Tunnelling and

Underground Space
Technology
incorporating Trenchless
Technology Research
Tunnelling and Underground Space Technology 22 (2007) 185–205
www.elsevier.com/locate/tust

A numerical approach to the hyperstatic reaction method for

the dimensioning of tunnel supports
P.P. Oreste *

Dept. of Ingegneria del territorio, dell’ambiente e delle geotecnologie (DITAG), Politecnico di Torino, Italy

Received 28 April 2005; received in revised form 24 April 2006; accepted 13 May 2006
Available online 10 August 2006

Abstract

The hyperstatic reaction method is particularly suitable for the dimensioning of support structures. This method simulates the inter-
action between the support and rock surrounding the tunnel through independent ‘‘Winkler’’ type springs. It requires the definition of
the active loads that are applied directly to the support structure by the rock mass. Further passive loads are due to the reaction of the
rock mass to the displacement of the support structure.
A numerical approach to the hyperstatic reaction method is presented in this paper. The parameters that condition the calculation and
the dimensioning techniques of the support structures, on the basis of the results of the method, are also presented. A specific code,
named FEMSUP, was developed using a FEM framework, to perform calculations with the HRM. This code is able to consider the
effective geometry of the support and the horizontal active loads that are different from the vertical ones; it is therefore able to analyse
the rock mass–structure interaction in detail.
An extensive parametric analysis performed using FEMSUP has made it possible to define the necessary support structures (steel sets
incorporated into a shotcrete lining) in a wide number of cases that cover the conditions that are generally encountered in excavation
practice. Ten design tables were drawn up to summarise the results of the parametric analyses. From an examination of the tables, it
is possible to verify the influence of the various calculation parameters on the dimensioning of the support structure and to obtain rough
indications on the entity of the support structures in the preliminary stages of a project.
 2006 Elsevier Ltd. All rights reserved.

1. Introduction both from the qualitative and quantitative point of view.

Furthermore, the times necessary to prepare the models,
Numerical methods that discretisize the entire medium carry out the calculations and interpret the results are usu-
around a tunnel are not commonly used for the dimension- ally very long.
ing of support structures due to the needed computational Simplified analytical methods, such as the convergence-
efforts. These structures are usually simulated through confinement method (Rechsteiner and Lombardi, 1974;
beam elements that are connected to the nodes of a numer- Hoek and Brown, 1980; Brown et al., 1983; Panet, 1995;
ical model, with the consequences that the displacements of Oreste, 2003), are able to stimulate the mean displacements
the structure are closely linked to the deformations of the of the support structure and the mean loads acting on it,
rock mass around the tunnel. If a correct interface, which but they do not supply the bending moments or the shear
does not permit the development of normal tensile stresses, forces that are necessary, together with the normal forces,
is not placed between the support and the tunnel perimeter, for its correct dimensioning. The relative stiffness solution
bending moment results are different from the true ones, (Einstein and Schwartz, 1979) is a very interesting and sim-
plified approach which allows one to obtain, for circular
tunnels and elastic ground, the bending moments and axial
*
Tel.: +39 0 11 5647608; fax: +39 0 11 5647699. forces in the support structure. This method, however,
E-mail address: [email protected]. considers the support realised in the tunnel without any

0886-7798/$ - see front matter  2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tust.2006.05.002
186 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205

previous convergence (relaxation) of the rock mass sur- bending stiffness) and the pressure-displacement relation
rounding the tunnel. that describes the interaction between the structure and
The hyperstatic reaction method (Duddeck and Erd- the rock mass.
mann, 1985; Bouvard-Lecoanet et al., 1988; Leca and The reaction produced by the rock mass on the defor-
Clough, 1992) (Fig. 1), which belongs to the numerical mations of the support is also a function of the stiffness
method category, is instead particularly suitable for the of the structure. This only depends on the mechanical
dimensioning of support structures. This method simulates and geometrical characteristics of the support. Its dimen-
the interaction between the support and rock surrounding sioning should therefore be carried out in a succession of
the tunnel through many independent ‘‘Winkler’’ type approximations, working on variations of input data and
springs: the reason it is named hyperstatic is due to the verifying the compatibility of the stress and strain state
great number of connections of the support structure with with the maximum admissible conditions.
the outside. The method requires definition of the active Steel sets incorporated into a shotcrete lining are one of
loads that are applied directly to the support structure by the most commonly used preliminary support methods.
the rock mass; these active loads can be estimated from Because of the bi-dimensional approach of the calculation
the technical literature (Barton et al., 1974; Barton, 2002; method and the composite nature of the structure, it is nec-
Unal, 1983; Goel et al., 1995, 1996; Singh et al., 1992, essary to determine the characteristics of an equivalent
1997; Bieniawski, 1989). They can also be estimated on continuous support that must be introduced into the calcu-
the basis of the in situ monitoring, using back-analysis pro- lation. The interpretation of the results relative to the
cedures. Further passive loads are due to the reaction of the equivalent support is a decisive step for the dimensioning
rock mass to the displacement of the support structure. of the different individual structural components and it is
The support–rock interaction influences the stress state in a step that requires particular care.
the structure to a great extent and this interaction depends A numerical approach to the hyperstatic reaction
on the mechanical characteristics of the rock mass. As these method is presented in this paper. The parameters that con-
are only generally known with a certain approximation, it is dition the calculation and the dimensioning techniques of
often necessary to carry out parametric or probabilistic type the support structures, on the basis of the results of the
analyses in order to be able to completely describe the method, are also presented.
uncertainty on the stress state of the support structure. A specific code, named FEMSUP, was developed using
These types of analyses need many calculations and the an FEM framework, to perform calculations with the
hyperstatic reaction method results to be particularly suit- hyperstatic reaction method (HRM). This code is able to
able for this purpose, due to the short time it requires. consider the effective geometry of the support and horizon-
Three parameters influence the stress state in a support tal active loads that are different from the vertical ones; it is
structure more than the others when a calculation with therefore able to analyse the rock mass-structure interac-
the hyperstatic reaction method is made: the applied active tion in detail.
loads, the stiffness of the structure (normal stiffness and The results of an extensive parametric analysis (4200
analyses) using this calculation code has made it possible
to prepare 10 design tables for the preliminary evaluation
of the type of support that is necessary, when the applied
external loads and the geomechanical quality of the rock
mass are known. Application examples on real cases show
how the calculation code and the indications that can be
obtained from the design tables, should be used.

2. The mathematical formulation of the method

The fundamental hypothesis on which the numerical

approach to the hyperstatic reaction method is based, is that
of subdividing the support structure into a finite number of
linear sub-dominions, called elements (for which it is possi-
ble to describe the stress–strain law in a simple way), which
are connected to the outside through stiffnesses distributed
Fig. 1. Calculation scheme of support structures with the hyperstatic over the nodes. The support structure is therefore repre-
method. Active loads are applied to the tunnel support by the rock mass in sented, in the calculation, by mono-dimensional elements
the roof (vertical loads, qv) and on the lateral sides (horizontal loads, qh). that are able to develop bending moments, axial forces
The active loads are independent of the displacements that develop in the
and shear forces (Fig. 2).
support and at the rock-support interface. Key: qv: vertical load; qh:
horizontal load; kn: stiffness of the interaction springs; R: tunnel radius; The beam element i is defined by the following parame-
E Æ J and E Æ A: bending and normal stiffness of the support. ters: the inertia modulus J and area A of the transversal
 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205 187

Gi and Ri give the axial forces in the initial node h and in

the final node j of the generic element i in the first two ele-
ments (Fig. 2), the transversal forces and the bending mo-
ment in the initial node h in the third and fourth elements,
and the transversal force and the bending moment in the
final node j in the fifth and sixth elements.
Eq. (1), re-written in explicit form, becomes:

Fig. 2. Scheme of the behaviour of a beam-type finite element, with

reference to the local Cartesian coordinates. Key: h: the initial node; j: the
final node; u: the axial displacement; v: the transversal displacement; u: the
rotation; x and y: the local Cartesian coordinates.

section, the elastic modulus E of the constituent material

and length Li (distance between the terminal connecting
nodes). In bi-dimensional methods, reference is made to a
unitary thickness and therefore the parameters J and A
should also be calculated considering the sum of the inertia
modules and the areas that are involved in the support for ð2Þ
a unitary length in the direction of the tunnel axis.
In the global Cartesian reference system, the nodal dis-
The unknown parameters of the problem are the dis-
placements q are connected to s through the following
placement components of the nodes of the discretized matrix expression:
structure. Once these unknown displacements are known,
it is possible to obtain the stress characteristics on the si ¼ ki  qhj;i ð3Þ
inside of each element and therefore also along the entire
length of the support structure. where
The evaluation of the unknown displacements is made
2 3
through the definition of the global stiffness matrix of the wh;i
entire structure and of its connections to the surrounding 6 7
6 zh;i 7
environment. The global stiffness matrix is obtained start- 6 7 " #
6 7
ing from the local stiffness matrices of each single element, 6 uh;i 7 qi
and then passing to their assembly. qhj;i ¼ 6 7
6w 7 ¼ q and
6 j;i 7 iþ1
The local stiffness matrix zi of the i element is obtained 6 7
6 zj;i 7
by placing the work produced by the inner forces of the 4 5
finite element equal to that produced by the external nodal uj;i
forces, which are evaluated according to a local reference 2 3
system (Huebner et al., 2001). cos ai sin ai 0 0 0 0
6 7
One therefore has: 6 0 0 0 cos ai sin ai 07
6 7
6 7
6 sin ai  cos ai 0 0 0 07
zi  si ¼ Gi þ Ri ð1Þ ki ¼ 6 7
6 0 0 1 0 0 07
6 7
si: the vector of the displacements in nodes h and j of ele- 6 7
6 0 0 0 sin ai  cos ai 07
ment i, evaluated according to the local reference 4 5
system: 0 0 0 0 0 1
si ¼ b uh;i uj;i vh;i uh;i vj;i uj;i c
qh‚j,i is the vector of the nodal displacements of element i; qi
u and v: the axial and transversal displacement in the is the vector of the nodal displacement of node i; w and z
local reference system; are the displacements along axis x and along axis y of
u: the rotation of the element in correspondence to the the global Cartesian reference system; ai is the angle that
nodes; forms the local Cartesian reference system of element i with
Gi and Ri: the external nodal forces and the nodal forces respects to the global Cartesian reference system.
applied by the neighbouring elements, which are evalu- In the same way, in the global Cartesian reference sys-
ated according to the local reference system; tem, the nodal forces are given by the following matrix
expressions:
[
GiT = Gxh ,i Gxj ,i G yh ,i Gzh ,i G yj ,i Gzj ,i ] = [G
T T
a ,i GbT,i ] Gi ¼ ki  F hj;i ð4Þ
RT
i = [R xh ,i Rxj ,i R yh ,i Rzh ,i R yj ,i Rzj ,i ] = [R
T T
a ,i
T
R
b ,i ] Ri ¼ ki  Qhj;i ð5Þ
188 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205

where the sub-vectors composed of the three external forces ap-

plied to each node.
Some of the elements of the stiffness matrix K are mod-
ified along the diagonal to consider the presence of the
interaction springs connected to the nodes of the structure.
a ai p
iþ1
K 3i2;3i2 ¼ K 3i2;3i2 þ k t;i  cos2 þ  ð9Þ
2 2 2
a ai p 
iþ1
K 3i1;3i1 ¼ K 3i1;3i1 þ k t;i  sin2 þ  ð10Þ
2 2 2
a ai p 
Making the necessary substitutions, it becomes possible iþ1
K 3i1;3i2 ¼ K 3i1;3i2 þ k t;i  sin þ 
to write in the global Cartesian reference system the follow- 2 2 2
a ai p
ing equation: iþ1
 cos þ  ð11Þ
2 2 2
zi  ki  qhj;i ¼ ðGi þ Ri Þ ¼ ki  ðF hj;i þ Qhj;i Þ a ai p 
iþ1
K 3i2;3i1 ¼ K 3i2;3i1 þ k t;i  sin þ 
2 2 2
And therefore: a ai p
iþ1
 cos þ  ð12Þ
2 2 2
ðkTi  zi  ki Þ  qhj;i ¼ F hj;i þ Qhj;i
where i is the number of the generic node; kt,i is the normal
which can be re-written in the following form: stiffness of the interaction spring connected to node i; ai
and ai+1 are the angle between the local and global refer-
k i  qhj;i ¼ F hj;i þ Qhj;i ð6Þ ence systems, for the element i and for the element i + l.
The constraints of the structure are placed by simply
where ki is the local stiffness matrix of the element i in the eliminating, from the stiffness matrix, the rows and col-
global Cartesian reference system: umns that refer to the nodal displacement that is prevented
by the constraint. The dimension of K therefore reduces in
k i ¼ kTi  zi  ki function of the number of degrees of freedom that have
been eliminated by the constraints.
The assembly of the local stiffness matrices in the global The vector of the nodal displacements q having been
stiffness matrix is made according to the criteria illustrated obtained from Eq. (7), it is then possible to evaluate the
in Huebner et al. (2001). stress characteristics at the nodes of the structure. In prac-
At the end of this stage, the matrix of the global stiffness tice, a conversion of the nodal displacements qh‚j,i is per-
K constituted by (3n + 3) · (3n + 3) elements, where n is formed at the local reference system of the generic
the total number of nodes, and the vectors of the unknown element i and, once the vector si is calculated, the character-
displacements q and the nodal forces F, both of them con- istics of stress Ci at the nodes can be determined immedi-
stituted by (3n + 3) elements, is obtained. ately through the local stiffness matrix:

K q¼F ð7Þ si ¼ ki  qhj;i ð13Þ

C i ¼ zi  si ð14Þ

3. The rock–support interaction

The rock mass interacts with the support in two ways:

through the normal springs connected to the nodes of the
structure (Fig. 3) and through applied active loads (qv
ð8Þ and qh in Fig. 1). The normal springs allow the reaction
produced by the rock to be simulated when the support,
where ki,a, ki,b, ki,c, ki,d is the sub-matrices of k, each of which deforms under the applied active loads, moves
them of 3 · 3 dimension: towards the rock.
The reaction of the rock can be described through the
  pressure–displacement relation (p–d), which, in the FEM-
k i;a k i;b
ki ¼ SUP calculation code, has been considered as being of a
k i;c k i;d
non-linear (hyperbolic) type (Fig. 4):
 
q1, q2, q3, . . . , qn+1 are the sub-vectors composed of the plim
p ¼ plim  1  ð15Þ
three displacements of each node; F1, F2, F3, . . . , Fn+1 are plim þ g0  d
 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205 189

where plim is maximum reaction pressure that the rock mass a thickness equal to the unitary study depth (along the tun-
can offer; g0 is the initial stiffness (for values of d close to 0) nel axis) (1 m) and a height equal to the depth of influence
of the rock mass. (k) inside the rock mass. k is the distance from the tunnel
This relation represents the simplest way to describe, perimeter at which the tensional effects produced by the ac-
without any artificial edge, the behaviour of the rock mass, tion of the support have ceased to exist: therefore, it can be
when the initial stiffness and the limit pressure are known assumed that, at a distance of k in the rock mass, the dis-
with a certain confidence. From the plate loading test in placements induced by the action of the support structure
the rock masses it is possible to note a curve load–displace- are nil and that a fixed point can be foreseen for the normal
ment that is very similar to the hyperbolic one (AFTES springs at this location.
guidelines on the plate loading test of the rock mass). The stiffness g0 can be estimated using the following sim-
The apparent stiffness g* of the rock is given by the ratio plified expression (United States Army Corps of Engineers,
p/d that can be obtained at each node of the support struc- 1997):
ture, that is, from:
E0
  g0 ffi 1:5  ð17Þ
 plim plim Deq
g ¼  1 ð16Þ
d plim þ g0  d
E0 is the initial elastic modulus of the rock mass; Deq: the
The evaluation of the initial rock stiffness g0 (bedding stiff- equivalent diameter of the tunnel.
ness) requires a ‘‘volume of influence’’ (v.o.i.) to be identi- The value of the maximum reaction pressure plim can be
fied in the rock mass for each spring (Oreste, 1999). The preliminary and prudently estimated starting from the
v.o.i. has the shape of a parallelepipedon with a width that cohesion values crm and the friction angle urm of the rock
is equal to the distance between the springs ((Li + Li+1)/2), mass, under the hypothesis of zero confining pressure:

2  crm  cos urm

plim ¼ ð18Þ
1  sin urm
node i+2
normal spring When the tunnel overburden is great, the tangential stresses
on the tunnel perimeter (confining pressure) cannot be neg-
element i+1 ligible and so to the limit pressure of Eq. (18) this second
term has to be added: (1 + sinurm)/(1  sinurm) Æ Drcon,
node i+1 where Drcon is the confining pressure on the tunnel
perimeter.
The stiffness kt,i of each spring is simply given by the
element i
product of gi (evaluated in the node to which the spring
is connected to) for the area of competence of that node:
node i
 
Fig. 3. Details of the rock–support interaction through Winkler springs  ðLi1 þ Li Þ
k t;i ¼ gi  1
connected to the support nodes. 2
 
plim plim ðLi1 þ Li Þ
¼  1  ð19Þ
di plim þ g0  di 2
plim As di are functions of the nodal displacements qi, the un-
reaction pressure of the rock p

knowns of the problem, it results that also the stiffnesses

kt,i of the springs connected to the nodes of the structure
are functions of the unknown nodal displacements. The
solution is reached through successive iterations, initially
α tgα = η0 placing gi ¼ g0 and reducing the value of gi and therefore
also of kt,i until the values of the nodal displacements ob-
tained from the calculation are compatible with the
hypothesised stiffnesses, less an error that can be consid-
ered negligible. As the displacements of the structure to-
wards the rock vary from node to node, the stiffnesses
Support normal displacement d will also be different for the hypothesised springs.
The springs disappear in zones where the support struc-
Fig. 4. Non-linear relation between the reaction pressure of the rock and
the displacement of the support (p–d). The value of the initial stiffness of ture moves towards the tunnel: this is generally the case of
the rock mass is equal to g0 and the maximum pressure plim is reached for the roof, but when the horizontal active loads are greater
very high values of d. a is equal to arctan(g0). than the vertical ones, it occurs at the sidewalls. Therefore,
190 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205

only compressive loads are possible in the normal direc- crushed rock). By analysing the potential instability of the
tion, where the tunnel support moves towards the rock rock mass on the tunnel sides and hypothesing a planar
mass: springs only work in compression. sliding surface, it is possible to evaluate the qh/qv ratio in
The shear stiffness at the interface rock–support was not function of the vertical load qv, the tunnel height H, the
considered: it is in fact not guaranteed that the interface specific weight c and the strength parameters crm and urm
can absorb shear stresses and it is very difficult to estimate of the rock mass:
the shear stiffness value. Using a numerical model was pos-    
sible to note that when the shear stiffness on the interface is qh cH 1  tan urm
tan a c
¼ 1þ  
neglected, the bending moments are greater both in the qv 2  qv 1 þ tan urm  tan a qv
crown and in the sidewalls. This assumed simplified  
1
hypothesis is then precautionary. 
sin a  ðcos a þ sin a  tan urm Þ
The active vertical load qv can be estimated using very
interesting direct correlations found by various authors
where a is the inclination of the sliding surface: a ¼ p4 þ u2rm .
(Barton et al., 1974; Barton, 2002; Unal, 1983; Goel
et al., 1995, 1996; Singh et al., 1992, 1997; Whickam
et al., 1972; Bieniawski, 1989) between the rock mass qual- 4. The equivalent support concept
ity index and the vertical loads on the roof of the tunnel
supports. In Fig. 5 are shown the graphic suggested by Apart from the interaction with the rock mass, the
Barton et al. (1974) and the nomogram of Goel et al. behaviour of the support is also influenced by its bending
(1995) to evaluate the vertical loads on the support struc- and normal stiffness and the restraining conditions that
tures. This method of evaluating the vertical loads are are assumed. The bending and normal stiffnesses depend
based on the monitoring of several case histories. on the elastic modulus of the support material and on the
The horizontal loads qh that are applied to the side walls geometry of its section. The type of constraint at the foot
are usually considered to be some percentage of the vertical can influence the support behaviour to a great extent and
ones. Generally, the ratio between the horizontal and ver- depends on the type of foundations and on the geometry
tical loads on the support structure is lower than the of the support.
in situ stress ratio K0 for a high quality rock mass, while The most commonly used type of preliminary support
it can be higher than K0 for very crushed rock masses structure in tunnels is that of steel sets inserted in a shotcrete
and for squeezing and swelling ground. For the Terzaghi’s lining. This is a composite system for which it is necessary to
Rock Load Classification, as modified by Deere et al. define a normal and a bending equivalent stiffness in order
(1970), the horizontal loads (side pressure) on the support to be able to use a bi-dimensional calculation method. For
structure are low when the geomechanical quality of the this purpose, it is considered that the equivalent support is
rock mass is medium or good and can become considerable made up of a homogeneous lining of a thickness s, along the
when the quality of the rock mass is very poor (completely entire axis of the tunnel, with an elastic modulus E. If the

Fig. 5. On the left: Barton et al. (1974) graphic for the evaluation of vertical loads on the support structures, when Q index and Jr parameter are known.
On the right: nomogram of Goel et al. (1995) for the estimation of vertical loads when the tunnel overburden H and radius a, with N index of the rock mass
are known.
 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205 191

stiffnesses of the real system are made equal to those of the Because of the variety of geometric forms and con-
equivalent support, the following is obtained: straints that the calculation code is able to consider, it is
  suitable for the dimensioning of both preliminary supports
Esteel Aset and concrete final linings.
normal stiffness : E  s ¼ Eshot  sshot þ  1  Eshot 
Eshot d For geometries no. 3 and no. 4, it is possible to foresee
mechanical characteristics of the support that are different
ð20Þ in the invert arch from the remaining zones on the perim-
    eter of the tunnel.
s3 s3 Esteel
bendings tiffness : E  ¼ Eshot  shot þ 1 The programme is also able to consider:
12 12 Eshot
J set  horizontal loads that are different from the vertical ones
 Eshot  ð21Þ
d (qh 6¼ qv);
 scale parameters, that are able to deform the support
where Eshot and Esteel are the elastic modulus of the shot- along the direction x and y (mx, my); the m coefficients,
crete and of the steel; sshot is the thickness of the shotcrete multiplied by the initial coordinates of the nodes, are
lining; Aset and Jset are the area and the moment of inertia able to modify their position and therefore the shape
of the section of individual steel sets; d is the spacing be- of the support;
tween the steel sets.  the mechanical parameters of the rock mass that condi-
By resolving the system made up of Eqs. (20) and (21), it tion its reaction to the deformations of the support
is possible to obtain: (crm, urm, Erm).
vhffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
  i
u Having defined the geometrical and mechanical param-
u Eshot  s3shot þ 12  Esteel  1  Eshot  J set
u E d eters that influence the problem, it is possible to obtain the
s ¼ t h  shot  i ð22Þ
Esteel
Eshot  sshot þ Eshot  1  Eshot  d Aset bending moments M, the normal force N and the shear
force T along the support profile, from the calculation.
h   i3 These parameters and the geometry of the section of the
Aset 2
Eshot  sshot þ EEsteel shot
 1  E shot  d support being known, the stresses induced inside the sup-
E ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
h   i ð23Þ port structures can finally be determined.
Eshot  s3shot þ 12  EEsteel shot
 1  E shot  J set
d To do this, it is usually preferred to hypothesise that the
bending moments and the shear forces in the preliminary
supports are entirely absorbed by the steel sets, while, as
It can be seen, for the usual type of supports adopted, far as the normal force N is concerned, it is divided between
how the thickness of the equivalent support is very similar the steel sets and the shotcrete lining on the basis of the
to the thickness of the shotcrete lining ðs ffi sshot Þ. The normal stiffnesses:
stiffness of the support system can therefore be repre-
sented only by the value of the elastic modulus of the
M set ¼ M  d ð24Þ
equivalent support (E).
For plane strain conditions, when the support structure T set ¼ T  d ð25Þ
is continuous along the tunnel axis (i.e. shotcrete lining
associated to steel sets), the elastic modulus of the equiva- Esteel  Aset =d Esteel Aset
N set ¼ N d ¼   N ð26Þ
lent support must be corrected in this following way: E  s E s
N set
N shot ¼N ð27Þ
 E d
E ¼
ð 1  m2 Þ
where N, M and T are the normal force, bending moment
and shear force in the equivalent support per metre of tun-
5. The FEMSUP calculation code and the dimensioning of nel, obtained using the FENSUP calculation code; Nset,
the supports Mset and Tset are the normal force, bending moment and
shear force in each single steel set; Nshot is the normal force
The FEMSUP calculation code, whose description is in the shotcrete per metre of tunnel.
given in Appendix A in detail, is able to consider four dif- The maximum stresses acting on the steel set (rmax,steel)
ferent support geometries (Fig. 6) and five different types of and in the shotcrete lining (rmax,shot) are calculated using
constraint at the base of the steel set (Fig. 7) in an auto- normal science of the construction techniques:
matic way. For the sake of simplicity, only half of the tun- in the maximum moment point (Tset = 0):
nel is considered, because of the symmetry of the problem.
The sign convention adopted by the numerical model is M set hset N set N shot
rmax;steel ¼  þ rmax;shot ¼ ð28Þ
shown in Fig. 8. J set 2 Aset sshot
192 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205

y y
final node final node

Rc Rc

π/2

Lp
1) initial node 2)
γ initial node

y y

final node final node

yc
Rc Rc

x π/2 x

Lp
3) 4) Ria
Ria

initial node initial node

Fig. 6. Four support geometries considered in the FEMSUP calculation code. The geometric parameters required for a univocal definition of the geometry
of the support structures are indicated in the figure. Key: Ria: the radius of the invert arch; Rc: the crown radius; Lp: the sidewall length.

1) 2) 3) 4) 5)
Fig. 7. Type of constraint foreseen: 1–4 for the initial node of geometries
no. 1 and no. 2; 5 for the initial node of geometries no. 3 and no. 4 and
for the final node of all 4 geometries. Key: (1) horizontal hinged-roller;
(2) fixed hinge; (3) horizontal clamped-roller; (4) clamped; (5) vertical
clamped-roller.

in the maximum normal force point (Tset 6¼ 0):

2 3
  sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2  2 Fig. 8. Sign convention adopted by the numerical model.
4 M set hset N set N set 3  T set  S set 5
rmax;steel ¼ max  þ ; þ
J set 2 Aset Aset bset  J set
The dimensioning of the support structure must be car-
N shot ried out in such a way that the maximum stresses induced
rmax;shot ¼ ð29Þ
sshot on its inside are lower than the maximum admissible values
for the materials that have been used.
where hset is the height of the steel set section; bset is the An analysis for the verification of a preliminary support,
width of the central zone of the steel set section; Sset is made up of steel sets and shotcrete lining, performed using
the static moment of the half-section of the steel set with the FEMSUP calculation code, is here given as an example.
respects to the barycentre axis. The case refers to a road tunnel (width 14 and height 11 m),
 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205 193

which is at present under construction in Piedmont (North A value of the maximum stress of 75 MPa in the steel set
Italy). The ground involved in the excavation has poor and of 0.90 MPa in the shotcrete lining can be found from
mechanical characteristics and the maximum depth of the an examination of Table 4, these being values that are
tunnel is 110 m. The geometrical parameters of the tunnel much lower than the maximum admissible ones. The max-
(scheme no. 3 of Fig. 6) are shown in Table 1, the geotech- imum stress in the steel set can be found at the centre of the
nical parameters of the ground are shown in Table 2 and crown, while the maximum stress in the shotcrete lining is
the mechanical and geometric characteristics of the obtained in correspondence to the invert arch.
adopted support structure are shown in Table 3. The esti- From the graph of Fig. 9 it is possible to note how the
mated vertical loads are 0.035 MPa, while the horizontal maximum moment is located at the centre of the crown
ones are 0.0175 MPa. The results that were obtained from and a consistent relative maximum is present in corre-
the calculation are given in Table 4 and Fig. 9. spondence to the back. Other relative maximums of the
moment, but of lower entity, can be found on the side-
Table 1 walls and in the invert arch. The normal forces N are
Geometric parameters of the examined tunnel almost constant in the invert arch and in the sidewalls,
Crown radius Rc (m) 6.15 but tend to diminish on the back and in the crown: they
Invert arch radius Ria (m) 10.82 reach a minimum at the centre of the crown. The shear
Height of the invert arch center yc (m) 6.15 forces show a peak in correspondence to the invert
Scale parameter in the x direction mx 1.14 arch-sidewall connection and a maximum relative between
Scale parameter in the x direction my 1.00
the back and the centre of the crown. The reaction pres-
sure supplied by the ground following the displacements
Table 2 of the support is very high in correspondence to the base
Geotechnical parameters of the ground of the sidewall, while it is obviously nil in the area of the
Cohesion crm (MPa) 0.25 crown where the structure tends to move towards the cen-
Friction angle urm 34 tre of the tunnel.
Elastic modulus Erm (MPa) 350
6. Validation of the proposed numerical model
Table 3
Mechanical and geometric characteristics of the support system In order to validate the proposed approach a simplified
Geometry of the steel set section example (Fig. 10) was studied also using the commonly
Moment of inertia Jset (m4) 4.276 · 105 employed numerical method FLAC (Itasca Consulting
Section area Aset (m2) 6.68 · 103 Group, 1994) which uses a difference finite approach. The
Section height hset (m) 0.20 comparison with analytical methods resulted to be difficult
Geometry of the support system because FEMSUP applies external active loads and reac-
Steel set spacing d (m) 1.00 tion loads on the nodes of the FEM model, while closed
Thickness of shotcrete lining sshot (m) 0.22 form solutions consider external loads as distributed pres-
Mechanical characteristic sures. FLAC code allows to compile routines with the lan-
Steel elastic modulus Esteel (MPa) 210,000 guage FISH; a specific fish program, which permits to
Shotcrete elastic modulus Eshot (MPa) 12,000 consider the same interaction principles of the FEMSUP
Equivalent support characteristic code (Fig. 3), was realized. The support structure was made
Elastic modulus E (MPa) 16,470 up of 40 elements and four different cases have been consid-
Thickness s (m) 0.24
ered in regard to the linkage in point A (Fig. 10): case (1):

Table 4
Calculation results
Crown center Back Sidewall Invert
Equivalent support
Bending moment M (MN · m/m) 0.0272 0.0227 0.0081 0.0135
Normal force N (MN/m) 0.2122 0.2695 0.2991 0.3075
Shear force T (MN/m) 0 0 0 0
Steel set
Steel set bending moment Mset (MN/m ) 0.0272 0.0227 0.0081 0.0135
Steel set normal force Nset (MN) 0.0751 0.0954 0.1059 0.1089
Maximum stress in the steel set rset,max (MPa) 75 67 35 48
Shotcrete lining
Normal force in the shotcrete lining Nshot (MN) 0.1371 0.1741 0.1932 0.1986
Maximum stress in shotcrete rshot,max (MPa) 0.62 0.79 0.88 0.90
194 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205

Invert arch Sidewall Crown Invert arch Sidewall Crown

norm. force N (MN)

mom. M (MN x m)
0.02 0.3
0 0.28
0.26
-0.02 0.24
-0.04 0.22
0 5 10 15 20 0 5 10 15 20
a b -3
x 10
shear force T (MN)

Normal displ. (m)

0.05 10

0 5

-0.05 0

0 5 10 15 20 0 5 10 15 20
c d
reaction press. (MPa)

0.1 0.05

rotations (˚)
0
0.05 -0.05
-0.1
0
0 5 10 15 20 0 5 10 15 20
e support progr. (m) f support progr. (m)

Fig. 9. Results obtained from the calculation for the equivalent support. Key: (a) bending moments along the support; (b) normal force; (c) shear force;
(d) normal displacements (positive towards the inside of the tunnel); (e) reaction pressure of the ground; (f) rotation of the support structure. The points
along the support in which the internal stresses in the steel sets and in the shotcrete lining were obtained are shown with circles.

Fig. 10. Simplified calculation example used for the validation of the proposed approach with a commonly used numerical method (FLAC):
qv = 3.33 MPa, qh = 1.87 MPa, g0 = 200 GPa/m, E = 30 GPa, s = 0.4 m, R = 4 m.

horizontal hinged-roller constraint; case (2): hinged; case Always the same results were obtained by FLAC and
(3): clamped; case (4): horizontal clamped-roller. A vertical FEMSUP code for all the four studied cases. In Fig. 11
clamped-roller constraint was assumed in point B. A hori- are shown some of these results for the case 2 (simple
zontal load equal to half of the vertical load was applied to hinged constraint assumed in point A). This type of con-
the support. straint forces nil normal displacements and bending
 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205 195

2.0 20.0

bend. mom. (MNxm/m)

axial forces (MN/m)

1.5 femsup
15.0
1.0 flac
10.0
0.5
femsup
0.0 5.0
flac
-0.5 0 2 4 6
0.0
-1.0 0 2 4 6
support length (m) support length (m)

1.0 0.025
shear forces (MN/m)

0.5 0.020 femsup

normal displ. (m)

0.015 flac
0.0
0.010
-0.5 0 5
0.005
-1.0 femsup
0.000
-1.5 flac
-0.005 0 5
-2.0 -0.010
support length (m) support length (m)

Fig. 11. Comparison results (bending moments, axial force, shear force and normal displacement along the support) of the simplified example of Fig. 10
for the case 2 (linkage in point A: fixed hinge). Support length (that is the circumferential distance along the support, measured from point A) is equal to 0
for the point A and reach its maximum value for the point B.

moments in point A. While the clamped-roller constraint in Four rock masses were also considered; these were char-
point B forces only nil shear forces (and rotations). acterised by their geomechanical qualities: RMR = 25, 35,
45 and 55. The geomechanical parameters of the rock mass
7. Parametric analysis and design tables that intervene in the support-rock interaction were
obtained for each RMR value: cohesion, friction angle
Two extensive parametric analyses were developed in and elastic modulus (Table 7). The analysis relative to
order to better investigate the effects of the geomechanical RMR = 55 was not performed for geometry (a) as invert
quality of the rock mass, the stiffness of the support struc- arches are not usually installed for preliminary supports
ture and the applied loads on the value of the maximum in rock masses with this geomechanical quality.
moments and the axial forces in the equivalent support. Vertical loads of between 0.025 and 0.25 MPa were
The first analysis (a) was developed for the geometry of applied to the support, for intervals of 0.025 MPa. Three
the support given as an example in the Section 5 (Table different qh/qv ratios were analysed: 0.33, 0.66 and 1.
1), while the second (b) was for a geometry that was similar Tunnel dimensions equal to 50% (b = 0.5), 75%
to the first, but without the invert arch. The geometrical (b = 0.75), 125% (b = 1.25), 150% (b = 1.5) of the original
parameters of the second geometry that was studied (b) ones, identified from the geometrical parameters in Tables
are given in Table 5 and they refer to scheme 2 of Fig. 6. 1 and 5, were also investigated for each of the two consid-
The constraint that was adopted at the base of the sidewall ered geometries.
in geometry b is that of a hinge (type of constraint no. 2 of The two parametric analyses that were developed (for
Fig. 7). This type of constraint results the more suitable geometries a and b in Fig. 12) calculated 1800 and 2400 dif-
when the steel set is founded in a groove in the ground ferent cases, respectively, using the FEMSUP code, in such
(the typical case) later filled with shotcrete. In this case a way as to be able to cover all the possible situations that
the steel set foot can not move but only rotate. could be encountered in design practice.
Four support systems (steel set + shotcrete lining) were The values of the moments and the normal forces at the
used in the calculation: light (L), medium (M), heavy (H) centre of the crown (1), on the back (2), in the side walls (3)
and massive (MV). The mechanical characteristics and and in the invert arch (4), where the maximums relative of
the geometries of each of these systems are reported in the bending moment were obtained (Fig. 9), were identified
Table 6. for each of the examined cases. In these points the shear
forces are obviously nil (T = 0).
Table 5 In order to simplify the interpretation of the results, the
Support parameters for geometry b bending moment and the normal force have been expressed
Crown radius Rc (m) 7.00 as function of the load qv and of the width of tunnel B:
Height of the sidewall Lp (m) 2.50
Scale parameter in the x direction mx 1.00 M ¼ v  qv  B2 ð30Þ
Scale parameter in the y direction my 1.00 N ¼ l  qv  B ð31Þ
196 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205

Table 6
Mechanical and geometrical characteristics of the four support systems adopted in the parametric study: (1) light (L); (2) medium (M); (3) heavy (H); (4)
massive (MV)
Type L Type M Type H Type MV
Geometry of the steel set section
Moment of inertia Jset (m4) 9.34 · 106 2.138 · 105 4.276 · 105 7.784 · 105
Section area Aset (m2) 2.28 · 103 3.34 · 103 6.68 · 103 7.82 · 103
Section height hset (m) 0.16 0.20 0.20 0.24
Geometry of the support system
Steel set spacing d (m) 1.25 1.00 1.00 1.00
Thickness of shotcrete lining sshot (m) 0.20 0.22 0.22 0.25
Mechanical characteristic
Steel elastic modulus Esteel (MPa) 210,000
Shotcrete elastic modulus Eshot (MPa) 12,000
Type L Type M Type H Type MV
Equivalent support characteristic
Elastic modulus E (MPa) 13,600 14,200 16,470 15,900
Thickness s (m) 0.20 0.23 0.24 0.28

Table 7  v in observation points 1 and 2 tends to increase with qv,

Geomechanical parameters of the rock mass for the considered RMR
diminish with the RMR and more slightly with k, when
values
b < 1; it also increases with the stiffness of the support
RMR = 25 RMR = 35 RMR = 45 RMR = 55
structure for b < 1; when b P 1 it is basically constant
Cohesion 0.23 0.33 0.46 0.66 with respects to qv, k, the RMR, the type of support
crm (MPa)
used and b itself; the values of v in observation points
Friction angle 30 36 40 44
urm 1 and 2 basically result to be the same;
Elastic modulus 1500 2650 4700 8350  v in observation point 3 tends to increase with qv and k,
Erm (MPa) to diminish with the RMR, when b 6 1; it also results to
increase with the stiffness of the support structure for
y y b 6 1; when b > 1 it is basically constant with respects
final node final node to qv, k, the RMR, the type of support used and b itself;
1 1  v in observation point 4 tends to increase with qv, above
yc Rc
all for low values of the RMR, k and b; it increases with
2 2
Rc the stiffness of the support structure above all for b 6 1;
when b > 1 it is basically constant with respects to qv, k,
π/2
x x the RMR, the type of support used and b itself;
 the value of l in the four observation points varies quite
3
3 Lp considerably, from 35% to 67%, according to the case
Ria
(in general it assumes higher values on the sidewalls
4 and in the invert arch);
initial node initial node
geometry a) geometry b)  l in observation point 1 is independent of qv and k,
while it increases slightly with a diminishing of the stiff-
Fig. 12. Critical points in the support structure in which the values of the ness of the support for b 6 1; it varies quite remarkably
bending moment and the normal force were identified for each of the cases
with the dimension of the tunnel for low values of b:
examined in the parametric analyses.
l = 35–42% for b = 0.5, l = 40–45% for b = 0.75,
The analyses of v and l, with a variation of the various l = 42–47% for b P 1;
parameters that influence the calculation, has made it pos-  l in observation point 2 is not very dependent on qv and
sible to draw the following conclusions: k, while it increases slightly with a diminishing of the
Geometry (a) of Fig. 12: stiffness of the support for b P 1; it assumes higher val-
ues than those obtained in observation point 1 (l = 50–
 the value of v in the four observation points vary quite 55%);
remarkably, from 0% to 1% according to the case (in  l in observation point 3 is usually not very influenced by
general it assumes higher values in the centre of the qv, while it depends on the stiffness of the support and
crown and on the back): 0.1–1% per b = 0.5, 0–0.6% on k when b < 1: it diminishes with an increase in k
per b = 0.75, 0–0.4% per b P 1; and in the stiffness of the support; it is sensitive to the
 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205 197

dimensions of the tunnel: l = 50–60% for b = 0.5, The graphs of the calculation results (for v and l
l = 55–64% for b = 0.75, l = 59–65% for b = 1; respectively) in the centre of the crown (critical point 1),
l = 61–66% for b = 1.25; l = 62–67% for b = 1.5; for geometry (a) of Fig. 12 and for a ratio qh/qv = 1,
 l in observation point 4 is not influenced by qv or by k, are reported in Figs. 13 and 14 as an example. A certain
while it depends on the stiffness of the support when influence of load qv, the type of support and the geome-
b < 1: it diminishes with an increase in the stiffness of chanical quality index RMR can be noted both for v
the support; it is sensitive to the dimensions of the tun- and for l.
nel: l = 55–61% for b = 0.5, l = 60–64% for b = 0.75, Once v and l are known, the type of support is chosen
l = 60–65% for b = 1; l = 60–65% for b = 1.25; on the basis of a comparison of the maximum stresses in
l = 61–67% for b = 1.5; the steel and in the shotcrete, obtained from the calculation
according to the criteria explained in Section 5, with the
Geometry (b) of Fig. 12: maximum admissible values for these materials.
If, for example, it is necessary to dimension the support
 the value of v in the three observation points varies from structure of a tunnel with a shape and dimensions like
0 to 0.9% according to the case (in general it assumes those of type (a) reported in Fig. 12 (b = 1), in a rock mass
higher values on the sidewalls for k = 1); v is practically which has an RMR index = 35, in which a vertical load qv
independent of the value of qv; on the preliminary supports equal to 0.05 MPa and a ratio
 v in observation point 1 tends to diminish with the qh/qv = 1 have been estimated, the following v and l
RMR and increase with the stiffness of the support parameters can be obtained for the centre of the crown
structure when b < 1; when b P 1 it is basically constant from Figs. 13 and 14.
with respects to the RMR, the type of support used and
b itself; v results to be influenced by k, above all for
k P 0.66;
 v in observation point 2 tends to slightly diminish with
the RMR and increase with the stiffness of the support RMR = 25
0.0040
structure when b < 1; when b P 1 it is basically constant
with respects to the RMR, the type of support used and 0.0035 support L
M / (q v x B 2)

b itself; v results to be influenced by k: it shows mini- support M

0.0030 support H
mum values for k = 0.66 and maximum values for k = 1;
support MV
 v in observation point 3 is only influenced by the value 0.0025
of k and b; for k = 0.33 and 0.66 v is independent of b
and equals 0.17 and 0.39, respectively; when k = 1 the 0.0020
0.00 0.05 0.10 0.15 0.20 0.25
following values are obtained: v = 0.8–0.92 for b = 0.5,
qv (MPa)
v = 0.7–0.8 for b = 0.75, v = 0.6–0.75 for b = 1,
v = 0.6–0.7 for b = 1.25 and for b = 1.5; RMR = 35
 the value of l in the three observation points varies quite 0.0040

remarkably, from 37% to 72%, according to the case (in

0.0035
M / (q v x B 2 )

general it assumes higher values on the back and the support L

sidewalls); l is practically independent of qv and only support M
0.0030
support H
slightly dependent on the type of support;
support MV
 l in observation point 1 is influenced by k for k > 0.66: it 0.0025
increases quite remarkably when passing from k = 0.66
0.0020
to k = 1; for k 6 0.66 l equals 40I42 independently of 0.00 0.05 0.10 0.15 0.20 0.25
b, while for k = 1 l increases with an increase in b and qv (MPa)
passes from l = 37–38% for b = 0.5 to l = 52–58% for
b = 1, and then remains practically constant for b > 1; RMR = 45
0.0040
 l in observation point 2 is influenced by k: it increases
with an increase in k; it is equal to about 50%, indepen- 0.0035
M / (qv x B 2 )

support L
dently of b for k = 0.33 and reaches maximum values support M
for k = 1: 56–60% for b = 0.5, 60–64% for b = 0.75, 0.0030
support H
63–68% for b = 1, 64–70% for b = 1.25, 62–72% for 0.0025
support MV
b = 1.5;
 l in observation point 3 is very influenced by k: it is min- 0.0020
0.00 0.05 0.10 0.15 0.20 0.25
imum for k = 0.66 and reaches a maximum value for
qv (MPa)
k = 1; it is equal to 53–58% for k = 0.33, it diminishes
at 48–54% for k = 0.66 and reaches 68% for k = 1; with Fig. 13. Development of v = M/(qv Æ B2) with a variation of the vertical
equal k it tends to increase with b. load qv for different types of support and for an RMR = 25, 35 and 45.
198 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205

RMR = 25 v4;L ¼ 0:00117 l4;L ¼ 0:644

0.48
v4;M ¼ 0:00130 l4;M ¼ 0:628
v4;H ¼ 0:00130 l4;H ¼ 0:627
N / (q x B)

0.46 support L
support M v4;MV ¼ 0:00153 l4;MV ¼ 0:632
support H
0.44 support MV Once the v and l parameters are known, the bending mo-
ments M and the axial forces N in the equivalent support
0.42
can be derived from Eqs. (30) and (31) (qv = 0.05 MPa
0.00 0.05 0.10 0.15 0.20 0.25 and B = 14 m) (Table 8).
qv (MPa) From these values it is possible to derive the bending
moment in the single steel set Mset (MN · m), the axial
RMR = 35 forces in the single steel set Nset (MN) and in the shotcrete
0.48
Nshot (MN/m), and the maximum stresses in the steel
rmax,steel (MPa) and in the shotcrete rmax,shot (MPa) (Tables
support L
N / (q x B)

0.46
support M
9 and 10).
support H From an examination of the stresses in the support
0.44 support MV structure, it can be seen that, for the example under exam-
ination, the maximum values in the steel are reached in cor-
0.42 respondence to observation points 1 (crown) and 2 (back),
0.00 0.05 0.10 0.15 0.20 0.25 while the maximum values in the shotcrete are in corre-
qv (MPa) spondence to observation point 4 (invert arch). As a value
of 220 MPa can be considered as the maximum admissible
RMR = 45
0.48 stress value in the steel, because of the provisional charac-
ter of the preliminary supports, and a value of 5 MPa in the
support L
shotcrete, the adequate support for the examined case
N / (q x B)

0.46
support M results to be of type M (medium) (Table 6).
support H Working in the same way for all the 4200 cases of the
0.44 support MV parametric analyses, it was possible to define indicative
tables of the type of support that is necessary (Tables 11–
0.42 20), with a variation of the applied load, of the coefficient
0.00 0.05 0.10 0.15 0.20 0.25 k, of the geomechanical quality (RMR) and of the dimen-
qv (MPa) sions of the tunnel (b) for the two different geometries of
the support structure. In the cases where it is not possible
Fig. 14. Development of l = N/(qv Æ B) with a variation of the vertical
load qv for different types of support and for an RMR = 25, 35 and to identify a support type that is able to support the loads
45. transmitted by the rock mass, it is necessary to intervene
with reinforcement and consolidation works of the rock
mass that lead to a reduction of the loads acting on the sup-
v1;L ¼ 0:00229 l1;L ¼ 0:463 port structure.
The slenderness of the sidewalls was taken into account
v1;M ¼ 0:00303 l1;M ¼ 0:445 in the evaluation of the stress conditions on the sidewalls
v1;H ¼ 0:00315 l1;H ¼ 0:442 for the geometry (b). In more detail, the normal stress in
v1;MV ¼ 0:00305 l1;MV ¼ 0:448 the steel and in the shotcrete where the bending moment
reaches a maximum value was obtained by the following
In the same way, it is possible to obtain the following Eq. (32), which substitute the Eq. (28):
parameters for observation points 2 (back), 3 (sidewalls) 2 3
and 4 (invert arch): M eq h N M h N
rmax;steel ¼ max 4 5
set set max set set
  þ xset  ;  þ xset 
J set  1  1:5  N set 2 Aset J set 2 Aset
N cr
v2;L ¼ 0:00208 l2;L ¼ 0:552
v2;M ¼ 0:00298 l2;M ¼ 0:544 N shot
rmax;shot ¼ xshot  ð32Þ
v2;H ¼ 0:00306 l2;H ¼ 0:542 sshot
v2;MV ¼ 0:00262 l2;MV ¼ 0:547 where Meq is equivalent bending moment in the steel set
(sidewalls): Meq = 1.3 Æ Mmed with 0.75 Æ Mmax 6 Meq 6
v3;L ¼ 0:00044 l3;L ¼ 0:635 Mmax; Mmed and Mmax mean and maximum bending mo-
v3;M ¼ 0:00053 l3;M ¼ 0:618 ment in the sidewall steel set; Ncr is Eulerian critical axial
2
v3;H ¼ 0:00059 l3;H ¼ 0:610 force: N cr ¼ p Ekst2 Aset ; kset is slenderness of the sidewall steel
set
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
v3;MV ¼ 0:00067 l3;MV ¼ 0:615 set: kset ¼ lp = J set =Aset ; lp: sidewall length; xset is a factor
 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205 199

Table 8
Bending moments and normal forces in the equivalent support, in correspondence to the 4 observation points where the bending moment shows a relative
maximum
Observation points Type of support
L (light) M (medium) H (heavy) MV (massive)
1 (crown center)
M (MN · m/m) 0.0224 0.0297 0.0309 0.0299
N (MN/m) 0.324 0.312 0.309 0.314
2 (back)
M (MN · m/m) 0.0204 0.0292 0.0300 0.0257
N (MN/m) 0.386 0.381 0.379 0.383
3 (sidewalls)
M (MN · m/m) 0.0043 0.0052 0.0058 0.0066
N (MN/m) 0.445 0.433 0.427 0.431
4 (invert arch)
M (MN · m/m) 0.0115 0.0127 0.0127 0.0150
N (MN/m) 0.451 0.440 0.439 0.442

Table 9
Bending moments and normal forces in the steel set and in the shotcrete lining, in correspondence to the 4 observation points where the bending moment
shows a relative maximum value
Observation points Type of support
L (light) M (medium) H (heavy) MV (massive)
1 (crown center)
Mset (MN · m) 0.0281 0.0297 0.0309 0.0299
Nset (MN) 0.0562 0.0662 0.1095 0.1132
Nshot (MN/m) 0.2791 0.2453 0.1999 0.2004
2 (back)
Mset (MN · m) 0.0255 0.0292 0.0300 0.0257
Nset (MN) 0.0670 0.0809 0.1343 0.1382
Nshot (MN/m) 0.3328 0.2999 0.2451 0.2447
3 (sidewalls)
Mset (MN · m) 0.0054 0.0052 0.0058 0.0066
Nset (MN) 0.0771 0.0919 0.1512 0.1554
Nshot (MN/m) 0.3828 0.3407 0.2758 0.2751
4 (invert arch)
Mset (MN · m) 0.0143 0.0127 0.0127 0.0150
Nset (MN) 0.0782 0.0934 0.1554 0.1597
Nshot (MN/m) 0.3883 0.3462 0.2835 0.2827

Table 10
Maximum stresses in the steel and in the shotcrete, in correspondence to the 4 observation points where the bending moment shows a relative maximum
value
Observation points Type of support
L (light) M (medium) H (heavy) MV (massive)
1 (crown center)
rmax,steel (MPa) 265 159 89 61
rmax,shot (MPa) 1.40 1.12 0.91 0.80
2 (back)
rmax,steel (MPa) 248 161 90 57
rmax,shot (MPa) 1.66 1.36 1.11 0.98
3 (sidewalls)
rmax,steel (MPa) 80 52 36 30
rmax,shot (MPa) 1.91 1.55 1.25 1.10
4 (invert arch)
rmax,steel (MPa) 157 88 53 44
rmax,shot (MPa) 1.94 1.57 1.29 1.13
200 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205

Table 11
Support typology necessary for geometry (a) of Fig. 12 and for b = 0.5 (B = 7 m), with a variation of qv, k and RMR
qv (MPa) RMR = 25 RMR = 35 RMR = 45
k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00
0.025 L L L L L L L L L
0.050 L L L L L L L L L
0.075 M M M L L L L L L
0.100 M M M M M M M M M
0.125 H M M M M M M M M
0.150 H H H H H H M M M
0.175 H H H H H H H H H
0.200 H H H H H H H H H
0.225 MV MV H H H H H H H
0.250 MV MV MV MV MV H H H H

Table 12
Support typology necessary for geometry (a) of Fig. 12 and for b = 0.75 (B = 10.5 m), with a variation of qv, k and RMR
qv (MPa) RMR = 25 RMR = 35 RMR = 45
k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00
0.025 L L L L L L L L L
0.050 L L L L L L L L L
0.075 M M M M M M M M M
0.100 H H M M M M M M M
0.125 H H H H H H H H H
0.150 H H H H H H H H H
0.175 MV MV MV H H H H H H
0.200 – MV MV MV MV MV MV MV MV
0.225 – – – – MV MV MV MV MV
0.250 – – – – – – – – –

Table 13
Support typology necessary for geometry (a) of Fig. 12 and for b = 1 (B = 14 m), with a variation of qv, k and RMR
qv (MPa) RMR = 25 RMR = 35 RMR = 45
k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00
0.025 L L L L L L L L L
0.050 M M M M M M M M M
0.075 M M M M M H H H H
0.100 H H H H H H H H H
0.125 H H H H H H MV H H
0.150 MV MV MV MV MV MV MV MV MV
0.175 – – – MV MV MV MV MV MV
0.200 – – – – – – – – –
0.225 – – – – – – – – –
0.250 – – – – – – – – –

Table 14
Support typology necessary for geometry (a) of Fig. 12 and for b = 1.25 (B = 17.5 m), with a variation of qv, k and RMR
qv (MPa) RMR = 25 RMR = 35 RMR = 45
k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00
0.025 L L M L M L M L L
0.050 M M H M M M H M M
0.075 H H H H H H H H H
0.100 H H H H H H MV MV H
0.125 MV MV MV MV MV MV – – MV
0.150 – – – – – – – – –
0.175 – – – – – – – – –
0.200 – – – – – – – – –
0.225 – – – – – – – – –
0.250 – – – – – – – – –
 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205 201

Table 15
Support typology necessary for geometry (a) of Fig. 12 and for b = 1.5 (B = 21 m), with a variation of qv, k and RMR
qv (MPa) RMR = 25 RMR = 35 RMR = 45
k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00
0.025 M M L M L M M M M
0.050 H H H H H H H H H
0.075 H H MV MV MV H MV H H
0.100 MV MV – – – MV – MV MV
0.125 – – – – – – – – –
0.150 – – – – – – – – –
0.175 – – – – – – – – –
0.200 – – – – – – – – –
0.225 – – – – – – – – –
0.250 – – – – – – – – –

Table 16
Support typology necessary for geometry (b) of Fig. 12 and for b = 0.5 (B = 7 m), with a variation of qv, k and RMR
qv (MPa) RMR = 25 RMR = 35 RMR = 45 RMR = 55
k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00
0.025 L L L L L L L L L L L L
0.050 L L L L L L L L L L L L
0.075 L L M L L M L L M L L M
0.100 L L M L L M L L M L L M
0.125 M M H L M H L M H L M H
0.150 M M H M M H M M H M M H
0.175 M M H M M H M M H M M H
0.200 M M MV M M MV M M MV M M MV
0.225 H H MV M H MV M H MV M H MV
0.250 H H MV H H MV H H MV H H MV

Table 17
Support typology necessary for geometry (b) of Fig. 12 and for b = 0.75 (B = 10.5 m), with a variation of qv, k and RMR
qv (MPa) RMR = 25 RMR = 35 RMR = 45 RMR = 55
k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00
0.025 L L L L L L L L L L L L
0.050 L M M L M M L M M L M M
0.075 L M H L M H L M H M M H
0.100 M M H M M H M M H M M H
0.125 M H MV M H MV M H MV M H MV
0.150 M H – M H – M H – M H –
0.175 H H – H H – H H – H H –
0.200 H MV – H MV – H MV – H MV –
0.225 H MV – H MV – H MV – H MV –
0.250 H MV – H MV – H MV – H MV –

Table 18
Support typology necessary for geometry (b) of Fig. 12 and for b = 1 (B = 14 m), with a variation of qv, k and RMR
qv (MPa) RMR = 25 RMR = 35 RMR = 45 RMR = 55
k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00
0.025 L L M L L M L L M L L M
0.050 L M H L M H L M H M L H
0.075 M H MV M H MV M H MV M M MV
0.100 M H – M H – M H – H H –
0.125 H MV – H MV – H MV – H MV –
0.150 H MV – H MV – H MV – H MV –
0.175 H – – H – – H – – MV – –
0.200 H – – H – – MV – – MV – –
0.225 MV – – MV – – MV – – – – –
0.250 MV – – MV – – MV – – – – –
202 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205

Table 19
Support typology necessary for geometry (b) of Fig. 12 and for b = 1.25 (B = 17.5 m), with a variation of qv, k and RMR
qv (MPa) RMR = 25 RMR = 35 RMR = 45 RMR = 55
k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00
0.025 L M H L M H L M H L L H
0.050 M H MV M H MV M H MV M L MV
0.075 M H – M H – H H – H M –
0.100 H MV – H MV – H MV – H H –
0.125 H – – H – – H – – MV H –
0.150 H – – H – – MV – – MV H –
0.175 MV – – MV – – – – – – MV –
0.200 – – – – – – – – – – – –
0.225 – – – – – – – – – – – –
0.250 – – – – – – – – – – – –

Table 20
Support typology necessary for geometry (b) of Fig. 12 and for b = 1.5 (B = 21 m), with a variation of qv, k and RMR
qv (MPa) RMR = 25 RMR = 35 RMR = 45 RMR = 55
k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00 k = 0.33 k = 0.66 k = 1.00
0.025 L M H L M H M L H L L H
0.050 M H – M H – M M – M M –
0.075 H MV – H MV – H H – H M –
0.100 H – – H – – H – – H H –
0.125 MV – – MV – – – – – – H –
0.150 MV – – MV – – – – – – MV –
0.175 – – – – – – – – – – – –
0.200 – – – – – – – – – – – –
0.225 – – – – – – – – – – – –
0.250 – – – – – – – – – – – –

that is function of the type of the steel set section and of the necessary to install the invert arch foreseen in geometry
sidewall steel set slenderness: xset ffi 9  105  k2set  (a) or to intervene with reinforcement and consolidation
0:0013  kset þ 1:0402 when kset > 20; xshot is a factor that techniques of the rock mass.
is function of the sidewall shotcrete slenderness: xshot ffi
2  104  k2shot  0:0239  kshot þ 1:5857 when kshot > 50; kp ffiffiffi
shot 8. Conclusions
is slenderness of the sidewall shotcrete: kshot ¼ 2  3
ðlp =sshot Þ. The hyperstatic reaction method is particularly suitable
The previously reported design tables permit indications for the dimensioning of the most widely used support struc-
to be made on the type of support that is necessary, in a tures in tunnels: steel sets incorporated in a shotcrete lining.
simple and quick way, having defined the geometry of the The method allows the results of the bending moment, the
tunnel, the geomechanical quality of the rock mass, the ver- shear forces and the normal force along the support profile
tical loads foreseen on the support and the ratio k between to be obtained, in reduced calculation times.
the horizontal and vertical loads. Using these tables, it is A numerical type approach to the hyperstatic reaction
also possible to verify the influence of the various calcula- method has been presented in this paper and the adopted
tion parameters on the dimensioning of the support struc- mathematical procedure, the parameters that condition
ture, in order to decide whether it is convenient to study the the calculation and the techniques used to dimension the
unknown parameters in more detail. They are also useful support structures, on the basis of the results of the
during the dimensioning analyses as they help one to start method, are also given. The FEMSUP code is described.
with the numerical calculations considering types of sup- This code is able to consider the actual geometry of the
port that are close to the final ones, thus saving precious support and the horizontal active loads that are different
time. from the vertical ones; it is therefore able to analyse the
From an examination of the tables it can be seen how rock mass-structure interaction in detail.
geometry (b) in Fig. 9 is more influenced by coefficient k An extensive parametric analysis has made it possible to
than geometry (a). When k = 1, the support structures define the necessary support structures in a wide number of
are able to bear modest loads with geometry (b), above cases that cover the conditions that are generally encoun-
all for large dimensions of the tunnel. In these cases, it is tered in excavation practice: four different geomechanical
 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205 203

quality values (described through the RMR index), ten ver- Appendix A. Detailed description of the FEMSUP
tical load values, three coefficient k values, two support calculation code
structure geometries (with and without the invert arch), five
different tunnel dimensions (from a width of 7 m to a width The FEMSUP calculation code has been written in
of 21 m) and four support types, characterised by the thick- MATLAB programming language and it foresees the fol-
ness of the shotcrete lining and by the foreseen type of steel lowing calculation stages:
set, were all considered. The moment and the normal force
were evaluated for each case in different critical points of  Definition of the elements that make up the structure;
the support (the centre of the crown, the back, the sidewalls  Automatic determination of the nodes on which the
and the invert arch), where a maximum in the development applied external loads act;
of the bending 1moment can be observed. The maximum  Construction of the global stiffness matrix, starting from
stresses in the steel and in the shotcrete were then calcu- the local stiffness matrix;
lated and compared with the maximum admissible stresses  Modification of the global stiffness matrix in order to
of the two materials. Finally, the most suitable type of sup- take the constraint conditions into consideration;
port was chosen, on the basis of these comparisons, for  Definition of the nodal forces in the zone in which the
each studied case. applied external loads act;
Ten design tables were drawn up to summarise the  Iterative calculation that foresees a progressive reduc-
results of the parametric analyses. tion (up to the annulment) of the stiffness of the
From an examination of the tables, it is possible to ver- springs which work under traction (because of the dis-
ify the influence of the various calculation parameters on placement of the support towards the tunnel) and the
the dimensioning of the support structure in order to be continuous modification of the stiffness of the springs
able to decide whether it is convenient to perform an in which work under compression (because of the dis-
depth study of the parameters that are not precisely known. placement of the support towards the rock), in function
These tables are also useful during the dimensioning anal- of the displacement value reached in the previous cal-
yses as they allow one to start with the numerical verifica- culation step;
tions from configurations that are similar to the final  Resolution of the system of linear equations and obtain-
calculation ones, thus saving precious time. They can also ing the nodal displacement vectors, for each step of the
be adapted to supply rough indications on the entity of iterative calculation;
the support structures in the preliminary stages of the pro-  Once the iterative process has finished, the nodal dis-
ject or to suggest variations, during the tunnel construc- placements under the global coordinates system are first
tion, of the characteristics of the supports foreseen in the converted considering the local coordinates system of
design phase, on the basis of the actually encountered rock each element and then multiplied by the local stiffness
mass geomechanical quality. matrix (expressed under the local coordinates system)
In order to carry out the calculation with the FEM- to obtain the nodal forces and therefore the stress states
SUP code or to use the design tables presented in this of each element;
paper, it is necessary to know the entity of the loads  Plotting of the results.
applied to the support structure. In particular, it is neces-
sary to know the vertical load qv and the coefficient k, the The data of the problem that are necessary for the calcu-
ratio between the applied horizontal and vertical loads lation are directly inserted into the programme, attributing
(qh/qv). These two important parameters can be evaluated the respective value to each variable: the name of the var-
initially on the basis of empirical relations and geome- iable in the calculation programme is here given together
chanical indexes of the rock mass. Different empirical cor- with each datum:
relations are available in literature that are able to supply
indications on qv in function of the geomechanical quality  Type of support geometry (Fig. 6) (‘‘geom’’): this can be
index (RMR or Q) or of the geomechanical parameters of equal to 1, 2, 3 or 4;
the rock mass. Vertical and horizontal loads can also be  Number of elements in the support structure (for half of
obtained backwards from the convergence measurements the tunnel) (‘‘num_el’’);
during the construction of tunnels. In this case, FEMSUP  Crown radius (‘‘Rcal’’);
code can be useful to verify static conditions of the sup-  Stretching coefficient of the structure in the horizontal
ports which have been defined in the design phase. There direction (x) (‘‘mx’’): this coefficient is multiplied by the
is, however, a lack of a rational organisation of knowl- coordinate x of each node of the structure in order to
edge in this field; this lack of knowledge on the vertical obtain a modification of the support geometry (for values
loads acting on the support structures requires further of ‘‘mx’’ above 1, a widening of the structure is obtained,
in depth scientific studies so as to be able to estimate their while for values below 1, a diminishing is obtained);
value, with a certain precision, under the various condi-  Stretching coefficient of the structure in the vertical
tions that can be encountered during the excavation of direction (y) (‘‘my’’): as for ‘‘mx’’, a deformation of
a tunnel. the structure occurs, but now in the vertical direction;
204 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205

 Elastic modulus of the equivalent support in the crown Asost3 = Asost1;

and sidewall zones (‘‘Esost1’’); Jsost3 = Jsost1;
 Area (thickness) of the equivalent support in the crown elseif geom == 4
and sidewall zones (‘‘Asost1’’); Rar = 10.9;
 Inertia moment of the equivalent support in the crown Lp = 2.2;
and sidewall zones (‘‘Jsost1’’); Esost3 = 16470;
 Initial angle of the crown development (only for geom- Asost3 = .24;
etry 1 in Fig. 6) (‘‘gamma_in’’); Jsost3 = .24^3/12;
 Type of constraint on the initial node (Fig. 7) (‘‘vinc’’): end
this can be equal to 1, 2, 3 or 4 (constraint type 5 is auto-
matically imposed in the initial node of geometries no. 3
pv = .025;
and no. 4 of Fig. 6);
ph = .666*pv;
 Length of the rectilinear sidewall (only for geometries
no. 2 and no. 4 of Fig. 6) (‘‘Lp’’);
 Radius of the invert arch (only for geometries no. 3 and c = .33;
no. 4) (‘‘Rar’’); fi = 36/57.297;
 Distance between the centres of the crown and the invert Eroccia = 2650;
arches (only for geometry no. 3) (‘‘yc’’);
 Elastic modulus of the equivalent support in the invert The measurement units that have been used are those of
arch zone (geometries no. 3 and no. 4) (‘‘Esost3’’); the international system (SI): m for the lengths, MPa for
 Area (thickness) of the equivalent support in the invert the pressures and rad for the angles. A bi-dimensional type
arch zone (geometries no. 3 and no. 4) (‘‘Asost3’’); calculation (plain strain analysis) has been made and there-
 Inertia moment of the equivalent support in the invert fore a depth of the tunnel equal to 1 m has been considered.
arch zone (geometries no. 3 and no. 4) (‘‘Jsost3’’); At the time of writing, the FEMSUP calculation pro-
 Vertical pressure applied to the crown of the support gramme is available on the following website:
structure (‘‘pv’’); http://staff.polito.it/pierpaolo.oreste/index.htm
 Horizontal pressure applied to the sides of the support
structure (‘‘ph’’);
References
 Cohesion of the rock mass (‘‘c’’);
 Friction angle of the rock mass (‘‘fi’’); AFTES, 1997. Guidelines on the plate loading test of the rock mass,
 Elastic modulus of the rock mass (‘‘Eroccia’’). Tunnel et Ouvrages Souterrain.
Barton, N., 2002. Some new Q-value correlations to assist in site
The first lines of the FEMSUP programme, dedicated to characterisation and tunnel design. International Journal of Rock
Mechanics and Mining Sciences, 185–216.
the input of the data of the problem, are given in the
Barton, N., Lien, R., Lunde, J., 1974. Engineering classification of rock
following masses for the design of tunnel support. Rock Mechanics (6), 183–236.
Bieniawski, Z.T., 1989. Engineering Rock Mass Classification. John
geom = 2; Wiley, New York.
num_el = 50; Bouvard-Lecoanet, A., Colombet, G., Esteulle, F., 1988. Ouvrages
souterrains: conception, réalisation, entretien. Presses de l’école
Rcal = 7;
nationale des Ponts et chaussées, Paris.
mx = .75; Brown, E.T., Bray, J.W., Ladanyi, B., Hoek, E., 1983. Ground Response
my = .75; Curves for Rock Tunnels. Journal of Geotechnical Engineering (109),
15–39.
Deere, D.U., Peck, R.B., Parker, H., Monsees, J.E., Schmidt, B., 1970.
Esost1 = 13600; Design of tunnel support systems. High Res. Rec. (339), 26–33.
Asost1 = .2; Duddeck, H., Erdmann, J., 1985. On structural design models for tunnels
Jsost1 = Asost1^3/12; in soft soil. Underground Space (9), 246–259.
Einstein, H.H., Schwartz, C.W., 1979. Simplified analysis for tunnel
supports. Journal of the Geotechnical Engineering Division (April),
if geom == 1 499–518.
Goel, R.K., Jethwa, J.L., Paithankar, A.G., 1995. Indian experience with
gamma_in = 90/57.297;
Q and RMR systemTunnelling and Underground Space Technology,
vinc = 1; Vol. 10. Pergamon, pp. 97–109.
elseif geom == 2 Goel, R.K., Jethwa, J.L., Dhar, B.B., 1996. Effect of tunnel size on
Lp = 2.5; support pressure, Technical note. International Journal of Rock
vinc = 1; Mechanics and Mining Science and Geomech. 33 (7), 749–755.
Abstract.
elseif geom == 3
Hoek, E., Brown, E.T., 1980. Underground Excavations in Rock. The
Rar = 10.82; Institution of Mining and Metallurgy, London, 527 p.
yc = 6.15; Huebner, K.H., Dewhirst, D.L., Smith, D.E., Byrom, T.G., 2001. The Finite
Esost3 = Esost1; Element Method for Engineers. John Wiley and Sons, inc., New York.
 P.P. Oreste / Tunnelling and Underground Space Technology 22 (2007) 185–205 205

Leca, E., Clough, W., 1992. Preliminary design for NATM tunnel support Congress of the International Society for Rock Mechanics, pp. 1049–
in soil. Journal of Geotechnical Engineering 118 (4), 558–575. 1054.
Oreste, P., 1999. Important aspects of the analysis and dimensioning of Singh, B., Jethwa, J.L., Dube, A.K., Singh, B., 1992. Correlation between
tunnel supports through numerical methods. Gallerie e Grandi Opere observed support pressure and rock mass qualityTunnelling and
Sotterranee, April 1999, 57:39–50. underground space technology, Vol. 7. Pergamon, pp. 59–75.
Oreste, P., 2003. Analysis of Structural Interaction in Tunnels using the Singh, B., Goel, R.K., Jethwa, J.L., Dube, A.K., 1997. Support pressure
Convergence-Confinement Approach. Tunnelling and Underground assessment in arched underground openings through poor rock
Space Technology. Pergamon Press, Oxford. massesEngineering Geology, Vol. 48. Elsevier Science, 59–81.
Panet, M., 1995. Le calcul des tunnels par la méthode convergence- Unal, E., 1983. Design guidelines and roof control standards for coal mine
confinement. Presses de l’école nationale des Ponts et chaussées, Paris. roofs, Ph.D. Thesis, Pennsylvania State University.
Rechsteiner, G.F., Lombardi, G., 1974. Une methode de calcul elasto- U.S.A.C.E. (United States Army Corps of Engineers), 1997. Tunnels
plastique de l’état de tension et de deformation autour d’une cavite and Shafts in Rocks: Engineering and Design, chapter 9. Washington
souterraine. In: Advances in rock mechanics, Proceedings of the Third D.C.