Ä.F.T.E.S.

GROUPE DE TRAVAIL N° 7
TUNNEL SUPPORT AND LINING

Recommendations for use of convergence-confinement method

Draft prepared by P. GESTA, Chairman, in consultation with :

Mme BOUVARD and MM G. COLOMBET J. COSTE A. GUENOT
GUILLAUME LESOT J.-P. MAGNAS F. NELVA
M.PANET M.RAUZY

1. - PURPOSE OF RECOMMENDATIONS 4.2. - Convergence and Virtual Support Pressure

Versus Distance to Working Face
2. - SUMMARY OF GENERAL PRINCIPLES
4.2.1. - Elastic behaviour
2.1. - Self-support Capability of Ground
4.2.2. - Elastoplastic behaviour
2.2. - Acceptable Confinent and Convergence
4.3. - Practical Guidelines
2.2. 1. - Definition of convergence
4.4. - Time Effect
2.2.2. - Maximum confinig pressures of different types of
5. - PLOTTING CHARACTERISTIC CURVE OF L1NING
support
OR SUPPORT
2.2.3. - Large or small convergence and corresponding
5.1. - General
support
5.2. - In-situ Concrete and Shotcrete support
2.3. - Description of Method
5.3. - Support by Grouted Rockbolts
3. - PLOTTING CHARACTERISTIC CURVE OF GROUND 5.4. - Support by blocked Steel Ribs
3.1. - Simplifying Assumptions 5.6. - Combined Support System
3.2. - Analysis 6. - APPLICATION TO DESIGN
3.2. 1. - General-Notation 7. - MODIFICATION FOR ACTUAL CONDITIONS
3.2.2. - Elastic medium 7.1. -Introduction
3.2.3. - Ideal elastoplastic medium 7.2. - Purposes and Requirements for Monitoring du-
3.2.4. - Elastoplastic medium with brittle fai/ure behaviour ring Construction
3.2.5. - Pure cohesive medium 7.3. - Monitoring .Programme
3.3. - Finite Element Analysis 7.4. - Interpretation and Comparaison with Design
3.4. - Time factor Predictions
4. - ANAL YSIS OF CONVERGENCE VERSUS 7.4.1. - General
DISTANCE TO WORKING FACE 7.4.2. - Interpretation of convergence measurements
4.1. - Strain and Stress fields inside Ground Near 7.4.3. - Other measurements
Working Face 8. - CONCLUSION

FOREWORD these methods, too, also demands the same circumspec-

tion in interpreting the results of theoretical analysis and
Tunnel support design is still a highly controversial sub- an awareness of the need for comparing these results
ject. with actual conditions as recorded during construction.
Recent advances in soil and rock mechanics now enable The AFTES recommendations must not be considered as
engineers to take a less unrepresentative approach to infallible rules guaranteeing strict agreement between
analysing the processes involved in the response of soil
theory and reality, but simply as guidelines making it ea-
and rock to man-made structures in contact with them
sier for engineer to use these methods correctly.
(such as retaining walls, diaphragm walls, tunnel linings,
Success will depend chiefly on the representativeness of
building foundations, anchoranges) providing the appro-
priate measurements are performed before, during and the parametres. in other words, in the quality of enginee-
alter construction. ring measurements.

From this viewpoint, modern methods of tunnel support The assistance of an experienced specialist is still crucial
design are as representative as comparable methods in interpreting these findings and adapting the support
used for analysing other substructures. But the use of design to the conditions thus revealed.

206
 I-I
1. - PURPOSE OF RECOMMENDATIONS - Thereby, control decompression of the surrounding
ground (which always leads to a serious deterioration in
The convergence-confinement method and its applica- its mechanical properties) ;
tion in predicting and analysing the behaviour ol an un-
- Optimize support quantities and cost by applying
derground structure with its support and/or lining has
only enough confinig pressure to keep convergence wi-
been discussed in many publications, listed in the
thin acceptable limits.
Bibliography. The method has come into much more wi-
despread use in recent years. AFTES has described the
basic principles and general rules ol application in 2.2. - Acceptable Confinement and Convergence
Thoughts on the Usual Methods of Tunnel Lining Design It is extremely dilficult to define what is meant exactly by
(working group N° 7, Tunnels et Ouvrages Souterrains « keeping convergence within acceptable limits » since
W 14, March-April 1976) and a number of conlerences the amount ol convergence is not only factor to be consi-
at the 26 th October 1978 symposium on the method in dered. The rate at which movements occur is generally
Paris (Tunnels et Ouvrages Souterrains N° 32, March- more important, the usual safety criterion being that it
April 1979). should slow down gradually with time, as the working
face recedes.
Booklet N° 69 ol the French General Specifications for
Construction Contracts deals with works of underground
2.2. 1. - Definition of convergence
engineering construction , and some ol the clauses cover
the design ol support and lining (appendices 3 and 7), Driving a tunnel or excavating achamber causes
including the development of support methods which changes in the original stress field in the surrounding
« help the ground to stand up by itselt». ground, accompagnied by a strain field whose three-di-
mensional extent depends on these original stresses
These recommendations have been drafted with refe-
and the mechanical properties of the ground.
rence to all these publications. Their purpose is to state
the practical approach to using the method during the de- In the most general case, each particIe of ground under-
sign and construction stages ; and promote more ready goes a movement which can be resolved along three
interpretation of results. axes : longitudinal, radial and tangential. This movement
is usually referred to by its radial cmponent, which is the
most significant one. The quality usually measured is
2. - SUMMARY OF GENERAL PRINCIPLES convergence, ie, the change in the distance between two
points on the exposed face. When these two points are
2.1. - Self-support Capability of Ground diametrically opposed, convergence can, in most cases,
be considered as the sum of the two radial displace-
The convergence-confinement method ol tunnel design ments. It is usually expressed in millmeters, the refe-
was formulated in parallel with the use ol methods of rence base being the distance between the two points at
support (shotcrete, rockbolts, lightweight ribs) which the time the suds were placed in the face. Relative
help the ground to mobilize its own ressources, over the convergence and the length of this reference base. lt is
last twenty years or so. usually considered as being the same as the ratio bet-
ween the radial displacement and the radius (Cv/loF-Ur/r).
These construction method are known generically under lt is a dlrnensionless number.
the name of the New Austrian Tunnelling Method
(N.A.T.M.) or in France by the initials P.T.S.S. method. Other measurements characteristic of the deformation of
They both have the same starting point, in yhat the ob- the surrounding ground include expansion (radial defor-
jective is to place more emphasis on the strength of the mation in the ground) and subsidence (surface settleent)
surrounding ground in the ground/support complex and above shallow tunnels (see Section 7.3.).
give due weight to the lact that excavation is a process
which proceeds gradually, and the fact that the problem 2.2.2. - fy1aximum confining pressures of different
is a three-dimensional one. Unlike the « hyperstatic types of support
reaction » approach which examines the performance of The maximum confining pressure the support can reach
the support under applied loads representing the sur- belore breaking or collapsing is known from the usual
rounding ground, convergence-confinement coneen- technological conditions lor the use of different types of
trates on the behaviour ol the ground under the eelect of support. Experience shows that, in practice, the range is
confining pressures representing the support. A basic quite narrow.
tenet is that it is not generally the purpose of support to
prevent elastic deformation of the surrouding ground or For rockbolts, conditions relating to minimum bolt spa-
evn the formation of a pseudo-plastic zone in wich the cing, bond or anchor strength, bar sizes, and steel
failure criteria are exceeded. grades usually lead to a maximum conlining pressure ca-
pability of between 0.05 MPa and 0.2 MPa (0.5-2 bar),
This appears a more realistic stance since this strain regardless of diameter.
state is very often reached even near the working face
before it is possible to place any type of support at all. Lightweight steel ribs used alo ne usually give confining
pressures between 0.05 MPa and 0.1 MPa, with heavy
Convergence-confinement accepts that support aims es- ribs going up to 0.2 MPa if their inertia is properly chosen
sentially at limiting the deformations of the lace and ex- to suit the diameter of the excavation (cl. AFTES recom-
tension ol the pseudo-plastic zone to arrive at a state of mendations on the Use ol Steel Ribs, Tunnels et Ou-
equilibrium which complies, if possible, with the lollowing vrages Souterrains, Special Issue, July 1982, pp. 80-97).
three requirements :
Shotcrete, in the standard thicknesses of 0.05 - 0.20 m,
- Limit convergence to acceptable values. compatible can develop higher contining pressures in smaller tun-
with excavation and the ultimate purpose ol the nels. up to 0.5 MPa lor a diameter ol 4 rn, and 0.20 MPa
structure : for a diameter ol 10 m.
 It is noticeable that, despite the variety ol temporary
support methods available (ribs, bolts, shotcrete), the
maximum conlining pressures are all ol the same order
of magnitude, not usually exceeding 0.2 - 0.3 MPa ; the
corresponding deformations on the orther hand vary
widely with different between them. Only voussoirs are
capable of applying any substantially higher pressure
(1 - 2 MPa).

2.2.3. - Large or small convergence and correspon-

ding support

In most cases relative convergence Ur/r is generally bet-

ween 103 and 4.10'2 when it is contained by installing
a suitable type of support promptly. Support type,
convergence and confining pressure are all mutually
interdependent.

"Stiff or semi-flexible" support (heavy ribs, shotcrete with

or without lightweight ribs and rockbolts) reduces
Figure2
convergence for a given confining pressure. Recorded
Characteristic curves of support (a) and ground (b)
relative convergence is usually less than 5.103.

There is a danger of the ribs collapsing in ground sub-

The graph of O"rversus Ur, curve (b) on figure 2, is cha-
ject to large deformation or high initial stresses.
racteristic of the behaviour of the ground. It is the cha-
"Flexible support", especially rockbolts, can accompany racteristic curve of the ground (the convergence curve)
the movement without failure. Recorded relative conver- at a point on the tunnel wall. lts shape depends on the
gence can be as much as 10.2. type of response of the tunnel ground, the cross-sectio-
nal shape of the tunnel and the initial stress state.
The convergence-confinement method offers a better
approach to acceptable deformation and measures re- Specific points on this graph represent the successive
quired to adapt the design to actual conditions. stages of tunnelling and installation of the support.

- Point F (coordinates Urf and (1 - 1..f)0"0) shows the

2.3. - Description of Method deformation at the section as the working face passes
In the convergence-confinement method, analysis of the though;
three-dimensional response of the ground around the - Point S (coordinates Urs and (1- 1..s)0"0) represents
excavation is reduced to a problem of plane strain by the deformation on the section after the support has
examining a typical cross section to determine the effect been installed at a distance. d from the working face.
of a continuous reduction in the radial stress OTacting
Obviously, Urs ;:::Urf
on the perimeter of the tunnel from its initial value 0"0 to
zero. The zero value of the radial stress represents the Once the support is in place, it restrains subsequent
condition along the tunnel wall excavation, with no sup- convergence.
port, the tunnel being assumed to be of indefinite length.
The performance of the support located at the rlevant
It has been shown to a good degree of accuracy that the
cross section is defined by curve (a) in figure 2, which
nearby front face (figure 1) is equivalent, in terms of me-
chanics, along the wall equal to (1- 1..)0"0' The initial represents the confining pressure it applies versus its ra-
dial displacement. This is the characteristic curve of the
state far enough ahead of the front face for it to have no
support (or confinement curve) at this point.
influence corresponds to À. = O. As the face approaches
and reaches this section and then recedes from it, À. in- The shape this curve obviously depends on the mechani-
creases gradually from 0 to 1 ; the radial displacement cal properties and geometry of the structures forming the
of the bare tunnel wall Ur also increases : the tunnel support (cross section, inertia, modulus and yield point,
converges. post-yield behaviour, ultimate strength).

In order to plot th is graph, one must also know the posi-

I tion of its zero point on the ordinate, in other words, one
must know the proportion of convergence Ura which has
I occured at the cross section by the time the support be-
gins to be loaded. It may be equal to Urs, but with some
L types of support, it may be necessary to allow for the ini-
I tial gap Su between the support and the rock, which re-
quires some extra radial deformation of the rock before
I the support can come into play. With steel ribs, this gap

e
is mean sum of the gaps rneasured radially between the

8
outside face of the rib and the rock. The value of Su de-
"Tl' pends on the type of support, the quality of werkmanship
f-o,-O"-j
.,{J...'>- in packing the gaps, and the compressibility of the pac-
king material.
,,-0 o ( x ci ).-1

In the general cases: Ura = Urs + Su

To determine Urf and Urs one must know how much the
Figure 1 decompression percentage varies with the distance ol
Virtual support pressure near the working face the lront lace from the section considered.

:>OR
The point of intersection of the two characteristic curves
(a) and (b) thus detined yields the ultimate equilibrium
parameters far enough behind the tront face for it not
to have an influence, but without the time effect. These
parameters are the maximum confining pressure th at
can be reached by the support at equilibrium and the /
corresponding total convergence: /
/
I
(1 = Àq) (Jo and Urq
t
I
This description shows that the th ree critical stages in
\
constructing the characteristic curves in the conver- \
gence-confinement method are: \

" ..•.. ......

- plotting the characteristic curve of the ground,
----
- examining the influence of the approach of the front
face to know the relative position of the two curves, Figure.3
Cireular tunnel in iso tropie homogeneous medium
- plotting the characteristic curve of the suppport and
Iining. UI

U-a
I
Certain widely accepted restrictive assumptions enable
relatively simple analytical solutions to be found for U3 - 3
these problems, provided one is aware of the limitations
and knows how to interpret the results accordingly. This t
UI
is discussed in the following sections.

3. - PLOTTING CHARACTERISTIC CURVE

OF GROUND

1- Pseudo-elastie behaviour E1
3.1. - Simplifying Assumptions 2 - Peak strength
3 - Strength loss with dilataney
The use of analytical formulae capable of giving the
4 - Residual strength
equation for the charasteristic curve of the ground
usually makes use of the assumption that the excavation Figure4
can be considered as equivalent to a circular tunnel in Ground parameters
an isotropic, homogeneous material. It is also assumed
v poisson's ratio for ground in elastic phase
that the initial stresses are isotropie and that the tunnel
is doop enough for there to be no significant differences (Je uniaxial compressive strength of ground
in the initial stress 0"0 around the tunnel (this requires a (Je = 2 c cos cp in coarse-grained ground)
minimum cover of the order of 3-4 diameters). 1 - sin <p

With this cylindrical symmetry and the other assump- (c et <pcohesion and internal friction angle (peak
tions, the charcteristic curves remain the sarne along the strength characteristics)
whole length of the wall of the tunnel. Cr = 0 et <pr
residual strength characteristics
Despite their very restrictive nature, these assumptions
are acceptable in many practical cases and enable a Ii- kp = t92 (1Ç + CP) = 1 + sin cP
4 2 1 - sin <p
mited number of parameters to be used to gain sufficient
insight for designing the support system and estimating passive earth pressure
the probable order of magnitude of convergence. 'Y: unit weight of ground

3.2. -Analyse 3.2.2. - Elastic medium

3.2. 1. - General-Notation If one assumes an isotropie distribution of initial stresses

and a circular tunnel shape, all strain will be elastic if the
The equations used in the analysis depend chiefly on initial stress (Jois less than half the uniaxial compressive
the type of ground response considered, the most fre- strength of the ground :
quent assumptions being elastic, ideal elastoplastic, and 0"0 < O"e
elastic-brittle and strain softening behaviour. 2

The following parameters and notations are used : The characteristic curve of the ground for the elastic
phase is a straight line intercepting the Ur axis at a point
radius of excavation
Uro such that :
(Jo initial homogeneous isotropie stress
Uro = 1 + v 0"0 r (1)
(Jr virtual radial support pressure along tunnel wall E
À expansion proportion such that : 1 - À = ~
When 0"0 is less than (Je/ 2 all strain is elastic.
a"
rd relaxation radius (boundary of plastic zone) When it is higher than ae / 2 strain will lirst be elastic so
long as the virtual support pressure (Jr is higher than (Jre
U, amplitude of convergence at lace the point at which the stresses in surrounding ground
E strain modulus ol ground in elastic phase reach the yield point.

I
 3.2.3. - Ideal elastoplastic medium

@T
C1.
The behaviour of this medium is determined by :
0'
o
q. - its deformation modulus under elastic strain ;
- the peak strength criterion, eg, Mohr-Coulomb type
criterion :
01 = kp 0., + Oe (4)
which can also be written as a function of c and lp
Ol = 0, 1 + sin <p + 2 c cos <p (5)
1- sin lp 1 - sin lp
~ A further parameter needed is some relationship for
strain amplitude. This may be the total expansion for the
who Ie volume of material undergoing plastic strain (ie,
where the peak strength resistance is reached), or it can
be a relationship between strains in two orthogonal direc-
tions in the section considered (eg. radial and tangentiai).

Depending on which alternative is chosen, the equations

will of course be different, and so, toe, will be parameters
o needed.

a) Labasse equations

cr <Qs. In the first case, the designer starts off setting some pre-
o 2 determined, uniform volurne increase for the whole zone
concerned, taking an expansion coefficient k such that :
k =' 1t rd2 - 1t (r - Ur)2 (6)
1t (rd2 - r2)

We th us arrive at the Labasse equations relating the

virtual support pressure Or to convergence Ur along the
exposed wall by means of parameters C and lp and the
expansion coefficient k.

With th is notation, the Labasse equation can be written,

for lp *' 0: (*)
Or =-C cot lp + (1 - sin <p)
(00 + C cot lp) ( 1 ) sin <P (7)
1 + 2 Ur - Ure 1 - sin <p
(k - 1)r

\ \
Plastic stram

Iro
ur;.

Etastic strain

Figure5
Ground characteristic curve (elastic deformation)

Associated with Ore is a relation percentage Àe such that

Oe = (1 - Àe) 00 and it can be shown that :
Àe = _1_ (kp_1+QC) (2)
kp + 1 00

If 00 > Oe / 2 only that part of the line representing À va-

lues between 0 and Ie is considered.

The coordinates of point E, which represents the elastic

limit of the ground (À = Àe) are:
Ore = (1 - Àe) 00 Figure 6
Standard elastoplastic equilibrium approach
Ure = Àe Ure = 1...:LY Àe 00 r (3)
E
• For <p = O. see 3.2.5

210
Ure represents convergence along the tunnel wall at the with:
end of elastic phase, as defined and calculated in the Ep = - 1..:t....Y À (Jo + i\Ep ,I
e
previous sectien (Eq. 3). E
The radius rd of the boundary of the zone in plastic fS = - 1 + v Àe (Jo + i\Ee
strain is : E

(8) We find :

(11 )
In this way, we can calculate (Jr and the radius of the re-
laxation zone rd for all values of Ur greater than Ure·
In an ideal elastoplastic medium, radius rd is given, for
This allows us to plot the characteristic curve of the lp#- 0 by : (*1
ground and the graph showing the position of the boun-
x (kp - 1) (Jo + (Jç ]'/kP"
dary of the plastic strain zone versus the virtual support
pressure.
r-
rd _ [ 2
kp + 1 (1 - À)(kp - 1) (Jo + (Jç
(12)

Of the various parameters which must be determined, Thus, every value of (Jr corresponds to a value of À such
the peak streng th characteristics can be obtained from that (Jr = (1 - À) (Jo ; using À therefore, we can calculate
simpte tests. The expansion coefficient on the other rd then Ur defining a point on the characteristic curve,
hand, which depends on dicontinuities opening under the complete curve being built up from such points.
tensile stresses and dilatancy under shear, is much It will be noted that the definition of rd is independent of
more difficult to quantify. Experience seems to indicate the value of (J, and that, tor a = 1, the relationship bet-
that plastic strain takes place with increasing expansion ween U. and rd is written :
and tnat shear movements are ottert the predominant
factor. (13)

It must be remembered that the Labasse equations as-

su me that, during plastic strain, radial displacement Urd c) Comparison between Labasse and Panet equations
is zero along the circular boundary of radius rd, meaning
The Labasse equation is the more widely-used one. lt
that elastic strain bevond the relaxation zone is (sorne-
started to be used when the convergence-confinement
times wrongly) ignored.
method was first appearing. The uncertainties arise from
b) Panet equations (1976) the problem of determining the expansion coefficient,
which does not always represent a physical reality.
These are derived from equations defining the radial Ep Recent studies seem to show that expansion of the
and tangential Ea strains with cylindrical symmetry ground is not usually the predominant factor in conver-
gence, which is apparently caused chiefly by shear rno-
Ep = du
vements. The Panet equations should therefore give a
dp
better representation of actual processes, even though,
Eo = .IJ. through the a coefficient, they obviously stil rely on the
P proper choice of some prior assumption concerning the
dilantancy of the surrounding ground.
and the boundary condition for p = rd is
Results are less sensitive to changes in a within the
o
up = lI.e -E-
1 + v (Jo rd
usuallimits than changes in k in the Labasse equation.

If we attempt an analytical comparison of the two sets of

Various assumptions can now be examined.
equations, it is found that the Labasse equations for k = 1
Assumption 1: would be in perfect agreement with the Panet equations
for a = 1 if the Labasse equation did not disregard elastic
No volume change in plastic zone.Thus, in plane strain : strain Ure. To allow for this, we must take:
Ep + ES= 0 ka = 1 + 1 + v À.e (Jo
E
In this case:
This value of k expresses the continuity of the detorma-
(9)
tion process in the elastic and plastic domains. The tan-
gent to the characteristic curve at point Ure (1 - Àe) (Jo is
Assumption 2 : the line representing elastic stress relief.

Strains in plastic zone accompanied by positive volume When using the Labasse equation Eq. 7, one must
change, assuming that : consider th is value ka as being the minimum value of k
for which there is no dilatancy.
Ep+aEs=O
d) Gravity effect
therefore :

U r=lI.e-- 1 + v
0 (rd)"
(J 0- + 1 xr (10)
Decompression is accompanied by the formation of an
unstable zone above the crown which might eventually-
E r
collapse (figure 7).
Assumption 3 :
As convergence increases, the shear strength of the
The strain in the decompressed zone occuring alter ground around the perimeter of this zone progressively
peak strength has been reached, Arp and M:o must be decreases, so that a greater part of the weight of the
such th at :
i\fp + uL\ro = 0 • Far 'P = 0, see 3.2.5

211
 Unstable wedge
..,.-- ;.,..- ---// or
.
--- //;,.-::; / <,
/' V..//? / -, -c-

/
/
/
/'
~(~
;'; " \
/ \
I \
I \
I I
I I
\ I
\ I
\ /

-
\ /
"" '- /
..-/ ,/ /
-------
.•.....•.

Figure 7 FigureB
Gravity effect on decompressed zone Gravity effect in plotting ground characteristic curve

material must be cartled directly by the engineered sup- The theory behind the adding of these two pressure
port. This is in fact observed in many cases, but it is a terms is not strictly justified, but the approach does give
feature which des not appear from the shapes of the a more faithful representation of what happens in ground
characteristic curves plotted with the above equations, that is highly deformable or has low cohesion.
which suggest that the support pressure decreases
continuously as convergence increases. 3.2.4. - Elastoplastic medium with britt/e (ai/ure be-
haviour
If there is a Iikelihood of substantial convergence occu-
ring, it is wise for the <Jraxes of these curves to add an As in the ideal elastoplastic medium, behaviour is defi-
ned by the deformation modulus under elastic strain,
extra pressure term <Jrgto represent the stability of the
and the peak strength criterion :
decompressed zone over which gravity is acting. Unlike
the support pressure term, it increases as convergence <J1= kp <J3+ <Je (16)
increases. of the Mohr-Coulomb type.

<Jrgis usually taken as equal to the weight of the wedge However, unlike in the previous case, it is assumed that
of decompressed ground between the exposed wall of cohesion forces are zero after failure, and the residual
the tunnel and the boundary of the plastic strain zone. strength is provided only by internal friction, ie ;
The following simplifying assumptions are used.
<J1= kp <J3 (16 bis)
<Jrg= y (rd - r) at crown, As in the previous case, post-failure strain mayor may
not be accompanied by dilatancy.
<Jrg= 0 at sidewalls.
Using the Panet equations, rd is given in terms of <Joand
The favourable effect of gravity acting at the invert is so-
<Jeby the equation :
metimes included, but this is not considered as being
fully justified in the general case. rd = r ( 2 - <Je/<Jo) _1_ (17)
(kp + 1)(1 -A) kp-1
With the Terzaghi equatlons, frictional resistance along
slip surfaces can be included, so that only a portion of Using this value of rd, Ur can be calculated as above, by :
the weight of the plastic-strain wedge need be conside-
red. This means :
Ur= Ure (2(~f+l +a-1) (18)
a+1 r
<Jrg= yB - 2C (14) 3.2.5. - Pure cohesive medium (<p = 0, C i' 0)
2 tg <p
In which In th is case, the equations discussed so far are usually
indeterminate. It is however possible to overcome this
B = b + 2 h tg (..1L - -!lL) problem and write equations suitable for this case.
4 2 These equations are as follows.
for tunnel of width band height h at a depth H below The Labasse equation Eq. 7 becomes :
ground level.
o- = - Cu + 00 - Cu loge (1 + 2Ur - Ure) (7 bis)
Lastly, still using the same notation, there is the equa- (k -·1)r
tion given by Caquot : Panet equation Eq. 12 becomes :
rd = e AGo - Cu (12 bis)
(15)
r 2 Cu

212
 time before installing support). and secondly through the
Or-
rheological properties of the material, usually characteri-
zed by the relaxation time T.

A dimensionless parameter, the viscosity index K, corn-

oines these two èftects :

(19)

In which 0 is the tunnel diameter, V is the rate of ad-

vance of the working face, and T is the characteristic
time of the time-dependent behaviour of the ground ; in
a visco-elastic material, T is the relaxation time.

Time can be a very important factor tor some values of

this parameter. When K > 0.10, it can be considered as
having a not insignificant effect.

Plotting an accurate convergence curve is not however

•• easy. Research is still going on, chiefly to determine the
o U
r in-situ relaxation time of the material.

We do however know that the curve lies inside a zone

Figure 9 bounded by a pair of "ideal" curves (figure 10) curve
Ground characteristic curve (elastoplastic behaviour) AA'A" represents the instantaneous response (curve B)
considered as representing the deformability of the walls
if the tunnel wre excavated instantaneously. Curve AA"
The Terzaghi equation Eq. 14 becomes : represents very long term behaviour (curve BI), taken as
showing the deformarbility of the walls, if the tunnel were
O"rg= H ( Y - ~) with B = b+2 h (14 bis) driven slowly enough for all these strains to occur in
B each phase of equilibrium, especially before installation
of the support or lining.
The Caquot equation Eq. 15 becomes :
There is a curve for each different time history of tunnel
O"rg= y (rd - r) - O"elog e ..!:Q. (15 bis)
driving and support installation within these two bounda-
ries. Figure 10 is an example of a convergence curve for
an unsupported tunnel where the working face stops be-
3.3. - Finite Element Analysis fore it had receded beyond the zone of influence. In sec-
tion AB, driving proceeded at a constant rate, work stop-
In some cases, the simplifying assumptions used in the ped for segment BC, and then driving resumed at a
analytica I equations are too far from reality to be se- constant rate for CO.
riously considered, either because tunnel shape is far
from being circular (as with high caverns) or because the OA" is the convergence arising purely from the rheologi-
tunnels are driven by headingand benching or very near cal properties of the ground, which occurs outside the
to the ground surface, or again because the surrounding zone of influence of the working face.
ground or the initial stresses are severely anisotropic.
In view of our lack of knowledge on this particular as-
In each of these cases, the numerical methods now pect, it is recommended as a first approach that the ex-
available (fine element methods) can incorporate most treme curves should be used when considering the time
of the relevant assumptions provided that the represen- factor in tunnel design.
tative parameters, especially those governing ground
response, can be included.

A set of different equilibriurn states for decreasing to de-

tremine the deformations of the surrounding ground in
each intermediate equilibrium state. Convergence at the
wall is taken as the most important factor. This makes it
possible to plot points on the characteristic curves of the
ground at critica I points on the tunnel perimeter (eg,
crown, springings, invert).

J- -------~~
3.4. - Time Factor (figure 10)

The ground does not deform instantaneously and there

are almost always time-dependent strains. The conver-
gence process can continue for a very long time, even
l:> 0
after the front face has receded to a consideral distance,
or has ceased to advance.

The time factor affects the convergence graph of the

ground in two different ways, firstly through the time his- Figure 10
tory of construction (speed of excavation and elapsed Ground characteristic curves with time effect
 II 1 J J l--+--l
+ + + + +-f- -f. 'i- X + + +
-t+ + ++ -I- + Jl. x )(-\'-\- + + +
+++ ++-I-'l.x)( -\-+ +
-t+ + ++ -I- -f....)( x ~-\-+ ++ +
+ + -++ +- 'f.. X x).;' + + +
1+ +++ +-f....X x -r+++++ +
++ +++ +-f..X~++ + +
x
Figure 13
Advancinq workinq face
Deformation of tunnel longitudinal profile

Figure 11
Principal stress trajectories

4. - ANAl YSIS OF CONVERGENCE VERSUS

DISTANCE TO WORKING FACE Ij.
1 \

f,
1 \
/ I
This analysis in an essential part of both the preliminary
~ngl~~~':!
__~
" I

and final design processes, to determine the relative po- --------/ 0

sitions of the characteristic curves of the support and the
. 0.1
ground. It is a matter of examining the trend in conver-
gence or virtual support pressure along a length of the 02
tunnel, for a given position of the working face, assu-
ming all others parameters, including the mechanical 03
properties of the ground, to remain constant. 0<:
The accuracy of the method in reproducting the actual 0.5
response of the tunnel walls depends largelyon kno-
wing, at least qualitatively, how much convergence has 0.6
al ready occured by the time the support is instalied. 0.7
Ignoring this time lapse and assuming, as is sometimes
done, that the origins of both curves coincide, is a se- 0.8
rious infedelity to the actual processes in play, leading to 0.9
false conclusions on the final equilibrium of the structure. Radial
1.0
This section discusses the results of calculations made
u
with three-or-two-dimensional numerical models with cir-
UC1
cular symmetry, backed up by observations of actual -3 -2 -1 o 2 3 x
tunnel behaviour near the working face, and then recom-
r
mends some practical rules to help in using the conver-
gence-confinement method.

4.1. - Strain and Stresses Fields inside Ground

Near working Face
The response of a tunnel near the working face is basi- Figure 14
cally three-dimensional. Figure 11 shows the principal Radial and longitudinal components of displacement near face

stress field in a place on the tunnel axis. the principaf di-

rections of the stress tensor, initially parallel to the axes
2Go Ge =2Go of the model, rotate through 90° as thetunnel advances.
Thus, the shear stresses on a unit element perpendicu-
lar to the tunnel centreline are negligible some distance
in front of and behind the working face (which is justifo-
cation of the plane approach to the problem), but peak
just at this face.
û. ux (Je: Go
Figure 12 shows the crown stresses on an axis aa, pa-
o
c..«, rallel to the tunnel centreline. lt clearly shows this
change in the shear stress (o«): the disppearance of the
radial stress (o-) and the gradualdoubling of the tangen-
ti al stress (<Je). The longitudinal stress (àx) remains
mainly the same.

The strain field is iIIustrated in figures 13 and 14. The

first figure shows the deformation of the characteristic
profiles, especially at the working face. Figure 14 is an
accurate plot on the radial and longitudinal strain corn-
Figure 12 ponents along axis aa. Both Figures show the relatively
Stresses on both sides of face at crown high value of the longitudinal component at and just be-

214
 2 x/r

Figure 15
Figure 17
Expansion proportion Ä vs. distance x to face
Expansion confinement curve Ä (elastoplastic material)
(elastic behaviour)

hind the working face which tends to « bulqe » towards But the A I x function itself depends on the deformation
the free side. behaviour behaviour of the ground.

Observations such as these of tunnel behaviour near the 4.2.1. - Elastic behaviour
working face incited Lombardi and Egger to suggest
ways of analysing its effect. Lombardi focused on the When 00 < Oe I 2, ie, when drop in the confining stress
decompression of a disk, postulating radial volume produces only elastic deformation, the curve of A versus
forces in the material to account for the observed shear distance x to the working face is as shown on figure 15.
at the workinf face. Egger considers the working face as
a hemisphere in order to examine the behaviour of a ma- It can be seen on this graph that the reflexion percen-
terial with spherical symmetry just behind the face. tage at the front face is 0.265. The points representing
These two approaches, which were discussed in the 10 % and 90 % relaxation are, respectively, at X10 (be-
October 1978 AFTES symposium, are certainly nearest hind the face) such that X10 = - 0.9 and X90 (in front of
to three-dimensionnal reality, but they are not easy to the face) such that X90 = 2 r
use in practice. lt is preferabie to use a method based
on the virtual radial support pressure concept, which, al- Convergence can be obtained immediatly from A, since
though of being much more convenient to use. we can write :
Ur = A Uro
4.2. - Convergence and Virtual Support Pressure in which Ure, is the ultimate convergence given by Eq. 1.
Versus Distance to Working Face
In this case, more than 25 % of the total deformation oe-
From the definition of the relaxation percentage A, the curs just in front of the working face.
virtual support pressure Or at any point on the exposed
tunnel wall is : 4.2.2. - Elastoplastic behaviour
Or = (1 - A) 00
When 00 > Oe I 2, the drop in the confining stress most
The convergence Ur asociated with this relaxation per-
often leads to a stress field around the tunnel in wich the
centage can be derived from Or with the equations in
strength of the material is exceeded localiy. This pro-
Section 3.2.1 above, which depend on the deformation
duces what is known as a decompressed zone, subject
behaviour of the ground (elastic, elastoplastic with or wi-
to plastic strain, and which may include the working face
thout strain softening).
itself, whose contribution to support is thus diminished.
Therefor, knowing the function relating A to the di stance
x between the section considered and the front face, we lt is easily understandable in theseconditions that the
can derive the variation of Or or U. versus this distance. shape of the graph of the relaxation percentage A versus
the di stance x to the working face is different than when
all strain is elastic. As noted by Egger at the 1978
AFTES symposium, the shape of the A = f(x) curve de-
pends on the geomechanical properties of the ground.

Studies by Panet & Guenot (1982) have yielded curves

showing the virtual support pressure versus x for diffe-
rent conditions of plastification in a purely cohesive ma-
terial (figure 17). Parameter Ns shown on the curves is
the « stability number » definedas :
I
I
I -' Ns = 00 I Cu (20)
in which 0"0 is the initial stress, and Cu is the undrained
I
I
cohesion of the materia!.
Days
There is a different curve for each value of the stability
parameter. The relaxation percentage at the working face
varies over a very large range, depending on the value of
Figure 16
N s ; it ranges from 0.25 wilh Ns = 1, 10 0.60 wilh Ns = 3.

I
Extent of teilure zone on both sides of face
(after Lombardi, Montreux 1979) Egger even menlioned an exlreme value of 0.85.

215
 However, this research shows that, des pi te the large the ground must include terms for the behaviour of the
range of variation in results on the relaxation percentage material from which these members are made (elastic
of the working face, the proportion of convergence that phase, yield point, plastic phase, ultimate load) and the
has al ready occured in front of the working face remains various other factors which can affect strain magnitude:
relatively constant, regardless of the deformation beha- yielding of wedges, deformation of bearing plates, slip-
viour of the ground. It is between 20 % and 30 % of final page at ancnors. yielding of ribs, etc.
convergence.
With some types of support (shotcrete or in-situ
At a distance X back from the working face, at least concrete, steel ribs, voussoirs), it is usually only neces-
equal to the radius of the section, the set of relaxation sary to consider elastic strain. But with other types, used
curves (figure 17) covers a much smaller range, and the where convergence is large, one must also consider the
relaxation percentage is much Iess dependent on Ns plastic phase of their deformation.

For example : If we consider the elastic portion of radial strain any de-
formation L'>urbetween a pair of successive states cor-
for X / r = 1 0,81 < À. < 0,90 responds to an increase in the confining pressure !l<Jr
such that :
for X / r = 2 0,90 < À. < 0,95
(21 )
4.3. - Practical Guidelines

In brief, the procedure for evaluating the relaxation per- in which which ks represents the stiffness of the support
centage À. can be summarized by means of the decision during th is type of strain.
table below, based on the position of the section consl- If Ura is the initial radial displacement of the ground
dered with reference to the workinq face, and the beha- bevond which load begins to be applied to the support
viour of the ground. (Ura = Urs + su), the equation for the confinement curve
(a straight line) for the elastic part of the strain is :
Elastic Purely Cohesive Ground with
Behaviour Elastoplastic Behaviour Ur = Ura + _r_ ar (22)
(<Jo< <JeI 2) (<Jo><JeI 2) ks
X/r=O 1..=0.27 Ur, I Ur~ = 0.25 - 0.30 • This equation is valid up to the yield point of the mate-
rial. Beyond this, it is usually assumed that the confine-
XI r= 1 1..= 0.81 1..1= 0.81 - 0.90 depending on Ns
ment pressure aru remains constant during subsequent
XI r= 2 1..=0.90 1..2= 0.90 - 0.93 depending on Ns strains. This leads to a confinement curve of the type
shown in figure 18.
Other cases Figure 17 Figure 17 or numerical model
The term aru thus defines the ultimatè support pressure.

A numerical model must be used if a high degree of pre- The following sections discuss the various types of sup-
cision is required, or plastic behaviour is very important port available, and the particular equations applicable.
factor.
Most of this information is taken from underground
Excavations in Rocks by E. Hoek & E.T. Brown, publi-
4.4. - Time Effect
shed in 1980.
The above rules for determining the abscis sa of the ori-
gin of the confinement curve remain valid when the time 5.2. - In-situ Concrete or shotcrete Support
factor is included. Uncertainty arises when attempting to
plot the precise convergence curve, as discussed in A formed concrete or shotcrete lining, of thickness e, is
Section 3.4. applied to a tunnel of (excavated) radius r.

As a first approach to determine Urs the shortime The support pressure ar developped by this lining is
convergence curve Bi can be used. This usually overes- given by the equation :
timate the loads on the support, since the initial conver- ar = ks Ur - Ura
gence Urs will be minimal for a given value of the dis- r
tance x at which the support is installed behind the wor-
king face.

5. - PLaTTING CHARACTERISTIC CURVE

E
OF L1NING OR SUPPORT

5.1 . - General

The characteristic curve of the lining or support (confine-

ment curve) depends chiefly on the geometrical and me-
Ur
chanical characteristics of these structural members. o
The equation used for relating the radial displacements
of these members to the confining pressure applied to

•. From Uro i Ur,:, and the characteristic curve of the ground, de-
nve correspondmg value of 1..0.Present state of research seems
to indicata (Tunnelling 1982) that these results are also valid for Figure 18
frictional materiais. Typical confinement curve (elastoplastic material)

216
 Eiestic etonqetion of bar

Tb.' breaklng Slrenglh

Figure 19
• Shotcrete or formed concrete Iining
o 2 3 4
BalI elongalion (cm)

in which the lining stiffness ks is :

Figure20
k = Eb (r2 - rj2) (24) 80lt load.elongation curve from pull-out test on 25 mm dia.
s
(1 + v) [(1 - 2 v)r2 + ri2] bolt, lenght 1.80 m, Rawplug expansion shell anchor
in which Eb is the elastic modulus ol the concrete, v is
the Poisson's ratio of the concrete,and rj is the lining in-
side radius. 5.3. - Support by Ungrouted Rockbolts
II the lining thickness is small compared with the radius, The effectiveness ol support by means ol ungrouted me-
this equation can be written in simplilied lorm : chanically or chemically anchored roc1<bolts depends
equallyon the delormation characteristics ol the anchor,
kb= ~ (24 bis) load-distribution plate and bolt head as on elongation
(1 - v2)r behaviour of the bar.
in wich r is the tunnel radius (ie, the radius to the rock /
The linear-elastic elongation ol the bar is given by the
concrete interface).
equation:
This equation is only applicable to a lull-circle lining. The öL = 4L Tb (26)
general analysis can only be extended to non-circular or 1td2 Ea
part-eirele linings il proper allowance is made lor the re-
sulting reductions in bearing capacity in the design. in which L is the Iree length ol the bolt between anchor
and head, d is the bar diameter, Ea is the Young 's mo-
When heavyweight reinlorcement is incorporated in the dulus ol the bar material, and Tb is the load acting on
lining (e.g. concrete over steel ribs), the contribution ol the bolt.
each coponent must be considered in evaluating the
stillness of the composite system. In addition to th is elastic elongation, we must also consi-
der the displacement due to the delormation characteris-
The behaviour of such composite support systems will tics ol the component parts ol the bolt (anchor, load
-be examined later. plate, bolt head) :

Light reinlorcement (e.g., libre or small-size reinlorcing (27)

bar) has little effect on lining stiffness, and can be ignored.
is which Q is related to the load-delormation characteris-
The linal support pressure which can be developed by in tics of these items.
situ concrete Iining can be calculated lrom the theory ol
hollow cylinders subject to an external radial pressure, This quantity, Q, can determined Irom the load-exten-
ie: sion curve obtained lrom a pull-out test:

Q = (U2 - Ueb2) - (Ul - Uebl) (28)

O"ru = -L c'b [1 -K ] ou O"ru = c'b ~ (25)
T2 - Tl
2 r2 r
in which (Ul, Tl) and (U2, T2) are two points on the
(This applies to thinlinings with a low e / r ratio);
straight section ol the elongation curve as shown in li-
In these equations, O"'b is the uniaxial compressive gure 20.
strength ol the in situ concrete or shotcrete.
In order to eliminate the non-linear response ot the bolt
The equation applies only to circular linings around the resulting lrom initial bedding-in of the components (an-
full perimeter ol the tunnel section. chor and load palte), the bolt is usually preloaded imme-
diately after installation -, In terms ol the rock-support in-
If buckling is not analysed separately, the support or li- teraction analysis, this has the effect of shifting the load-
ning system is usually designed to keep the concrete extension curve to the position shown by the dashed line
stress to below 6 MPa approximately. in ligure 20.

217
 \ \ d
\
\, Block

\i \

Figure21 Figure22
Rockbolt support characteristics Blocked rib support characteristics

Without this preloading of a few ton nes, the characteris- 5.5. - Support by Blocked Steel Ribs
tic curve of the support system would be much less stiff
in its response to deformations of the rock. The stiffness of a blocked steel rib is given by :

Preloading must not however be exaggerated, it must be ...L = ..m:..+ --M3 [a ( a +sin a cos al _ 1 ] + 2.e.Jkl (31)
properly controlled to leave an adequate margin of bolt Kc Ea.S Ea.1 2 sin2 a Ec b2
strength for subsequent displacements. in which :
An ungrouted mechanically or chemically anchored rock- = tunnel radius
bolt can fail suddenly if the ultimate strength of the bolt
e = longitudinal spacing of ribs
system is exceeded.
a = half angle between blocks (in radians)
The stiffness ks of a set of ungrouted rockbolts is given b = width of rib (and blocks)
by the equation :
S = rib cross-sectional area
...L = ~I [ -AL + al (29) I = moment of inertia of rib
ks r 1td2 - Ea
Ea = elastic modulus of steel
in which ec is circumferential rockbolt spacing, and el is
d = bloc thickness
the longitudinal bolt spacing.
Ec = elastic modulus of block material
The ultimate support pressure which can be developed
by ungrouted rockbolting is : The blocks are assumed to be square in plan with a side
of dimension b which is the same as the rib width. .
(Jru = ..-I!L (30)
ec el The maximum support pressure which can be developed
by a steel rib is :
in which Tbr is the maximum load which can be borne by
each bolt as determined from a pull-out test in a rock si- 3 S I (Ja
(Jru = (32)
milar to the one for which the bolt system is being desi- 2e r a[ 31 + h S [r - (d + 1/2 h)] (1 - cos a)]
gned.
in which (Ja is the ultimate strength of the steel and h is
5.4. - Support by Grouted Rockbolts the radial thickness of the rib.

In th is type of support system, it is not possible to diffe- 5.G. - Combined Support Systems
rentiate between deformations in the rock and those in
the rockbolt. It is assumed that the stiffness of a combined support
system is equal to the stum of the component stiffness :
At the present time, only a qualitative approach is pos-
sible. k's = ks1 + ks2 (33)

Some authors consider that grouted rockbolts supply in which ks1 is the stiffness of the first system, and ks2 is
support in that they internally reinforce the rock. Such the stiffness of the second system.
bolts limit the effect of dilatancy.
If the two systems are installed one after the other, and
The net effect is to improve the residual characteristics c if we denote the radial displacement which has occured
and <p of the rock, which changes the characteristics when the first system begins to be loaded as Ura1, and
curve 01 the ground. the radial displacement required for load to be applied to

218
 /
x/r U

:-:::;~
-2 +'1 C.5 0 r

-u.
o - 1 \
\
\

Figure23 0
~
Combination of simultaneous or successive ·0.)
support elements +',

the second system as Ura2, then the support curve is de-

fined by :

(34)
j +L
J
::r
o
Ur = Ura1 + _r_ o-, for Urat < Ur < Ura2
ksl
Ur = Ura2 + r o-, tor Ur > Uraz
kS1 + kS2

The maximum deformation which can be tolerate is the

deformation causing faiture of the support system which
is the first to reach its maximum strain limit.
Figure24
The ultimate support pressure at this time is the sum of Predicted convergence curve with support
the support provided by this system, and the partial
confining pressures provided by the other systems, at have occured by the time the support is installed, using
the corresponding points of their deformation (figure 23). the decision table in Section 4.3, the curves in figures 17
or 19, or a numerical model.
6. - APPLICATION TO DESIGN c) Determine graphically the value of Urs associated with
this Às value.
The designer is usually interested in selecting the best
type of support system with reference to the geometry of In the case of a rock with elastoplastic behaviour and
the structure as dictated by its intended use and the installation of support at the working face (x / r = 0), step
geomechanical properties of the ground, as determined b) can be bypassed, as described in Section 4.3, by ta-
from the exploratory drilling and geophysical surveys un- king Urs such that :
dertaken up to that time.
Urs / U= = 0,3
An initial analysis of the project should enable a first
choice to be made among the various types of support in which U= is the radial displacement at an infinite dis-
available (see Recommendations, AFTES Group 7, on tance behind the working face, calculated iJl step a).
choice of support types, Tunnèls et Ouvrages Souter- d) Allow for the gap (if any) between the support and the
rains, Special issue, July 1982). rock su.
The convergence-confinement method can then be used e) Plot the support pressure curve.
for comparing the response of various possible types of
support. By examining different specifications (rockbolt One can attempt to further supplement the analysis as
size and spacing, shotcrete thickness, steel rib size and shown in figure 24, by plotting the following on the same
spacing) and installation procedures (advance of wor- figure :
king face before installing support), one can arrive at the
optimum support system that is compatible with maxi- - In the top right-hand corner, the characteristic curves
mum acceptable convergence and direct and indirect of the ground and of the support as described above ;
costs associated with the relevant working methods.
- In the top left-hand corner, the curve showing the re-
A typical sequence of operations might be as follows laxation percentage versus distance to working face.
(see figure 24) : This curve is one of those shown in figure 17, selected to
suit the characteristics of the ground and the natural loa-
a) Construct characteristic curve of ground by one of the ding rate (Ns = (Jo / Cu). lt enables f,s to be determined
methods described in Section 3. graphically versus the distance x at which the support is
instalied behind the face (in this example, x !r = 0.5) ;
b) Having decided the di stance by which the working
face advances before support is installed (x Ir), deter- - In the bottom right-hand corner, the predicted
mine the value of the relaxation percentage ")es which will c:onvergence curve (U. versus x I r) which is derived

219
 from the other two curves by transferring the coordinates
point- by point.

The chart can be further refined by plotting the long-term

characteristic curve of the ground. The position of its
point of intersection with the characteristic curve of the
support gives an insight into the ultimate equilibrium
(pressure and convergence), and also indicates the pro-
bable shape of the long-term convergence curve.

Insofar as the characteristics of the ground are relatively

uniform and weil known, one must obtain a set of predic-
tions which can guide the initial choice of support type,
and will serve as a basis for checking response in the
course of construction. This procedure does however
only set orders of magnitude, because of the uncertainty
on must of the parameters. Research is continuing to
perfect the method, and improve accuracy.

Figure25
7. - MODIFICATION FOR ACTUAL CONDITIONS Convergence measurement with in var wire

7.1. - Introduction
One of the advantages of the convergence-confinement
method is speed and the ease with which the basic de-
sign can be adapted to suit actual conditions during - Precision (the important point being to measure the
construction, often leading to substantial savings on the speed and acceleration ol time-varying processes).
support installed.
- Quick reading and immediate interpretation (graphs
Continuous monotoring of structural response is there- must be updated immediately the information is available).
fore a vita I factor for the designer. Measurements and re-
cords are interpreted to adapt the support system to suit - Minimum interference with construction work.
actual ground conditions, recorded response being a cru- - The physical quantities measured must not be inhe-
cial criterion in the judgements formed by the engineer rentiy subject to excessive scatter.
on site.

Monitoring during construction for adapting the design to 7.3. - Monitoring Programme
site conditions is most important when the support sys-
The basic measurements requir.ed are convergence re-
tem consists of shotcrete and/or rockbolts, as described
cords on studs cemented into the exposed walls of the
in appendix 7 to the commentary on bulletin 69 tunnel. This is a direct measure ol deformation between
Underground Construction Issued as part of the Cahier two points on the tunnel wall, which summates all the
des Clauses Techniques Générales (1982 edition). types ol deformation of the rock at these points, so that
scatter in the readings is not excessive. The measure-
7.2. - Purposes and Requlrernents for Monitoring ment is usually made by means of an invar wire connec-
during Construction ted to a readout instrument incorporating a precision dy-
namometer so that wire tension is always the same.
Section 5.5.b of the Guidelines in appendix 7 mentioned There may be three studs (sidewalls and crown) or more
above state the measurements to be undertaken in the (sidewalls, springings, crown, etc. see figure 25), depen-
course of construction. ding on the size of the tunnel.
The purpose of such measurements is : System making use of rellected coherent radiation
(radar or laser) are being developed,and should over-
- To check the effectiveness of the support, by veri- come the problems of access to the crown and interfe-
fying that the general trend is towards stabie equilibrium, rence with the tunnelling work associated with the invar
- To obtain early warning of distress, by modifying the wire instrument.
support system where the trend is not towards stability, Convergence measurements may be supplemented by
the following records, il opportunities are available :
- To analyse the impact of any changes in the ground
encountered, support type or working methods, - Topographic survey at ground level,
- If applicable, to keep watch on the impact of the work - Levelling survey, which is useful in the event or f100r
on the surrounding environment. heave or swelling.
The instruments used tor monitoring the works during - Displacement measurements within the rock (rod-
construction must therefore satisfy the following criteria: type strain rosettes or borehole extensometers),

- Ease and speed of installation or setting up. - Endoscopic examination (examination ot the decom-
pressed zone),
- Strength and protection (which is more important
nearer the working face). - Measurement óf vertical displacements from ground
level (settlement recording systems) ror shallow tunnels,
- Sensitivity (to detect the very first signs of incipient
events). - Clinometer measurements (transverse delormations),

220
- Stresses in concrete (vibrating wire strainmeters), - Predicting final convergence from the lirst measure-
ground and/or concrete / rock interface (flat stresse ments, in order to organize additional measurements as
cells), quickly as possible in the event of need.

- Stresses in rockbolts and ribs. If it has been possible to plot the predicted convergence
curve as explained in Section 4, the two sets of curves
The advantage of making several types of measurment can be compared from the point of origin of the instru-
is to improve interpretation by the cross-checking made mental curve, assuming that convergence prior to that
possible between, say, stress and strain records. time has been as predicted.

Different measurements are usually made at a number of To predict final convergence from initial measurements
« standard » profiles (eg, convergence and topographic in a ground which is relatively uniform over the length of
survey), plus a few « reinforeed " profiles for endoscopie the tunnel, it is possible to try to find empirically a gene-
examination, stresses, absolute displacements, etc. ral equation for the convergence curves, lor this type of
ground, and extrapolate to later instrumented sections.
The distance between such test sections depends on the Panet & Guenot * suggest an equation of the following
frequency of changes in ground conditions along the type for an elastoplastic material :
length of the tunnel. Under ec normal » conditions, for
example, standard sections may be 15-40 mapart, with (35)
rei nfo reed sections at 100-500 mintervals.

The frequency at which measurements are made de-

in which C is convergence versus distance to working
pends on the type of ground involved, but most impor-
face (and versus time if the ground exhibits viscous be-
tantly, on the distance from the working face (or the haviour), C~ is final convergence, x is the distance to
elapsed time behind the working face, which eernes to the working face, r is the tunnel radius, and A is a
the same thing). when the working face is near (Iess constant which depends on the mechanical properties of
than 3-4 diameters away), measurements must be made the ground.
at least daily, subsequently being spaeed out gradually
to weekly or more, except in the event of anomalies (eg, This equation agrees weil with the elastoplastic result for
an increasingly rapid rate of movement), in which case x> 0.5 r.
measurements must be repeated at shorter intervals.
In theory, if one knows two values of C for distance X1
and X2 from the working face, it is possible to calculate A
7.4. -Interpretation and Comparison with Design and C~. But such an extrapolation does require some
Predictions degree of confidence in the uniformity of the measure-
ments. When in doubt, one should make an empirical
7.4.1. - General plot of the most probable curve from all the measured
points. If they are obtained at frequent intervals, it is ea-
Instrument records are interpreted as a function of time. sier to plot convergence rates in this way, rather than
The time-dependent trend in a quantity (eg, deformation convergence rates in th is way, rather than convergence
of a given zone of ground) is a fundamental indicator tor itself.
the engineer. It can be said that a quantity which does
not vary with time indicates mechanical stability, regard- In th is way, it is possible to estimate the final conver-
less of its magnitude. But conversely, extrapolating a genee C~ from the time the first measurements are
trend is not a straightforward process. It can only be made.
done with the aid of a mechanical model, which is itself Panet & Guenot (same ref.) have shown that A is related
« calibrated » from previous in-situ measurements.
in an approximate way to the radius of the plastic zone
rd by A = 0.84 rd / r. Determining A enables this radius to
It is often that an acceleration in deformations eventually
be estimated. Whatever may be learned from such com-
leads to failure. In practice, however, it would seem that
parisons, the essential criterion in the engineer's judge-
failure can always be avoided if there is no « accele-
ment will be how smoothly the convergence rate de-
ration » in a graph plotted with a logarithmic time scale.
creases. Any increase in the rate of convergence (ex-
In other words, there must in fact be some degree of cept if tunnel driving resumes after prolonged interrup-
deceleration. tion) requires appropriate action, usually consisting of
installing more support or changing tunnelling method
The following sections discuss the interpretation of diffe-
and support type.
rent types of measurement.
Similar action is required in the event of an anomaly on
7.4.2. - Interpretation of convergence measure- the amplitude or rate of convergence which cannot be
ments immediately explained.

Convergence records are used to plot curves of conver- 7.4.3. - Other measurements
genee versus distance from the working face.
Measurements in the reinforeed test sections other than
In most cases, test sections cannot be set up any closer convergence of the exposed tunnel faces are used to
than 0.2-0.5 r from the face, so that the measurements conlirm the convergence records, but can also yield ad-
will not cover convergence ahead of, or immediately at, ditional informations.
the face.
Measurements of settlement from ground level can yield
There are two types ol problem involved in comparing inlormation on the amount of initial displacement that oe-
curves from instrumental data and predictions : curs as the working face passes a point ; measurement

- Selecting the starting point ol the instrumental curve • M. Panel, A. Guenot .. Ana/ysis of Convergence Behind the Face
on the full convergence curve, and of a Tunne!", International Symposium, Srighlon, 1982.

221
 of displacements in boreholes and endoscopic examina- 8. - CONCLUSION
tion may indicate the extent of the plastic zone, which
can then be compared with the design predictions.

Stain-meters and flat stress-measuring cells have the di- These recommendations have attempted to provide a
sadvantage of providing discrete measurements, and sa state-of-the-art review of the practical possibilities of
suffer from the fact that soil and rock is not a continuous using the convergence-confinement method tor predic-
medium; in addition, they modify the stress distribution ting and analysing the response and behaviour of tunnel
around them, sa that there is aften much scatter in the support and linings.
readings. Stress changes are more important than stress
levels.
There is still room tor progress in improving the practical
Measurements of stresses in the ground or support sys- reliability of the method. Rather than perfecting the theo-
tem can be used to estimate the support pressure, which retical mode Is or the mathematics, the important area for
can then be compared to the theoretical pressure for the research is to improve our knowledge of the physical
same degree of convergence, as taken from the charac- processes, especially in connection with time-dependent
teristic curve of the ground or of the support. deformations of the ground.
In interpreting stress readings from the vibrating wire
strain-meters embedded in a concrete Iining, suitable Such progress will only be made possible by the collec-
corrections must be made tor temperature effects and tion of extensive instrumental data and a sustained effort
concrete shrinkage. at its interpretation.

BIBLIOGRAPHY

E. Hoek & E.T. Brown - Underground Excavations in 13 O. Carvalho & K. Kovari - Displacements
Rock. The Inst. Mining and Metallurgy, London 1980. Measurements as a Mean tor Safe and Economical
Tunnel Design. Int. Symp. Field Meas. Rock Mech.,
2 W.A. Amberg & G. Lombardi - Une méthode de celcut
Zurich 1977, pp. 709-722.1977.
élastoplastique de /'état de tension et de déformation
autour d'une cavité souterraine. 3rd ISMR Conf., Denver 14 P. Londe - Mesures in situ dans les tunnels. Rev. Fr.
1974 Vol. 11 B, pp. 1055-1060. 1974. Géotechnique, W 4, June 1978, pp. 49-62.
15 G.G. Wadel - Tunnel relaxation tor determining the ini-
3 M. Panet & B. Gaudin - La mécanique des roeties ap-
tial and long-term, Déformation around and underground
pliquées aux ouvrages de Génie civil. Annales de
opening. 12th U.S., Rock Mech. Symp., Rolla, nov.
I'Ecole nationale des Ponts et Chaussées, pp. 145-169.
1970, pp. 903-931.
1976.
16 W.R. Judd & W.H. Perloff - Strain distribution around
4 G. Lombardi & W. Amberg - L'influence de la méthode underground opening, Comparison Between predicted
de construction sur /'équilibre final d'un tunnel. 4th ISMR and measured displacements. U.S. Dept. of the Army,
Conf., Montreux 1979, Vol. I, pp. 475-484.1979. March 1971, NTIS-AD 723 532, 120 p.
5 Y. Niwa & T. Fukui - Stresses and Displacements 17 H. Albrecht - Das Zeit - Deformations Verhalten am
Around an Advancing Face of a Tunnel. 4th ISMR Conf., Ausbruchrand eines unter-irdischen Felshohlbaues
Montreux 1979, Vol. I, pp. 703-710.1979. (Deformation vs. Time in Rock Cavity). Geol. J.B. C,
N° 7,1973, pp. 19-54.
6 M. Panet - Les déformations différées dans les ou-
vrages souterrains. 4th ISMR Conf., Montreux 1979, 18 L. Rabcewicz & E. Hack - Die Bedeutund der
Vol. 111, pp. 291-301. 1979. Messung im Hohlraumbau (111) - Erfahrungen bein
Tauerntunnel (Importance of Measurements in
7 P. Panet & A. Guenot - Analysis of Convergence Underground Engineering - Experience at Tauern
Behind the Face of a Tunnel. Tunnelling 82, The Inst. Tunnel). Bauingenieur, N° 10, 1975, pp. 369-379.
Mining and Metallurgy, London, pp. 197-204. 1982.
19 W. Amberg - Auswertung von Messergebnissen aus
8 P. Egger - Einfluss der Post-failure-Verhalten von Fels dem Tauerntunnel (Analysis of Measurements from
an den Tunnelansbau (Influence du comportement du Tauern Tunnel). Schweizerische Bauzeitung, N° 42,
roetiet après la rupture sur Ie soutènement des tunnels). 20.10.1979, pp. 755-761.
Thesis, Karlsruhe Univ., Veröff, Inst. Badenmechanik 20 P. Berest, P. Habib, N. Minh Duc - Tentative d'inietpré-
N° 57,1973, Trans. B.R.G.M. W 540. tation des mesures faites aux tunnels du Fréjus et du
9 C.W. Schartz & H.H. Einstein - Improvements of Grand Sasso. Rev. Fr. Géotechnique, N° 12, Aug. 1980,
Ground Support Performance by Full Considerations of
pp.44-45.
Ground Displacements. Transportation Res. Record, 21 J.R. Beau et al. - Tunnel routier du Fréjus : mesures
USA, N° 684, pp. 14-20. 1978. géotechniques effectuées sur Ie enentier français et leur
application pour la détermination et /'adaptation du sou-
10 W. Steiner, H.H. Einstein et at. - Improved Design for
tènement provisoire. Rev. Fr. Géotechnique, N° 12, Apr.
Tunnel Support, U.S. Dept. of Transportation 1980, pp. 57-82.
DOT/RSPA/DPB. 50.79.10, May 1980, 330 p. 1980.
22 G. Comes & P. Debreuille - Convergence
11 P. Gesta, M. Kerisel, P. Londe, C. Louis, M. Panet - Measurements in a Deep Tunnel. Int. Symp. Field Meas.
Stabilité des tunnels par la méthode convergence-confi- Rock Mech., Zurich 1977, pp. 723-734.
nement. General Report, A.F.T.E.S. Symposium 23 P. Duffaut & J. Piraud - Soutènement des tunnels pro-
October 1978. Pub. in Tunnels et Duvrages Souterrains, fonds autrefois et aujourd'hui. Industrie Minérale, Dec.
W 32, March-April 1979, pp. 70-143. 1979. Special Issue. In cahier 7, Fr. Com. Rock Mech., 1975,
12 French Committee on Rock Mechanics - R61e de /'aus- 17 p.
cuttntion dans la conception et /'exécution des ouvrages 24 P. Duifaut - Stabilité des cavités souterraines. In La
souterrains. 4th ISMR COllI., Montreux 1979, Vol. I, pp. pratique des sols et fondations. Filliat ed., Editions du
773-786. 1979. Moniteur, Paris, 1981, pp. 787-823.

222