1.

INTRODUCTION
The Mandai st orage compl ex i s t he fi rst
underground rock cavern project in Singapore.
Among other advantages, by building the depot
underground instead of on surface, more than 300
Ha of land will be made available for other use,
which is equivalent to 400 basketball courts. This
is very beneficial to the land-scarce Singapore.
Lasting for about two years, the final explosive
blast on 27/07/2001 marked the end of excavation
stage of Phase One of the project. A typical dig-
completed cavern is shown in Figure 1.

The tunnels/caverns are built in granite that
belongs to Triassic period and is about 220 million
years old. It is heavily fractured by three major
rock joint sets, in which one is sub-horizontal and
two are sub-vertical. A few weak zones are also
encountered in the area. The granite has a high
horizontal in situ stress that is favourable to the
stability of caverns with wide span especially when
the vertical joints are dominant. The project
consists of many rock tunnels and caverns with
various sections. The rock support design is based
on the recommendation of Q-chart in which
rockbolts and shotcrete are the main rock support
to the tunnels/caverns.

This paper is to present the rock support design in
Mandai project, Singapore. The rock condition is
firstly briefed. The rock support design is based on
the recommendations of Q-system, which is
improved by taking account of the actual rock
condition. The rock support design is then
evaluated by using convergence-confinement
method.

2. ROCK CONDITION
The rock condition is basically obtained from site
investigation done before the construction (Wu et
al. 2000). More information is updated from the
rock mapping during the construction. Site
investigation showed that downward from the
ground surface, t he format i on i s generally
composed of three layers of residual soil,
weathered granite and fresh granite. Residual soil
(heavy sandy loam) widely overlays on the Bukit
Timah Granite with thickness varying from 0 to
70m with the average of 29.8m. The top rock mass
is slightly or heavily weathered. The thickness of
weathering layer varies from zero to 27.7 m, which
is normally related to the bedrock elevation. The
surface profile of the bedrock is generally flat
except at soil trenches . The principal materials of
the granite are quartz (29%), alkali-feldspar (29%),
plagioclase (36%), biotite (5%) and hornblende.
Six weak zones were detected and two of them
were observed during the construction of the Phase
one, which caused water inflow 20~200 l/min and
considerable volume of cement were grouted.

Q-val ues wer e es t i mat ed bas ed on s i t e
investigation as shown in Figure 2. It can be
observed that good and very good rock occupies
75%. Q-values collected from the rock mapping
during the construction are also shown in the same
figure for a comparison. It indicates a good
consistency between the estimation (75%) and the
site collection (68.6%) for sum of good and very
good rock. However, the estimation shows a very
higher percentage of 23.2% for very good rock
relative to the site collection of 2.0%. This might
be because those dominant rock joints are vertical
and could not be fully detected by vertical
boreholes.

Favourable stresses have been discovered in the
concerned area. The magnitude ratio between the
average horizontal stresses and the vertical stress is
about 2.5:1. The stress field is favourable for the
stability of underground opening but rock spalling
Rock support design in Mandai project, Singapore
Chen S.G., Tan K.H. & Ong H.L.
ST Architects & Engineers, Singapore


ABSTRACT: This paper is to present the rock support design in Mandai project, Singapore. The rock
support design is based on the recommendations of Q-system, which is improved by taking account of the
actual rock condition. The rock support design is then evaluated by using convergence-confinement method.
The evaluation indicates that, in general, the rock support design is reasonable.
was observed in site during the construction in a
few spots.














Figure 1. Typical dig-completed cavern (Fang
2001)


















Figure 2. Comparison on Q-value between
estimation and collection

3. ROCK SUPPORT DESIGN
The rock support design in the project is based on
the recommendation of the Q-chart proposed by
Barton and his co-workers (Barton et al. 1974,
Barton 1976; Barton 1983) on the basis of an
analysis of some 200 tunnel case histories from
Scandinavia. Grimstad and Barton (1993) updated
it based on studies of 1,050 case records and
successfully utilised it in Gjovik Olympic Cavern
(Barton et al. 1994). In practice, geologists carry
out rock mapping on site after every blasting,
assess the rock quality, and determine the Q-value,
which is calculated using the empirical formula by
Barton (1983):




(1)


where: RQD is the rock quality designation, J
n
is
the number of joint sets, J
r
is the joint roughness,
J
a
is the degree of alteration or filling along the
weakest joint, J
w
is the joint water reduction factor
and SRF is the stress reduction factor. The ratings
of various input parameters to the Q-value are
given in Table 1.
Once the Q-value is calculated, the rock support
can be determined from the Q-chart as shown in
Figure 3. The y-axis represents the equivalent
dimension of the underground opening that is
obtained by dividing the span, diameter or wall
height of the excavation by the excavation support
ratio (ESR). The ESR depends on the type or use of
the underground opening (Grimstad & Barton
1993). It should be noted that the Q-value in the
Q-chart is related to the total amount of support
(temporary and permanent) in the roof. Wall
support can also be found using the same figure by
applying the wall height and the following
adjustments to Q. For, Q > 10, Q
wall
= 5Q. For 0.1
< Q < 10, Q
wall
= 2.5Q. For Q < 0.1, Q
wall
= Q.

The Q-chart on determining rock support is based
on experience from numerous underground
projects. Being statistically based, a support chart
can never replace or accurately represent the
ground conditions at site. A main reason is that all
the actual geometrical features of rock joints (e.g.,
joint strike and dip angle) can not be included in a
suppor t char t . Therefore, havi ng a good
characterisation of the rock mass and the ground
conditions and having a good understanding of the
geological conditions at site is of utmost important
to employ Q-chart for rock support design. During
the rock support design on site, efforts were put to
further optimise the rock support and some
adjustments were made according to the rock
condition in the current project. One of the areas
in optimisation is to change the rock supports on
the walls, as dominant rock joints are vertical and
the walls are not so high. Only one or two rows of
rockbolts are applied to the upper walls and no
rockbolt on the rest of the wall unless at locations
where spot rockbolt is required due to the rock
condition. The another is to apply shotcrete only
for the arch up to 2 m below the arch base for rock
mass wi t h Q-value larger than 10. As t h e
tunnels/caverns in the project have several
different section sizes, rock support is classified for
each section size to make rock support design and
installation more efficient and effective.

SRF
J
J
J
J
RQD
Q
w
a
r
n

23.2
51.8
13.6
5.8 5.6
2.0
66.6
23.5
8.0
0.0
0
10
20
30
40
50
60
70
V. good
Q>40
Good
Q=11~40
Fair
Q=5~10
Poor
Q=1~4
V. poor
Q<1
ROCK CLASS
P
E
R
C
E
N
T
A
G
E
(
%
)
Estimation based on site
investigation
Collection in Phase one
during the construction
4. EVALUATION OF THE ROCK SUPPORT
DESIGN
As mentioned earlier, Q-system is basically an
empirical method coming from statistic analyses of
numerous underground projects. For verification
of the rock support design recommended,
eval uat i on was done usi ng convergence
confinement method.
4.1 Convergence-confinement method
The convergence-confinement method was
formerly suggested in May 1977 in France during
Ratings for the rock quality designation (RQD) Description and ratings for the parameter Jn (joint set number)
A. Very poor RQD = 0 - 25% Massive, no or few joints Jn = 0.5 - 1
B. Poor 25 - 50 One joint set 2
C. Fair 50 - 75 One joint set plus random 3
D. Good 75 - 90 Two joint sets 4
E. Excellent 90 - 100 Two joint sets plus random 6
Notes: Three joint sets 9
(i) Where RQD is reported or measured as < 10 (including 0), Three joint sets plus random 12
a nominal value of 10 is used to evaluate Q Four or more joint sets, heavily jointed, "sugar-cube", etc. 15
(ii) RQD intervals of 5, i.e. 100, 95, 90, etc. Crushed rock, earthlike 20
are sufficiently accurate Note: For tunnel intersections, use (3.0 x Jn); for portals, use (2.0 x Jn)
Desciption and ratings for the parameter Jr (joint roughness number)
Large scale waviness of joint plane
Planar Slighty Strongly Stepped Interlocking
undulating undulating (large scale)
Very rough 3 4 6 7.5 9
Rough 2 3 4 5 6
Slightly rough 1.5 2 3 4 4.5
Smooth 1 1.5 2 2.5 3
Polished 0.75 1 1.5 2 2.5
Slickensided
*)
0.6 - 1.5 1 - 2 1.5 - 3 2 - 4 2.5 - 5
Notes: (i) For filled joints (no rock wall contact) Jr = 1
(ii) Jr = 0.5 can be used for planar, slickensided joints having lineations, provided the lineations are oreintated for minimum strength
Descriptions and ratings for the parameter Ja (joint alteration number)
JOINT WALL CHARACTER Condition Wall contact
Healed or welded joints filling of quartz, epidote, etc. Ja = 0,75
Fresh joint walls no coating or filling, except from staining (rust) 1
Slightly altered joint walls non-softening mineral coatings, clay-free particles, etc. 2
Friction materials sand, silt calcite, etc. without content of clay 3
Cohesive materials clay, chlorite, talc, etc. 4
Partly wall contact No wall contact
FILLING OF: Type Thin filling (< 5 mm) Thick filling
Friction materials sand, silt calcite, etc. without content of clay Ja = 4 Ja = 8
Hard cohesive materials compacted filling of clay, chlorite, talc, etc. 6 10
Soft cohesive materials medium to low overconsolidated clay, chlorite, talc, etc. 8 12
Swelling clay materials filling material exhibits swelling properties 8 - 12 12 - 20
Description and ratings for the parameter Jw (joint water reduction factor)
Dry excavations or minor inflow, i.e. < 5 l/min locally pw < 1 kg/cm
2
Jw = 1
Medium inflow or pressure, occasional outwash of joint fillings 1 - 2.5 0.66
Large inflow or high pressure in competent rock with unfilled joints 2.5 - 10 0.5
Large inflow or high pressure, considerable outwash of joint fillings 2.5 - 10 0.3
Exceptionally high inflow or water pressure at blasting, decaying with time > 10 0.2 - 0.1
Exceptionally high inflow or water pressure continuing without noticeable decay > 10 0.1 - 0.05
Note: (i) The last four factors are crude estimates. Increase Jw if drainage measures are installed
(ii) Special problems caused by ice formation are not considered
Description and ratings for parameter SRF (stress reduction factor)
Multiple weakness zones with clay or chemically disintegrated rock, very loose surrounding rock (any depth) SRF = 10
Single weakness zones containing clay or chemically disintegrated rock (depth of excavation < 50 m) 5
Single weakness zones containing clay or chemically disintegrated rock (depth of excavation > 50 m) 2.5
Multiple shear zones in competent rock (clay-free), loose surrounding rock (any depth) 7.5
Single shear zones in competent rock (clay-free), loose surrounding rock (depth of excavation < 50 m) 5
Single shear zones in competent rock (clay-free), loose surrounding rock (depth of excavation > 50 m) 2.5
Loose, open joints, heavily jointed or "sugar-cube", etc. (any depth) 5
Note: (i) Reduce these valued of SRF by 25 - 50% if the relevant shear zones only influence, but do not intersect the excavation
c / 1
/ c SRF
Low stress, near surface, open joints > 200 < 0.01 2.5
Medium stress, favourable stress condition 200 - 10 0.01 - 0.3 1
High stress, very tight structure. Usually favourable to stability, may be except for walls 10 - 5 0.3 - 0.4 0.5 - 2
Moderate slabbing after > 1 hour in massive rock 5 - 3 0.5 - 0.65 5 - 50
Slabbing and rock burst after a few minutes in massive rock 3 - 2 0.65 - 1 20 - 200
Heavy rock burst (strain burst) and immediate dynamic deformation in massive rock < 2 > 1 200 - 400
Note: (ii) For strongly anisotropic stress field (if measured): when 5 < 1 / 3 <10, reduce sc to 0.75 sc. When s1/s3 > 10, reduce sc to 0.5sc
(iii) Few case records available where depth of crown below surface is less than span width. Suggest SRF increase from 2.5 to 5 for low stress cases
/ c SRF
Squeezing Plastic flow of incompetent rock Mild squeezing rock pressure 1 - 5 5 - 10
rock: under the influence of high pressure Heavy squeezing rock pressure > 5 10 - 20
Swelling Chemical swelling activity Mild swelling rock pressure 5 - 10
S
m
a
l
l

s
c
a
l
e

s
m
o
o
t
h
n
e
s
s

o
f

j
o
i
n
t

s
u
r
f
a
c
e
C
o
n
t
a
c
t

b
e
t
w
e
e
n

j
o
i
n
t

w
a
l
l
s
P
a
r
t
l
y

o
r

n
o

w
a
l
l

c
o
n
t
a
c
t
CLEAN JOINTS:
COATING OR
THIN FILLING
a) Rock wall contact, and
b) Wall contact before 10 cm shear
C
o
m
p
e
t
e
n
t

r
o
c
k
,

r
o
c
k

s
t
r
e
s
s

p
r
o
b
l
e
m
s
W
e
a
k
n
e
s
s

z
o
n
e
s

i
n
t
e
r
s
e
c
t
i
n
g

e
x
c
a
v
a
t
i
o
n
Table 1. Ratings of the input parameters of the Q-system ( Grimstad and Bar ton, 1993)
a meeting devoted to tunnel calculations by the
AFTES (LAssociation Francaise de Travaux en
Souterrain French Association for Underground
Works). It is an attempt to evaluate tunnel stability
conditions by means of a mathematical model
(Gesta et al. 1980, Kerisel 1980, Duddeck 1980).
The convergence confinement method is based on
the assumption that the rock mass is an equivalent
continuous homogeneous medium and the tunnel is
circular. It also assumes constant in situ stress and
uniform support pressure. It involves calculating
the ground reaction curve and the support reaction
curve, and determining their intersection as shown
in Figure 4. The ground reaction curve describes
the relationship between radial convergence and
radial pressure. The support reaction curve
expresses the relationship of support displacement
versus support pressure. The intersection (Point A
in Figure 4) implies the balanced s tate with a radial
displacement of the tunnel u
a
and a support
pressure to the tunnel p
a
. Point B in Figure 4
presents an ultimate state of tunnel stability which
implies that p
b
is the minimum support pressure
required for maintaining the tunnel stability and u
b

is the maximum allowable convergence of the
tunnel.
4.1.1 Determination of ground reaction curve
Considering a circular tunnel subjected to a
constant in situ stress
0
and a uniform support
pressure p
a
, the rock mass surrounding the tunnel
will deform and the stress in the rock mass will be
redistributed due to the excavation of the tunnel.
When the stress exceeds the strength, the rock
mass will be yielded subjected to Mohr-Coulomb
criteria and a plastic zone occurs with radium of.




(2)


where




a is the radium of tunnel,
c
is the uniaxial
compressive strength of rock mass,
0
is the in situ
stress, p
a
is the support pressure, c is the cohesion
and is the friction angle.

The radial stress at the interface between elastic
zone and plastic zone can be calculated by

Figure 3. Q-Chart for estimating rock support (from Grimstad and Barton, 1993)
1
1
0
0
) 1 (
) 1 (
1
2

1
]
1

+
+

c a
c
p
a r

sin 1
cos 2

c
c

sin 1
sin 1

a
c
r
p
a
r
a
r

,
_

+ 1
1

,
_

1
0
1
0
1
1

100
50
20
10
5
2
1
0.001 0.004 0.01 0.04 0.1 0.4 1 4 10 40 100 400 1000
G F E D C B A
Exceptionally
poor
Extremely
poor
Very
poor
Poor Fair Good
Very
good
Ext.
good
Exc.
good
20
11
7
5
3
2.4
1.5
B
o
l
t

l
e
n
g
t
h

i
n

m

f
o
r

E
S
R

=

1
S
p
a
n

o
r

h
e
i
g
h
t

i
n

m
E
S
R
Rock mass quality Q = x x
ROCK CLASSES
9
CCA
8
RRS Sfr
7 6
Sfr Sfr
5
B(+S)
4
3
B
2
sb
1
2
5
cm
1
5
c
m
1
2
c
m
9

c
m
5
c
m
4
c
m
1.0 m
1.3 m
1.6 m
2.0 m
3.0 m
4.0 m
1.0 m
1.3 m
1.5 m
1.7 m
2.1 m
2.3 m
2.5 m
B
o
l t s
p
a
c
in
g
i n
u
n
s
h
o
tc
r
e
te
d
a
re
a
s
Bo
l t spaci ng in shotcreted a
rea
RQD Jr Jw
Jn Ja SRF
REINFORCEMENT CATEGORIES:
sb
B
B(+S)
1) Unsupported
2) Spot bolting,
3) Systematic bolting,
4) Systematic bolting,
(and unreinforced shotcrete, 4 - 10 cm),
5) Fibre reinforced shotcrete and bolting, 5 - 9 cm,
6) Fibre reinforced shotcrete and bolting, 9- 12 cm,
7) Fibre reinforced shotcrete and bolting, 12 - 15 cm,
8) Fibre reinforced shotcrete, > 15 cm,
reinforced ribs of shotcrete and bolting, ,
9) Cast concrete lining,
Sfr + B
Sfr + B
Sfr + B
SfrRRS+B
CCA



(3)





















Figure 4. Convergence-confinement description


Finally the convergence (displacement) of the
tunnel is obtained from


with no plastic zone (4)



with plastic zone (5)

where G is the shear modulus.

Varying the support pressure p
a
, a series of points
(u
a
and p
a
) can be calculated by using above
formulas and the ground reaction curve can then be
obtained.
4.1.2 Determination of support reaction curve
The support reaction curve is subjected to not only
the support stiffness and strength (maximum
supporting pressure) but also the rock support
installation time. A good design should just
mobilise the maximum support capacity by
adjusting the installation time so that the support
reaction curve intersects with the ground reaction
curve at minimum supporting pressure.

The maximum support pressure of rockbolts can be
calculated by



(6)


where F is the ultimate capacity of rockbolts, s
a

and s
p
are rockbolt spacing, respectively, in
longitudinal and transverse directions.

The maximum support pressure of shotcrete can be
calculated by



(7)


where is the reduction coefficient subjected to
actual tunnel profile and t is shotcrete thickness.
The total maximum support pressure, p
a
, is sum of
that of rockbolts and shotcrete as



(8)
4.2 Application in the current project
The convergence-confinement method is applied to
check the effectiveness of the support design in
Phase One of the current project. The caverns are
not circular and neither the in situ stress is uniform
in the current project. However, with closer
evaluation, the case in the current project could be
considered as equivalent to a circular tunnel, with a
diameter of the span subjected to a constant in situ
stress that is the same as the vertical in situ stress.
This is because that the principal of optimal design
is to ensure the displacement distribution along the
roof be as uniform as possible. On the other hand,
it was observed on site that the actual excavation
profile is rough due to the presence of rock joints
so that the shotcrete failure is governed by tension.
Therefore, tensile strength is used in calculating
the shotcrete support capacity.
4.2.1 Determination of parameters
To apply the convergence-confinement method,
parameters including Youngs modulus, rock mass
cohesion and friction angle must be determined
which can be derived from the Q-value. The RMR
(rock mass rating) and Yongs modulus are then
calculated from (Bieniawski 1984)



,
(9)

0
a
0
r0
0
0
pa
A
Ground reaction
curve
Support reaction curve
u
ua
pa
P, pressure
ub
pb
B
u0
) (
2
0 a
e
a
p
G
a
u
( )
a
r
G
u
r y
p
a
2
0
0
2
1

p a
b
s s
F
p

1
]
1

+

2
2
) (
1 5 . 0
t a
a
p
c s

s b a
p p p +
44 ln 9 + Q RMR
40
10
10 ) (

RMR
GPa E
The cohesion and friction angle of rock mass can
be obtained from




(10)

4.2.2. Assessment of the support design
The support design for pattern rockbolt and
shotcrete is assessed by the factor of safety (FOS).
The factor of safety is defined as the ratio of the
allowable maximum convergence divided by the
actual convergence (FOS=u
b
/u
a
, u
b
and u
a
refer to
Figure 4). The act ual convergence u
a
i s
determined by the intersection of ground reaction
curve and support reaction curve. The allowable
maximum convergence u
b
is related to the rock
cavern damage criteria. Bieniawski (1984)
discussed the criteria for evaluation of the rock
cavern stability based on cavern deformation
monitoring. He pointed out that the instability of
the rock cavern may be caused by the failure of the
rock joints, shotcrete and rock bolts. He found that
joint movement and loosening occurs when the
displacement is more than three times the elastic
displacement. Thus, in this study, an allowable
convergence of three times the elastic displacement
is adopted.

Figure 5 shows the factor of safety versus Q-value
for various tunnel types. It can be seen that the
FOS ranges from 1.25 to 2.66 and it is increasing
with increasing Q-value. By comparing with the
normally adopted FOS of 1.5 ~ 2.0 used in
engineering practices, it can be seen that the
support design for Class low D (Q=1) has slightly
lower FOS of 1.25~1.29. On the other hand the
support design for Class high B (Q = 40) has
slightly higher FOS of 2.65~2.66. It suggests that
the support design for the project is, in general,
reasonable.
5. DISCUSSION AND CONCLUSIONS
The rock support design in Mandai underground
rock cavern project is presented. The rock support
design is based on Q-chart and improved according
to the actual rock condition. The convergence
confinement method is employed to evaluate the
rock support design. The evaluation indicates that
the rock support design for good rock is slightly
conservative and more rock support would be
applied to the poor rock. Nevertheless , it can be
concluded that the support design for the project is,
in general, reasonable since there are many
uncertainties in rock tunnelling.

6. REFERENCES
- Barton N., Lien R. & Lunde J. (1974)
Engineering classification of rock masses
for the design of tunnel support. Rock
Mechanics, Vol. 6, No. 4, pp. 183-236.
- Barton N., (1976) Recent experiences with
the Q-system of tunnel support design.
Proceedi ngs of t he Symposi um on
Expl orat i on f or rock engi neeri ng,
Johannesburg, November, Vol. 1, pp. 107-
117.
- Barton N. (1983) Application of Q-system
and index tests to estimate shear strength
Figure 5. Factor of safety versus Q-value
5 . 4 5 . 0 ) ( + RMR
o

44 ln 9 + Q RMR
1.27
1.92
2.25
2.64
0.0
0.5
1.0
1.5
2.0
2.5
3.0
1 4 10 40
Q-value
F
a
c
t
o
r

o
f

s
a
f
e
t
y
Tunnel type I
Tunnel type II
Tunnel type III
Trendline
and deformability of rock masses. Proc.
Int. Symp. Eng. Geol. Underground Const.,
Lisboa, Vol. 1, pp. 51-70.
- Barton, N., By, T.L., Chryssanthakis, P.,
Tunbridge, L, Kristiansen, J., Loset, F.,
Bhasin, R.K., Westerdahl, H. & Vik, G.
( 1994) Pr e di c t e d a nd Me a s ur e d
Performance of the 62m Span Norwegian
Olympic Ice Hockey Cavern at Gjovik Int.
J. Rock Mech. Min. Sci. & Geomech.
Abstr., Vol. 31, pp. 617-641.
- Bieniawski Z.T. (1984) Rock mechanics
design in mining and tunneling. Balkema,
Rotterdam, pp. 272.
- Duddeck H. ( 1980) On t he bas i s
requirements for applying the convergence-
confinement method. Underground Space,
Vol. 4, No. 4, pp. 241-247.
- Fang Joyce (2001) Blast off. The Straits
Times, Satueday, July 28, 2001.
- Gesta P, Kerisel J., Londe P. Louis C. &
Panet M. (1980) General report: Tunnel
stability by the convergence-confinement
method. Underground Space, Vol. 4, No.
4, pp. 225-232.
- Grimstad E. & Barton N. (1993) Updating
of the Q-System for NMT. Proceedings of
the International Symposium on Sprayed
Concrete Modern Use of Wet Mi x
Sprayed Concrete for Underground
support , Norwegian Concrete Association,
Oslo.
- Kerisel J. (1980) Commentary on the
general report on t he convergence-
confinement method. Underground Space,
Vol. 4, No. 4, pp. 233-240.
- Wu Y. K., Teo T.Y., Zhou Y.X. & Seah
C . C . ( 2 0 0 0 ) Ge o l o g i c a l s i t e
char act er i sat i on and geot echni cal
information system. DTG Technology
Seminar 2000 Building & Infrastructure,
Singapore.

FILE C: \TEMP\DOCS\PO_42.DOC | 19 FEBRUARY 2002 | PAGE 8