See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/265623079

Numerical modelling of ground-tunnel support interaction using bedded-

beam-spring model with fuzzy parameters

Article in Gospodarka Surowcami Mineralnymi - Mineral Resources Management · January 2011

CITATIONS READS

25 2,572

3 authors:

Fabrizio Barpi Monica Barbero

Politecnico di Torino Politecnico di Torino
65 PUBLICATIONS 740 CITATIONS 85 PUBLICATIONS 660 CITATIONS

SEE PROFILE SEE PROFILE

Daniele Peila
Politecnico di Torino
147 PUBLICATIONS 2,739 CITATIONS

SEE PROFILE

All content following this page was uploaded by Daniele Peila on 07 January 2015.

The user has requested enhancement of the downloaded file.

 GOSPODARKA SUROWCAMI MINERALNYMI

Tom 27 2011 Zeszyt 4

FABRIZIO BARPI*, MONICA BARBERO*, DANIELE PEILA**

Numerical modelling of ground-tunnel support interaction using

bedded-beam-spring model with fuzzy parameters

Introduction

In order to design a tunnel lining following the usually applied codes and methods, the
use of a computational tool to evaluate the underground and surface displacements, the
plasticized zones around the void and the forecasted stresses acting inside the lining is
required. On the other hand, the construction procedure could produce tunnel shapes
different from the designed ones. For this reason, designers must always carry out para-
metrical analyses to asses the feasibility of the design. The British Tunnelling Society (2004)
clearly states that the most important goal of a tunnel design is to provide an understanding of
the rock mass and lining behaviour during tunnelling, including the evaluation of risks. Risk
analysis is the essential way for producing a robust and safe design. Finally the design
process should provide the basis for interpreting the monitoring results during construction.
Due to the uncertainties concerning the properties of the rock mass and the loads induced in
the lining the precision of the available analytical and numerical methods is often greater than
the reliability and the accuracy of the data obtained from site investigations. Furthermore, the
imperfections that unavoidably affect the tunnel construction have a great influence on the
answer of the support system. Therefore, in order to understand the influence of input
parameters on the analysis results, sensitivity analyses with the ground support interaction
model have to be carried out.
The widely used lining design methods in tunneling practice are (AFTES 1976; USACE
1997; BTS 2004):

** Assistant Professor, Dept. of Structural and Geotechnical Engineering

** Professor. Dept. of Land, Environment and Geo-engineering, Politecnico di Torino
72

— empirical methods, usually based on rock mass classification;

— analytical solutions, which are usually developed using:
– continuum analytical models (Muir Wood 1975),
– convergence-confinement method (Hoek and Brown 1980; Brown et al. 1983;
Panet 1995),
– limit equilibrium methods, to evaluate the stability of rock blocks around a tunnel
and the stability of tunnel face (Hoek and Brown 1980; Goodman and Shi 1985),
– bedded-beam-spring models (USACE 1997; Oreste 2007; Barpi and Peila 2011)
where the tunnel lining is modelled as a series of beams connected to each other
and to the ground by radial and tangential springs that simulate the ground support
interaction;
— two and three-dimensional numerical analyses, which can be carried out using the
finite element, the finite difference methods or the distinct element method with the
ability to model complex geometrical, geological and geotechnical structures (BTS
2004).
The bedded-beam-spring model (also known as the hyperstatic reaction method) presents
some disadvantages, such as the difficult to correct evaluate the spring stiffness (that is the
object of the present investigation) and the loads to be applied to the lining, which has to
be evaluated independently from the deformation of the system using, e.g., dead load
approaches or rock mass classifications. Despite these drawbacks, this approach is widely
applied in design practice thanks to its ease of use, and the possibility of obtaining a quick
and simple evaluation of the actions inside the structural elements by varying the acting
loads, and thus carrying out sensitivity analyses, that is, as mentioned before, of primary
importance.
Tunnel designers should always take into account that every model could be affected
by many error sources that could lead to poor predictions (Kalamaras 1996; BTS 2004;
Peila 2009; Borio and Peila 2011) such as the theoretical shape of the tunnel, which can
be different from the reality due to the construction method. For this reason, the present
research analyses the tunnel boundary conditions after excavation and its influence on
a ground-support system and on the stresses inside the lining.
The study of the ground-tunnel interaction is performed by introducing a predetermined
degree of variation (fuzziness) into the input ground parameters. Given a certain degree
of uncertainty in the parameters, the fuzzy set theory makes it possible to evaluate the
uncertainty in the results, thereby avoiding the difficulties associated with stochastic ana-
lysis, since this method does not require any knowledge of the probability distribution
functions.
Here it is assumed that model parameters are affected by a certain degree of uncertainty
(defined by the so-called membership functions) and the computational results are calculated
by solving the fuzzy equations, thus quantitatively estimating the influence of a given change
in the model parameters.
 73

1. Fuzzy logic

Fuzzy logic was introduced by Zadeh in the 60’s in an attempt to formalize approximate
knowledge and approximate reasoning (Zadeh 1968; Zadeh 1988). It is a superset of
conventional Boolean logic extended to handle the concept of partial truth (i.e., truth values
between “completely true” and “completely false”). Fuzzy logic recognizes more than true
and false values: propositions can be represented with degrees of truthfulness and falsehood.
This logic can be employed to develop complex systems, systems that are controlled by
human experts, systems that use human observations as input and systems that are naturally
vague (behavioural or social sciences).
It must be noticed that there are many similarities between fuzzy theory and probability
theory; for example, both of them express uncertainty and have their values in the [0, 1]
range, A detailed description of this technique is beyond the scope of this paper: the reader is
referred to Dubois and Prade (1980) for a general treatment of this approach and to Schulz
and Huwe (1997) for an example of application.

2. A new computational model for tunnel lining design

A new computational model has been developed in order to couple the bedded-beam-
-spring model with a fuzzy set analysis tool to take into account the imprecision in the data
as outlined below.
This model is able to take into account a generic number of imprecise parameters and to
solve the correspondent fuzzy equations. It performs the proper calculations (solution of the
fuzzy equations) and output the membership function of the chosen parameters.

3. The bedded-beam-spring model

The bedded-beam spring model describes the tunnel lining as a series of beams connected
at their nodes to a series of radial and tangential springs that are designed to model the ground
reactions (Fig. 1).
The code used for the numerical simulations is based on the finite element concepts.
The tunnel lining is discretized by using beam elements (with three degree of freedom)
connected each other with nodes that are respectively connected to fixed point by normal and
tangential springs. This procedure allows to take into account the interaction between ground
and arch when the arch is affected by deformations induced by the applied loads (the ground
action). The calculation procedure is the displacement method (Bathe 1996; Zienkiewicz,
Taylor 2000), usually used in FEM models. The loads acting on the lining are evaluated by
using an empirical formulation taking into account the properties of the rock mass and the
geometry of the tunnel.
74

Fig. 1. Layout of the model, horizontal (qh) and vertical loads (qv) and node numbers (R = 5.75 m)

Rys. 1. Szkic modelu – obci¹¿enie poziome (qh) i pionowe (qv), zaznaczono numery wêz³ów
(promieñ R = 5,75 m)

The structural elements (i.e., beams) are usually modelled as linear elastic; their stiffness
is a function of the thickness and the elastic modulus of the constituting materials. Since the
tunnel first lining is made of shotcrete and steel ribs it is necessary to define the equivalent
tunnel cross section and a modulus of deformability that should take into due account the
different properties of shotcrete (continuous) and steel ribs (discontinuous). The procedure
used in this study is the one proposed by Hoek et al. (2008).
The stiffnesses of the springs Kn and Kt (Fig. 2) are usually evaluated from the rock mass
data (Table 1) using very simple relationships as those derived from Winkler theory (DAUB
2005; AFTES 1976; USACE 1997). For example, USACE (1997) suggests to use:

E rockmass 1 (1)
Kn = and Kt = Kn
R eq × (1+ n ) 3

while Oreste (1999) suggests:

E rockmass (2)
Kn = (1.33¸2)
2× R eq

where Req is the equivalent radius of the tunnel, E and n the elastic modulus and the Poisson
coefficient of the ground, respectively.
 75

Fig. 2. Constitutive relationship of the springs representing the ground reaction in the bedded-spring model
pn and pt represent the normal and tangential load, dn and dn the normal and tangential displacement and Kn
and Kt the normal and tangential stiffness

Rys. 2. Zale¿noœæ konstytutywna opisuj¹ca relacjê oœrodka w modelu pow³okowo-belkowo-spre¿ystym

pn i pt stanowi¹ normalne i styczne obci¹¿enia, dn i dn normalne i styczne przemieszczenia oraz Kn i Kt
normalna i styczna sztywnoϾ

TABLE 1
Stiffness Kn for different RMR classes
Three values of damage D are assumed: 0 (excellent quality controlled blasting),
0.3 (relatively shallow disturbance of the rock mass) and 0.7 (controlled blasting)
TABELA 1
Sztywnoœæ Kn dla ró¿nych klas
Trzy wartoœci wspó³czynnika stopnia zniszczenia D: 0 – prace strza³owe doskona³ej jakoœci,
0,3 – stosunkowo p³ytkie naruszenie masywu skalnego, 0,7 – kontrolowane prace strza³owe

Ed Req Kn Ed Req Kn Ed Req Kn

RMR
Rock mass class (D = 0.0) (D = 0.3) (D = 0.7)
(aver.)
GPa GPa GPa GPa GPa GPa

I 90 74.49 62.49 63.70 53.08 48.70 40.58

II 70 23.70 19.75 20.16 16.80 15.40 12.83

III 50 7.50 6.25 6.37 5.31 4.87 4.06

IV 30 2.37 1.98 2.01 1.68 1.54 1.28

Using Eqs. 1 and 2, the values of Kn for rock mass can be obtained from the RMR rock
mass classification of Bieniawski as reported as an example in Table 1. The rock mass
modulus of deformation is calculated using the formula suggested by Hoek et al. (2002) and
the average values of RMR for each class of rock mass quality are used.
76

æ Dö s (3)
E d = ç 1- ÷ ci 10(GSI -10 ) 40
[GPa]
è 2 ø 100

where D represents the degree of disturbance of the rock mass, sci the intact rock uniaxial
compressive strength and GSI the geological strength index (Hoek, Brown 1997).
Recently, Oreste (2007) proposed more refined formulations with an hyperbolic con-
stitutive relationship but, for the present work, a simpler formulation has been used due to
the difficulty of determining the parameters suggested by the author from the in-situ
investigation and, therefore, affected by the related uncertainties.
This design approach even if very simple is still widely used in tunnelling design
also because it permits an easy parametric calculation; it is suggested by DAUB (2005)
for the design of the shield in a mechanized tunnel.
In rock tunnels the lining is usually made of steel ribs and shotcrete and in the design
the steel ribs are usually considered in contact with the rock along its whole length.
This hypothesis is often far from the reality due to the irregularity of the rock boundary
induced by blasting and by the natural joints pattern distribution. Actually, the steel ribs are
in contact with the rock mass only in a discrete number of points (Fig. 3) and the empty
spaces between rock and ribs are filled with shotcrete that is more deformable than the
rock mass.

Fig. 3. Lining made of shocrete and steel ribs; ideal (on the left), real (on the right)

Rys. 3. Obudowa wykonana z torkretu i ³uków stalowych;

wyidealizowany (po lewej), realny (po prawej)

The bedded spring model has therefore to take into account the presence of this shotcrete
bed that has a crucial importance in the Kn and Kt evaluation. These parameters are
influenced by the stiffness and thickness of the shotcrete bed since the stiffer element of the
lining (the steel rib) interacts with the ground through the shotcrete bed that is thicker with
the tunnel shape is more irregular.
The model used in the paper takes into account this irregularity using a fuzzy logic (Barpi,
Peila 2011): the more irregular is expected to be the tunnel boundary after drill and blast
excavation the largest will be the membership functions of Kn and Kt, thus the imprecision of
the used values is increased.
 77

4. Application of fuzzy set analysis to ground-tunnel interaction

The influence of the uncertainty or imprecision affecting three parameters of the pre-
sented numerical model is assessed through a fuzzy approach. The parameters Kn and Kt
(spring stiffness) and Es (elastic modulus of the lining, due to the possible irregular in-
stallation of the shotcrete) are considered to be fuzzy, i.e., associated with a certain degree of
uncertainty, as defined through the appropriate membership functions m (triangular in this
study). Such functions represent the level of confidence (or imprecision) of a variable;
a typical shape could be triangular, trapezoidal, Gaussian or bell-shaped. Compared to
stochastic analysis, this approach is a simpler and more intuitive way to allow for the
uncertainty in some parameters, since it does not call for the determination of probability
distribution functions. Given a value of the membership function it is possible to obtain
a closed interval within which lies the variable of interest. The projections of a-cut (cut at
level a) on the axis of the variable define the left and right boundaries of this interval.
For example, a-cut = 0.5, as in Figure 5, defines intervals from 6.4 · 105 kN/m3 to
9.6 · 105 kN/m3 for Kn; from 2.08 · 105 kN/m3 to 3.12 · 105 kN/m3 for Kt and from
1.14 · 107 kN/m2 to 1.71 · 107 kN/m2 for Es.
As a result, the variables of interest (output of the model), for instance the internal forces
(bending moment M(q), normal force N(q) and shear force T(q)) are expressed by fuzzy
numbers (q is represented in Fig. 1). The same applies to any other output of the model, i.e.,
stresses in the boundary elements or in the structural elements. It means that M(q), N(q)
and T(q) are associated with a certain membership function to be determined by finding,
for each level 0 £ a £ 1:

min max M ( q; K na ; K ta ; E sa ;... ) a Î [ 0,1] (4)

min max N ( q; K na ; K ta ; E sa ;... ) a Î [ 0,1]

min maxT ( q; K na ; K ta ; E sa ;... ) a Î [ 0,1]

with:

Kn £Kn £Kn (5)

Kt £Kt £Kt

Es £ Es £ Es

where the underline indicates the left boundary of a variable and the over-line the right
boundary of the same variable, corresponding to a certain (given) value of a level.
78

The solution is found by minimizing and maximizing the fuzzy variables (subjected to the
constraints in Eqs. 5) and calculating displacements, stresses, internal forces (or any other
parameters of interest) for each level of a. The range of variation of the fuzzy variables
represents their level of imprecision due to the imprecision in the input data.
The determination of the membership functions of internal forces, i.e., their level of
imprecision, is achieved by repeating the calculation for different values of a. This problem
can be viewed as an optimization problem, as explained in Dubois and Prade (1980).

6. Numerical simulations

In order to explain the proposed methodology, a case of a road tunnel excavated by drill
and blast in the North of Italy was analyzed. The rock mass is constituted of dolomitic
limestone (Fig. 4) with a RMR varying between classes II and III of Bieniawski (1989).
No water was observed in the tunnel. The rock mass modulus of deformation was calculated
with the Hoek-Brown formula (Hoek et al. 2002), thus the average values of Kn and Kt,
to be used for the calculations were obtained, considering a thickness of the shotcrete
bed equal to zero.
The applied loads qv and qh were evaluated using the Terzaghi formulation with a drained
cohesion of 0.13 MPa and a drained friction angle of 23° for the rock mass. The obtained
loads are qv = 0.37 MPa and qh = 0.5 · qv = 0.185 MPa.
The thickness of the lining was determined by homogenising (Hoek et al. 2008)
the tunnel lining made of shotcrete (0.22 m thick) and steel ribs (two IPN180 ribs –
standard European I beams, DIN 1025 – are installed side by side with a distance of 1 m,
see Table 2).
The bedded-spring elements stiffnesses Kn and Kt (Fig. 1) were chosen to be equal
to Kn = 8 · 105 Pa/m and Kt = 2.6 · 105 Pa/m while the elastic modulus of the lining
Es = 1.42 · 107 N/m2. To avoid introducing any further uncertainties, the springs were
designed to remain in the elastic range by raising the plasticization level.

TABLE 2
Material properties for shotcrete (0.22 m thick) and steel (two IPN180 ribs placed side by side)
used as a first phase lining for the studied tunnel
TABELA 2
W³aœciwoœci torkretu (gruboœæ 0,22 m) i stali (dwa dwuteowniki IPN180 umieszczone obok siebie),
stosowanych jako pierwszy etap podparcia dla badanego tunelu

Elastic modulus Poisson ratio Tensile strength Compressive strength

Material
GPa – MPa MPa

Shotcrete 10 0.15 0.5 10

Steel 210 0.25 355.0 355

 Fig. 4: Simplified geological longitudinal section of the studied tunnel

Rys. 4. Uproszczony geologiczny przekrój pod³u¿ny analizowanego tlenku

79
80

As mentioned above, parameters Kn, Kt and Es are considered to be fuzzy and, hence,
associated with given membership functions.
To assess the influence of their shape, two different cases are examined. Each of them
is described by the membership functions with symmetrical triangular shape with respect
to the reference values showed in Figs. 5 and 6.

Fig. 5. Membership functions for Kn, Kt and Es for case 1 (left) and case 2 (right)

Rys. 5. Funkcja przynale¿noœci dla Kn, Kt i Es dla przypadku 1 (po lewej) i przypadku 2 (po prawej)
 81

Fig. 6. Membership functions for bending moment, normal force and shear force (element 48) for case 1

Rys. 6. Funkcja przynale¿noœci dla momentu zginaj¹cego,

si³y normalnej i si³y œcinaj¹cej w elemencie 48 dla przypadku 1

7. Discussion of the results

The results of the analyses in terms of M, N and T on element 48 (located at the base of
the arch, i.e., for q = 0 in Fig. 1) are shown in Figs. 6, 8 and 9 (case 1) and Figs. 7, 10
and 11 (case 2). Each diagram shows the values obtained for different a-cuts and allows to
investigate the effects of parameters imprecision on the computed values and to identify the
parameters with the most adverse influence on the response. For instance, in the element 48,
for a-cut = 0, the bending moment ranges between 4.732 Nm/m and 3.308 Nm/m. With
respect to the value 4.078 Nm/m that refers to a-cut = 1, the values of the bending moment
that can be present in the lining are 18.9% smaller and 16.0% larger. Therefore, this range of
variation should be considered in the interaction diagram.
It should be noted that, despite a linear membership functions of the input data is adopted
(Fig. 5) the membership functions of the results (internal forces in the lining) are not linear.
Furthermore the imprecision in the data results in a much higher imprecision in the values of
T than in N or M, as shown in Fig. 6 (case 1). Similar consideration can be drawn for results
82

of case 2 (Fig. 7), where the membership functions are “narrow” comparing with to those
of case 1.
Compared to a sensitivity analysis, the presented method allows to take into account an
entire set of imprecise parameters and to determine the level of imprecision in the results.
A different level of imprecision in the data can be obtained by changing the membership
functions.
The results presented here show the relevance of taking into account properly the
imprecision in model parameters and to quantitatively evaluate its influences on the results.
The interaction diagrams for shotcrete and steel are shown in Figs. 9, 10 and 11 according
to the procedure outlined in Hoek et al. (2008) and Carranza-Torres and Diederichs (2009)
(the material properties are listed in Table 2).
The diagrams for steel present two limit domains for the material strength: the larger
curve is obtained using the maximum strength criterion i.e. the maximum principal stresses
acting in the considered cross section (Hoek et al. 2008) while the smaller one is based on the
well known Tresca criterion, i.e the material is considered yielded when the acting tangential
stress reaches a pre-defined limit value.

Fig. 7. Membership functions for bending moment, normal force and shear force (element 48) for case 2

Rys. 7. Funkcja przynale¿noœci dla momentu zginaj¹cego, si³y normalnej i si³y œcinaj¹cej
w elemencie 48 dla przypadku 2
 83

By using a safety factor of 1.5, some points for shotcrete fall outside the diagrams, which
means a bending failure (Fig. 8). Similar considerations apply to Figs. 9, 10 and 11 based on
the membership functions represented in Fig. 5. It must be pointed out that the data
represented in Fig. 8, 9, 10 and 11 are related to the values of a = 0 and 1 (the minimum
and the maximum a in the triangular used membership function of input parameters). The
computed results (bending moments, normal force and shear force) are therefore represented
by two different values: the minimum and the maximum that correspond to the chosen a-cut.
Since the chosen membership function of input parameters is of triangular shape, for a = 0

Fig. 8. Bending moment vs. normal force for shotcrete (left) and steel ribs (right) (case 1)

Rys. 8. Zale¿noœæ miêdzy momentem zginaj¹cym a si³¹ œcinaj¹c¹ dla torkretu (z lewej) i ³uków stalowych (z prawej)
dla przypadku 1

Fig. 9. Shear force vs. normal force for shotcrete (left) and steel ribs (right) (case 1).
In the right figure the dotted ellipse represent the Tresca criterion while the solid curve represent the maximum
strength criterion

Rys. 9. Zale¿noœæ miedzy si³¹ œcinaj¹c¹ a si³¹ normaln¹ dla torkretu (z lewej) i stalowych ³uków (z prawej)
w przypadku 1. Na rysunku z prawej strony lini¹ przerywan¹ zaznaczono kryterium Tresca,
a lini¹ ci¹g³¹ kryterium maksymalnej wytrzyma³oœci
84

Fig. 10. Bending moment vs. normal force for shotcrete (left) and steel ribs (right) (case 2)

Rys. 10. Zale¿noœæ miedzy momentem zginaj¹cym a si³¹ normaln¹ dla torkretu (z lewej) i stalowych ³uków
(z prawej) w przypadku 2

Fig. 11. Shear force vs. normal force for shotcrete (left) and steel ribs (right) (case 2).
In the right figure the dotted ellipse represent the Tresca criterion while the solid curve represent the maximum
strength criterion

Rys. 11. Zale¿noœæ miêdzy si³¹ œcinaj¹c¹ a si³¹ normaln¹ dla torkretu (z lewej) i stalowych ³uków (z prawej)
w przypadku 2. Na rysunku z prawej strony lini¹ przerywan¹ zaznaczono kryterium Tresca,
a lini¹ ci¹g³¹ kryterium maksymalnej wytrzyma³oœci

there are the two values of the computed results while for a = 1 (the reference values)
the two values are coincident.
The results clearly show that if the average thickness is not considered, a wrong design
could be obtained since the designed lining could be not adequate to support the applied
loads, and cracks and damages could occur.
Based on this example, it is clear the contribution of the proposed methodology that
permits an easy calculation of the effects of the imprecision and to find the most adverse
combination of parameter.
 85

Conclusions

A new method for tunnel lining design is presented in the paper. It is based on a fuzzy
approach, and applied to a well-known tunnel lining design method to take easily into
account a degree of imprecision in model parameters. Furthermore the larger is the im-
precision of the geometrical data (due to excavation procedure that cannot be completely
forecasted at the design stage) the larger is the effect on the computed forces acting in the
lining. The influence of this imprecision can lead to an underestimation of the structure
strength that leads to cracks and damages. The considered imprecise parameters are
the stiffness of the bed spring due to the irregular shape of the tunnel boundary induced
by blasting and the deformation module of the beam that discretize the lining (Kn, Kt
and Es).
The presented example permits to clarify the influence of the imprecision of input
parameters on the results. Given this set of imprecise parameters (different membership
functions), the influence of the imprecision affecting model results (M, N and T) is quan-
titatively evaluated.
It was clearly demonstrated that if the imprecision vary from +20% to + 40% from
the average chosen value (the one that should be used in a deterministic analysis) the
obtained results in term of acting forces go outside the yield criterion (i.e. the lining is
fractured). Since the membership functions can be determined by using an expert judgment,
the contribution of an expert in the field is easily included in the model and a complex
statistical analysis can be avoided. Furthermore the set of parameters with the greatest
influence on the answer of the system can be identified and properly taken into account at the
design stage.

REFERENCES

Association Française des Tunnels et de l’Espace Souterrain, 1976 – Consideration on the usual method of tunnel
lining design, Tunnels & Ouvrages Souterrains, vol. 14, mars/avril.
B a r p i F., P e i l a D., 2011 – Influence of the tunnel shape on shotcrete linings stresses after excavation.
Computer-Aided Civil and Infrastructure Engineering, on line version,
DOI: 10.1111/j.1467-8667.2011.00728.x.
B a t h e K.J., 1996 – Finite Element Procedures. Prentice Hall.
B i e n i a w s k i Z., 1989 – Engineering rock mass classifications: a complete manual for engineers and geologists
in mining, civil, and petroleum engineering. Wiley-interscience publication, Wiley.
B o r i o L., P e i l a D., 2011 – Laboratory test for EPB tunnelling assessment: results of test campaign on two
different granular soils. Gospodarka Surowcami Mineralnymi, Vol. 27, No. 1, pp. 85–99.
B r o w n E., B r a y J., L a d a n y i B., H o e k E., 1983 – Ground response curve for rock tunnels. Journal of
Geotechnical Engineering, ASCE, Vol. 109, pp. 15–39.
British Tunnelling Society (BTS), 2004 – Tunnel Lining Design Guide. Thomas Telford (London).
C a r r a n z a -T o r r e s C., D i e d e r i c h s M., 2009 – Mechanical analysis of circular liners with particular
reference to composite supports. For example, liners consisting of shotcrete and steel sets. Tunnelling and
Underground Space Technology, Vol. 24, Elsevier, London, pp. 506–532.
86

DAUB (German Commitee for Underground Construction), 2005 – Recommendation for static analysis of shield
tunnelling machines. Tunnel, vol. 7, pp. 44–58.
D u b o i s D., P r a d e H., 1980 – Fuzzy Sets and Systems: Theory and Application. Academic Press, San
Diego.
G o o d m a n R., S h i G., 1985 – Block Theory and its Application to Rock Engineering. Prentice-Hall, New
York.
H o e k E., B r o w n E., 1997 – Practical estimates of rock mass strength. International Journal of Rock Mechanics
and Mining Science, vol. 34, No. 8, pp. 1165–1186.
H o e k E., B r o w n E., 1980 – Underground Excavation in Rock. The Institution of Mining and Metallurgy,
London.
H o e k E., C a r r a n z a -T o r r e s C., C o r k u m B., 2002 – Hoek-Brown failure criterion-2002 edition. 5th North
American Rock Mechanics Symposium, Toronto (Canada), pp. 267–273.
H o e k E., C a r r a n z a -T o r r e s C., D i e d e r i c h s M., C o r k u m B., 2008 – The 2008 Kersten Lecture.
Integration of geotechnical and structural design in tunnelling. 56th Annual Geotechnical Engineering
Conference, Minneapolis (USA), pp. 1–53.
K a l a m a r a s G., 1996 – A probabilistic approach to rock engineering design: application to tunnelling. Mile-
stones in Rock Engineering: Bieniawski Jubilee Collection, Balkema, Rotterdam, pp. 113–135.
M u i r W o o d A., 1975 – The circular tunnel in elastic ground. Geotechnique, vol. 25, No. 1, Thomas Telford,
London, pp. 115–127.
O r e s t e P., 1999 – Aspetti notevoli dell’analisi e dimensionamento dei sostegni di gallerie attraverso i metodi di
calcolo numerici. Geoingegneria Ambientale e Mineraria, Vol. 57, Patron Editore, Bologna, pp. 39–50.
O r e s t e P., 2007 – A numerical approach to the hyperstatic reaction method for the dimensioning of
tunnel supports. Tunnelling and Underground Space Technology Vol. 22, No. 2, Elsevier, London,
pp. 185–205.
P a n e t M., 1995 – Le Calcul des Tunnels par la Méthode Convergence-Confinement. Presses de l’École
Nationale des Ponts et Chaussées, Paris (France).
P e i l a D., 2009 – Indagini preliminari nella costruzione di gallerie: analisi della letteratura tecnica. Geoingegneria
Ambientale e Mineraria, Vol. XLVI, No. 3, Patron Editore, Bologna, pp. 37–59.
S c h u l z K., H u w e B., 1997 – Water flow modelling in the unsaturated zone with imprecise parameters using
a fuzzy approach’. Journal of Hydrology, Vol. 201, pp. 211–229.
US Army Corps of Engineering (USACE), 1997 – Tunnels and shafts in rock. Engineering Manual 1110-2-2901,
Washington (USA).
Z a d e h L., 1968 – Fuzzy algorithms. Information and Control, Vol. 12, pp. 94–102.
Z a d e h L., 1988 – Fuzzy logic. Institute of Electrical and Electronic Engineers Computer, Vol. 1, No. 14,
pp. 83–93.
Z i e n k i e w i c z O.C., T a y l o r R.L., 2000 – Finite Element Method: Volume 1, 2, 3, Butterworth-Heinemann.

MODELOWANIE NUMERYCZNE WSPÓ£PRACY OBUDOWY TUNELU Z OŒRODKIEM Z WYKORZYSTANIEM MODELU

POW£OKOWO-BELKOWO-SPRʯYNOWEGO ORAZ METODY ZBIORÓW ROZMYTYCH

S³owa kluczowe

Obudowa modelu, model numeryczny, zbiory rozmyte

Streszczenie

Przedstawiono i przeprowadzono dyskusjê wyników badañ wspó³pracy obudowy tunelu z oœrodkiem z wy-
korzystaniem okreœlonego stopnia zmiennoœci niektórych parametrów wybranego modelu. Badania te zwi¹zane
s¹ z rozwa¿aniami, ¿e parametry i geometria modelu tunelu s¹ obarczone pewnym stopniem niepewnoœci,
wynikaj¹cym z niedok³adnoœci konstrukcji i du¿¹ zmiennoœci¹ parametrów masywu skalnego. Badania zosta³y
 87

przeprowadzone z u¿yciem teorii zbiorów rozmytych zak³adaj¹c, ¿e trzy parametry modelu s¹ obarczone pewn¹
niepewnoœci¹ (zdeterminowan¹ przez funkcjê przynale¿noœci). Obliczenia numeryczne wykonano rozwi¹zuj¹c
równanie funkcji losowych dla zró¿nicowanych funkcji przynale¿noœci. W celu sprawdzenia wp³ywu modelu
parametrów i opracowania uproszczonej procedury i narzêdzi dla projektantów, przeprowadzono badania oparte
na zbiorach losowych z u¿yciem znacznej i szeroko stosowanej metody wykorzystuj¹cej model pow³okowo-
-belkowo-sprê¿ynowy.

NUMERICAL MODELLING OF GROUND-TUNNEL INTERACTION USING BEDDED-BEAM-SPRING MODEL

WITH FUZZY PARAMETERS

Key words

Tunnel lining, numerical model, fuzzy logic

Abstract

The study of the ground-tunnel interaction by introducing a predetermined degree of variation (fuzziness)
in some parameters of the chosen model is presented and discussed. This research comes from the consideration
that tunnel model parameters and geometry are usually affected by a degree of uncertainty, mainly due to
construction imprecision and the great variability of rock mass properties. The research has been developed by
using the fuzzy set theory assuming that three model parameters are affected by a certain amount of uncertainty
(defined by the so-called membership functions). The response of the numerical model is calculated by solving
the fuzzy equations for different shapes of the membership functions. In order to investigate the effects of some
model parameters, and to provide a simple procedure and tool for the designers, a study on the effect of tunnel
boundary conditions, based on a fuzzy model, has been carried out using a simple but well known and widely used
design method such as the bedded-beam-spring model.
View publication stats