4
D u t 1-V Quartzittc 0.90 25 5--10 yrs. 3.1
sandstone
Dut 2-H Quartzitie 2.04 21 10 y r s . 4.3
sandstone
1.09 21 5 - 1 0 yrs. 3.3
D u t 2-V Quartzitic 0.41 25 6 - 1 2 mths. 2.2
sandstone
Quartzitic
sandstone
D u t 3-V 1.1 21 5--10 yrs. 3.3 1.02 21 5--10 yrs. 3.2
w e a t h , and
brecc.
Granite
D u t 4-H 1.7 21 10 yrs. 4.0
sheared
0.47 25 6--12 mths. 2.4
Granite
Dut 4-V 0.09 33 1 day--1 wk. 1.2
sheared
D u t 5-H Granite 7.70 17 10 yrs. 7.2 9.74 17 10 yrs. 8.0
Dut 6-H Granite 89.70 9 10 y r s . 19.3
33,32 13 I0 yrs. 13.0
D u t 6-V Granite 33.20 13 10 yrs. 13.0

the relationship proposed by Barton et al. (1974). Similarly the stand-up


time was estimated from the diagram given by Barton (1976b).
To evaluate the results obtained, each classificationsystem was firstcon-
sidered separately. Comparisons were made between the in-situand borecore
classificationvalues and, when applicable, between the in-situand borecore
ratings for individual parameters. Since rock strength data were only obtained
for the borecores, this parameter could not be used for the purpose of com-
parison. The joint water data from the drilling records were insufficient to
33

lOmin 1 hour lOhourslda l w e e k 1month 6mon. 1 cear 5 10 20


70 70
50- ~ 50
t,O /.0
30 D 3o
2o '~--- 20
~s !s n ].---"" is
10"
8 -8

6 4 t n -6
5 ~ ~ , i , ~0 ,.~ -5
"" 80
3 -3

2 2o~ 1.2

£~~ ¢ I 1,1
o,.= .q5
1 10 10' D3 1

STAND UP TIME, hours


Fig.6. Geomechanics Classification System; median line used to estimate the stand-up
time and unsupported span.

allow a meaningful comparison between the borecore and in-situratings in


either classification system.
The results are presented graphically in the firstquadrant of a rectangular
coordinate system with the in-situ and borecore values along the abscissae
and ordinates, respectively. The points plotted are compared with the ideal
linear relationship of borecore values equal to in-situvalues, depicted by a
straight line through the origin with a 45 ° slope angle. In order to indicate
the dispersion of data measured, lines were constructed at equal distances
on each side of the ideal line defined above. These are analogous to the
standard error of estimate about a regression line and are drawn parallel to
the ideal line. Since the two lines in the present case are associated with
certain limiting values of the rock classes or parameter categories for the
appropriate classificationsystems discussed, they will be referred to as limits.
The magnitude of limits for each rock class or parameter category was
selected as described in the following sections of this paper.
34

The Geomechanics Classification System

Rock mass rating


The borecore and in-situ R M R values for each site are plotted in Fig.7. As
can be seen from this illustration, the selected values for the limits are ± 19.
This figure was selected after considering the application of the RMR values
in defining the various rock mass classes. Since the accepted difference
between rock mass classes is 20, a difference of up to 19 between related
borecore and in-situ R M R values can be tolerated without yielding different
rock classes. Results plotting within these limits will, in addition, never
differ b y more than one rock class.
In Fig.7, 82% of the results lie within the selected limits. This implies that,
notwithstanding the errors due to the relatively small number of observations,
there is an 82% probability of a borecore classification of a rock mass being
correct, or in error by only one rock class.
The histogram, presented in Fig.8, shows the frequency of the n u m b e r of
rock class differences obtained for each set of related borecore and in-situ
R M R values. It also indicates whether the borecore rock classes are equal to,
greater, or less than the in-situ rock classes. The results shown in this diagram
differ from those implied in Fig.7, in that no consideration has been given to
the previously selected limiting values. In some instances, therefore, bore-
core and in-situ RMR values differing by more than 19, b u t still yielding
only one rock class difference, have been included. This diagram shows that
18% of the borecore results are the same as the in-situ results, 61 and 14%
are one class lower and higher, respectively, and 7% are t w o classes lower.
Differences exceeding this were n o t recorded.

Joint spacing
The limits selected for this parameter, as well as for the other parameters
for the Geomechanics Classification System, were based on the minimum
t00

go
//~
80
/" A /"
/
70

~ 60
m
~ so
40

20 / ~+~ / A Bos

10 ® Dut

17 2'0 3'o ~'o s'o io 7'0 8b 9'o 16o


RMR (INS)

Fig.7. Geomechanics Classification System; borecore and in-situ RMR values.


35

t=1
14.
75%

50%

S=/o

-3 -2 -1 0 1 2 3
[~IFFERENCE BETWEEN INS 8, BC ROCK CLASSES

Fig.8. Oeomechanics Classification System; histogram showing the frequency of rock class
differences between the l~recore and i.-situ ~ values.

difference between the ratings for each category. This approach was used
because the difference between the ratings for the various categories is not
always constant for a particular parameter, and it is important to assess the
borecore and in-situ ratings for all rock conditions.
The in-situ and borecore ratings for the joint spacing are presented in
Fig.9. 71% of the points lie within the selected limits. When considering all
the points with respect to the ideal line, there is a tendency for the borecore
ratings to be lower than the in-situ ratings. 64% of the points lie below the
ideal line, 25% plot above it and 11% fall on it.

30 //~

J ~ Q // /- / / / "
20 "1- 20

C " o
C J:/. j j:
10 ,,, ,," ,-// 0 A BOS ~ ' "
.," -I- D e l ,'" -I- ,,'" -I- Del
," ~ , o Dut 6 " Dut
~ // (BC =,~(i NS) * 5

/ 0 t:
5 lo 2'o 2's 3'o 6 ~ 2"o ~s
Rating (INS) Rating (INS)
Fig.9. Geomechanlcs Cla~fication System; borecore and in-situ joint spacing rating
values.

Fig. 10. Geomechanics Classification System; borecore and in-situ joint condition rating
values.
36

Joint condition
The ratings for this parameter are presented in Fig.10. 75% of the points
fall within the selected limits and 29% plot on the ideal line. 21 and 50%
lie above and below this line, respectively. These results suggest a slightly
better correlation between the borecore and in-situ ratings. A tendency for
the borecore ratings to be lower than the in-situ ratings is again apparent.

Join t orien ration


The ratings for the joint orientation are presented in Fig.ll. Only 61% of
the points fall within the limits selected for this parameter. With respect to
the ideal line, 14% of the points plot on the line, whilst 18 and 68% lie
above and below it, respectively. The distribution of points indicates that,
with respect to the ideal relationship, there i s n o real correlation between
the borecore and in-situ ratings.

Q ClassificationSystem

Q values
The borecore and in-situQ values are plotted in Fig.12. Owing to the
logarithmic intervalsdefining the rock classes for this classificationsystem,
it was not feasible to use a constant value for the limits as was done for the
Geomechanics System, and some other method of selectinglimits was neces-
sary. One way was to combine appropriate rock classes,according to simil-
aritiesbetween their given support recommendations, in order to obtain a
constant class interval.
Barton et al. (1974) related their support categories to the Q value, and to
the equivalent dimension of an underground excavation. The latterwas

-12 I000- ~ / ~ //
// /'"/"// , /'
-I0 100- ,/ ® /

[OgO(eC}=log O(INS)±1
Um 10- ,, " A ,'"
//G Q ,

C5
~-s
..'" ~../~.~.+ A Bo,
," ~'7" -- ," + Del
/" ~ .,'" A Bos ," cS'~ /,'" A Q Dut
-2 .. / \..,- +_De, o,ol. " ~ , , . / " + ....

0,001 Ir /
-2 -~ -io 42 o,ool o.m o~ i ib ~6o 10b0
Rating (INS) O. (INSI

Fig.11. Geomechanics Classification System; borecore and in-situ joint orientation rating
values.
Fig.12. Q Classification System; borecore and in-situ Q values.
37

determined by dividing the span of the excavation b y its excavation support


ratio (ESR), where the ESR was a value assigned to different types of exca-
vations and reflected their required degree of safety. If one examines the
support recommendations given b y Barton et al. (1974), it can be seen that
for low values of Span/ESR there are certain similarities in the recommenda-
tions for rock classes related to higher Q values, i.e. classes I to 5. Since the
tunnels considered in this study were all small diameter excavations with
similar ESR values, it was reasonable to group rock classes 1--6 into pairs,
with Q value intervals of a factor of 10, similar to classes 7 - 9 . This gave six
modified rock classes with constant intervals which also provided a simpler
basis for comparison with the rock classes from the Geomechanics System.
Owing to the logarithmic relationship between the borecore and in-situ Q
values, the selected limits are given by ±1.
In Fig.12, 82% o f the results lie within the selected limits. This again
implies that, notwithstanding the errors due to the small number o f observa-
tions, there is an 82% probability of a borecore Q value being equal to, or
within a range of a factor of 10 of the in-situ Q value from the same site. For
Q values less than 10, this means a difference of up to one rock class between
related borecore and in-situ results, b u t for higher Q values it could mean a
difference of up to two rock classes.
The histogram presented in Fig.13 is similar to that shown for the Geo-
mechanics Classification System and illustrates the frequency of the number
of rock class differences obtained for each set of related borecore and in-situ
Q values. This diagram does also n o t take cognizance of the limits selected
for this classification system and includes borecore and in-situ results which
display only one rock class difference b u t whose Q values differ b y more
than a factor of 10. The diagram shows that 46% o f the borecore results are
the same as the in-situ results, 36 and 11% are one rock class lower and

75°/o
d
Lt.I
U_
.50°/o

-3 -2 -1 0 1 2 3
DIFFERENCE BETWEEN INS & BC ROCK CLASSES

Fig.13. Q Classification System; histogram showing the frequency of rock class differences
between the borecore and in-situ Q values.
38

higher, respectively, and 7% are t w o classes lower. Differences greater than


t w o were n o t recorded.

Joint set number


The most c o m m o n difference between the ratings for the various categories
of joint sets was selected as the limits for the parameter. In this case higher
ratings indicate poorer rock mass conditions and therefore all points plotting
below the ideal line represent examples where fewer joint sets (i.e. better
rock conditions) were estimated from the borecore than from the in-situ
survey.
The results presented in Fig.14 show that 71% of the points lie within the
selected limits, and 43% plot on the ideal line with 14 and 43% above and
below it, respectively. There was thus a general tendency for fewer joint sets
to be recognized in the borecore than were recorded from the in-situ survey.

Joint roughness
For this parameter, higher ratings are related to more stable rock mass
conditions. The limits selected were based on the generalized m e t h o d Osed
to assess the joint roughness characteristics. The results presented in Fig.15
show that 71% of the points plot within the selected limits, while 39% plot
on the ideal line, and 47 and 14% lie above and below it, respectively. The
results suggest that there was a tendency for the borecore estimates of the
joint roughness to indicate better rock mass conditions than the in-situ
estimates.

Joint alteration
The limits selected for this parameter were the most c o m m o n rating
difference for the range of conditions encountered. Larger rating differences

20- ,~, BOS ," / //"


+ Del // Jr(BC)=Jr[IN$].*I~--'--~-~//
O (but // / // /

is. + ;~f'" ,"'" A ,~ //" //


l.~///"~..~////
12"
m
en g. //" -I" c/" //.
--F / /" A /"
6. // // % :
• p " / + Del
;:/" // e
o,s
2.
i
dI

asl 2 3 a 6 9 lz is 2o
Jr (INS}
Jn (INS)

Fig.14. Q Classification System; borecore and in-aitu joint set number rating values.
Fig.15. Q Classification System; borecore and in-situ joint roughness rating values.
39

are generally associated with higher rating values related to more intense
alteration conditions, and wider limits would therefore probably be m o r e
appropriate for such cases.
The results, presented in Fig.16, show that 64% of the points lie within
the selected limits while 36% plot on the ideal line. 43 and 21% plot above
and below the line, respectively. The borecore and in-situ results for this
parameter correlated quite well with the ideal relationship although there was
a slight tendency for the borecore estimates to indicate poorer joint alter-
ation conditions that were noted in-situ.

Stress reduction factor


The results for this parameter are presented in Fig.17. The selected limits
are again the most c o m m o n difference between the ratings for the various
categories. 93% of the results lie on or within the limits and 68% plot on the
ideal line. 28 and 4% of the points plot above and below the line, respec-
tively. There was again a slight tendency for the borecore estimates to
indicate poorer rock mass conditions than were encountered underground.

Rock quality designation


Because the rock quality designation is c o m m o n to both rock mass
classification systems, the actual borecore and in-situ RQD values were
compared. Limits of 25% were selected as appropriate, because, with the
" g o o d " and "excellent" categories, it is the most c o m m o n range used for
differentiating between the various rock qualities. The results presented in
Fig.18 illustrate t h a t 89% of the points lie within the selected limits, 4%
plot on the ideal line and 7 and 89% lie above and below it, respectively.
Very little correlation was apparent between the two sets of results, but the
borecore RQD values were generally lower than the in-situ values.

2°1
15

/-" ~
SRF(BC)=SRF~N~*-2,1~*_2,////~

J f LL 10" ~. Jr :~(¢" . ~ .,~"


133 n-
O jl j/ f / -
& Bos
to ?,5" ~ : f ~cij
• , Bos
6 + Del + Del
5' =SRF[INS}
4~ e Dut e Dut

2o ~51 22,s. g "~'s lb 15 i0


Ja (INS) SRF (INS)

Fig.16. Q Classification System; borecore and in-situ joint alteration rating values.

Fig.17. Q Classification System; borecore and in..situ stress reduction factor rating values.
40

100- A Bos &.~


+ Del
® Dut
80-

60-
RaDISC):ROmI~=Z~ f / ®///'"+
/ "";"
/// / A
• /

-/•//~"
0
~,~ ~,0~ / /" + -t-
o
20" ® ® A

i0
A-I=CI-&
¢/
io io 8'0 16o
ROD °/o (INS)
Fig.18. Rock quality designation; borecore and in-situ values.

Comparison Of the Geomechanics and Q Systems

To facilitate a comparison of the results from the Geomechanics and Q


Classification Systems, the stand-up times, unsupported spans and support
recommendations for both systems are presented in Tables I and H. The
borecore and in-situ classification values have been plotted in Figs.19 and
20, respectively.
Linear regression analyses were undertaken in order to assess a possible
correlation between each classification system. The linear relationships
calculated for the borecore and in-situ results, respectively, are given by the
following expressions:
RMR(BC) = 4.6 InQ(BC) + 55.5
RMR(!NS) = 5.0 InQ(INS) + 60.8
tO0

90

8O f
A
90"/, confidence timit~ ~ ~' "¢=
7O
I
I
b--£ ~o I
-90°/° confidence limit
~ so
~0 In situ rel=tionship ~ 1 ..=-"

30
A Bos
20 ~ ,,-,-" ~ , I " 4- Del
e Dut
10 ~ =~"
=-~ ~iemawski'$ relationship
0 .t
o,ooi ooi 0,1 1o 100 1000
Q(BC)
Fig.19. Comparison between the borecore RMR and Q values.
41

100 ~.

90 f A //--~.+/
~ieniowski ~ r elot ior+sl+iJ~

80 I"
--Borecore retQl[on+hip

60
I
t
Z 50 t

"*r . . . . - +% I ~
40
I3d
30
~ ~ ~ - -9O=/o ¢onfiaence limit A Bos
20 • "f -t- Del
o Dut

qo~ qot o,1 lO 100 1000

Q (INS)
Fig.20. Comparison between the in-situ RMR and Q values.

The 90% confidence limits, calculated from the standard errors o f estimate
about these regression lines, were +18.6 for the borecore results, and +26.8
for the in-situ results. The relationship between RMR and Q, after Bieniawski
(1976), is also shown on each diagram.
The scatter of points about the regression lines is greater for the in-situ
values than for the borecore values. In both cases, however, it is probably
t o o great to indicate any meaningful correlation between the two classifica-
tion systems. It is nevertheless interesting to note that the respective bore-
core and in-situ regression lines are similar.

DISCUSSION

The Geomechanics Classification System

Rock mass rating


The results from the present study suggest that borecore measurements
tend to indicate poorer rock mass conditions than in-situ measurements.
With respect to the selected limits of +19, 82% of the borecore and in-situ
results differed by less than this value. This was a general evaluation and
disregarded those borecore and in-situ RMR values differing by more than
19, but by no more than one rock class.
The histogram evaluation of the results was more specific and was
related to the rock class differences only, regardless of the magnitude o f the
difference between the RMR values. The percentage Of results not exceeding
one rock class difference was thus greater than the percentage obtained in
terms of the selected limits. There was a marked tendency for the borecore
rock class to be lower than the in-situ rock class.
Generally one might anticipate less variation between the borecore and in-
situ results for better quality than for poorer quality rock. This was, however,
not borne o u t by the present study because the high RMR values displayed
42

as much scatter as the low values. This may be due to the relatively small
number o f observations as well as to the limited range o f rock classes
examined.
The R M R values correlated reasonably well with the various lithologies
studied. The hard lavas from the Bushkoppies Tunnel generally gave high
R M R values, whilst lower values were obtained from the weathered sediments
in the Delvers Street Tunnel. The RMR values from the granite in the Du
Toitsldoof Tunnel were similar to those obtained from the lavas. The quartz-
itic sandstones and the faulted and sheared rocks associated with the Du
Toitskloof Fault all gave low values.

Joint spacing
The borecore evaluation of the joint spacing generally indicated more
closely jointed rock conditions than were actually found in-situ. The results
from the Delvers Street and Du Toitskloof Tunnels deviated most signifi-
cantly from the ideal relationship. No satisfactory explanation can be given
for this because, for example, results from the Delvers Street Tunnel, which
represented similar sheared and closely jointed rock mass conditions,
displayed approximately equal b u t opposite deviations from the ideal
relationship. Widely jointed quartzite from the same tunnel, also gave a
similar deviation. The largest discrepancy at the Du Toitskloof Tunnel was
from relatively widely jointed rock with an average joint spacing o f 1--3 m,
which appeared to be more closely jointed in the borecore, where average
spacings of 5 0 - - 3 0 O m m and 0.4--I m were indicated.
The average joint spacing in a rock mass is a function of both t h e spacing
of each joint set as well as the number of joint sets present. If all the joint
sets can be observed and measured, a good estimate o f the average spacing
should be possible. If they are n o t all intersected, as may occur in a bore-
core, an incorrect estimate would be obtained. The error would, however, be
Opposite to that noted for the present results, i.e. wider joint spacings would
be indicated from the borecores. This suggests that the core recovery is an
important factor to consider when evaluating this parameter.
No results were obtained from joint sets with spacing values within the
two lower spacing categories and no meaningful c o m m e n t can therefore be
made concerning the correlation between the in-situ and borecore ratings for
all the categories. Regarding the various lithologies, the lavas were generally
more widely jointed than the weathered and sheared sediments. The Du
Toitskloof Tunnel results, particularly from the quartzite sandstones and
sheared rocks, display more scatter.
The joint spacing rating values contribute to the low borecore R M R
values. The discrepancy noted between the borecore and in-situ resttlts is
apparently related mainly to the core quality. Better quality core would
probably improve their correlation with respect to the ideal relationship. In
reality one would expect to obtain good quality core recovery from widely
jointed rocks, deteriorating with an increase in the degree of jointing. This
is evidently n o t always the case, however, and may be due to other factors,
43

such as brittleness, which could cause the harder rocks to break up if n o t


drilled carefully.

Joint condition
The reasonably good correlation between the borecore and in-situ results
is surprising, considering that the joint continuity cannot be assessed at all
from the borecore. In p o o r quality rock, the joint separation is also very
difficult to determine. The results from the Bushkoppies Tunnel displayed
the largest deviation from the ideal relationship. These were due mainly to
incorrect joint separation estimates or, as at the Du Toitskloof Tunnel, an
incorrect assessment of the joint roughness.
It is suspected that the joint condition estimates for a particular borecore
do n o t necessarily give a true indication of the in-situ conditions at that
particular site. Nevertheless, the generally fair correlation is possibly due to a
tendency for one to select an average rating value for a parameter which
cannot easily be assessed. The overall trend towards low borecore ratings
could be indicative of a tendency to err on the conservative side, in a similar
situation,

Joint orientation
Although there is no real correlation between the borecore and in-situ
ratings for this parameter, there is a tendency for them to contribute towards
the low borecore R M R values. Owing to the lack o f information regarding
the strike and dip directions of joints in borecores, the joint orientation -was
generally assumed to be "fair" with respect to the tunnel axis.
The problems involved in correctly evaluating this parameter from a bore-
core are n o t related to core recovery, b u t rather to the physical difficulties
o f orientating the core. Some form of core orientation technique, for
example, could greatly assist in obtaining a better correlation. A good knowl-
edge of the local geology would also be of great value in obtaining a picture
of the regional joint orientations.

The Q Classification System

Q Values
The results for the Q Classification System also suggest that borecore
measurements tend to indicate poorer rock mass conditions than in.situ
measurements. 82% o f the borecore and in-situ results plotted within the
limits selected for this system. As noted previously, however, these limits
excluded borecore and in-situ Q values differing b y more than a factor of
10, b u t b y no more than one rock class.
Similarly to the Geomechanics Classification System, the histogram evalua-
tion o f the Q values yielded a higher percentage, with up to one rock class
difference, than did the evaluation with respect to the selected limits. Results
showing no class differences were most c o m m o n for this system. The
remaining results showed a tendency for t h e borecore rock class to be lower
than the in-situ rock class.
44

The majority of Q values obtained during this study varied between 0.I
and 10.0 and did not, therefore, cover a sufficiently large range to assess the
correlation of borecore and in-situQ values over the entire range possible.
One would expect, however, that a better agreement between the results
would be obtained from better quality rock.
The results obtained from the lavas of the Bushkoppies Tunnel suggest that
the Q values reflect the rock mass conditions rather than the lithological
characteristics of the rock. This view is supported, to some extent, by the
results from the D u Toitskloof Tunnel where the sheared granite and porphy-
ritic granite displayed different Q values, whilst the former rock type gave
values similar to those obtained from the sheared argillaceous quartzite at the
Delvers Street Tunnel. Houghton (1975), also, found very littlelithological
differentiation between Q values calculated for sandstone, mudstone and
l~mestone at the Kielder Experimental Tunnel in England.

Joint set n u m b e r
The confidence with which the number of joint sets present in a rock mass
can be estimated, is related to the method used to record the joints. In the
case of a borecore or in-situ survey, the estimated number of joint sets will
be a function of the orientation and spacing of each set with respect to the
orientation and length, respectively, of the borecore or in-situ survey line.
The method used to estimate the number of joint sets from the borecores, in
this study, assumed that all joints with similar dip angles (inclination angles
with respect to the core axes) also had similar dip directions. Obviously, this
was n o t necessarily correct and could have resulted in underestimates of
the n u m b e r of joint sets present. The results obtained from the present study
are consistent with this expected trend. This parameter tended, therefore, to
contribute to a better borecore estimate of the rock mass condition than
was noted in-situ.
The points which plotted outside the limits selected for this parameter
were not always related to sites with poor rock mass conditions. Examples of
good rock mass quality, where all the joint sets present were not intersected
by the borecore, were noted at both the Bushkoppies and Du Toitskloof
Tunnels.

Joint roughness
The borecore evaluation of this parameter generally indicated similar or
better roughness conditions than the in-situ evaluation. This is contrary to
what might be expected since it implies that planar joint surfaces, for
example, were classified as undulating. One would in fact expect the opposite
error to be more likely. A possible reason for this c o u l d be that the larger
asperities on planar surfaces were assumed to be undulations.
Considering the small joint surface area exposed in a rock core, the
correlation between the borecore a n d in-situ ratings is quite good. This is
probably due to the simple classification adopted during the present investi-
gation. With respect to the final Q values, this parameter does n o t contribute
to the lower borecore than in-situ values.
45

Joint alteration
The reliability of borecore joint alteration estimates is related to the
quality of the drilling operation, and the care taken by the drill operator to
record changes in the colour of the flushing water and bit penetration rate.
This latter information can indicate zones o f soft clay which may be totally
or partially lost during drilling, b u t which are important to the strength
characteristics of the rock mass. A general knowledge of the geology can also
be a great assistance in providing some indications of the possible alteration
conditions which m a y be expected.
No drilling information was available to supplement the borecore obser-
vations for the present investigation. Although there appeared to be a slight
tendency for the borecore results to indicate more advanced joint alteration
than the in-situ results, the general correlation was good. The range of alter-
ation conditions examined was limited, however, and it is likely that the
correlation would be significantly poorer for joints with no direct wall rock
contact.
Observations at only t w o sites from the Bushkoppies and Du Toitskloof
Tunnels deviated significantly from the ideal line. At the former tunnel, thick
clay gouge, which had evidently been washed away during drilling, was
revealed b y the in-situ survey. At the Du Toitskloof Tunnel, a borecore
intersected several zones of clayey grit above the tunnel, which were n o t
present at the tunnel level.

Stress reduction factor


The categories defined b y Barton et al. (1974) for this parameter are well
defined and are related to the rock mass rather than to the individual joint
sets. The correlation between the in-situ and borecore results was better than
for any other parameters evaluated from the Q System. There was a slight
tendency for unfavourable borecore estimates and the results contributed
only minimally to the conservative borecore Q values.

Rock quality designation

The borecore and in-situ R Q D values showed a p o o r correlation with


respect to the ideal relationship. The measurement of joints along a borecore
or in-situ survey line is the same t y p e of procedure and should theoretically
yield similar R Q D values. One problem, and possibly the most relevant one
in the present context, is related to the different directions of each survey
line. F o r example, a vertical borehole and a horizontal in-situ survey line
could yield entirely different RQD values if the rock mass displayed a prefer-
ential joint direction. The fact that the structurally more homogeneous or
blocky rock types, such as the granite and lavas, display a somewhat better
correlation than those with an inherent structure, tends to support this idea.
Other factors, such as the core quality, which can be dependent on t h e care
and experience of the drill operator, and the accuracy of in-situ measurements,
can also affect the RQD estimates. In some cases, particularly with rocks
46

which possess an inherent structure, it may be extremely difficult to decide


if a break in a core is natural or induced by drilling.
This parameter has contributed very significantly to the low borecore
classification values for both classification systems. It is possibly more
important in the Q System, however, where most of the other parameters
evaluated indicated better rock conditions from the borecore measurements.
However, if the previous points regarding the different borecore and in~situ
survey directions are valid, then unless multi<iirectional drilling is practised
very little can be done to obtain better borecore RQD estimates, other than
ensuring good core recovery.

Comparison of the RMR and Q values

The linear regression lines,determined for the borecore and in-situR M R


and Q values, respectively, are very similar and could indicate some correla-
tion between the two classificationsystems. The scatter of points about the
lines is large and, as pointed out by Barton (1976c) with reference to
similar results presented by Bieniawski (1976a), is probably too great to
support the suggestion that a good correlation exists between the two systems.
The relationship between R M R and Q given by Bieniawski is very different
from those calculated from the present results. Bieniawski's relationship is
based on a far greater number of data points, and is thus more likely to be
generally applicable. It is nevertheless felt by the present authors that the
results presented here are representative of the types of rock mass conditions
studied in this investigation.
The results of the comparisons between the borecore and in-situ classifica-
tion values for each system individually were identical, with 8 2 % of the R M R
and Q values plotting within their respectively selected limits.The histogram
evaluations of both sets of results also yielded similar values, with 93%
showing no more than one rock class difference. The distribution of the rock
class differences was, however, different for both systems. The Geomechanics
Classification System displayed a marked tendency for the borecore rock
class to be lower than the in-siturock class.The Q System, however, showed
a trend towards a more normal distribution about the zero class difference
line. There appears, therefore, to be a better chance of obtaining a correct
borecore rock class from the Q Classification System than from the Geo-
mechanics System.
With a few exceptions, the correlation between the borecore and in-situ
stand-up times and unsupported spans for both classification systems is very
poor. The differences are so marked in some cases that, although the Geo-
mechanics stand-up time is less than the Q System stand-up time, the unsup-
ported span predicted by the former is greater than the limiting span predicted
by the latter. The support predictions from each system appear to correlate
slightly better, although when examined in detail, are very often quite
different.
47

Houghton (1975) considered the Geomechanics System to be the easiest


to apply in practice, while Bieniawski (1976b) found that the Q System was
"relatively easy to apply once the principles were understood". In the present
study, neither system was found to be particularly difficult to use once an
appreciation of the geology of a particular site had been attained. Advantages
and disadvantages were found for both systems.
In order to classify a rock mass from a borecore, the prerequisite for both
classification systems is good core recovery. To obtain average rating values
for each parameter pertinent to the Geomechanics System, it is necessary to
try to evaluate each joint set present. Some idea of the number of joint sets
is therefore required. This can be achieved visually or by using some form of
statistical method. Once the n u m b e r of sets has been defined, their individual
.spacings can be estimated. The joint continuity, which is part of the joint
condition parameter, cannot be determined from the borecore, and without
specialised core orientation techniques, the joint orientation rating is likely
to be based on very generalized information. The water and rock strength
parameters are not directly related to the quality of the borecore.
For the Q System, it is also necessary to estimate the number of joint sets
present, as well as the roughness and alteration condition of the set considered
to be most likely to cause instability. The joint roughness is not always easy
to evaluate in a borecore. The greatest difficulty with this parameter is in
differentiating between undulating and planar surfaces. As already discussed,
the reliability of the joint alteration assessment is related to the core recovery
and drilling procedure, and with the appropriate information can be evaluated
relatively easily. The most difficult aspect of this system, when classifying
the rock mass from a borecore, is selecting which joint set is most significant
to the stability o f the planned excavation. The set with the poorest roughness
and alteration conditions is usually chosen. The stress reduction factor is a
generalized parameter which was not found difficult to assess, provided the
data relevant to the rock strength and in-situ stress was available. The joint
water parameter is not related to the borecore.
For the in-situ classification of the rock mass, it is necessary to determine
the RQD from measurements or visual estimates o f the joint spacing and
joint frequency. Some care is needed, particularly in laminated rocks where
all structures are not necessarily actual parting planes. The other parameters
in both classification systems can be estimated visually and by approximate
measurements, or from detailed joint surveys. Neither system has any
particular advantage in this respect. Rock strength data is more difficult to
obtain if no rock core is available, although with the development of the
irregular lump point load test, this information can be obtained relatively
easily. Strength data is n o t always necessary for the Q System. Owing to the
generalized subdivision used for the joint orientation parameter of the Geo-
mechanics System, visual estimates are usually sufficient.
Both classification system.s suffer from similar problems regarding the bore-
core evaluation of the classification parameters. The impossibility of deter-
mining the joint continuity for the Geomechanics System, however, possibly
48

gives the Q System a slight advantage in this respect. Minor difficulties were
found with both systems for in-sltu rock classification, with the RQD esti-
mates being the most problematical.

Comparison of predicted and ~nstalled supports

In view of the fact that one of the main objectives of the rock mass
classification systems used in this investigation is to indicate the support
requirements in underground excavations, some comments on the correlation
between the predicted supports and those actually adopted is appropriate.
Owing to differing approaches to support problems by the contractors at
each tunnel, however, it is difficult to be general in this regard. The support
measures predicted from the borecore and in-situ results respectively,
together with those actually installed at each site, are presented in Tables III
and IV.
For the Q Classification System, the limiting span and stand-up time were
used to decide whether or n o t support was required. When the limiting span
exceeded the tunnel dimension and was associated with a stand-up time of
10 years, it was assumed that the rock mass could remain permanently unsup-
ported. For the Geomechanics System, the stand.up time never greatly
exceeded one year even if the unsupported span was significantly larger than
the span of the tunnel. In these cases it was therefore n o t possible to decide
whether the rock mass could permanently remain unsupported.
At the Bushkoppies Tunnel, the in-situ support predictions, based on the
Q Classification System, compared favourably with the supports actually
installed. The agreement is n o t as good for the borecore predictions, although
general similarities can be noted. The comparison between the installed
supports and predictions from the Geomechanics borecore and in-situ results
was n o t good, however.
At the Du Toitskloof Tunnel, the correlation was again better for the Q
System than for the Geomechanics System. For the latter the predicted
supports were generally more conservative than those installed. The Delvers
Street Tunnel results were n o t amenable to a comparison of this nature,
because, with minor variations in spacing, steel arches, and lagging when
necessary, were used throughout the entire tunnel. A 300 mm thick concrete
lining was also installed soon after excavation of the face.

CONCLUSIONS

The investigation described in this paper was undertaken to establish


correlation trends between rock mass classification values, derived from bore-
core and in-situ measurements, with a view to determining the reliability of
classifying rock masses for engineering purposes from borecore data alone.
The data were obtained from various geological environments, and although
it is realized that these do n o t cover the full range of conditions which m a y
be encountered in underground excavations, the results are nevertheless
considered to be generally applicable to other similar environments.
49

TABLE III

Comparison of installed and recommended supports on the basis of the Geomechanics


Classification System*

Sample S u p p o r t installed Recommended support


site
f r o m in-sltu m e a s u r e m e n t s from borecore measurements
number

Bos 1 B(t), r a n d o m , B 2.5 m + c l m ( o c c ) B 1.5---2 m + c l m (c)


+ 1 - - 2 m ave. s p a c i n g S 50 m m (c if r e q d . ) S 5 0 - - 1 0 0 m m (c)
3 0 m m (s)
Bos 2 None B 2.5 m + e l m ( o c c ) B 2.5 m + c l m ( o e e )
S 5 0 m m (c if r e q d . ) S 5 0 m m (c if r e q d . )
Bos 3 None B 2.5 m + e l m ( o c c ) B 1--1.5 m + elm
S 5 0 m m (c if r e q d . ) S 1 0 0 - - 1 5 0 m m (e)
1 0 0 m m (s) + L R 1 . 5 m
Bos 4 None sB (if r e q d . ) B 2.5 m + c l m ( o c e )
S 50 m m (c if r e q d . )
Bos 5 None sB (if r e q d . ) B 2.5 m + c l m ( o c c )
S 5 0 m m (c if r e q d . )
Bos 6 None sB (if r e q d . ) B 2.5 m + e l m ( o c c )
S 5 0 m m (c if r e q d . )
Bos 7 B(t) 1 - - 1 . 5 m B 2.5 m + clm (occ) B 1 . 5 - - 2 m + c l m (c)
S 5 0 m m (c if r e q d . ) S 5 0 - - 1 0 0 m m (c)
3 0 m m (s)
Bos 8 sB B 2.5 m + e l m ( o e e ) B 2.5 m + e l m ( o c c )
S 5 0 m m (c if r e q d . ) S 5 0 m m (c if r e q d . )
Bos 9 None B 2.5 m + c l m ( o e c ) B 2.5 m + c l m ( o e c )
S 50 m m (c if r e q d . ) S 5 0 m m (c if r e q d . )
Bos 1 0 None B 2.5 m + c l m ( o c c ) B 1 . 5 - - 2 m + c l m (c)
S 5 0 m m (c if r e q d . ) S 1 0 0 - - 1 5 0 m m (c)
1 0 0 m m (s)
Bos I I sB sB (if r e q d . ) B 1 . 5 - - 2 m + c l m (c)
S 1 0 0 - - 1 5 0 m m (c)
1 0 0 m m (s)
Bos 1 2 None sB (if r e q d . ) B 2.5 m + e l m ( o c c )
S 5 0 m m (c if r e q d . )
Dut 1 None B 2.5 + c l m ( o c c ) B 1 . 5 - - 2 m + c l m (c)
S 5 0 m m (c if r e q d . ) S 5 0 - - 1 0 0 m m (c)
50 m m (s)
Dut 2 None B 2.5 m + c l m ( o c c ) B 1 . 5 - 2 m -t- c l m (c)
S 5 0 m m (c if r e q d . ) S 5 0 - - 1 0 0 m m (c)
5 0 m m (s)
Dut 3 S(mr) 10--15 cm B 1 . 5 - 2 m + c l m (c) B 1 - - 1 . 5 m + c l m (c)
S 5 0 - - 1 0 0 m m (c) S 1 0 0 - - 1 5 0 m m (c)
30 m m (s) 100 m m (s)
L R 1.5 m ( i f reqd.)
Dut 4 B(tg) 1 - - 2 m B 1 - - 1 . 5 m + c l m (c) B 1 . 5 - - 2 m + c l m (c)
S 1 0 0 - 1 5 0 m m (e) S 50---100 m m (e)
1 0 0 m m (s) 5 0 m m (s)
Dut 5 sB B 1 . 5 - - 2 m + e l m (c) B 2.5 m + c l m ( o e c )
S 1 5 0 - 1 0 0 m m (c) S 5 0 m m (e if r e q d . )
50 m m (s)
Dut 6 None sB (if r e q d . ) B 2.5 m + e l m ( o c c )
S 5 0 m m (c if r e q d . )

* E x p l a n a t i o n o f s y m b o l s : B = s y s t e m a t i c bolting w i t h spacing indicated; clm = chain link


mesh; (c) ffi c r o w n ; (s) ffi side; sB = s p o t bolting; S = shotcrete w i t h t h i c k n e s s indicated;
(occ) = occasional; (mr) = m e s h reinforced; L R = light steel sets.

The investigation consisted o f the classification o f the rock mass from


borecore measurements alone, and the subsequent examination o f each bore-
core site underground, where the procedure was repeated from the in-situ
5O

TABLE IV

Comparison of installed and recommended supports on the basis of the Q Classification


System*
Sample S u p p o r t installed Recommended support
site
f r o m in-sltu m e a s u r e m e n t s from borecore measurements
number

Bos 1 B(t), r a n d o m , B(tg) 1 m + S(mr) 5 c m S(mr) 7.5--15 cm


+ 1 - - 2 m ave. s p a c i n g
Bos 2 None None None
Bos 3 None S 2.5--5 cm S(rnr) 1 5 - - 2 5 c m
Bos 4 None None None
Bos 5 None None None
Bos 6 None None B(utg) 1 m + S 2--3 c m
Bos 7 B(t) 1 - - 1 . 5 m B(utg) 1 m + S(mr) 5 cm B(utg) 1 m + S(mr) 5 cm
Bos 8 sB None B(utg) 1 m + S(mr) 5 e m
Bos 9 None None None
Bos 10 None B(tg) 1 m + S(n~) 5 c m B(utg) 1 m + S(mr) 5 c m
Bos 11 sB B(tg) 1 m + S(mr) 5 c m B(utg) 1 m + S(inr) 5 c m
Bos 12 None None None
Dut 1 None S 2.5---5e m B(utg) I in + S(n~) 5 e m
Dut 2 None S 2.5--5 e m B(tg) 1 m + S(n~) 5 e m
Dut 3 S(mr) 10--15 e m S 2.5--5 e m S 2.5--5 e m
Dut 4 B ( t g ) 1 - - 2 In B ( u t g ) 1 In + S ( m r ) 5 c m S ( m r ) 5---10 e m
Dut 5 sB None None
Dut 6 None None None

* E x p l a n a t i o n of symbols: B ffi systematic bolting with spacing indicated; (utg) = untensioned,


grouted; S ffishotcrete with thickness indicated; (mr) = mesh reinforced; (tg) ffitensioned,
grouted post-tensioned; sB ffispot bolting; (t) ffi tensioned.

measurements. The borecore and in-situ classification values were then com-
pared. In general, it m a y be concluded that for both classification systems,
borecores can be used to classify rock masses for engineering purposes. Bore-
core measurements, however, tend to indicate poorer rock mass conditions
than in-situ measurements.
The tendency for lower borecore than in-situ classification values is
opposite to t h a t indicated by results presented by Barton (1976a) from a
similar type of investigation. Barton's borecore Q values were about twice
his in-situ values. This discrepancy appears to be related to the different rock
conditions examined in each case. The present results are based on measure-
ments from a variety of geological environments which included jointed rocks
predominantly, although massive rocks were also examined. Barton's data
were obtained from "quite massive biotite gneiss" only. This suggests that
the relationship between the borecore and in-situ classification values may be
linked to the rock mass condition, with lower borecore than in-situ values
being associated with jointed rocks, and the opposite being the case in more
massive varieties.
With regard to site investigations for underground excavations in rock,
both the Geomechanics and Q Classification Systems can be most useful.
Their value must be seen in perspective and their limitations always recognized.
Their best application would seem to be towards providing a general picture
of the anticipated rock conditions, and an initial assessment of the likely
51

support requirements in a planned underground excavation. However, the


data should never be regarded as the final result in this respect. Since bore-
core results are generally expected to be directly applicable to the rock mass
in the immediate vicinity of the borehole only, great care is necessary when
attempting to extrapolate between boreholes. When possible, other investi-
gatory techniques should also be used.

The Geomechanics Classification System

The results obtained using the Geomechanics Classification System indicate


that borecore measurements provide reasonably reliable estimates of the rock
classes anticipated in planned underground excavations. The borecore and in-
situ RMR values often differed significantly, however. When evaluated with
respect to selected limits, the results suggest that there is approximately an
80% probability of a borecore rock class being equal to or within one class
of a related in-situ value. Borecore and in-situ estimates of the unsupported
span were similar, but the associated stand-up times and support recommen-
dations did not compare as favourably.
A general trend towards lower borecore than in-situ RMR values was
noted. This is a reflection of similar trends displayed by all the classification
parameters evaluated. The rock quality designation and joint spacing par-
ameters were most important in this regard. The joint condition and joint
orientation parameters are difficult to assess accurately from borecores, but
their influence on the observed trend was not as marked as the other two
parameters. This is apparently due to a tendency for one to select inter-
mediate rating values when the borecore is difficult to assess.
In its application, the Geomechanics System was found to be simple for
both borecore and in-situ classifications. The generalized descriptions used
for the individual parameters assisted in this regard.

The Q ClassificationSystem

The results obtained from the Q Classification System indicate that bore-
core data m a y be used with some certainty to classifyrock masses with Q
values exceeding approximately 0.1. Values significantly lower than this were
seldom encountered during the present study. With respect to selected limits,
the results suggest that there is approximately an 8(Wo probability of the
borecore and in-situ Q values being within one rock class of each other. The
borec0re and in-situlimiting spans and stand-up times were also generally
similar.
A tendency toward lower borecore than in-situ Q values was also noted
for this system. The borecore assessment of the individual classificationpar-
ameters correlated reasonably well with the in-situratings. The joint set
number and joint roughness parameters indicated better rock conditions
from borecore measurements than from in-situmeasurements, whilst the
joint alteration parameter and stress reduction factor displayed an opposite
52

trend. The rock quality designation probably influenced the present results
most significantly with a marked tendency to lower borecore than in-situ
values.
The Q ClassificationSystem was found simple to apply for both borecore
and in-siturock classification.The joint roughness parameter is possibly the
most difficultparameter to assessaccurately from a borecore, and the joint
alteration parameter could prove problematical in cases of advanced joint
alteration.

Comparison of the Geomechanics and Q Classification Systems

The correlation between the borecore and in-situ rock classes was found
to be better for the Q System than for the Geomechanics System. A similar
result was obtained for the unsupported or limiting spans, and stand-up times,
calculated from the classification values.
Neither classification system showed a good correlation between the bore-
core and in-situ support predictions, although a slightly better result was
again obtained for the Q System. This result appears to be due, to some
extent, to the greater number of support categories, and more detailed
support recommendations for this system, together with the fact that some
allowance can be made for excavation dimensions.

ACKNOWLEDGEMENTS

The writers are deeply gratefulto the Johannesburg City Engineers


Department for allowing access to the Bushkoppies and Delvers Street
Tunnels and for itspermission to use the data obtained. They are also
indebted to the Department of Transport (Division of National Roads) for
its permission to utflisethe data from the D u Toitskloof Pilot Tunnel in this
publication.
The willingnessof Messrs Van Niekerk, Kleyn and Edwards, Consulting
Engineers, to provide the opportunity for carrying out the investigationand
making the relevant testresults,borecores and log sheets freely available
from the Du Toitskloof Tunnel is gratefully acknowledged.
The authors also wish to express their appreciation to Mr D.R. Bassarab,
formerly a lecturer in the Department of Geology, Witwatersrand University,
Johannesburg, under whose direction the earlier classification works were
carried out.
The present publication is a partial extract from a dissertation submitted
by the first author to the University of the Witwatersrand, Johannesburg.

REFERENCES

Barton, N., 1976a. Recent experiences with the Q System in tunnel support design.In:
Z.T. Bieniawski (Editor),Proceedings of the Symposium on Exploration for Rock
Engineering. Johannesburg, Vol.1, pp.107--115.
53

Barton, N., 1976b. Unsupported ground openings. In: Rock Mechanics Meeting, Swedish
Rock Mechanics Research Foundation, Stockholm, pp.61--94.
Barton, N., 1976c. Session report on rock mass classifications. In: Z.T. Bieniawski
(Editor), Proceedings of the Symposium on Exploration for Rock Engineering.
Johannesburg, Vol.2, pp.167--172.
Barton, N., Lien, R. and Lunde, J., 1974. Engineering classification of rock masses for
the design of tunnel support. Rock Mech., 6: 189--236.
Bergrnan, M., 1974. Rock mass investigation in depth. Reliability of different methods
for drill hole investigations. In: Advances in Rock Mechanics, Proc. Congr., Int.
Soc. Rock Mech., 3~1, Denver, Colo., Vol.2, Pt.4, pp.15--20.
Bieniawski, Z.T., 1976a. Rock mass classification in rock engineering. In: Z.T. Bieniawski
(Editor), Proceedings of the Symposium on Exploration for Rock Engineering.
Johannesburg, Vol.1, pp.97--106.
Bieniawski, Z.T., 1976b. Session report on rock mass classifications. In: Z.T. Bieniawski
(Editor), Proceedings of the Symposium on Exploration for Rock Engineering.
Johannesburg, Vol.2, pp.167--172.
Brink, A.B.A. and Associates, 1974. Geotechnical Report, Borehole Profiles and Geo-
logical Map of the Bushkoppics Outfall Sewer Tunnel, Phase 1, Vol.4, City of
Johannesburg, Contr. No.3450.
Cockcroft, T.N. and Loudon, P.A., 1976. The Du Toitskloof T u n n e l - A case history of
the geotechnical investigations. Proc. Sym. Expl. Rock Eng., pp.301--312.
Franklin, J.A., Broch, E. and Walter, G., 1970. Logging the Mechanical Character of Rock
Imperial CoIl. Rock Mech. Res. Rept., No.D14, London.
Houghton, D.A., 1975. The Assessment of Rock Masses and the Role of Rock Quality
Indices in Engineering Geology with Reference to Hard Rock. M.Sci. Thesis, Imperial
College, London.
Payne, T.G., 1976. Assessment of the Use of Rock Classification Systems in Practice.
M.Sci. Thesis, Imperial College, London.
Piteau, D.R., 1979. Geological Factors Significant to the Stability of Slopes Cut in Rock.
In: Planning Open Pit Mines. South A_rTicanInstitute of Mining and Metallurgy,
pp.33--54.
Priest, S.D. and Hudson, J.A., 1976. Discontinuity spacings in rock. Int. J. Rock Mech.
Mining Sei., 13: 135--148.
Rankilor, P.R., 1974. A suggested field system for logging cores for engineering purposes.
Bull. Assoc. Eng. Geol., XI (3): 247--258.
Rocha, M., 1971. A method of integral sampling of rock masses. Rock Mech., 3: 1--12.
South African Section of the Association of Engineering Geologists (Core Logging
Committee), 1976. A guide to core logging for rock engineering. In: Z.T. Bieniawski
(Editor), Proceedings of the Symposium on Exploration for Rock Engineering.
Johannesburg, Vol.1, pp.71--86.
Terzaghi, R.D., 1965. Sources of errors in joint surveys. Geotechnique, 15: 287--304.
Van Niekerk, Kleyn and Edwards and P.A. de Villiers in association with Electro-Watt
Engineering Services Limited, Ziirich, 1978. Du Toitskloof Tunnel Pilot Bore;
Unpublished Test Results and Geological Data.

You might also like