Support for very weak rock associated

with faults and shear zones

Evert Hoek

Distinguished lecture for the opening of the International Symposium

on Rock Support and Reinforcement Practice in Mining, Kalgoorlie,
Australia, 14-19 March, 1999
Hoek - Support for very weak rock associated with faults and shear zones Page 2 of 20

Distinguished lecture for the opening of the International Symposium on Rock Support and Rein-
forcement Practice in Mining, Kalgoorlie, Australia, 14-19 March, 1999

Support for very weak rock associated with faults and shear
zones

Evert Hoek
Consulting Engineer, 3034 Edgemont Boulevard, North Vancouver, British Columbia, Canada,
V7R 4X1

SYNOPSIS: Controlling the stability of excavations in very weak rock associated with faults and
shear zones requires the innovative use of combinations of support systems such as rockbolts,
shotcrete, forepoles and, in some cases, yielding steel sets. Drainage and the sequence of excava-
tion play critical roles in reducing the potential for failure. These approaches are illustrated by
means of practical examples.

INTRODUCTION

Faults and shear zones present special challenges in tunnelling because they can lead to sudden
and uncontrolled collapses unless appropriate action is taken as soon as they are encountered.
The very weak and highly deformable nature of the materials and, in some cases, the presence of
large volumes of high pressure water trapped behind the impermeable fault materials results in
squeezing or flowing ground conditions. In order to control this behaviour, support must not only
have sufficient capacity but it must be installed in a sequence that does not allow uncontrolled
deformation of the tunnel.
Early detection of the presence of faults or shear zones is a very important component in the
overall process of support design. When the presence of faults is suspected in the rock mass
through which the tunnel is being excavated, a probe hole ahead of the advancing face should be
made mandatory. This hole can be percussion drilled and the rate of penetration, colour and vol-
ume of return water and the character of the chippings should be monitored during the drilling.
This will indicate significant differences in the rock mass character ahead of the face and, where
very weak rock is indicated, a diamond-drilled probe hole can be used to explore the rock in more
detail. In civil engineering tunnel driving these probe holes are typically drilled during weekend
maintenance shifts and their length should be approximately the total advance during the week
plus one tunnel diameter. This ensures that the rock has been explored for at least one tunnel di-
ameter ahead of the advancing face.
When the presence and the approximate extent of a fault or shear zone has been confirmed,
steps have to be taken to design a support system and a sequence of excavation and support in-
stallation to deal with the anticipated conditions. It is essential that all the required support ele-
ments should be available close at hand before the fault itself is exposed to any significant extent.
Depending upon the nature and extent of the fault and whether water is present, a variety of
support systems and excavation sequences can be used. Before discussing these alternatives and
considering some practical examples, it is necessary to consider some fundamental issues related
to tunnelling through weak rock.
Hoek - Support for very weak rock associated with faults and shear zones Page 3 of 20

ROCK MASS STRENGTH / IN SITU STRESS

In the context of this discussion, a rock mass is considered to be weak when its in situ uniaxial
compressive strength is less than about one third of the in situ stress acting upon the rock mass
through which the tunnel is being excavated. This can be demonstrated by means of the plot of
tunnel convergence versus the ratio of rock mass strength to in situ stress given in Figure 1. This
plot shows a sudden increase in convergence for a strength/stress ratio of less than about one
third. The plot was generated from a closed-form analysis of the development of rock mass fail-
ure surrounding an unsupported circular tunnel subjected to equal stresses in all directions. The
analysis used follows that described by Duncan-Fama (1993) and by Hoek, Kaiser and Bawden
(1995).
A Monte Carlo simulation was used to carry out this analysis for 2000 iterations for uniform
distributions of the rock mass properties, tunnel radius and in situ stress level. The rock mass
properties were varied from fair to extremely poor, corresponding to the properties of weak sand-
stones and mudstones down to material that can almost be classed as soil. The in situ stresses
were varied from 2 to 20 MPa, corresponding to depths below surface from 75 to 750 m, and the
tunnel diameters were varied from 4 to 16 metres.

30
Strain = (Tunnel convergence / tunnel diameter) * 100

20

10

2%
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Rock mass strength / in situ stress

Figure 1: Plot of tunnel convergence against the ratio of rock

mass strength to in situ stress.

CRITICAL STRAIN

Sakurai (1983) has suggested that the stability of tunnels can be assessed on the basis of the strain
in the rock mass surrounding the tunnel. The strain is defined by the ratio of tunnel convergence
to tunnel diameter. A critical strain of approximately 2% represents the boundary between ‘sta-
ble’ tunnels that require minimal support and ‘unstable’ tunnels that require special consideration
in terms of support design.
Hoek - Support for very weak rock associated with faults and shear zones Page 4 of 20

The application of this concept to practical tunnel problems is illustrated in Figure 2 that
shows the percentage strain observed during the construction of three tunnels in Taiwan1. It can
be seen that those tunnels categorised as requiring special support consideration fall above a line
that is well defined by Sakurai’s critical strain concept.
Note that all of the tunnels included in Figure 2 were constructed successfully, including those
that suffered strains of approximately 10%. In some of these cases the tunnels had to be re-mined
since the profiles were no longer adequate to accommodate the service structures for which they
were designed.
Sakurai’s critical strain of 2% has been plotted in Figure 1. It can be seen that this corre-
sponds well with the earlier conclusion that tunnels, excavated under conditions where the rock
mass compressive strength is less that about one third of the in situ stress level, will suffer serious
stability problems unless adequately supported.

100

Sakurai's critical strain condition

10
Strain - ε pc percent

0.1

Tunnels requiring special support consideration

Tunnels with no significant support problems
0.01
0.01 0.1 1 10 100
Uniaxial compressive strength of rock mass σcm - MPa

Figure 2: Percentage strain for different rock mass strengths. The

points plotted are for the Second Freeway, the Pinglin and the
New Tienlun Headrace tunnels in Taiwan.

ROCK MASS STRENGTH ESTIMATES

As a first approximation, the in situ stress can be assumed to equal the product of the depth below
surface and the unit weight of the rock mass. The uniaxial compressive strength of the rock mass
can be estimated from the Geological Strength Index (GSI) proposed by Hoek and Brown (1997)
and extended by Hoek, Marinos and Benissi (1998). This descriptive index, which depends upon
the structural characteristics and the surface conditions of discontinuities in the rock mass, is de-
fined in Table 1.
1
Information in this plot was supplied by Dr J.C. Chern of Sinotech Engineering Consultants
Inc., Taipei.
 Hoek - Support for very weak rock associated with faults and shear zones Page 5 of 20

Table 1: Table for estimating GSI (Hoek and Brown 1997, Hoek, Marinos and Benissi 1998)

Slickensided, highly weathered surfaces with

GEOLOGICAL STRENGTH INDEX

Smooth, moderately weathered and altered

Slickensided, highly weathered surfaces

Very rough, fresh unweathered surfaces

coatings or fillings of angular fragments

From the description of structure and surface conditions of the rock

Rough, slightly weathered, iron stained

mass, pick an appropriate box in this chart. Estimate the average
value of the Geological Strength Index (GSI) from the contours. Do

with soft clay coatings or fillings

not attempt to be too precise. Quoting a range of GSI from 36 to 42
is more realistic than stating that GSI = 38. It is also important to
recognize that the Hoek-Brown criterion should only be applied to

SURFACE CONDITIONS
rock masses where the size of the individual blocks or pieces is
small compared with the size of the excavation under considera-
tion. When individual block sizes are more than approximately one
quarter of the excavation dimension, failure will generally be struc-

VERY GOOD

VERY POOR
turally controlled and the Hoek-Brown criterion should not be used.

surfaces
surfaces
GOOD

POOR
FAIR
STRUCTURE DECREASING SURFACE QUALITY

INTACT OR MASSIVE – intact rock specimens or 90

massive in situ rock with very few widely spaced dis- N/A N/A N/A
continuities

80
DECREASING INTERLOCKING OF ROCK PIECES

BLOCKY - very well interlocked undisturbed rock mass 70

consisting of cubical blocks formed by three orthogonal
discontinuity sets
60

VERY BLOCKY - interlocked, partially disturbed rock

mass with multifaceted angular blocks formed by four 50
or more discontinuity sets

BLOCKY/DISTURBED - folded and/or faulted with 40

angular blocks formed by many intersecting discontinu-
ity sets

30

DISINTEGRATED - poorly interlocked, heavily broken

rock mass with a mixture of angular and rounded rock
pieces
20

FOLIATED/LAMINATED – Folded and tectonically

sheared foliated rocks. Schistosity prevails over any
N/A N/A 10
other discontinuity set, resulting in complete lack of
blockiness
5
Hoek - Support for very weak rock associated with faults and shear zones Page 6 of 20

An approximate relationship between the ratio of rock mass strength σcm to laboratory rock
strength σci and the value of GSI is given by:

σ cm = 0.019 σ ci e 0.05 GSI (1)

As an example of the application of the information presented above, consider a rock mass which
has been assigned a GSI = 23. The laboratory compressive strength of the rock mass is 5 MPa and
hence, from equation 1, the rock mass compressive strength is estimated at 0.3 MPa. A tunnel
being excavated at a depth of 150 m below surface in a rock mass with a unit weight of 0.025
MN/m3 will be subjected to an in situ stress of approximately 3.8 MPa. Hence, the ratio of rock
mass strength to in situ stress is approximately 0.08 and, from Figure 1, this is clearly in the cate-
gory where serious stability problems can occur unless appropriate support is installed.

CHARACTERISTIC LINE CALCULATION

Fenner (1938) introduced the concept of calculating the convergence associated with the forma-
tion of a ‘plastic’ zone, or zone of damaged rock, surrounding an advancing tunnel. The basic
elements of this concept are illustrated in Figure 3 which shows that the size of the plastic zone
depends upon the equivalent support pressure pi. In an unsupported tunnel the value of pi reduces
from the in situ stress value po to zero with distance from the face. The rock starts reacting to the
oncoming tunnel about one-half tunnel diameter ahead of the face. At the face about one third of
the total deformation has occurred and the final total deformation and complete formation of the
plastic zone occurs about 1.5 tunnel diameters behind the face. The relationship between the sup-
port pressure pi and the inward deformation δ of the tunnel walls is known as the characteristic
line and a typical example is shown in Figure 4. Included in this are plots of the thickness of the
plastic zone and of the reaction of a support system installed in the tunnel. This support reaction
curve is discussed in the next section of this paper.
The equations used to calculate the curves presented in Figure 4 together with a sample
spreadsheet are included in Appendix 1.

Figure 3: Assumed support pressure pi at different positions relative to the ad-

vancing tunnel face. (Not to scale).
Hoek - Support for very weak rock associated with faults and shear zones Page 7 of 20

3 35

30

Support pressure pi (MPa)

Plastic zone thickness

Critical stress
25
2 Characteristic line
20

Thickness of plastic zone 15

1
10
Support reaction
5
Equilibrium condition
0 0
0 5 10 15 20 25 30
Tunnel convergence (percent)

Figure 4: Example of characteristic line analysis including a plot

of plastic zone thickness and of a support reaction curve for a 10
m diameter tunnel at 150 m depth.

The characteristic line calculations included in Appendix 1 are based upon the assumption that
the rock mass surrounding the tunnel fails with no increase in volume. This is an appropriate as-
sumption for very weak rock masses of the type associated with faults and shear zones in which
the rock mass is likely to crush rather than to fail in a dilatant manner. A number of analyses in
which dilation and also time-dependent failure of the rock mass have been published (Brown et al
1985, Panet 1993). These analyses can give a very useful insight into the behaviour of different
types of rock masses. However, users should beware that the sophistication of the equations does
not seduce them with an illusion of precision. In fact, all these analyses are only crude approxi-
mations of the actual tunnel behaviour, because the simplifications required to allow the equa-
tions to be solved seldom reflect the actual in situ conditions. For example, all of these analyses
assume that the tunnel is circular and that it is subjected to a hydrostatic stress field in which
stresses in all directions are equal.
In spite of its limitations, the characteristic line method has one very important attribute that
makes it a very useful engineering design tool. The closed form of the equations used in these
analyses makes it possible to assign probability distributions to each of the significant variables
and to carry out Monte Carlo analyses that give distributions of the output variables. This is the
process that was used to generate the points plotted in Figure 1. This type of probability analysis
has also been used by Grasso et al (1997) to obtain an equivalent ‘factor of safety’ for tunnel sup-
port designs.

SUPPORT CHARACTERISTICS

When support is installed in a tunnel where plastic failure has occurred in the surrounding rock
mass, the support pressure pi provided by the support depends upon the stiffness of the support,
its maximum load bearing capacity and the distance from the face when it was installed. The sup-
port acts very much like a system of springs and the support pressure increases with increasing
deformation until the capacity of the system is exceeded. Hoek and Brown (1980) and Brady and
Brown (1990) have published equations for calculating the stiffness and capacity of different
support systems and these have been used to estimate the support characteristics given in Figure
5. Note that these characteristics are based upon the assumption that the support system is sym-
Hoek - Support for very weak rock associated with faults and shear zones Page 8 of 20

metrical around the tunnel. In other words, it is assumed that the steel sets and concrete or shot-
crete linings are completely circular and that the rockbolts are installed in the roof, sidewalls and
floor of the tunnel. These assumptions do not reflect typical support installations in the field and
so, as in the case of the characteristic line calculations, they should be used to explore behaviour
patterns rather than to calculate support characteristics to three decimal places.
As an example of the application of the information contained in Figure 5, consider the case of
the tunnel behaviour illustrated in Figure 4. The total strain of the tunnel without support is ap-
proximately 28%, which means that the 10 m diameter tunnel will converge 2.8 m. From experi-
ence, I would suggest that the rock mass surrounding the tunnel would not be capable of sustain-
ing this amount of convergence and that, without immediate support, this tunnel would collapse.
Immediate support in this case means the installation of a suitable support system immediately
behind the advancing face. As stated earlier, approximately one third of the final convergence has
already occurred at the tunnel face. Consequently the earliest that conventional support can be
installed is at a convergence of about 9%. It is probable that the face itself would also require
support in the form of grouted fibreglass dowels or an umbrella of forepoles.
The plot in Figure 4 shows that the thickness of the plastic zone at a convergence of 5% is
about 12 m. As a general rule, rockbolts or cables should have 1 to 2 m of anchorage in undis-
turbed rock outside the plastic zone. This means that rockbolts or cables of about 14 m would be
required to provide support in this case and installation of these systems in a 10 m diameter tun-
nel is not a very efficient operation. In view of the uncertainty associated with the reliability of
the anchorage in this poor quality rock mass, I suggest that rockbolts or cables are not an appro-
priate support system for this case and that steel sets or lattice girders should be considered.
From the information given in Figure 5, the support capacity of full-circle 200 x 200 mm wide
flange ribs, 150 x 200 mm I section ribs, 124 x 108 mm TH section ribs and three-bar lattice gird-
ers at a spacing of 1 m in a 10 m diameter tunnel have support capacities of approximately 0.5
MPa. The maximum elastic strain that can be sustained by these girders is about 1% and, since
they are installed at a convergence of 9%, this gives a total convergence of 10%. This support
behaviour is plotted in Figure 4 as the support reaction curve.
Equilibrium conditions are achieved when the convergence of the tunnel and the support sys-
tem are equal and, as shown in Figure 4, these conditions are defined by the intersection of the
characteristic line and the support reaction curve.
The ‘factor of safety’ of the support system can be defined as the ratio of the maximum sup-
port capacity to the support pressure required for equilibrium. In this case, this factor of safety is
approximately 2. While this may seem to be excessive, it must be remembered that it has been
assumed that the full-circle steel support systems function perfectly. This may be too optimistic
an assumption. As will be shown in the practical examples discussed later, there are times during
the excavation of the tunnel and the installation of support when the support elements are sub-
jected to unfavourable loading conditions and when they can be over-stressed. Hence, some re-
serve capacity is appropriate.
Note that all of the support pressures in Figure 5 have been plotted for steel set or rockbolt
spaced at 1 m and that, in order to determine the support pressures for other spacings, the equa-
tions given for each support type should be used.
When support types are combined, the total available support pressure can be estimated by
summing the maximum allowable pressures for each system. However, in making this assump-
tion it has to be realised that these support systems do not necessarily act at the same time and
that it may be necessary to check the compatibility of the systems in terms of deformation. For
example if lattice girders embedded in shotcrete are installed immediately behind the tunnel face,
they will accept load immediately while the shotcrete will accept an increasing amount of load as
it hardens (compare curves 24, 25 and 26 in Figure 5). Depending on the rate of advance of the
tunnel, it is necessary to check that the capacity of the lattice girders is not exceeded before the
shotcrete has hardened to the extent that it can carry its full share of the load.
 Hoek - Support for very weak rock associated with faults and shear zones Page 9 of 20

Section depth - mm
Flange width - mm

Thickness - mm

Curve number
Maximum support pres- Maximum support pres-

Weight – kg/m
Curve number

UCS - MPa
Age - days
sure pimax (MPa) for a sure pimax (MPa) for a
Support type tunnel of diameter D (me- Support type tunnel of diameter D
tres) and a set spacing of s (metres)
(metres)

1m 28 35 20 pi max = 57.8D −0.92

−1.23
305 305 97 1 pi max = 19.9D s 300 28 35 21 pi max = 19.1D −0.92
−1.3
203 203 67 2 pi max = 13.2D s 150 28 35 22 pi max = 10.6D −0.97
−1.4
150 150 32 3 pi max = 7.0D s 100 28 35 23 pi max = 7.3D −0.98
Wide flange rib
50 28 35 24 pi max = 3.8D −0.99
Concrete or shotcrete
lining 50 3 11 25 pi max = 1.1D −0.97
203 254 82 4 pi max = 17.6D −1.29 s
50 0.5 6 26 pi max = 0.6D −1.0

152 203 52 5 pi max = 11.1D −1.33 s

Concrete and shotcrete linings

section rib

Grouted bolts and cables

pi max = 15.5D −1.24 s

Split sets and Swellex

Wide flange steel ribs
171 138 38 6

TH section steel ribs

Anchored rockbolts
section steel ribs
pi max = 8.8D −1.27 s

Lattice girders
124 108 21 7

TH section rib

220 190 19
10.00
8 pi max = 8.6D −1.03 s 20

140 130 18
1

3 bar lattice girder 5.00

4
Maximum support pressure pimax - MPa

2 9
6
21
5
220 280 29
2.00 8
9 pi max = 18.3D −1.02 s 3 7
22
140 200 26 23
1.00

4 bar lattice girder

0.50 24
2
34 mm rockbolt 10 pi max = 0.354 s 10
17
11
2 16'
25 mm rockbolt 11 pi max = 0.267 s 12 19
25
2 18
19 mm rockbolt 12 pi max = 0.184 s 0.10
14
13
26
17 mm rockbolt 13 pi max = 0.10 s 2
0.05 15
SS39 Split set 14 pi max = 0.05 s 2

EXX Swellex 15 pi max = 0.11 s 2

Rockbolts or cables
spaced on a grid of 20mm rebar 16 pi max = 0.17 s 2 0.01
s x s metres 2 3 4 5 6 7 8 9 10 15 20
2
22mm fibreglass 17 pi max = 0.26 s Tunnel diameter D - metres
Plain cable 18 pi max = 0.15 s 2
Figure 5: Approximate maximum capacities for different
Birdcage cable 19 pi max = 0.30 s 2 support systems installed in circular tunnels. Note that
steel sets and rockbolts are all spaced at 1 m.
Hoek - Support for very weak rock associated with faults and shear zones Page 10 of 20

NUMERICAL ANALYSIS OF SUPPORT

The characteristic line calculation presented above is a very useful tool that is adequate for many
practical support design problems. However, since it only gives an estimate of the final support
capacity, it cannot be used to investigate the details of the excavation and support installation se-
quence required to deal with very difficult tunnelling problems such as those under consideration
in this paper. In such cases, a numerical analysis will provide a more complete analysis.
Fortunately, there are several excellent programs available commercially that make it possible
to carry out these analyses quickly and efficiently. Two of the best-known programs are FLAC2, a
very powerful finite difference program, and PHASE2W3 a simpler and more user-friendly finite
element program. The program PHASE2W (Windows 95 version) was used to carry out a more
detailed analysis of the problem discussed in the first part of this paper.
The tunnel to be analysed has a modified horse-shoe shape of 10 m span and it is being exca-
vated, as part of a drill-and-blast tunnel driving operation, through very weak rock at a depth of
150 m below surface. The properties of the rock mass, from the spreadsheet given in Appendix 1,
are as follows:

Geological Strength Index GSI 23

Rock mass compressive strength σcm 0.28 MPa
Friction angle φ 22.15°
Cohesive strength c 0.1 MPa
Rock mass deformation modbulus E 473 MPa
Rock mass Poisson’s ratio ν 0.3

It is assumed that the in situ stress field is hydrostatic, in other words the stresses are the same in
all directions. This is a reasonable assumption for very weak rock such as that in a fault or shear
zone, since this type of rock has already undergone failure and is incapable of sustaining signifi-
cant stress differences. Hence, even if the far field stresses are asymmetrical, the stresses within
the fault zone are likely to be approximately hydrostatic.
The first issue to be checked is the stability of the tunnel face. It has already been established,
by means of the characteristic line calculation, that the ‘plastic zone’ surrounding the tunnel is
very large and that substantial support is required to maintain its stability. What about the stabil-
ity of the face and what support measures need to be taken to keep this stable?
Figure 6 shows the results of a three-dimensional analysis carried out using the axi-symmetric
option in PHASE2W. This shows that the displacements of the face are of a similar order to those
of the tunnel walls and this suggests that special measures will be required in order to prevent
collapse of the face during excavation. These issues are discussed in detail in the following sec-
tion.

TUNNEL FACE STABILITY

Water pressure and drainage

The first practical question to be considered is whether there is likely to be water pressure be-
hind the face. Fault zones are generally less permeable than the surrounding rock mass and tend
2
Available from the ITASCA Consulting Group Inc., Thresher Square East, 708 South Third
Street, Suite 310, Minneapolis, Minnesota 55415, USA, Fax + 1 612 371 4717. Internet:
http://www.itascacg.com.
3
Available from Rocscience Inc., 31 Balsam Avenue, Toronto, Ontario, Canada M4E 3B5, Fax +
1 416 698 0908. Internet: http://www.rocscience.com.
Hoek - Support for very weak rock associated with faults and shear zones Page 11 of 20

to act as dams. Consequently, there is a reasonable chance that a large volume of high-pressure
water may be trapped behind the face. Driving the full face of a 10 m span tunnel into a fault zone
with water trapped behind it is an invitation to disaster.
It is essential, when working in such ground, that a percussion-drilled probe hole be advanced
well ahead of the face at all times. This will give warning of the presence of high pressure water
and allow time for drainage measures to be set up before the fault material is exposed in the face.
In general, drainage of the water is the most satisfactory solution and it may be necessary to in-
stall additional pumping capacity to deal with the water volume. In some cases, for example when
mining or tunnelling under the sea or under a lake, drainage may not be the best option and grout-
ing of the rock ahead of the face may have to be considered. The purpose of this grouting is to
create a zone of impermeable rock in which the water pressure and flow can be controlled during
tunnel driving.

Deformed profile

Figure 6: Vertical section through a three-dimensional finite

element model of the failure and deformation of the rock mass
surrounding the face of an advancing circular tunnel.

Face support

Once the issue of water has been taken care of, the next question is how to prevent collapse of
the face. Depending upon the severity of the problem, several options are available and these are
reviewed below.

Grouted fibreglass dowels

The simplest option for face support is to reinforce the face with grouted fibreglass dowels which
provide excellent support but are easy to mine through as the face is excavated. Experience sug-
gests that these work best when the rock mass has a low clay mineral content and where friction
Hoek - Support for very weak rock associated with faults and shear zones Page 12 of 20

rather than cohesion control its shear strength. This is because efficient operation of the dowels
requires a high strength bond between the grout annulus and the rock mass and this is difficult to
achieve in clay-rich cohesive materials. In the fault zone under consideration here, the cohesive
strength is low (0.1 MPa) and the friction angle reasonably high (22°). Hence there is a good
chance that fibreglass dowels would work well. However, given the magnitude of the ‘plastic’
zone in this case, it would be unwise to rely on such dowels as the only method of face support
and I recommend that they be used to supplement other support systems rather than to replace
them.

Partial face excavation

In tunnelling through weak ground it is generally accepted that the stability of the face depends
upon the area exposed. Consequently, one commonly used technique for maintaining stability is
partial face excavation in which the tunnel is driven in stages such that the area of each face is
small enough to control. One of the many alternative ways of doing this is illustrated in Figure 7
where a technique favoured by the German and Austrian tunnel engineers is used.

Figure 7: Partial face excavation method using a side drift fol-

lowed by top heading and bench.

In this method, a side drift is excavated first and this is supported by installing the final support,
such as steel sets embedded in shotcrete, against the outside wall and temporary support on the
vertical wall and on the floor. Where possible, this temporary support should consist of fibre-
reinforced shotcrete since this is easy to excavate when the tunnel is enlarged. However, in very
heavy squeezing conditions, heavy weld mesh or steel ribs may have to be embedded in the shot-
crete to provide sufficient support capacity. In many cases it is advantageous to drive the side
drift all the way through the fault or shear zone, since it can be used to establish efficient drainage
and ventilation arrangements before the main drive is attempted.
Hoek - Support for very weak rock associated with faults and shear zones Page 13 of 20

The excavation of the side drift is followed by opening the top heading to full span and this
involves destruction of the temporary wall and the extension of the temporary invert to full span.
This process should be carried out in steps of a few metres since one side of the top heading is
effectively unsupported during the excavation process.
Excavation of the bench to form the full tunnel profile should also be carried out in steps of a
few metres. The removal of the temporary invert leaves the top heading support ‘suspended’ until
the lower legs of the steel sets can be installed and the final invert placed. In heavy squeezing
conditions, a vertical bench parallel to the face is a good method of excavating the full profile
since it allows systematic installation of the set and closure of the invert to be carried out with the
benefit of support from the bench.
Placing steel sets and the application of shotcrete in a large tunnel requires heavy equipment to
operate close to the face. This means that placing the final invert can present significant practical
problems. Many tunnellers will attempt to advance the tunnel as far as possible and to leave the
invert to be completed as an off-line activity tens of metres behind the face. In squeezing ground
such as that under consideration here, this is always a serious mistake since it will allow floor
heave and severe inward deformation of the installed roof and sidewall support. Failure to close
the invert in time is probably the most common cause for the failure of support systems in
squeezing ground.
The complete process of side drift, top heading and bench excavation has been simulated by
numerical analysis and the results are presented in Figure 8. In spite of the approximations re-
quired in carrying out this analysis, the results provide a very useful guide to the adequacy of the
excavation sequence and support systems.
Comparison of the results of the characteristic line calculation (Figure 4) and the finite element
analysis (Figure 8) shows that the stepwise excavation and support installation sequence provides
a significant improvement on the full-face excavation process depicted in Figure 4. Convergence
is limited to about 5% and the maximum depth of the plastic zone in the roof is about 15 m. Note
that this finite element model has been gravity loaded and the self-weight of the broken rock re-
sults in more failure in the rock above the roof than in that below the floor.
In the finite element analysis it was assumed that the final support consists of steel sets, typi-
cally 200 x 200 mm wide flange sections, embedded in shotcrete. While it is difficult to incorpo-
rate the properties of this composite system into currently available numerical models, it is possi-
ble to make a reasonable approximation based upon a consideration of when different elements of
the support system are activated.
On the basis of such assumptions it was found that some minor yield of the support occurred,
particularly at the connections between the upper and lower portions of the sidewall support and
at the junctions of the sidewall support and the invert. This yield, which is associated with the
sequence of excavation and support installation, is relatively common in this type of support sys-
tem and has no major practical significance. It normally shows up as minor spalling in the shot-
crete and this can easily be repaired by chipping out the damaged material and applying fresh
shotcrete. Unless there is on-going time-dependent deformation, it is unlikely that this damage
will recur once the damage has been repaired.

Advancing under a forepole umbrella

As an alternative to the partial face method described above, the umbrella arch method is some-
times used, particularly by Italian tunnellers, for advancing through difficult ground (Carrieri et al
1991). In a 10 m span tunnel of the type being considered here, the method would typically in-
volve installing 12 m long 75 mm diameter grouted pipe forepoles at a spacing of 300 to 600 mm.
These forepoles would be installed every 8 m to provide a minimum of 4 m of overlap between
successive umbrellas. A sketch of a typical forepole umbrella is given in Figure 9.
 Hoek - Support for very weak rock associated with faults and shear zones Page 14 of 20

Deformed boundary
Temporary support

a. Excavation boundaries defined but no rock re- e. Installation of support in the form of steel sets em-
moved, model allowed to consolidate. bedded in shotcrete on tunnel walls and temporary
shotcrete invert. Removal of internal pressure.
Internal pressure 0.5 MPa
Yield of rock
Internal pressure
Deformed boundary 0.4 MPa

b. Excavation of side drift and application of an in-

ternal pressure of 0.5 MPa to simulate support f. Excavation of lower bench and application of in-
provided by the advancing face. ternal pressure of 0.4 MPa to simulate support
provided by face.
Temporary support

Deformed boundary

c. Installation of support in the form of steel sets em-

bedded in shotcrete against the tunnel wall and
temporary shotcrete invert and vertical partition.
Removal of internal pressure. g. Completion of sidewall support and casting of
concrete invert. Removal of internal pressure to al-
Internal pressure
0.5 MPa low full load on the completed support.

Figure 8 : Finite element analysis simulating the ex-

cavation sequence for mining a 10 m span tunnel
through a fault at a depth of 150 m below surface. Fi-
nal displacements for the roof are 250 mm, sidewalls
d. Excavation of top heading and application of inter- 275 mm and floor heave is 190 mm.
nal pressure of 0.5 MPa to newly excavated right
hand side to simulate support from face.
Hoek - Support for very weak rock associated with faults and shear zones Page 15 of 20

A first step in this method usually involves drilling holes, up to 30 m ahead of the face, for drain-
age. This is followed by the drilling of the 12 m long holes and installation of the pipe forepoles
to form the umbrella arch. In some cases, depending upon the nature of the rock mass being tun-
nelled through, jet-grouted columns are used rather than the grouted pipe forepoles. The tunnel is
then advanced 8 m before the cycle is repeated to create another protective umbrella.
This method is frequently used in combination with other support systems such as steel sets
embedded in shotcrete, face stabilisation by grouted fibreglass dowels and the use of a temporary
invert to control floor heave.

Figure 9: Sketch of tunnelling under the protection of a forepole umbrella.

Analysis of the performance of this system, particularly when used in combination with other
support systems, is extremely difficult. Simplified finite element analyses, using axi-symmetric
models, have been described by Grasso et al (1993) and similar studies, using FLAC3D have
been described by Itasca (1997). However, these analyses are certainly outside the type of studies
that could be considered part of routine tunnel engineering design and a great deal of reliance has
to be placed on judgement and experience. It is unlikely that this situation will change any time
soon.
In spite of the lack of design guidelines and analytical tools, the umbrella arch is a very power-
ful tunnelling tool which, in the hands of an experienced operator, can be used to excavate
through very difficult ground conditions.

Yielding steel sets

A method that has been used successfully in many tunnels involves the installation of steel sets
fitted with sliding joints such as that shown in Figure 10.
Sánchez and Terán (1994) describe the use of yielding elements in steel ribs for the support of
the Yacambú-Quibor tunnel in Venezuela – regarded by many as one of the most difficult tunnels
in the world. This 5.5 m diameter water supply tunnel through the Andes is being excavated
though weak rock masses, including graphitic phyllites, at a maximum depth below surface of
1200 m.
In the weakest rock sections, the support consists of WF6x20 steel sets, at 1 m spacing, with
two sliding joints. These joints are set to lock when an additional tunnel closure of about 300 mm
has been achieved. The sets are installed immediately behind the tunnel face and they are embed-
ded in shotcrete, except for a 1 m wide ‘window’ that is left for each of the sliding joints. Once
the joints have moved and locked, usually between 5 and 10 m behind the face, the ‘windows’ are
Hoek - Support for very weak rock associated with faults and shear zones Page 16 of 20

closed to complete the shotcrete lining. This support system has proved to be very effective and
measurements of tunnel convergence, carried out over several years, have shown that the tunnel is
completely stable.

Figure 10: Assembly of a sliding joint in a Toussaint-

Heintzmann or Top Hat section steel rib.

The installation of these sets immediately behind the face provides security for the men work-
ing at the face, in spite of the fact that the sets apply relatively little active support pressure to the
rock mass during the sliding stage. The settings of the amount of sliding allowed in the joints is
judged on the basis of the amount of tunnel convergence that can be tolerated before the full sup-
port reaction is activated (see Figure 4). Correct setting of the joints will achieve equilibrium be-
tween the tunnel convergence and the support reaction at much lower support pressures than for
rigid steel sets. Consequently, provided that large tunnel convergence is acceptable, much lighter
section support can be used than would otherwise be required.

Multiple tunnels

In one hydroelectric project in India, tunnelling through a large fault at depth proved to be ex-
tremely difficult and several collapses of the tunnel occurred as attempts were made to drive the
tunnel at full size. Eventually a decision was made to split the tunnel into three smaller tunnels
that would provide the same overall cross-section area for transmission of the water. Driving
these three smaller tunnels, while still difficult was ultimately successful because, as shown in the
case of the partial face excavation, dealing with smaller cross-sections has many practical advan-
tages when tunnelling in difficult ground.

CONCLUSION

Tunnelling through very weak rock associated with faults and shear zones is a difficult problem,
particularly when carried out under high in situ stress conditions. Conventional tunnel support
such as rockbolts and shotcrete are seldom adequate to deal with heavy squeezing conditions that
can occur when the rock mass surrounding the tunnel fails to a depth of several tunnel diameters.
Several alternative methods for maintaining stable face and tunnel stability have been explored
in this paper. In some cases, limited theoretical analyses of the support systems are possible
while, in other cases, reliance has to be placed on judgement and experience.
Tunnel driving costs, under the conditions described, are typically about three time average
tunnel driving costs and advance rates seldom exceed about 1 m per day. Attempting to save
money and time by adopting short-cuts or inadequate solutions invariably lead to even costlier
failures.
Hoek - Support for very weak rock associated with faults and shear zones Page 17 of 20

Faults and shear zones exist in almost every rock mass and so it is inevitable that a tunnel engi-
neer will be faced with the situations described in this paper at least once in his or her career. It is
as well to attempt to learn from the experience of others rather than to attempt to sort out the
problems when faced with a tunnel collapse. Reading about these problems is not a substitute for
visiting a tunnel in which mining through a fault is in progress. Tunnel engineers should take
every available opportunity to visit such tunnels.

REFERENCES

Brady, B.H.G. and Brown, E.T. (1985). Rock mechanics for underground mining. Allen and Un-
win, London.
Brown E.T., Bray J.W., Ladanyi B. and Hoek E. 1983. Characteristic line calculations for rock
tunnels. J. Geotech. Engng Div., Am. Soc. Civ. Engrs 109, 15-39.
Carrieri, G., Grasso, P., Mahtab, A. and Pelizza, S. (1991). Ten years of experience in the use of
umbrella-arch for tunnelling. Proc. SIG Conf. On Soil and Rock Improvement, Milan 1, 99-
111.
Duncan Fama, M.E. (1993). Numerical modelling of yield zones in weak rocks. In Comprehen-
sive rock engineering, (ed. J.A. Hudson) 2, 49-75. Pergamon, Oxford.
Fenner, R. (1938) Undersuchungen zur erkenntnis des gebirgsdruckes. Gluckauf.
Grasso, P., Mahtab, M.A., Rabajoli, G. and Pelizza, S. (1993). Consideration for design of shal-
low tunnels. Proc. Infrastructures souterraines de transport, Toulon, 19-28.
Hoek, E., and Brown, E.T. (1980). Underground excavations in rock. Instn Min. Metall., London.
Hoek, E. and Brown, E.T. (1997). Practical estimates or rock mass strength. Int. J. Rock Mech. &
Min. Sci. & Geomech. Abstracts. 34(8), 1165-1186.
Hoek, E., Kaiser, P.K. and Bawden. W.F. (1995). Support of underground excavations in hard
rock. Balkema, Rotterdam.
Hoek, E., Marinos, P. and Benissi, M. (1998) Applicability of the Geological Strength Index
(GSI) classification for very weak and sheared rock masses. The case of the Athens Schist
Formation. Bull. Engg. Geol. Env. 57(2), 151-160.
Itasca (1997). Excavation and support for a shallow tunnel. Company brochure on FLAC3D ap-
plications.
Panet, M. (1997). Understanding deformations in tunnels. In Comprehensive rock engineering,
(ed. J.A. Hudson) 1, 663-690. Pergamon, Oxford.
Sakurai, S. (1983). Displacement measurements associated with the design of underground open-
ings. Proc. Int. symp. field measurements in geomechanics, Zurich 2, 1163-1178.
Sánchez Fernández, J.L. and Terán Benítez, C.E. (1994). Túnel de Trasvase Yacambú-Quibor.
“Avance Actual de los Trabajos de Excavación Mediante la Utilización de Soportes Flexibles
Aplicados a Rocas con Grandes Deformaciones”. Proc. IV Congreso Sudamericano de Me-
canica de Rocas, Santiago 1, 489-497.
Hoek - Support for very weak rock associated with faults and shear zones Page 18 of 20

APPENDIX 1: ANALYSIS OF PLASTIC FAILURE

It is assumed that the onset of plastic failure, for different values of the effective confining stress
σ 3', is defined by the Mohr-Coulomb criterion and expressed as:

σ1'= σ cm + kσ 3' (A.1)

The uniaxial compressive strength of the rock mass σ cm is defined by:

2c 'cos φ '
σ cm = (A.2)
(1 − sin φ ')

and the slope k of the σ1' versus σ 3' line as:

(1 + sin φ ')
k= (A.3)
(1 − sin φ ')

where σ1' is the axial stress at which failure occurs

σ3' is the confining stress
c' is the cohesive strength and
φ' is the angle of friction of the rock mass

In order to estimate the cohesive strength c 'and the friction angle φ 'for an actual rock mass,
the Hoek-Brown criterion (Hoek and Brown 1997) can be utilised. Having estimated the parame-
ters for failure criterion, values for c 'and φ 'can be calculated.

Analysis of tunnel behaviour

Assume that a circular tunnel of radius ro is subjected to hydrostatic stresses po and a uniform
internal support pressure pi as illustrated in Figure A.1. Failure of the rock mass surrounding the
tunnel occurs when the internal pressure provided by the tunnel lining is less than a critical sup-
port pressure pcr , which is defined by:

2 po − σ cm
pcr = (A.4)
1+ k

If the internal support pressure pi is greater than the critical support pressure pcr, no failure oc-
curs, the behaviour of the rock mass surrounding the tunnel is elastic and the inward radial elastic
displacement of the tunnel wall is given by:
Hoek - Support for very weak rock associated with faults and shear zones Page 19 of 20

ro (1 + ν)
u ie = ( p o − pi ) (A.5)
Em

s modulus or deformation modulus and ν is the Poisson'

where Em is the Young' s ratio.

Figure A.1: Plastic zone surrounding a circular tunnel.

When the internal support pressure pi is less than the critical support pressure pcr, failure occurs
and the radius rp of the plastic zone around the tunnel is given by:

1
2( p o (k − 1) + σ cm ) ( k −1)
r p = ro (A.6)
(1 + k )((k − 1) pi + σ cm )

For plastic failure, the total inward radial displacement of the walls of the tunnel is:

2
r (1 + ν ) rp
u ip = o 2(1 − ν )( p o − p cr ) − (1 − 2ν )( p o − pi ) (A.7)
E ro
Hoek - Support for very weak rock associated with faults and shear zones Page 20 of 20