Code - Aster: How To Dig A Tunnel: Methodology of Excavation

Version
Code_Aster default
Titre : Comment creuser un tunnel : méthodologie d’excavat[...] Date : 23/10/2015 Page : 1/26
Responsable : GRANET Sylvie Clé : U2.04.06 Révision :
b51375ecdb6c
How to dig a tunnel: methodology of excavation
Summary:
This note proposes a methodology to simulate the digging of an underground gallery with Code_Aster. The
basic method is a method usually used in this kind of studies: method “convergence – containment”.
After a recall on the principle of the method, the principal stages of the command file Code_Aster are described.
Various digital examples make it possible to validate the procedure.
Warning : The translation process used on this website is a "Machine Translation". It may be imprecise and inaccurate in whole or in part
and is provided as a convenience.
Copyright 2019 EDF R&D - Licensed under the terms of the GNU FDL (http://www.gnu.org/copyleft/fdl.html)
Version
Code_Aster default
b51375ecdb6c
Contents
1 How to simulate the digging of a tunnel?.............................................................................................. 3
1.1 Principle of the method, put in work and validation........................................................................ 3
2 Introduction........................................................................................................................................... 4
3 A method to simulate the digging of a gallery starting from a model 2D: the method convergence-
containment......................................................................................................................................... 5
3.1 Principle general............................................................................................................................ 5
3.2 Application of the method for a digital calculation by finite elements.............................................. 7
4 Before attacking card-indexing of order Code_Aster… ........................................................................ 8
4.1 How to define the models starting from a simple grid?................................................................. 10
4.2 How to initialize the constraints?.................................................................................................. 13
4.3 How to calculate the nodal reactions at the edge of “the future” gallery? ..................................... 14
4.4 How to simulate the creation of a “vacuum” in the solid mass and the installation of the concrete?
..................................................................................................................................................... 15
4.4.1 Method A............................................................................................................................. 16
4.4.2 Method B............................................................................................................................ 16
4.5 Summary of the methods suggested............................................................................................ 16
5 Examples of command files................................................................................................................ 17
5.1 With the dealt problem................................................................................................................. 17
5.2 Case n° 1: excavation without supporting with initialization of the constraints by a calculation and
“softening” of the “excavated” elements....................................................................................... 18
5.3 Case n°2: excavation with supporting with initialization of the constraints by call to
CREA_CHAMP and déconfinement according to method A........................................................ 18
5.4 Case n°3: excavation with supporting with initialization of the constraints by call to
CREA_CHAMP and déconfinement according to the method B.................................................. 19
6 Validation of Code_Aster on an example of excavation in linear springy medium .............................. 20
7 As a conclusion: advices and prospects............................................................................................. 21
8 Bibliography........................................................................................................................................ 22
Annexe 1 Analytical formulas to apply the method convergence-containment to the case of a rock and
a supporting rubber bands and linear solid mass.......................................................................... 23
Annexe 2 Flow chart of synthesis on the methods allowing to simulate an excavation ......................... 25
Annexe 3 Comparison of the constraints obtained by digital calculation and the analytical solution .....26
Version
Code_Aster default
b51375ecdb6c
1 How to simulate the digging of a tunnel?

1.1 Principle of the method, put in work and validation
Context
The studies of géomechanics are generally based on a simulation of underground roadway drivage.
Examples of application can be quoted:
1. to evaluate the zone damaged by excavation (EDZ) around a gallery of storage;

2. to study the resaturation of an cell of storage by water of the site.
A certain number of studies were already conducted by department AMA on this subject, with
Code_Aster. However, few practical elements are available in documentations to reproduce this kind of
calculation. Department MMC undertook such a modeling with Code_Aster, in order to adapt the
procedure of application of the method classically used for this kind of calculation: method
“convergence – containment”. It comes out from this experiment which this application is not
completely commonplace that it is necessary to raise some practical technology matters of
implementation. To capitalize this experiment for the future users seemed rather important, in the
collective interest of the studies on storage in particular.
Objective
This note aims main aim to provide some preliminary technical advices to the users of Code_Aster
wishing to model an underground excavation.
Methodology
This note presents an application to a command file of Code_Aster method convergence –

containment. After a short recall on the principle of the method, a practical and operational description
of the orders to be used is given. The method is illustrated by calculations of validation of Code_Aster,
of which the command files are provided in appendix.
Result
Thanks to the application of the protocol suggested, two calculations of validation of Code_Aster were
implemented. The relative difference between digital results and analytical solution is lower than 2%.
Outlines
The method can be extended to the calculations nonlinear (plasticity, damage) and coupled in THM, in
particular within the framework of studies intended for the storage of nuclear waste.
Version
Code_Aster default
b51375ecdb6c
2 Introduction
For several years, studies have been carried out with Code_Aster in order to model the behavior of
works geotechnics (earth dams, tunnels, barriers worked for the storage of waste…).
Code_Aster was already used in particular to simulate the well or roadway drivage, within the
framework of the project Geological storage of nuclear waste HAVL (T4-01-10) or at the time of former
studies on major storage. The reports written until now (for example [6], [7] or [4]) are focused naturally
on the results, in order to answer the precise technology matter which justified the study. However to
simulate an excavation using a code finite elements is not inevitably an easy thing, and even if the
principles generals are recalled in the documents referred to above, one finds finally few elements on
the structure of the command files which were used as support with calculations.
In order to help the engineers in charge of the future studies of underground excavation with
Code_Aster, this note indicates some practical advices to begin in the realization from this kind of
calculation. Indeed, within the framework of the Storage project, MMC decided to adapt completely the
approach put in work by AMA in 2000 and in 2001. For that purpose, all the approach was reproduced
with version 6 of Code_Aster, on the basis of grid new and by exploring some alternatives. MMC also
profited from the assistance of the agents of AMA. In addition, this work led to a validation of
Code_Aster according to classical analytical formulas in linear elasticity (Kirsch formulas and method
convergence-containment, [5]).
This report thus presents:
1) classical method of simulation of an underground excavation in 2D by means of

a code finite elements;
2) different the option available to apply this method with Code_Aster ;
3) two CAS-tests of validation of Code_Aster for the problems of underground
excavations.
Thenecessary one with an advantageous reading of this note is the basic training with the use of
Code_Aster like a minimum of familiarization with the software package. The detail of the various
orders used is given by the user's documentation of Code_Aster (http://www.code-aster.org).
Version
Code_Aster default
b51375ecdb6c
3 A method to simulate the digging of a gallery starting

from a model 2D: the method convergence-containment
3.1 Principle general
This part is inspired largely by [5]. Let us announce that the CIH and TEGG also conducted a certain
number of studies with this method (for example, [2]). It is advised with the reader to refer to these
documents for more information on the principle of the method. The paragraphs which follow
summarize only the essence of the approach.
The method convergence-containment is usually used in the engineering of the underground works. Its
objective is to obtain an order of magnitude of displacements of the walls of the tunnel as well as the
efforts taken again by the rock and supporting. This method makes it possible to simplify the
calculation of a three-dimensional work by a two-dimensional calculation, by the introduction of an
adimensional parameter  called “rate of déconfinement”. It rests on the following assumptions:
1) plane deformations with assumption of small disturbances;

2) the tunnel is supposed of circular section and horizontal axis;
3) homogeneous ground of infinite extension;
4) solid mass following a linear or elastoplastic elastic behavior;
5) initial state of the constraints presumedly isotropic and homogeneous;
6) deep tunnel: no significant variation of constraints on the height of the gallery. In
practice, if H is the average depth of the work and R its ray, this assumption is
supposed to be satisfied if H / R10 ;
7) quasi-static balance (not of terms of acceleration).
One is interested in a section located in a plan perpendicular to the axis of the tunnel and one wishes
to carry out a two-dimensional calculation. The parameter  is supposed to take into account the
mechanical influence of the proximity of the coal face to this section, i.e. of a phenomenon whose
origin is out of plan considered by calculation.  depends on several parameters (rock, supporting,
length of nonconstant tunnel behind the coal face…) and its determination is not inevitably immediate
(many publications on the subject, for example [1]). This problem of analytical determination of the rate
of déconfinement leaves the framework of this document.
In fact, one introduces  to consider a fictitious tensor of the constraints  in the ground, which is a
fraction of the initial constraint  0 :
0
 = 1−  . with 0≤≤1
Version
Code_Aster default
b51375ecdb6c
[the Figure 3.1-a] below the evolution illustrates of  and of the radial constraint  R for a
nonconstant tunnel.

=1 1 R=0
R=0 R=(1-0
Figure 3.1- has: Evolution of the rate of déconfinement  and of the radial constraint  R
in the case of a nonconstant tunnel
Let us notice that =1 corresponds to déconfinement total of the rock: the influence of the coal face
on the behavior of the slice of tunnel disappeared and the tunnel is comparable to a very thick tube.
Since a part, even the totality of the constraints initially present within the solid mass disappear (it is
precisely the phenomenon of déconfinement), the walls of the excavation will tend to approach to reach
a new mechanical balance. It is the phenomenon of “convergence”. This phenomenon can lead to the
ruin of the work if the structure does not manage to find a state of steady balance following the
excavation.
If, for safety reasons or on stability, one decides to pose a supporting or a coating with the wall of the
tunnel, those go, from their mechanical stiffness, to be opposed to the natural phenomenon
convergence. In this case, final balance thus depends on the mechanical interaction between the rock
and the coating. Generally, this balance does not make it possible the constraints in the rock solid
mass to be cancelled like in the case of the nonconstant tunnel. It is said whereas the ground is
confined, from where the name of the method “convergence-containment”.
Version
Code_Aster default
b51375ecdb6c
Graphically, the application of this method amounts searching the point of intersection of the curve of
convergence, deduced from the behavior of the ground, and the curve of containment, deduced from
the behavior of supporting [Figure 3.1-b].
Figure 3.1-b: Example of curves of convergence and containment
The equations of the method “convergence-containment” in the case of a linear elastic solid mass are
provided in [§Annexe 1].
That it is for analytical or digital calculations, this method allows, using a simple model 2D, to deal the
3D problem which the simulation of an excavation constitutes.
3.2 Application of the method for a digital calculation by finite elements

A characteristic of calculations of excavation by finite elements is the need for implementing several
models (in the broad sense).
Indeed, a classical course of modeling can be summarized by the following stages:
1) stage 1: initialization of the constraints in situ ;

2) stage 2: calculation of the nodal reactions on the level as of walls of the excavation;
3) stage 3: déconfinement solid mass to simulate the progressive excavation and the
distance of the coal face;
4) stage 4: possible installation of a supporting/coating and end of déconfinement.
If the study requires it, the sequence of stages 2.3 and 4 can be repeated (case of an excavation in
divided sections, for example).
Version
Code_Aster default
b51375ecdb6c
In most case, the sequence of calculations is thus done on the basis of four configuration [Figure 3.2-
a].
Massif de sol Massif de sol

Excavation de la Pose du
Initialisation des galerie revêtement béton
contraintes Calcul des réactions
nodales
1 2 3 4
Figure 3.2-a : Typical example of sequence of a calculation of excavation with a computer code
The first configuration is used for:
1. to initialize the constraints of origin geostatics;

2. to initialize the hydrostatic pressure due to the possible presence of water and the temperature
(the present note does not discuss this precise item in detail);
The second configuration makes it possible to calculate the reactions to the nodes representing the
edge of the excavation.
At these stages of modeling, all the elements of the grid thus correspond to a material of type ground
or rock. One thus obtains a solid mass of ground in which reign a state of stresses corresponds to the
state of stresses in situ in the plan perpendicular to the axis of the gallery. One also knows the nodal
reactions at the edge of the excavation, which will allow déconfinement partial or total of the solid mass
in the stages which follow.
The third configuration is dedicated to déconfinement: one decreases the nodal reactions at the edge
of the excavation in order to simulate the digging of the tunnel. At the time of the realization of this
stage, the finite elements in the area corresponding inside the gallery should not take part any more
in the rigidity of the model. As it further will be seen, this can be taken into account in several ways in
practice.
One possibly passes to a fourth stage if one wants to simulate the installation of a supporting concrete
in the course of déconfinement for example. In this case, one adds elements with characteristics of
concrete and one continues reduction in the nodal reactions calculated in the stage n°1 to complete
calculation.
One thus notices that certain parts of the initial model will be seen affecting properties of ground
successively, of “concrete vacuum” then. In this sequence the source of some intrinsic difficulties is
at this kind of calculation.
The application of this approach by means of Code_Aster fact the object of the following chapters. It is
based on a simple case.
4 Before attacking card-indexing of order Code_Aster…

This chapter relates to some particular points of modeling which it seems important to comment before
being interested in the command files themselves. It is made up of a succession of paragraphs treating
Version
Code_Aster default
b51375ecdb6c
each one a question which an engineer can put when he carries out a classical calculation of
excavation using a standard code finite elements like Code_Aster.
Version
Code_Aster default
b51375ecdb6c
4.1 How to define the models starting from a simple grid?

The grid chosen in this study represents a quarter of model representing a cylindrical gallery in infinite
medium. The ray of the gallery is of 1,50 m , the thickness of concrete is 0,30 m and the grid is a
square of 20 m on side. According to the usual rules of modeling, the relationship between the
excavated ray R and dimension characteristic of the grid L is sufficient to consider that the boundary
conditions do not disturb the behavior of the excavation ( L10×R ).
Groun
d
Ground,
vacuum or
concrete
Ground or
vacuum
Figure 4.1-a: Grid used and materials
Version
Code_Aster default
b51375ecdb6c
From the point of view of the models (to the direction Code_Aster), it is necessary to distinguish some
particular zones from the grid (besides the other more classical zones, like the edges of the grid) and to
create the following objects (the names refer to the command files presented in Appendix):
1) the excavated edge where will be applied the nodal reactions to simulate
déconfinement (called BORD ) ;
2) the two points which are located at the ends of this curve, which are concerned
at the same time by the loading of déconfinement and the boundary conditions at
the edge of the solid mass.
Figure 4.1-b: Points and of points particular to identify together in the models Code_Aster
One can thus define (for example, because several configurations are possible):
1) a model SOL , in which all the grid is affected finite elements;

2) a model SOL_REST who does not understand the meshs which correspond
to the excavated part (they are not affected finite elements);
3) a model SOL_REST0 who understands SOL_REST and affected meshs
corresponding to the concrete pavement of finite elements.
Note:
The use of such a geometry to do a real calculation of excavation is partially criticizable, because
symmetry suggested risk to generate a nonphysical loading. In the event of application of the actual
weight for example, this one would be directed upwards in the lower part of the tunnel!
Version
Code_Aster default
b51375ecdb6c
Pesanteur
Partie maillée
Tunnel
Pesanteur induite par les

conditions aux limites
Partie non maillée mais

simulée par symétrie
Figure 4.1-c: Example of aberration which the use of a quarter of model can generate
in the simulation of a tunnel
Version
Code_Aster default
b51375ecdb6c
For a realistic study where one would wish to initialize the constraints using a loading in actual weight,
it would be thus necessary “to net the ground” until the rigid substratum (rock considered
indeformable), or all at least until a sufficient depth to free itself from the problem mentioned above.
One thus nets a half-model in this case there.
However, within the framework of our study, this constraint of grid did not obstruct us, since we took
into account neither the actual weight, nor couplings THM. The simulated loading is completely
compatible with the analytical solutions tested.
4.2 How to initialize the constraints?

Constraints in situ are generally represented by a tensor of order 2 whose principal directions
correspond to the vertical and the horizontal one. The vertical constraint is generally equal to the
weight of the various formations located above point considered and the horizontal constraint is
proportional to the vertical constraint:
 v = . z
 h=K 0 .  v
with  the voluminal weight of the overlying ground (in kN /m3 for example) and K 0 a coefficient
without dimension. K 0 can be given by measurements in situ or estimated by more or less empirical
relations. In the case of a semi-infinite solid mass subjected to an external constraint on its higher edge
or to its actual weight, the theory of linear elasticity provides a value of K 0 according to the Poisson's
ratio  :

K 0=
1−
Two methods were tested with Code_Aster to initialize the constraints in the country rock:
1. realization of a calculation (order STAT_NON_LINE) with a fictitious material equipped with a

Poisson's ratio allowing to obtain the report K 0 wished. This calculation is carried out on the
model which takes again all the grid of the study (for example, the model called SOL in the
preceding chapter). In this case, K 0≤1 (case of linear elasticity). There exist many cases where
K 0≥1 (if the ground is subjected to tectonic constraints, for example). In this case, the following
method becomes obligatory;
2. to directly assign the constraints to all the elements of the grid by the order
CREA_CHAMP (OPERATION = ‘AFFE’) ;
The first solution requires to define a fictitious material and to implement a calculation moreover.
However, if the loading is the actual weight (what is not the case of the CAS-test that we propose), this
method appeared at the same time intuitive to us and simple. In the case of a uniform stress field, the
use of CREA_CHAMP is unquestionably the most interesting method: it saves time computing and its call
is even simpler. For more complex distributions of constraints, CREA_CHAMP also function but we did
not use it (to refer to documentation [U4.72.04] who gives an example adaptable to our problem).
Version
Code_Aster default
b51375ecdb6c
4.3 How to calculate the nodal reactions at the edge of “the future”
gallery?
To calculate the nodal reactions at the edge of the gallery, it is necessary to block this part of the grid.
This operation should not generate nonphysical constraints by incompatibility with the loading applied
at the time of the phase of initialization. constraints. An opportunity given consists in imposing the same
loading as at the time of the initialization of the constraints by blocking the nodes of the edge of the
gallery only during this stage of calculation. This operation is without effect on the total result, which
remains identical to that of the preceding stage, but the “temporary” blocking of the nodes of the edge
of the gallery makes it possible to evaluate the nodal reactions there.
Même chargement
qu’à l’étape
précédente
Nœuds bloqués
u=0
Figure 4.3-a: Blocking of the nodes of the edge of the gallery to calculate there
nodal reactions
Concretely, this relative blocking of the edge of the gallery is possible thanks to the option DIDI (for
Differential DIrichlet) of the keyword EXCIT of the operator STAT_NON_LINE (Document [U4.51.03]).
The blocking of these nodes applies only to the increment of displacement considered and not to total
displacement (one imposes  u=0 and not u=0 ).
The initial state of this calculation (keyword ETAT_INIT of the operator STAT_NON_LINE) is defined
by the stress field obtained at the conclusion of the preceding stage.
Once this intermediate calculation carried out, the calculation of the nodal reactions is carried out
simply with the order CALC_CHAMP (FORCE = ‘REAC_NODA’). It is then advisable to provide to the
order CALC_CHAMP all loadings having produced the result from which the nodal reactions are
calculated, without omitting the loading voluminal if they exist (not taken into account in the examples
treated here).
One then builds a vector of loading by the recovery of the nodal reactions (CREA_CHAMP with the
keywords TYPE_CHAM = ‘NOEU_DEPL_R’ , NOM_CHAM = ‘REAC_NODA’ and OPERATION =
‘EXTR’). It should be noted that according to the paragraph [§3.1.1] of the user's documentation of
Code_Aster [U4.72.04], the option TYPE_CHAM = ‘NOEU_DEPL_R’ order CREA_CHAMP is in fact
without effect here (but nevertheless obligatory from the syntactic point of view), since an extraction is
carried out. This vector is then defined by the order AFFE_CHAR_MECA with the keyword VECT_ASSE
as a loading for the call following to the order STAT_NON_LINE (corresponding to the progressive
excavation of the gallery). This loading is associated with a function (operator DEFI_FONCTION)
describing the evolution of the rate of déconfinement as the progression of the digging.
Also let us notice that all the nodal reactions are extracted: those which act on the edge of the gallery
as those which act on the other edges of the model. Since these last act on blocked points with all the
stages of the calculation of excavation, their injection as a loading in STAT_NON_LINE according to is
without effect on the constraints and the deformations within the structure.
Version
Code_Aster default
b51375ecdb6c
4.4 How to simulate the creation of a “vacuum” in the solid mass and
the installation of the concrete?
Once carried out the calculation of the nodal reactions the question of the “elimination” of the
excavated part of the digital model arises so that its rigidity does not block the convergence of the
tunnel. For that purpose, we adopted two methods [Figure 4.4-a]:
1) method A : quasi-cancellation of the mechanical properties of the elements

located in the excavated zone (example: E=0,0001 Pa ), then introduction of
more realistic properties during the installation of supporting or the coating. This
method makes it possible to simplify the command file Code_Aster and gives
correct results for the simple case that we studied (small circular gallery,
excavated in only one section in an elastic solid mass). To conduct more
elaborate studies where the digital processing could be affected by the presence
of element with very low rigidity, it seems nevertheless preferable to us to be
based on the following method;
2) method B : initialization of the constraints directly by creation of fields at the
points of Gauss resulting from a calculation concerning a preceding stage.
Méthode A
Méthode B
Terrain
« Vide »
Béton 1 2 3
Figure 4.4-a: Various principles of modeling to simulate déconfinement solid mass
Other methods that we did not test can undoubtedly be applied to the studied problem (as the creation
of double nodes at the borders between materials who allow to bind or not the two structures).
Version
Code_Aster default
b51375ecdb6c
4.4.1 Method A
Method A do not call for a particular observation: it is enough to assign very weak mechanical
characteristics to the meshs becoming “empty” during calculation of déconfinement.
One proceeds in two times:

1. a first order STAT_NON_LINE, which makes it possible to reinject the loading made up of the
vector of the nodal reactions and the boundary conditions. The “empty” meshs thus correspond
to a very soft material;
2. a second call to STAT_NON_LINE who introduces supporting or the concrete pavement while
assigning to the corresponding meshs of the realistic characteristics for such a material.
With each call, the initialization of calculation takes again the entirety of the fields resulting from
preceding calculations (operand EVOL_NOLI for the keyword ETAT_INIT).
4.4.2 Method B
This procedure is based on the sequence of several models (to the direction Code_Aster). Calculation
is carried out by copying certain fields from one model to another.
Fields to be assigned to the model corresponding at the stage B−3 [Figure 4.4-a] are formally the
linear combination of two fields:
1. fields resulting from the preceding stage of calculation ( B−2 ) and which relates to only the
model corresponding to the solid mass of private ground of the excavated zone;
2. fields assigned to the elements of the group of mesh which represent the concrete voussoirs, in
the model which understands the solid mass and the gallery lining. In our case, these fields must
be initialized to 0 in B−3 . For that, one can for example assign a null weight to their contribution
in the linear combination. Thus these fields can in fact being obtained by an intermediate
calculation without real physical significance, for example the simple application of the boundary
conditions.
The order is used CREA_CHAMP with the option ADZE to assign to the points of Gauss of the third the
linear combination of fields resulting from preceding calculations models.
4.5 Summary of the methods suggested

To initialize the constraints, one can call on two methods:
1. Method I : to do a calculation (call to STAT_NON_LINE) on fictitious material;

2. Method II : to create the stress field wished by CREA_CHAMP.
To simulate the digging and the installation of the voussoirs, one has the choice between:
1. Method A : who consists in mechanically affecting “flexible” characteristics very in the

excavated zone;
2. Method B : who makes the use of several models which are connected and which is closer
to the physical reality of the modelled structure, the materials appearing and disappearing by
activation from one model to another. This method will be privileged.
A synthetic flow chart is proposed in [§Annexe2].
Version
Code_Aster default
b51375ecdb6c
5 Examples of command files

This part presents examples of structures of command file Code_Aster concerning a circular
excavation in infinite and elastic medium linear, within the framework of a purely mechanical study (not
of coupling THM). The cases presented here are the object of the cases test of validation SSLP105A,
SSLP105B and SSLP105C.
Three calculation cases are presented in this part:
1. an excavation without supporting with initialization by a bearing calculation on a fictitious
material to obtain the stress field wished (method I ) ;
2. an excavation with supporting, initialization of the constraints by a call to CREA_CHAMP and
followed by the method A for déconfinement and poses it voussoirs (methods II  A ) ;
3. an excavation with supporting, initialization of the constraints by a call to CREA_CHAMP and
followed by the method B for déconfinement and poses it voussoirs (methods II B ).
For cases 2 and 3, the scenario of digging is the following: excavation, déconfinement to 50% (
=0.5 ), poses voussoirs of 30 cm of thickness and end of déconfinement.
5.1 With the dealt problem

The geometry of the grid is listed in the paragraph [§4.1]. It contains 8477 nodes and 3304 elements.
The ray of the gallery is of 1,50 m , the thickness of concrete is 0,30 m and the grid is a square of
20 m of with dimensions. The other data are summarized in the following table.
Material Parameter Value
K0 1
Rock  v = h 5 MPa
E 4 GPa
 0,3
Concrete E 20 GPa
 0,2
Table 5.1-1: Data of the cases tests suggested
The boundary conditions and the loading are illustrated by the following figure:
Version
Code_Aster default
b51375ecdb6c
Pression des terrains sus jacents (5 MPa)
Ux = 0 Ux = 0
Déconfinement
Uy = 0
Figure 5.1-a: Boundary conditions imposed and loading
2. G
at the end of the process of déconfinement, = . U =0, 69 .
0 . R R
5.2 Case n° 1: excavation without supporting with initialization of the

constraints by a calculation and “softening” of the
“excavated” elements
This example is relatively simple: it is a question of simulating an excavation without installation of
supporting, with déconfinement total at the edge of the gallery. One thus uses one model for all
calculation.
The initial state is generated by a calculation (STAT_NON_LINE) who relates to the whole of the grid.
The properties of the elements are affected according to the state of stresses which one wants to reach
(here K 0=1 thus  =0,4999 , the value of 0,5 meaning the incompressibility of the rock not being
able to be used).
Following calculation relates to the nodal reactions at the edge of the future gallery. It is initialized
starting from the constraints resulting from the first call to STAT_NON_LINE.
The last call to STAT_NON_LINE is used to reinject the nodal reactions in a model where mechanical
properties of the excavated elements were very strongly weakened ( E tends towards 0. ). One
déconfine then completely the ground while making tighten these reactions towards 0 .
The command file corresponds to the case test SSLP105C.
5.3 Case n°2: excavation with supporting with initialization of the

constraints by call to CREA_CHAMP and déconfinement according to
method A
One follows the scenario of excavation described above. One uses that only one model for all
calculation. An order STAT_NON_LINE additional allows to introduce the voussoirs with a realistic
rigidity after déconfinement of 50%.
The command file corresponds to the case test SSLP105B.
Version
Code_Aster default
b51375ecdb6c
5.4 Case n°3: excavation with supporting with initialization of the

constraints by call to CREA_CHAMP and déconfinement according to
the method B
One always follows the scenario of excavation describes higher. This time, three models are used and
an intermediate calculation (without physical reality, called “can”) is necessary to transfer the fields
variables from one model to another at the time of the installation of the voussoirs after déconfinement
of 50%.
The command file corresponds to the case test SSLP105A.
Version
Code_Aster default
b51375ecdb6c
6 Validation of Code_Aster on an example of excavation in

linear springy medium
Validation of Code_Aster rest on the comparison of the digital results resulting from the cases n°1, 2
and 3 listed above with the analytical solution of [§Annexe 1]. For each calculation, one presents the
results got to the level of the keystone and the oven wall of the gallery, in term of radial constraint  R
constraint orthoradiale   and radial displacement U R ([Table 6-1], [Table 6-2] and Table 6-3]).
[§Annexe 7] presents two graphs describing the space evolution of the constraints along the vertical
axis of the model, with the right of the excavation. The good agreement between analytical solution and
digital results makes that the difference between these curves is hardly visible.
R (y)
 A
R
UR
B
Figure 6-a: Sizes compared for the validation of Code_Aster
Not A Not B
Variable Analytical Aster Relative Analytical Aster Relative
variation variation
 R ( Pa ) 0. −8.411 E3 It is checked 0. −1.625 E4 It is checked that
that ∣ R∣≪∣ ∣
∣ R∣≪∣ ∣
σ θ ( Pa ) −1. E7 −9.883 E6 1,2% −1. E7 −1.011 E7 1,1%
Ur ( m ) −0.0024375 −0.0024772 1,70% −0.0024375 −0.0023982 1,6%
Table 6-1: Case n°1, analytical comparison solution/results Code_Aster for

constraints radial and orthoradiale and for radial displacement in A and in B
Not A Not B
variation variation
 R ( Pa ) −1.52821 E6 −1.52974 E6 0.1% −1.52821 E6 −1.52652 E6 0,1%
  ( Pa ) −8.47179 E6 −8.40987 E6 0,7% −8.47179 E6 −8.52586 E6 0,6%
U r ( m ) −0.0016925 −0.0017218 1,7% −0.0016925 −0.0016664 1,5%

Version
Code_Aster default
b51375ecdb6c
Not A Not B
variation variation
 R ( Pa ) −1.52821 E6 −1.52943 E6 0,1% −1.52821 E6 −1.53171 E6 0,2%
  ( Pa ) −8.47179 E6 −8.40822 E6 0,8% −8.47179 E6 −8.52418 E6 0,6%
Ur ( m ) −0.0016925 −0.0017211 1,7% −0.0016925 −0.0016658 1,6%

The maximum difference between analytical and digital results is lower than 2%, except for the radial
constraint at the edge of the gallery excavated in the case n°1, where the theoretical value is 0. The
validity of calculation is checked by considering that the radial constraint is quite negligible in front of
the constraint orthoradiale.
Of course, all these variations can be reduced if the grid is still refined.
7 As a conclusion: advices and prospects

This note proposes a methodology which makes it possible to carry out calculations of excavation
using Code_Aster. Several scenarios of excavation reviewed and several methods are proposed.
The method and the software package are validated in the case of a circular gallery, dug in an infinite
solid mass consisted a linear elastic material. Code_Aster reproduced in a completely satisfactory way
the behavior of such an underground structure, with or without taking into account of supporting and/or
the coating.
From the point of view of the user, it seems more practical and faster to initialize the constraints by a
call to the order CREA_CHAMP rather than by a calculation on fictitious material.
If one seeks to model a purely mechanical behavior and if the phasage of the excavation is relatively
simple, to work with only one model appears to be the easiest method. It is enough to assign very
weak material properties to the meshs becoming “empty”. In the more complicated cases, the
implementation of several models used successively can prove more reliable from the point of view of
implementation the practical (error of modeling) and from the digital point of view (miscalculation), in
spite of the intermediate procedures of transfer of the fields (forced, internal displacements, pressures,
temperatures, variables…) from one model to another.
A later stage of validation of Code_Aster could be done on the linear coupled problems (THM in
medium saturated and elastic) or coupled and/or nonlinear (model CJS 1 being connected with the
model of Mohr-Coulomb, short-term excavation in not drained to compare with [3]).
Version
Code_Aster default
b51375ecdb6c
8 Bibliography
[1] D. BERNAUD and G. ROUSSET: “New implicit method” for the study of dimensioning of the tunnels,
French Review of Geotechnics n°60, pp 5-26, - 1992
[2] P. CATEL: Downstream of the Cycle – site of Bore-hole – Card 13 – Method convergence-containment,
notes EDF TEGG EFT GG/00 168 A – 2000
[3] A. GIRAUD: Couplings Thermo-Hydro-Mechanics in the not very permeable porous environments:
application to deep clays, thesis of the ENPC – 1993
[4] D. THE BOULCH: Comparison of modelings THM 3D and 2D of a work of storage with Code_Aster,
report Ajilon Technologies Cénergys 01-A – 2002
[5] MR. PANET: The calculation of the tunnels by the method convergence-containment, Presses of the
ENPC – 1995
[6] NR. SELLALI, C. CHAVANT and G. DEBRUYNE: Hydroplastic modeling of the excavation of an
underground gallery with Code_Aster, EDF MN HI-74/00/009/A notes – 2000
[7] NR. SELLALI, C. CHAVANT and G. DEBRUYNE: Modeling THM of an underground work of storage
with Code_Aster, EDF MN HI-74/01/014/A notes – 2001
Version
Code_Aster default
b51375ecdb6c
Annexe 1 Analytical formulas to apply the method convergence-

containment to the case of a rock and a supporting rubber
bands and linear solid mass
The medium is supposed to be elastic linear isotropic and subjected to an also isotropic stress field initial (
K 0=1 ).
Radial constraint, forced orthoradiale and radial displacement with the wall of the tunnel in springy
medium subjected toa rate of déconfinement 
{

 R= 1−
 . R2
r 2 . 0

 = 1
. R2
r 2
. 0
R2  0
U R =⋅ ⋅
r 2G
E
G is given by the following relation: G=
2⋅1 
Behavior of supporting:
That is to say K s the stiffness of supporting, it is given by the following relation if it is considered that
supporting is comparable to a thick or thin tube (  b is the Poisson's ratio of the concrete):
{
E b⋅e
si R10⋅e
 1− 2b ⋅R
K s=
E b⋅ R2e −R2i 
si R≤10⋅e
 1 b ⋅[  1−2⋅ b ⋅Re2R2i ]
That is to say P s confining pressure defined on the following figure
Version
Code_Aster default
b51375ecdb6c
One thus has:

P s⋅R= b⋅e
Ks
If k s= represent relative rigidity and  d the rate of déconfinement with the installation of supporting,
2⋅G
then the pressure of supporting and radial displacement in wall are given by:
{
ks
P s= ⋅ 1−d ⋅ 0
1k s
1d⋅k s  0
U R= ⋅ ⋅R
1k s 2⋅G
Version
Code_Aster default
b51375ecdb6c
Annexe 2 Flow chart of synthesis on the methods allowing to

simulate an excavation
Notations
The names of the objects are those of the command files presented in the appendices following.
SNL mean STAT_NON_LINE ; CC mean CREA_CHAMP ; CL mean boundary conditions
Stage 1: Initialization of the constraints

I : SNL1 with the loading of weight II : assignment by order CC desired field
actual or pressure wished and a material
equipped with a possibly fictitious Poisson's
ratio SOL
Stage 2: Recovery of the nodal reactions at the edge of the future gallery
CC to extract the constraints resulting from SNL1 with CL SNL1 with CL on
SNL1 on the object the object BORD in SOL_REST
SNL 2 with CL on the object BORD BORD in DIDI DIDI on model

in DIDI on model SOL SOL_REST
Recovery of the reactions Recovery of the Recovery of the reactions
BORD
reactions
Stage 3: Déconfinement
SNL3 with the SNL2 (model SNL 2 (only one
loading of the vector SOL_REST SOL ) with the material and model SOL_REST
of the nodal SOL_REST ) with the
loading of the vector
reactions and a “soft” loading of the vector of
of the nodal
material instead the nodal reactions
reactions and a
of “vacuum” BORD
BORD “soft” material
instead of “vacuum”
Eléments mous
Stage 4: Pose supporting

SNL 4 with 3 SNL3 with 3 CC to extract the
materials: rock, SOL_REST
materials: rock, results from SNL 2 SOL_REST
concrete and concrete and
vacuum (method vacuum (method SNL3 on model
SOL_REST + CONCRETE
A ) to complete A ) to complete ) for intermediate
déconfinement BORD déconfinement BORD
calculation
Combination of the fields
Eléments mous CC BETON
SNL4 to complete
déconfinement
Version
Code_Aster default
b51375ecdb6c
Annexe 3 Comparison of the constraints obtained by digital

calculation and the analytical solution
Case of the nonconstant tunnel
Evolution des contraintes selon l'axe vertical

r/R
0 2 4 6 8 10 12 14
0
-2
Contrainte (MPa)
-4
-6
Solution analytique Contrainte radiale
-8
Solution analytique Contrainte orthoradiale
-10 Calcul Code_Aster Contrainte radiale

Calcul Code_Aster Contrainte orthoradiale
-12
Case of the constant tunnel (from 50% of déconfinement)

Code - Aster: How To Dig A Tunnel: Methodology of Excavation

Uploaded by

Copyright:

Available Formats

Code - Aster: How To Dig A Tunnel: Methodology of Excavation

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Code - Aster: How To Dig A Tunnel: Methodology of Excavation

Uploaded by

Copyright:

Available Formats

Version

How to dig a tunnel: methodology of excavation

1 How to simulate the digging of a tunnel?

1. to evaluate the zone damaged by excavation (EDZ) around a gallery of storage;

This note presents an application to a command file of Code_Aster method convergence –

This report thus presents:

1) classical method of simulation of an underground excavation in 2D by means of

3 A method to simulate the digging of a gallery starting

1) plane deformations with assumption of small disturbances;

Figure 3.1-b: Example of curves of convergence and containment

3.2 Application of the method for a digital calculation by finite elements

Indeed, a classical course of modeling can be summarized by the following stages:

1) stage 1: initialization of the constraints in situ ;

Massif de sol Massif de sol

The first configuration is used for:

1. to initialize the constraints of origin geostatics;

4 Before attacking card-indexing of order Code_Aster…

4.1 How to define the models starting from a simple grid?

Figure 4.1-a: Grid used and materials

1) a model SOL , in which all the grid is affected finite elements;

Pesanteur induite par les

Partie non maillée mais

4.2 How to initialize the constraints?

1. realization of a calculation (order STAT_NON_LINE) with a fictitious material equipped with a

1) method A : quasi-cancellation of the mechanical properties of the elements

Figure 4.4-a: Various principles of modeling to simulate déconfinement solid mass

One proceeds in two times:

4.5 Summary of the methods suggested

1. Method I : to do a calculation (call to STAT_NON_LINE) on fictitious material;

1. Method A : who consists in mechanically affecting “flexible” characteristics very in the

A synthetic flow chart is proposed in [§Annexe2].

5 Examples of command files

5.1 With the dealt problem

Pression des terrains sus jacents (5 MPa)

Figure 5.1-a: Boundary conditions imposed and loading

5.2 Case n° 1: excavation without supporting with initialization of the

The command file corresponds to the case test SSLP105C.

5.3 Case n°2: excavation with supporting with initialization of the

The command file corresponds to the case test SSLP105B.

5.4 Case n°3: excavation with supporting with initialization of the

The command file corresponds to the case test SSLP105A.

6 Validation of Code_Aster on an example of excavation in

Figure 6-a: Sizes compared for the validation of Code_Aster

Table 6-1: Case n°1, analytical comparison solution/results Code_Aster for

Table 6-2: Case n°2, analytical comparison solution/results Code_Aster for

Table 6-3: Case n°3, analytical comparison solution/results Code_Aster for

7 As a conclusion: advices and prospects

Annexe 1 Analytical formulas to apply the method convergence-

That is to say P s confining pressure defined on the following figure

One thus has:

Annexe 2 Flow chart of synthesis on the methods allowing to

SNL mean STAT_NON_LINE ; CC mean CREA_CHAMP ; CL mean boundary conditions

Stage 1: Initialization of the constraints

SNL 2 with CL on the object BORD BORD in DIDI DIDI on model

Stage 4: Pose supporting

Annexe 3 Comparison of the constraints obtained by digital

Evolution des contraintes selon l'axe vertical

-10 Calcul Code_Aster Contrainte radiale

Case of the constant tunnel (from 50% of déconfinement)