ASSOCIATION
INTERNATIONALE DES TRAVAUX
ITA
EN SOUTERRAIN
INTERNATIONAL
TUNNELLING
AITES
ASSOCIATION
In Consultative Status, Category II with the
Towards an
United Nations Economic and Social Council
improved use
http://www.ita-aites.org
of underground
Space
Topic
CONVENTIONAL TUNNELLING
Title
Ideas on tunnel stability
Author
ITA WG Conventional Tunnelling, P. Duffaut, J. Piraud
published
in "Tribune",
Nr. 22, pp. 32 - 36, Year 2002
by ITA - AITES, www.ita-aites.org
WG 19 - "Conventional Tunnelling"
Working Group:
Open Session, Seminar, Workshop: -
Paper
Others:
Abstract: -
R�sum�: -
Remarks:
It is difficult to discuss "Ideas on Tunnel Stability" without their immediate association with the construction
methods that apply such ideas, or that were at the origin of such ideas. During tunnelling operations,
construction methods always need to be adapted to a natural environment which never is exactly as it was
imagined during the design process. Therefore a good s u p p o rting method has to fit the changing anatomy
and physiology of the rock mass, and should be evaluated for its adaptability to the laws and whims of Nature,
rather than for its theoretical qualities.
The following paper discusses tunnels driven with conventional methods and not those made with a shield or
tunnel-boring machine.
Secretariat : ITA-AITES c/o EPFL - B�t. GC  CH-1015 Lausanne - Switzerland
Fax : [phone] - Tel. : [phone] - e-mail : [e-mail] - www.ita-aites.org
I D E A S O N T U N N E L S T A B I L I T Y
Pierre DUFFAUT, Consulting Engineer, Paris & Jean PIRAUD, ANTEA, Orl�ans, France
It is difficult to discuss "Ideas on Tunnel Stability" Finally, if the rock mass is too strongly fractured or
without their immediate association with the construc- even soil like, self-support is no longer viable and the
tion methods that apply such ideas, or that were at the only answer is to strengthen the rock mass as a whole,
through extensive grouting or rock bolting.
origin of such ideas. During tunnelling operations,
construction methods always need to be adapted to a
THE TIME OF PASSIVE SUPPORT
natural environment which never is exactly as it was
During the 19th and the first part of the 20th Century,
imagined during the design process. Therefore a good
before the time of shotcrete and rock bolts, tunnel men
supporting method has to fit the changing anatomy
had no better way than dividing the full tunnel section
and physiology of the rock mass, and should be eva-
into smaller ones, bored one after the other, each of
luated for its adaptability to the laws and whims of
them heavily timbered. It should not come as a surpri-
Nature, rather than for its theoretical qualities.
se that the tunnel construction resulted at that time in
The following paper discusses tunnels driven with
many "national" methods, known then as English,
conventional methods and not those made with a
Belgian, French, German, Austrian and Italian tunnel-
shield or tunnel-boring machine.
ling methods (see for example Duffaut and Piraud,
1975).
THE PRINCIPLE OF SELF-SUPPORT
In those methods, the place of self-support was res-
tricted to so-called "good ground", and elsewhere to a
In the beginning, the self-support of holes and cavities
few tiny spaces between timber pieces. This timber
went without saying. This basic principle, also named
s upport was temporary and had to be replaced by
arch effect or arching, gave stable caverns to the pre-
thick vaults of masonry or brickwork, likely to stand for
historic men, even inside low strength rocks provided
many years. The major fault of such structures was
they harbour not too many joints.
that their contact with the surrounding rock mass was
Another type of self-supporting struc tu res are the
both late (it took weeks or months before the vault
caverns dug below a thick bed, like room and pillar
could be constructed) and discontinuous (a smooth
mining, which had been widely used for a long time at
vault cannot fit against a rough rock walls). This delay
the border of the limestone plateaux.
provided time and space for weathering and loosening
Early architects also used self-support in order to build
of the rock mass, which, little by little, will impose
vaults, for bridges as well as for churches or base-
increasing loads upon the vault and justify its thick-
ments. These vaults are stable even without any
ness a posteriori.
cement, provided that the building blocks are well fit-
After the Word War I, two major innovations appeared:
ted. Such arches face the gravity forces by progressive
As temporary support, steel ribs replaced timber, allo-
stressing of their elements. Above ground such arches
wing wider spans to be supported; this method, stan-
need heavy buttresses to provide for lateral abut-
dardised by Te rzaghi (1949), later became known as
ments; below the surface, however, full 3D abutments
AMSS, American Method of Steel Support;
are provided by Nature for free.
As definitive lining, cast-in-place concrete favourably
What is true for built arches is also true for tunnels and
replaced masonry, providing an intimate contact bet-
caverns, whether natural or man-made:
ween vault and rock mass, even if somewhat late.
If the rock is strong enough to bear the tangential
Both innovations introduced a type of industrialisation
stress along the perimeter (approx.: Rc > 3łH around a
in tunnelling, allowing faster boring with less manpo-
circular hole), and if there are no ill-oriented joint sur-
wer. But this progress hid two fundamental drawbacks
faces, the hole will stand without any support. Natural
of the passive support: an excessively long operational
caves as wide as 50 m are not uncommon, and the lar-
delay and a discontinuous contact between the sup-
gest known to date spans almost 400 m (Good Luck
port and the surrounding rock mass before concre te
Cave, Sarawak, Malaysia).
casting. As a consequence the final lining was desi-
When the strength of the rock is exceeded along the gned and justified as a bearing structure, submitted to
perimeter of the hole, and provided that the rock mass i l l - m a s t e red or unknown external loads; hence the
s tructural calculations by the well known "springs
is rather homogeneous, a plastic ring appears around
method", based on rather rough hypotheses, easily
the hole, accommodating the excess of stre s s .
adjustable to any presupposed results...
Excessive damage to the rock mass by plastic strain
may be prevented by a minimal skin support such as a These drawbacks, intrinsic to passive supports, were
thin shotcrete cover. tolerable for most shallow tunnels, as well as for some
TRIBUNE n�22 - ITA-AITES - June 2002 32
I DEAS ON TUNNEL S T A B I L I T Y
deep tunnels in hard rocks like the first Gotthard rail- M�ller. After successful applications to exceptional
way tunnel in 1875. However, they became prohibitive works such as the Frankfurt metro or the Wa ldeck
when the Simplon tunnel, about 1900, had to cross underground power station, in central Germany, the
crushed rock at great depth. No vault, whatever its method appealed to some interested French engi-
thickness, could withstand the thrust of 2000 m of neers, among them Claude Louis. He worked with the
mountain. One had to understand that the only way authors to make French contractors aware of the
was "to let the rock mass come", as coal miners used major progress brought by NATM, and of the potential
to do: being accustomed to the behaviour of rocks at savings brought to owners; he showed that in some
greet depth, they had learned how to tame soft or cru- cases this new method was the only technical answer
shed rocks like coal. (cf. Louis, 1972).
Such alpine experience was the basis for the Swiss Some relevant images were proposed at that time to
engineers Maillart and Andreae, who had the intuition illustrate the core of the message: the new technolo-
of the shotcrete efficiency some decades in advance: gies, shotcrete and rockbolts, could be compared to
Maillart, 1922: "masonry must tightly fit the rock-mass glue and nails as used in woodwork; the four main
contour", Andreae, 1949: "a light support, but quickly points of NATM may be associated with simple formu-
installed, is the most efficient mean against ro c k las:
thrust".
Excavating with care, thanks to a rounded cross sec-
As Rziha wrote in his Handbook for Tunnelling as early tion and a smooth blasting, may be compared to care-
as 1874, "the true art of the engineer is to prevent the fully dismantling the rock mass (instead of blasting it at
build-up of rock thrust instead of overcoming it after it full strength);
has built up". This fundamental target has been forgot-
Protecting the excavated rock surface with a shotcrete
ten during almost a century, through the lack of techni-
skin, equals dressing and bandaging a wound;
cal means likely to achieve it.
Rock bolting equals reinforcing the ground, as usually
done in reinforced concrete;
THE INVENTION OF NATM
Monitoring the ground behaviour implies relying on
measurements rather than on calculations.
Some years later, around 1960, the same acknowled-
gements as made in Switzerland led the Austrian engi- The complex formed by rock mass + rockbolts + shot-
crete may be com-
neer von Rabcewicz to
pared to a complex
put into practice the
of re inforced earth,
principles quoted
i.e. a granular soil
above. He implemented
mass + metallic
two newly available
strips + facing
technologies, i.e. ro c k
panels. In both
bolts borrowed fro m
cases, the main
coal miners and shot-
mechanical role is
crete coming from
played by the re in-
America, where such
f o rced gro u n d
material was used for a
mass; shotcret e
long time to rep a i r
acts like a skin
damaged stru c t u re s .
Analogy between a tunnel support with
shotcrete and rock bolts (below) and a only, whose role is
Comparison between a thick, passi- The target of the
vault of re inforced earth (above); experi-
ve support (on the left) and the "bea- method was clearly
vital for pro t e c t i n g
mental set-up studied by Behnia, 1973.
ring ring of re inforced rock" of the
the freshly exposed
posted: "help the rock
NATM (from Louis, 1972).
rock mass, but subordinate from the mechanical point
mass to support itself",
of view.
instead of striving to support the rock thrust with the
help of a lot of heavy members As a conclusion, NATM expressed in the art of tunnel-
ling the paradigm shifts that dominated geotechnics
Austrian engineers pointed out that the cross section
along the second part of the 20th Century: to improve
shape and the excavation method were crucial in order
to avoid, as far as possible, any damage to the sur- the properties of in-situ ground instead of trying to
make up for its weakness through costly artificial
rounding ground. They understood that they had to
structures. Plenty of ground is available around tun-
measure accurately the reactions of the ground with
nels, providing free 3D abutments; to obtain the best
time, just as did for a long time the dam engineers.
of it, one only needs to inject intelligence. The English
So began the "new" Austrian Tunnelling Method,
word "support" (like the French word "soutŁnement")
NATM, which was to become the spearhead of the
is misleading: the action of shotcrete and rockbolts is
" S a l z b u rg Club", under the aegis of Prof. Leopold
TRIBUNE n�22 - ITA-AITES - June 2002 33
I D E A S O N T U N N E L S T A B I L I T Y
to maintain, retain, or contain the ground, but never to soft rocks, any deformation must be prevented in
sustain or support it (Duffaut, 1994). order to avoid irreversible shearing in the ground and
limit surface settlements. In the 1980s, Austrian engi-
The main rese rvation, though, is that this method
neers had the idea of replacing the top heading with
absolutely requires a great rapidity of implementation,
two ogival sidewall drifts, whose shape is much better
a rigorous work organisation, and real-time monitoring.
suited to the low horizontal stress that prevails near
The lack of any one of these three conditions may
the ground surface. The subways of Vienna and most
create a dangerous situation; this is why it can only be
Ge rman towns, as well as 150km of German high-
operated with highly qualified and experienced staff at
speed railway tunnels, were constructed in this man-
all levels.
ner, if necessary after grouting or drawing down the
During the 1970s and 1980s hundreds of kilometres of
groundwater level. However, in order to avoid having
NATM tunnels were bored successfully, the more in
to pay the Austrians an export premium, the German
Austria, Germ a n y, Switzerland, Italy, and Japan.
engineers renamed the method as "Spritz betonbau-
Obviously, some failures have been reported. The most
weise" (literally: sprayed concrete lining method).
famous examples are in Munich (Germ a n y, 27th
Outside the Germanic world, the main tunnellers, i.e.
Sep.1994) and below the Heathrow airport (UK, 21st
Italians and Japanese, adopted NATM practices in the
Oct. 1994). The first case was associated with an
1970s and China did so in the 1980s. Since then,
unexpected local deepening of the alluvial gravel down
these countries have used just two tunnelling
to the tunnel crown; the second with a too optimistic
methods: either TBM, or NATM.
design for three parallel tunnels, a method proved valid
inside the chalk of the Channel tunnel but no longer so American and English engineers, however, ignored this
in the London Clay. A French advertising slogan sug- "continental" invention for a long time. England disco-
gests that "one glass of wine is harmless for car dri- vered it belatedly during work for the Channel Tunnel,
vers, but the damage begins with the second": this and brilliantly applied it to excavating an enormous
ought to be the same rule for tunnels in soils. cavern for the tunnels crossover at 40 m below the
seafloor. The major British engineering firms have tried
to rename NATM as SCL ("Sprayed Concrete Lining"),
THE TRIUMPH OF NATM
a name that places undue emphasis on the re latively
During the 1970s and 1980s, the concept of a "bearing
minor role played by shotcrete.
ring of reinforced rock" became the essential feature in
the world for conventional tunnelling. Still, its applica-
THE FRENCH EXCEPTION
tions varied depending upon the thickness of the over-
burden: NATM arrived in France at the beginning of the 1970s
and was successfully used in soft rocks under low
For deep tunnels, the major concern is to accompany
overburden, in particular for highway tunnels at Nice,
the elastic or elasto-plastic stress release of the rock,
preventing it from becoming disorganised, or, in milita- for the railway tunnel at Grigny near Paris and for
many projects of EDF (the
ry terms, by favouring its ordered retreat rather than a
French Electricity Board). As
rout. If the expected convergence is such that the
had happened in Germany,
shotcrete shell might break, it is necessary to create
the French tried to gallicise
longitudinal grooves that will close once the rock mass
the name of NATM with the
has converged. Such
short-lived French acronym
grooves delimit indepen-
PTSS (for Participation du
dent parts of the vault
terrain stabilis� au soutŁne-
that are literally suspen-
ment), introduced in 1982 in
ded from the rock mass
the handbook for
by means of rock bolts, in
Government contracts.
the same manner as a
"Berliner" wall is attached During the same period,
Characteristic curves of sup-
to the soil mass by pres- F rench engineers gave an
port and ground response
tressed anchors. The i n t e resting contribution to
(from Hudson, 1993)
Ta u e rn, Karawanken, the tunnel design by deve-
Arlberg and Inntal tunnels loping the "Convergence-confinement" method under
Variants of the NATM:above,
number among the most the guidance of Marc Panet. This theory uses mathe-
deep tunnel with longitudinal
d ifficult tunnels driven matical terms for expressing the behaviour of the ring
grooves,below, shallow tunnel
with ogival sidewall drifts. t h rough the Alps in this of rock around a tunnel: the stresses imposed on the
manner. support generally decrease with the continuing
convergence of this ring (Panet, 1995). This is the clas-
For shallow tunnels, commonly driven through soil or
TRIBUNE n�22 - ITA-AITES - June 2002 34
I DEAS ON TUNNEL S T A B I L I T Y
EMERGENCE OF THE NEW ITALIAN METHOD
In soft or weak ground, or under conditions where the
rock strength cannot withstand the overburd e n ,
experience had taught that not only the tunnel walls
but also the face have to be supported. This can be
done in various ways (cf. Pelizza et al., 1993):
By keeping a stabilising mass of ground (short bench)
in the middle of the face;
By spraying concrete on the tunnel face;
By reinforcing the face with horizontal fibreglass
bolts;
Variation of the radial displacement u for supports of various stiff-
By forepoling with steel bars or pipes ("umbre lla")
ness (from Panet, 1973).
along the future tunnel walls;
sic "Support supply and support demand" compari-
By creating resistant arches along these future walls,
son. Even though the method uses 2D calculations, it
either by mechanical pre-cutting and concrete "pre-
implicitly considers the 3D role played by the tunnel
vaults"(Perforex method), or by jet grouting.
face, using the "unloading factor".
All these pre-support methods have the same objec-
In practice, however, many French engineers conti-
tive: avoid collapse of the tunnel face, which would
nued to ignore the specifics of NATM, banned the
render useless all further wall support. The above
term from the French vocabulary, and continued the
mentioned techniques developed in many countries
application of the good old "passive support" sys-
during the 1980s, but without a global understanding
tem, which is more lucrative in poor ground condi-
of the mechanisms involved; in fact, practice largely
tions when the arches are paid by weight (!). This
preceded theory. In France, the major step was the
conformism contrasted with the rapid adoption of
Galaure TGV tunnel south of Lyons, which was suc-
TBM and shields by the French contractors after
cessfully excavated by a JV led by Bouygues in
1985.
1991-93; notwithstanding an exceptional cross sec-
In fact, the senior author was one of the few in the
tion of 150 m2, it was dug in full section, after mecha-
French Tunnelling Society (AFTES) who persisted not
nical pre-cutting and face re i n f o rcement by 18-m
only in defending NATM, but also in calling it by its
horizontal rock bolts.
proper name and in reminding all and sundry of its
Although the pre-
successes. At the same time, a bizarre controversy
vaults method
arose in some scientific circles concerning the exact
was patented in
definition and even the existence of this method (cf.
France, the deve-
Kovari, 1994). It is true that its many application
lopment and sys-
variants have disconcerted some theoreticians, but in
t e m a t i c
practice a NATM tunnel is immediately recognised by
application of
some common and clear characteristics:
p r e - s u p p o r t
Confidence in self-support and rounded shapes;
methods came
Immediate use of shotcrete and rock bolts;
from Italy, where
Rapid closure of the section and careful monitoring of
much of the tun-
deformation.
nelling ground is
Curiously, the French engineers have perfectly inte-
composed of
grated NATM practices in their intern a tion al
weak or soft
contracts. A striking example is the CERN caverns
rocks, often
that are being completed near Geneva by a joint
s trongly folded
Comparative analysis of the deformation
EDF-Knight Piesold engineering venture. A distinctive
and faulted.
response of a rock mass, using the
characteristic of these openings with a span of
Austrian and Italian methods (from Lunardi,
P i e t ro Lunard i
2000)
almost 20 m is that, for the first time, the shotcre te
had the merit of
shell is ignored as a structural element in the stability
unifying these techniques by proposing a tunnel-sta-
calculations. This is not to say that its role is negli-
bility theory that is based entirely on the equilibrium
gible; on the contrary, thanks to this shell the sur-
conditions of the "advance core". This theory favours
rounding rock does not loosen and retains its original
an axial-symmetric, rather than transverse, stability
p rope rties, or even improved ones thanks to rock
approach; the decisive parameter is the axial defor-
bolting (cf. Laigle et al., 2001).
mation ("extrusion") of the advance core, which pre-
TRIBUNE n�22 - ITA-AITES - June 2002 35
I DEAS ON TUNNEL S T A B I L I T Y
cedes and causes the convergence of the tunnel walls. tunnels of 135 m2 cross section for a total length of 73
km, is the most striking example of this new method;
This theory is known as ADECO-RS (Analisi delle
within year 2000, the site as a whole saw up to 30
deformazioni controllate nelle rocce e nei suoli). Like
NATM some 30 years before, it proposes a true "cultu- simultaneous advance faces, producing more than 3
km per month of
ral" revolution in that:
finished tunnels.
Everything revolves at the face: a tunnel with an uns-
At the same time,
table face will collapse, even though its walls may be
the pro g ress in
supported;
numerical 3D
The next step is to close the ring as rapidly as pos-
modelling has
sible: the cross section must not be divided, even and
clearly identified
especially in poor ground conditions!
the real gro u n d -
Where NATM seeks to create a bearing ring of reinfor-
s t a b i l i s a t i o n
ced rock around the excavated section, the New
m e c h a n i s m s
Italian Method seeks to create good ground ahead of
during excavation,
the tunnel face (and if necessary around the future
thus confirming the
walls), in order to be able to dig a stable tunnel.
intuitive conclu-
This new method can thus be considered as a 3D
Examples of standard cross sections used
sions of the inven-
for the new high-speed-rail tunnels
application of NATM principles, to be used in soft and
tors of the
through the Apennines (from Lunardi,
unstable ground where the latter would be impotent. In
2000)
p r e - s u p p o r t .
retrospect, we can see that the Austrian engineers did
Speaking in
somewhat neglect the mechanical role of the tunnel
images, the advance core can be assimilated to an
face. However, in both cases we are dealing with a
abutment of re i n f o rced earth, with the longitudinal
method that helps the ground in accepting the presen-
bolts playing the role of metallic strips; the last "spans"
ce of a hole, or prepares it in advance for the presence
of a tunnel are no more than a "provisional link" bet-
of a hole in the case of the Italian Method.
ween this abutment and the last ring of cast-in-place
In practice, the Italian engineers apply these precepts
concrete.
as follows:
There is little doubt that the New Italian Method will
The tunnel is to be driven full face, re g a rdless of
become universally accepted for tunnelling in unstable
ground quality; the main variable of adjustment is the
rocks at low and moderate depth, as did the New
type and "intensity" of the face reinforcement, which is
Austrian Method some 30 years earlier for tunnelling in
easily modulated without disturbing the tunnel advan-
jointed rock. The reasons are the same: both methods
ce;
are based on two geotechnical principles that always
In very poor ground conditions, the face reinforcement
lead to saving money:
is to be completed by pre-supporting means around
A better mechanical understanding of the gro u nd
the future walls, such as pre-vaults, forepoling or jet-
mass, enabling the optimum mobilisation of its own
grouting,
strength;
Most of the ground support and reinforcement is ins-
A systematic use of easily adaptable techniques for
talled horizontally, i.e. longitudinally, every 15 to 20 m
re i n f o rcing the ground as and where necessary, at
only and no longer at each round;
each step of tunnel advance.
Support of walls after excavation consists of steel ribs
linked by small shotcrete vaults; this is a relatively rigid
All the references for this article will be available on the ITA web site
support, as in the case where NATM is used at shallow
depths;
The concrete lining is cast in two stages only (first floor
and then vault), at less than a few diameters from the
face and a few days only after excavation; the concre-
te ring is almost always closed in the floor.
Italian engineers have invented specially adapted
machines for the different types of longitudinal sup-
port. Always available on site, they allow a true indus-
trialisation of tunnel construction in soft and weak
rocks, with advance velocities and an economy of
means that were hitherto unheard of. The new high-
speed rail connection through the Apennines, with 9
TRIBUNE n�22 - ITA-AITES - June 2002 36