Anda di halaman 1dari 6

Computational methods for deep tunneling design

S. Francia
SPEA Autostrade S.p.A., Milano, Italy
E. M. Pizzarotti, M. Rivoltini & C. Pecora
S.IN.C. S.r.l., Milano, Italy

ABSTRACT: In the deep tunneling design for the Milano Roma highway stretches known as Variante
di Valico, calculation procedures have been applied in which the analytical method of the Ground Reaction
Curve (GRC) is coupled to continuum discretization with the Finite Differences Method (FDM).

Analyses procedure is aimed to evaluate the strain distribution on the cavity boundary surface as a func-
tion of the distance from the front face considering also strengthening operations of the front face. Further-
more the analyses provide indications on the axial strain (extrusion) of the excavation front and on the exten-
sion of plastic areas as well as absolute and relative radial displacements of the natural cavity at any distance
from the front face and on the extensions of plastic zones.

Checks on the structural elements which are part of the first phase and final linings are carried out on the
basis of the stress states produced by the succession of the excavation and strengthening phases, simulated
through proper installation distances of these elements from the front face.

1 INTRODUCTION strengthening works of the front face, which are

simulated by increasing the rock mass strength pa-
This paper illustrates the computational proce- rameters or introducing structural elements in the
dures applied to deep tunneling design for the Mi- model. Furthermore the analyses provide indications
lano Roma highway stretches known as Variante on the axial strain (extrusion) of the excavation front
di Valico. and on the development of plastic areas as well as
In this case calculation procedures have been absolute and relative radial displacements of the
applied in which the analytical method of the natural cavity at any distance from the front face and
Ground Reaction Curve (GRC) is joined by contin- on the extensions of plastic zones.
uum discretization with the Finite Differences Checks of the structural elements which are part
Method (FDM) for the modeling of the supporting of the first phase and final lining are carried out fol-
structures and stabilization means. The constitutive lowing the stress states produced by the succession
model of the considered rock mass follows the plas- of the excavation and strengthening phases, simu-
ticity curve described by the Hoek & Brown formu- lated through proper installation distances of these
lation. elements in relation to the front face.
The verification analyses of the standard sec-
tions were carried out for tunnel stretches with ho-
mogeneous geomechanical conditions and sufficient
length to allow the application of the hypothesis of
plane strain state. Thus the analyses do not represent
local areas of intense fracture and limited length, for
which specific measures were disposed anyhow.
Procedure is aimed to evaluate the strain distri-
bution on the cavity boundary as a function of the
distance from the front face, considering also
2 INTERACTION BETWEEN GEOLOGICAL - ci: uniaxial compressive strength of the intact
STRUCTURAL DESIGN - mb; s; a: parameters which depend upon the rock
mass characteristics.
m/mi = exp (GSI-100)/28
2.1 Geologic characteristics
s = exp (GSI-100)/9 (3)
The geologic-structural shape, in which a part of

the layout of the new highway is designed, is the re- a=1/2 (4)
sult of complex geologic evolution of the Apenni-
nes. In order to use the Hoek & Brown criterion for
In application below shown, it is possible to char- estimating the strength and the deformability of
acterize two geomorphological zones, coupled to jointed rock masses, three properties of the rock
geological soil formation. mass have to be estimated:
The first one is characterized by hard morphol- - ci: uniaxial compressive strength of intact rock
ogy, frequently with steep slope, that are middle pieces;
mountain areas with sandstone-marl rock, called - mi: Hoek & Brown constant for the rock mass;
Macigno. In this zone the tipical landslide are by - GSI: value of Geological Strength Index for
collapse, by block movement and by detritus slide. rock mass.
The second one, clearly crumbly, is characterized
by allochthonous formation called Argille Scagli- The value of Geological Strength Index (GSI) is
ose. In this zone it is possible to examine corru- based on the value of Rock Mass Rating (RMR), by
gated elevation and little steep slope, subjected to in- the equation:
tense erosion, with flow slide and superficial GSI = URMR-5 (5)
gravitational distortion.
Below it is shown the analysis of a particular sec- where:
tion of tunnel when the mass rock is characterized URMR = RMR (6)
by Argille Scagliose.
evaluated by R5 = 15 and R6 = 0.
2.2 Geomechanical characteristics Moreover, to model the post peak behavior of the
The rock has been classified referring to the rock mass, it has been considered a residual value of
Rock Mass Rating (RMR) of Bieniawski. GSI once reached the frontier of the plastic criteria.
The geomechanical parameters, necessary for the About the Deformation Modulus (E) Serafim &
evaluation of the RMR value and of the consequent Pereira proposed a relationship between the in situ
rock qualitative class, are: modulus of deformation and Bieniawskis RMR
- R1: uniaxial compressive strength of the rock classification. Hoek & Brown proposed the follow-
material; ing modification:
- R2: Rock Quality Designation, RQD; GSI 10
- R3: spacing of discontinuities; E= 10 40
- R4: condition of discontinuities; 100
- R5: groundwater conditions;
Regarding the non associated flow rule, in the
- R6: orientation of discontinuities.
analytical solution there is a parameter f varying
from 1.0 (no dilatancy) to fa (associated flow rule):
By sum of absolute value of each parameter you
have numerical evaluation of RMR, by which you mb
assign one of the five rock mass classes of Bi- fa = 1+ 

eniawskis method. pcr
2 mb +s


The behavior of the fractured rock has been char- c

acterized by the Hoek & Brown constitutive model:

: where
pcr critical radial stress to have plastic zone

around the cavity.

1 = 3 + ci mb 3 + s



2.3 Analysis tools and algorithm

- 1: maximum effective stress at failure; The rock mass constitutive model is homogene-
- 3: minimum effective stress at failure; ous, isotropic, elasto-plastic, following a flow rule
non associated. Hoek & Brown model defines the has been carried out with or without that structures.
criteria. The strengthening phase is simulated by increasing
Instead the rock mass subject by excavation is geomechanical parameters just for the zones near by
non homogeneous and anysotropic one. This aspect front correlated to pressure applied.
simplifies the real behavior of the rock mass, which D) Elaborating the function displacements around
means a careful definition of the excavation and the cavity vs distance from the front in junction with
strengthening phases in order to contrast a poten- the function displacements around cavity vs internal
tially different behavior of the rock mass from re- pressure (results from A, B and C) you could get the
sults of analyses. function distance from front vs internal pressure.
The analyses, using the method of the characteris- The internal pressure is equivalent to strength of the
tic lines (Ground Reaction Curves, GRC, analytical front and it is known as excavation fictitious load
solution to the problem of a hole subject to variable (in italian,forza fittizia di scavo) (FFS below).
internal pressure in a continuous elasto-plastic, in E) For each proper section of tunnel the analy-
junction with the numerical solution of Panet for the sis are carried out by finite difference method (plain
behavior to a certain distance from the front) and fi- strain hypothesis), considering the real geometry of
nite difference model, estimate the stress and the
excavation and strengthening phases (radial and/or
displacement of the rock mass around the cavity and
umbrela rockbolts, steel frame support, spritz, rein-
therefore verify the dimensioning of the structures
The finite difference method is a numerical techi- forced concrete inverted arch). For each step of cal-
que to solve a set of differential equations, given ini- culus (excavation or installation) it has been evalu-
tial values and/or boundary values. The rock mass is ated the value of FFS correlated to the proper
represented by a mesh, where each element (or zone) distance of installation from front, to simulate the
has a particular constitutive model. real behavior in progress of the rock mass.
The calculus starts from geostatic condition, F) The least step, i. e. when compatibility and
where the model simulates the natural behavior. The equilibrium are obtained in all the rock mass zones
next steps introduce stress changes. and structures, the structural verify is carried out by
Excavation could be simulated by deleting some Ultimate State Limit method.
zone. It is possible to apply pressures on the bounda- The algorithm described is evaluated for each
ries of the mesh. Also it is possible to simulate a proper section of tunnel, i. e. for each length of tun-
strengthening phase by modifing the geomechanical nel where geomechanical conditions are near the
parametres or introducing some structure: a beam, if same and approximate the plain strain hypothesis.
the structures have flexural behaviour; cable, if it is The analysis cannot evaluate behavior for local
necessary to simulate radial rockbolts. fractured rock, for which specific works must be de-
The algorithm process is : signed.
A) For each set of geomechanical parameters
and overburden, it is carried out an analytical analy-
sis by GRC, considering a cavity with radius 1.00m. 3 APPLICATION
The results (absolute convergence and exten-
sion of plastic zone considering no pressure inside
the cavity) allows the first evaluation about stability
3.1 Variante di Valico
of cavity.
B) Carry out the finite difference analysis by It is illustrated the same calculus results on a
axialsimmetric model of a cylindrical hole (circular particular section of tunnel.
cavity with radius 1.00m) to evaluate the function In table 1 are represented the geomechanical data
convergence vs internal pressure. By that analyses it used to carry out the analysis:
is possible to modify locally the mesh (extension Table 1. Geomechanical data
and number of zones), because the results have to tipo
Mi [Gpa]
mb s
mr sr
referee to analytical ones. AS 35 1,00 35 25 8 5 1,19 0,30 0,49 0,000730 0,49 0,000730

C) Axialsymmetrical analysis to simulate exca- About Argille Scaglioses (AS, laminated clay)
vation progress. The results are: behavior, the swelling and viscoplasticity are con-
- to get displacements around cavity (without siderated usind a reduction of 10% for GSI value
strengthening structures) correlated to the distance with reduction of the geomechanical parameters.
from front; The section illustrated is PA-2/4/5, used in
- to evaluate strain and extension of plastic presence of Argille Scagliose and GSI value of 35.
zones, particullary near front. The overburden is 250m.
Regarding the rock mass where it is necessary In table 2 are illustrated: the strengthening
to use strengthening structures on front, the analysis structures (used in the analysis by FLAC), the proper
distance from the front face and the value of FFS JOB TITLE : GEOSTATICO - MESH INGRANDITA

FLAC (Version 3.40)


(evaluated at medium distance of installation). It has

LEGEND 1.200

been considered an over-excavation of 15cm. 17-Nov-99 17:29

step 547
-1.000E+00 <x< 2.000E+01
-6.000E+00 <y< 1.500E+01
Table 2. Structures and distance of installation. 0.800

Grid plot

STRUCTURE dmed[m] FFS 0 5E 0

Front 0.0 18,4% 0.400

Radial rockbolts 1.0 17,2%

Spritz and steel frame + 4.0 7.6%
inverted arch) (extra-excavation 0.000

(thickness = 100cm) 15cm)

Lining No bound- 0.0%
Medium thickness = 85cm ary -0.400

SEZ. PA 2/4/5
0.200 0.600 1.000 1.400 1.800

The analysis has been carried out using the curve


distance from the front vs FFS, calculated by the Figure 3. Mesh: particular of boundary of excavation.
axialsimmetrical numerical analysis including front
consolidation (figure 1). JOB TITLE : APPLICAZIONE FFS (*10^1)

FLAC (Version 3.40)


FLAC (Version 3.40) 2.250

17-Nov-99 17:30
step 572
-1.000E+00 <x< 2.000E+01
LEGEND 1.750
-6.000E+00 <y< 1.500E+01

28-O ct-99 17:21 Material model

step 14030 1.250 mohr-coulomb
-1.000E+00 <x< 4.000E+00 Net Applied Forces
-2.500E+00 <y< 2.500E+00 Max Vector = 3.408E+06
state 0 1E 7
At Yield in Shear or Vol.
Elastic, Yield in Past
Boundary plot

-0.250 0.000
0 1E 0

Exaggerated Boundary Disp.

Magnification = 4.516E+00 -0.750
Max Disp = 7.034E-02
Displacement vectors
Max Vector = 7.034E-02 -0.400

0 2E -1 SEZ. PA 2/4/5
-1.750 0.200 0.600 1.000 1.400 1.800

A.S. GSI=35-COP=250
-0.750 -0.250 0.250 0.750 1.250 1.750 2.250 2 .750 3.2 50 3.750 Figure 4. Applying FFS pressure along excavation boundary.

Figure 1. Axialsimmetrical analysis. Convergence at the front JOB TITLE : FASE 8:SCAVO E GETTO ARCO ROVESCIO (*1 0^ 1)

FLAC (Version 3.40)

and along the cavity.
LEGEND 1 .20 0

22-Nov-99 11:56
In the figures from 2 to 5 are illustrated the step 45947
-1.000E+00 <x< 2.000E+01

mesh and some steps of calculus. -6.000E+00 <y< 1.500E+01

Material m odel
0 .80 0

m ohr-coulom b
Beam plot

JO B TITLE : MESH CO MPLETA (*10^1) Cable plot

0 .40 0
FLAC (Version 3.40) 7 .00 0


5 .00 0
17-Nov-99 17:29 0 .00 0

step 547
-3.833E+01 <x< 1.083E+02
-6.833E+01 <y< 7.833E+01
3 .00 0

Grid plot
-0 .40 0

0 2E 1
1 .00 0 SEZ. PA 2/4/5
0.20 0 0.60 0 1 .0 00 1 .4 00 1 .80 0
(*10 ^1 )

-1 .00 0

Figure 5. Radial rockbolts.

-3 .00 0

-5 .00 0
In the figures from 6 to 9 are illustrated the re-
sults during the calculus process (extension of plas-
SEZ. PA 2/4/5
-0.2 00 0.0 00 0.200 0.400 0.600 0.800 1.0 00
tic zones, design loads for the structures).
(*10^2 )

Figure 2. Complete Mesh.


FLAC (Version 3.40) FLAC (Version 3.40)

1.200 1.200

22-Nov-99 11:56 23-Nov-99 14:05

step 45947 step 56947
-1.000E+00 <x< 2.100E+01 -1.000E+00 <x< 2.100E+01
-7.000E+00 <y< 1.500E+01 0.800
-7.000E+00 <y< 1.500E+01 0.800

Material model Material model

mohr-coulom b mohr-coulom b
Beam plot Beam plot
Cable plot Cable plot
0.400 0.400
# 3 (Cable) -4.051E+05 Moment on
# 4 (Cable) -4.129E+05 Structure Max. Value
# 5 (Cable) -4.201E+05 # 1 (Beam ) 2.021E+06
# 6 (Cable) -4.224E+05 # 2 (Beam ) -1.556E+06
# 7 (Cable) -4.242E+05
# 8 (Cable) -4.288E+05 0.000 0.000
# 9 (Cable) -4.316E+05
#10 (C able) -4.338E+05
#11 (C able) -4.346E+05
#12 (C able) -4.334E+05
#13 (C able) -4.345E+05
#14 (C able) -4.347E+05 -0.400 -0.400

#15 (C able) -4.347E+05

#16 (C able) -4.346E+05
#17 (C able) -4.347E+05
SEZ. PA 2/4/5 SEZ. PA 2/4/5
0.2 00 0 .60 0 1.000 1.400 1.8 00 0.2 00 0 .60 0 1.000 1.400 1.8 00
(*10 ^1 ) (*10 ^1 )

Figure 6. Axial sollecitation for rockbolts. Figure 9. Flexural sollecitation in final lining.


FLAC (Version 3.40)

(*1 0^1)

In figure 10 are illustrated the uniaxial interac-

tion failure surface of a reinforced concrete section

22-Nov-99 11:56

and the design loads by Ultimate Limit Analysis for

step 45947
-1.000E+00 <x< 6.400E+01 several steps of calculus.
-3.000E+01 <y< 3.500E+01 1.500

Elastic 2000
At Yield in Shear or Vol.
Elastic, Yield in Past 0.500
Arco Rovescio - Breve termine
Boundary plot Arco Rovescio - Lungo termine
1500 Arco Rovescio - Decadimento
Rivestimento di calotta
Exaggerated Boundary Disp. -0.500 Rivestimento di calotta - Decadimento
Magnification = 2.000E+00
Max Disp = 2.916E-01
Beam plot
Cable plot
M [kNm/m]

SEZ. PA 2/4/5
0 .500 1 .50 0 2.5 00 3.500 4.500 5.500
(*10 ^1 ) -1000


Figure 7. Extension of plastic zones.

0 2000 4000 6000 8000 10000 12000
N [kN/m]

FLAC (Version 3.40)

Figure 10. Sectional verify.

23-Nov-99 14:05
step 56947
-1.000E+00 <x< 2.100E+01
-7.000E+00 <y< 1.500E+01 0.800
It is important to note:
Material model
mohr-coulom b
Beam plot
- the maximum absolute displacement cavity of
Cable plot
Axial Force on
the boundary is nearly 29.2 cm, after the installation
# 1 (Beam )
Max. Value
6.965E+06 of primary linings. Because the radial displacement
# 2 (Beam ) 6.067E+06

at the front is 12.5cm, the relative displacement is
about 16.7cm, i. e. the same engineering measure of
over-excavation (15cm);

- the extension of plasti zones is 20m over the

SEZ. PA 2/4/5
cavity boundary (the overburden is 250m);
0.2 00 0 .60 0 1.000 1.400 1.8 00
(*10 ^1 )
- the stress for the structures are ever inside the
interaction failure design surface.
Figure 8. Axial sollecitation in final lining. At the final step, to simulate the swelling and the
viscoplasticity of the rock mass it has been consider-
ated a reduction of 10% for GSI value, that it has in-
volved an increment of the state of sollicitation on
the structures of definitive structures.
4 CONCLUSIONS FLAC (Fast Lagrangian Analysis of Continua) version
3.4, 1988 . Users Manual. Itasca Consulting Group,
In deep tunneling design it is only the sinergy Panet, M. & Guenot, A. 1983. Analysis of Convergence
between: behind the face of a tunnel. Laboratoire Central des
- geolocical characterization; Ponts et Chausses, Paris, France.
- geomechanical characterization and extima-
tion of geomechanical parameters of rock
- analytical solutions and numerical models
of excavation process, including installation
of linings;
- knowledge of constructions to define the
succession of the excavation and strengthen-
ing phases;
that allows the progress of geotecnichal and/or
structural design. That means marriage (including
disputes) of various technical knowledge and ex-
periences to design a particular class of works.
This synthesis concurs to delineate a calculation
process, that must be accompanied by the following
- the behaviour of the rock mass is homogene-
ous, isotropic, following an elasto-plastic criteria
with flow rule not associated;
- plastic model by Hoek & Brown criteria.
Furtherly it is shown an algorithm to geome-
chanical/structural design of deep tunnel. It is evi-
denced the geomechanical parameters (RMR,
URMR, GSI, Hoek & Brown constitutive model),
analytical tools (Ground Reaction Curve and Panets
solution) and numerical tools (finite difference
analysis by FLAC). The structural sectional analysis
completes the calculus of the linings.
Finally it is illustrated an example about a deep
tunnel with interaction of particular geomechanical


Bieniawski, Z. T. 1993. Engineering classification of

jointed rock masses. In South Africa Institution of
Civil Engineers, Transactions,.
Bieniawski ,Z. T. 1974. Geomechanics classification of
rock masses and its application to tunneling. In Proc.
3rd Int. Congr. Rock Mechanics, ISRM, Denver.
Bieniawski ,Z. T. 1978. Determining rock mass deform-
ability experience from histories.In Int. J. Rock Me-
chanics, Mining Sciences & Geomechanics Abstracts.
Bieniawski ,Z. T. 1984. Rock mechanics design in mining
and tunnelling. Balkema, Rotterdam.
Bieniawski ,Z. T. 1989. Engineering rock mass classifi-
cations. J. Wiley & Sons.
Hoek, E. & Brown, E. T. 1982. Underground excavation
in rock. Institution of Mining and Metallurgy, Lon-
Hoek, E. & Brown, E. T. 1982.: Pratical estimates of
rock mass strength International Journal of Rock
Mechanics and Minig Sciences.