3º Congresso Brasileiro de Túneis e Estruturas Subterrâneas

Seminário Internacional “South American Tunnelling – SAT 2012”

COMPARISON BETWEEN DIFFERENT DESIGN APPROACHES FOR STEEL PIPE

UMBRELLAS AT TUNNEL PORTALS.

SILVA, J. C. B. J. da (1); PEILA, D. (2), ASSIS, A. P. de (3),

(1) Universidade Federal da Bahia, Universidade Estadual de Feira de Santana – jcarlos@ufba.br

(2) Politecnico di Torino, Italy - daniele.peila@polito.it
(3) Universidade de Brasília – aassis@unb.br

Abstract - Steel pipe umbrellas have been extensively used for tunnelling under difficult conditions or at tunnel portals
since the early seventies. Despite the large number of applications throughout the world, several different design
approaches are usually used since different assumptions on how the umbrella pipes behave under loading have been
proposed. A comparison of the usually applied design methods: analytical approaches such as a simple beam with two
supports or a beam on multiple supports and two or three dimensional numerical approaches where the umbrella pipes
are modeled as a reinforced rock arch or individually; are presented and the results concerning a low overburden tunnel
are discussed. The feasibility of the methods to forecast the displacements and the stresses acting on the pipes has been
dealt with in particular.

Keywords: Steel pipe umbrella, numerical modelling, analytical design, tunnel stability.
INTRODUCTION

When adverse geotechnical conditions are encountered in tunnelling and the free span and self supporting
times are short, the technological possibilities for the designer are: to reduce the size of the excavation sections, and
increases the number of working attacks; to improve the rock mass quality or reinforce it or to pre-support the
excavation to apply a pressure to the tunnel face. Among the different ways of pre-supporting a tunnel ahead of the
tunnel face, steel pipe umbrellas have been widely used. The umbrella method, which consists of a closely spaced,
usually grouted, canopy of steel tubes, installed at the tunnel extrados, is effective in controlling deformations and
volume losses for a wide range of ground conditions as it improves face stability and increases the stand-up time. Steel
pipes are usually installed with a 5°-10° dip (with reference to the horizontal) in such a way as to form an umbrella of
pipes, with a size that ranges between 80mm and 220mm. The umbrella has a truncated cone shape which allows two
adjacent umbrellas be overlapped thus covering advancement lengths of 12-15m of which 9-12m are of excavation
(Figure 1). The diffusion of this method has been facilitated by the technological improvements on the installation
machines.

steel pipes

Figure 1 - Scheme of a steel pipe umbrella intervention for an excavation head and bench with face reinforcement and
underpinning of the steel stets (courtesy Geodata SpA, Turin) (left) and photo of an application in a granular soil (Turin
metro excavation) (right).

This technique has been used for: the construction of shallow tunnels in soft ground where a good control of
the ground displacements is necessary to reduce surface subsidence, the construction of tunnel portals because of low
overburdens; the construction of tunnels through weak ground with high overburdens or through fault zones, and the
crossing zones where the tunnel has already collapsed (Barisone et al., 1982; Pelizza and Peila, 1993; Shirakawa et al.,
1999; Carrieri et al., 2004; Volkmann, 2007). Despite the large number of applications there are still no generally
accepted methods or reliable means for designing a steel pipe umbrella since among the other factors it is difficult to
take into account: the high number of the involved geotechnical parameters; the three dimensional shape of the tunnel
near the face; the effect of the overlapping of the umbrella pipes and their connection to the steel sets, the stiffness of
the steel sets in comparison to the vertical loads and the geotechnical characteristics of the ground at the excavation face
and finally the influence of the position of the face during the excavation process. Some researches have supposed that
the pipes act by forming shell around the tunnel boundary that reduces the stresses acting on the rock core ahead of the
advancing tunnel face and that a two-dimension numerical model can also “model the physical behaviour of the
reinforced tunnel in a realistic manner” (Hoek, 2001), other authors believe that there is no significant mutual
interaction between the single pipes, thus each must be individually designed taking into account their longitudinal
direction (Max and Mattle, 2002; Oreste and Peila, 1998; Peila and Pelizza, 2003; Pelizza and Peila, 1993), while
others authors have used three-dimension numerical models to take into account the three dimensional shape of the
tunnel face and the presence of the pipes and face reinforcement (Max and Mattle, 2002 ; Peila, 1994; Uhtsu et al.,
1995; Oreste et al., 1999; Eclaircy-Caudron et al., 2005) and finally some authors have carried out physical modelling
to understand pipe behaviour more clearly (Kim et al., 2004; Takechi et al., 2000; Yoo et al., 2001; Shin et al., 2008) or
measurements in real tunnels (Volkmann et al., 2007; Ocak, 2008).
In order to provide a better understanding of the complex interaction between the pipes and the ground and of the
feasibility of the various design models, a comparison has been carried out of the results that can be obtained applying both
analytical methods: beams supported at both or and multi-supported beams with flexible supports (no mutual interaction
between the pipes is considered) and two and three dimensional numerical models (modelling the umbrella arch as a
reinforced shell around the boundary of the tunnel or as a system of individual elements).

NUMERICAL MODELS

The studied geometry

The studied tunnel geometry is based on a standard road tunnel size (Figure 1) full face excavated with
advancement steps of 1m at a low depth in drained conditions. The geometrical and geotechnical data of the various
models are summarized in Tables 1 and 2. The geotechnical parameters were chosen to represent a silty-sand with
different drained cohesion values and the Mohr-Coulomb yield criteria has been adopted. The minimum value of
drained cohesion was chosen to have a stable face for low overburden (Anagnostou and Serafeimidis, 2007), in order to
avoid influencing the results concerning the pipe stresses and deformations due to face instability.
The ratio between the horizontal and vertical undisturbed stresses, was set equal to 0.5 in all the models for simplicity
reason. When the improved material is used, its geotechnical properties of the material are evaluated by considering the
weighted average of the strength and deformation properties of the steel pipes and the original soil, as proposed by Hoek
(2001).

B
Reference pipe

H Pipe umbrella

Support
A
Tunnel axis

R
D

D1

Figure 1 - Geometric scheme of the tunnel

Table 1- Geometrical data of the tunnel

Model D R D1 H Pipe Pipe Pipe Pipe

number interax diameter thickness
[m] [m] [m] [m] [#] [m] [mm] [mm]
1 8.8 5.3 3.5 8.8 (1 D) 60 0.3 127 10
2 8.8 5.3 3.5 35 (4 D) 60 0.3 127 10
 Table 2 - Geotechnical characteristics of the natural (type a – b) and improved soil (type i).
Soil type Density Deformation Poisson Friction Drained Thickness
modulus modulus angle cohesion
[ kN/m3 ] [ MPa ] [-] [o] [ kPa ] [mm]
a 17 100 0,3 35 10 -
b 17 100 0,3 35 50 -
i 17 46000 0,3 45 200 350

The numerical and analytical models

The numerical models were set up by developing the calculations using: Flac 3D (ver. 3.10) where the pipes were
singularly modelled using 17 pile structural elements (two-nodded, straight, finite elements with six degrees of freedom per
node that are rigidly connected to the grid so that forces and bending moments develop within the pile as the grid deforms);
Plaxis 3D Tunnel and Plaxis 2D (ver. 8.1) where the pipes were modelled as an equivalent material (Figure 3, 4, 5).

L L1 Nodes Elements
Model
[m] [m] [#] [#]
1 120 60 6023 729
2 120 60 7631 933
L1
Face
Lateral Upper Bottom
Contraint
y x Fixed Free Fixed
y Free Free Fixed
x
L
Figure 3 - The two-dimensional Plaxis model mesh (n. 1) and general data of the two models.

L L1 L2 Nodes Elements
Model
[m] [m] [m] [#] [#]
1 120 70 100 15468 5175
2 120 70 100 16889 5832

L1
Face
Front Back Lateral Upper Bottom
Contraint
x Fixed Fixed Fixed Free Fixed
y Free Free Free Free Fixed

L yL2 z Fixed Fixed Fixed Free Fixed

z
x
Figure 4 - The three-dimensional Plaxis model mesh (n. 1) and general data of the two models.

L L1 L2 Nodes Elements
Model
[m] [m] [m] [#] [#]
1 51 31 30 38168 34986
2 51 55 30 78104 72522
L1
Face
Front Back Lateral Upper Bottom
Contraint
x Fixed Fixed Fixed Free Fixed
y Free Free Free Free Fixed
z Fixed Fixed Fixed Free Fixed
y
z
L2
L x

Figure 5 - The three-dimensional Flac model mesh (n. 1) and characteristic data of the two models.
 The three-dimensional computation is carried out by installing the steel pipes (pile elements or improved arches)
followed by a sequence of full face 1m long excavations with the installation of a very rigid lining (to minimize its influence
on the pipe behaviour ahead of the tunnel face) in the previously excavated advancement step. The excavation and lining
installation steps are repeated till an excavation of 9m is obtained, while in the numerical two dimensional computations are
carried out using the stress relief procedure and after the installation of the improved arch the support is installed at different
percentages of the initial forces ((λ=30%, 40% and 50%).
The analytical computations are carried out considering the total weight of the overburden for model 1 and the
Terzaghi load for model 2 while the spring stiffness (K) are those usually proposed for foundation design on sand (Bowels,
1982) (Figure 6).

Steel pipe

1m 0.25m deformable support (spring)

Fixed support
Model K [MN/m3]
9m I 1
II 5
III 10
Figure 6 - Geometry of the multi-supported beam and used properties of the support springs.

The obtained results

The surface displacement (point B – Figure 2), tunnel roof displacement (point A – Figure 2) and bending moments on
the reference pipe of the umbrella are compared (Figure 7). The bending moments of the beams are obtained from the
analytical approaches and they are compared with those obtained with Flac 3D. Table 3 shows the values of the maximum
bending moment considering all the adopted methods while Figures 8 and 9 report the bending moment diagrams in the pipe.

Table 3 - Maximum bending moment for the analysis methods.

Model Soil Maximum bending moment [KN.m]

Flac 3D Model I Model II Model III Beam 2
support
1 a 25,8 25,7 10,2 6,9 10,2
1 b 4,3 25,7 10,2 6,9 10,2
face 35,0 13,9 9,4 12,4
2 a collapsed
2 b 20,7 25,9 10,3 7,0 10,4

Model 1 soil a
Distance from the portal (m) Distance from the portal (m)
0 5 10 15 20 0 5 10 15 20
Displacement (mm)

0 0
Displacement (mm)

-2
-5
-4
-10 -6

-15 -8
-10
-20 -12

.
point A point B

Model 2 soil a
Distance from the portal (m) Distance from the potal (m)
0 5 10 15 20 0 5 10 15 20
Displacement (mm)

0
Displacement (mm)

0
-5
-10 -2
-15
-20 -4
-25
-30 -6
-35 -8
-40
-45 -10
-50
-55 -12

point A point B
 Model 1 soil b
Distance from the potal (m) Distance from the portal (m)
0 5 10 15 20 0 5 10 15 20
1
Displacement (mm)

Displacement (mm)
2
0 0
-2
-1
-4
-2
-6
-8 -3
-10 -4

point A point B

Model 2 soil b
Distance from the portal (m) Distance from the portal (m)
0 5 10 15 20 0 5 10 15 20
0 0
Displacement (mm)

Displacement (mm)
-10 -1
-2
-20
-3
-30
-4
-40 -5
-50 -6

point A point B

Figure 7 - Geometry of the multi-supported beam and the properties of the support springs. Key. Plaxis 3D: dotted
line; Flac 3D: bold line; Plaxis 2D λ=30% : ■; λ=40%: ▲; λ=50%:●

30
FLAC 3D C=50kPa
FLAC 3D C=10kPa
FLAC 3D C=50 kPa
20 model I
model II
model III
Bending moment (KN.m)

10 beam with two supports

-10 lining position

FLAC 3D C=10 kPa

beam supported at both ends
-20 tunnel face

-30
7 8 9 10 11 12 13 14 15
Distance from the portal (m)

Figure 8 - Bending moment. Model 1 - soil a and b.

30
FLAC 3D C=10 kPa
model I
20 model II
model III
beam with two supports
Bending moment (KN.m)

10

-10
lining position

beam supported at both ends

FLAC 3D C=10 kPa
-20
tunnel face

-30
7 8 9 10 11 12 13 14 15
Distance from the portal (m)

Figure 9 - Bending moment. Model 2 - soil b.

 From an analysis of the surface and the tunnel boundary displacements it emerges that the 3D codes give similar
values when the homogenizations scheme and the individually modelled single pipe approaches are used. The plastic zones
ahead of the tunnel face also have a similar extent (Figure 10) and show the influence of the presence of the pipe since without
the pipes the plastic zone reaches the surface. The 2D results with the homogenization approach are influenced to a great
extent by the value of λ and it is necessary to vary this relaxation factor when the surface displacement or the tunnel
boundary displacement have to be evaluated. Therefore the 2D computation does not appear to be a suitable design method.
The homogenization approach does not permit the bending moment in the pipes along the tunnel to be evaluated.
On the other hand, the multi supported beam method appears to be a valuable tool for the evaluation of the bending
moments in the pipe, since the results show the same trend as those computed using Flac 3D which shows that the maximum
value appears ahead of the face. The value of the maximum bending moment is closely connected to the value of the cohesion
when the Flac 3D code is used and to the spring stiffness when the beam on the multiple supports is used, but no literature
data link these two parameters. For a cohesion of about 50 kPa, when the Flac 3D code is used , the values of the maximum
bending moment are close to those computed with the spring stiffness equal to 10 MN/m3. The save can be seen for a
cohesion of about 10 kPa and K=1 MN/m3. When using a beam supported at both ends, the most important parameter is the
free span, which is directly linked to the value of the bending moment which is close to the maximum valore computed with
the beam on multiple support.

Model 2 – soil a Model 2 – soil b

Figure 10 - Plastic zones computed with the FLAC 3D code for model 2.

CONCLUSIONS

Despite the large number of steel pipe umbrella applications to date, there are no generally accepted methods for their
design. Many different approaches have been used in technical literature. This problem is more relevant near the tunnel portal
where the three dimensioned effect is particularly critical and when a low depth tunnel is considered. A comparison of various
computational methods has shown that the homogenization approach, when 2D numerical models are used, is not adequate
since the influence of the relaxation coefficient used for modelling the three-dimensional effect of tunnel face advancement,
has a great influence on the results and if the surface and tunnel displacements are considered it is necessary to used different
values. Furthermore, this analysis does not permit either the bending moment in the pipes and its safety factor to be evaluated
as requested by the used design codes such as, for example, Eurocode 7 “Geotechnical design” (EN 1997-1, 1997). From the
analysis of the displacements of the surface and the tunnel boundary, it appears that 3D codes give similar values both when
the homogenization scheme and the individually modelled single pipe approaches are used. The plastic zones ahead of the
tunnel face also have similar shape.
The multi supported beam method appears to be valuable tool for the evaluation of the bending moments in the pipe
since the values are similar to those computed with the Flac 3D. In conclusion, for the design of steel pipe umbrella it is
necessary to combine a 3D numerical computation (if the surface displacement is a relevant parameter) and a beam on
multiple support computation to calculate the bending moment thus obtaining the safety factors of the intervention.

REFERENCE
ANAGNOSTOU G.; SERAFEIMIDIS K (2007). The dimentioning of tunnel face reinforcement, Underground Space–
The 4th Dimension of Metropolises, Balkema, Rotterdam, Prague 2007, 289-296.
BARISONE G.; PELIZZA S.; PIGORINI B. (1982). Umbrella arch method for tunnelling in difficult conditions -
Analysis of Italian cases, IV International Congress of I.A.E.G.; New Delhi, 1982, 15-27.
BOWELS J.; Foundation. Analysis and design, McGraw-Hill, Singapore, 1982.
CARRIERI G.; GRASSO P.; FIOROTTO R.; PELIZZA S. (2004). Venti anni di esperienza nell'uso del metodo
dell'ombrello di infilaggi come sostegno per lo scavo di gallerie, Gallerie e grandi opere sotterranee, 72, (2004), 41-61.
EN 1997-1: Eurocode 7: Geotechnical design – Part 1: General rules, (2004).
ECLAIRCY-CAUDRON S.; DIAS D.; KASTNER R.; CHANTRON L (2005). Numerical modelling of a
reinforcement process by umbrella arch, Numerical modelling of construction process in geotechnical engineering for
urban environment 2005;
HOEK E. (2001). Big Tunnels in bad rock, Journal of Geotechnical and Geoenvironmental Engineering, September,
(2001), 726-740.
KIM C.; KIM K.; HONG S.; BAE G.; SHIN, H. (2004). Interpretation of field measurements and numerical analyses on
pipe umbrella method in weak ground tunneling, Eurock 2004, Schubert W. (ed.), Salzburg 2004, 167-170.
MAX J.; MATTLE B. (2002). Design of tube umbrellas, tunnel, 3, (2002), 4-11.
OCAK I. (2008). Control of settlements with umbrella arch method in second stage excavation of Istanbul Metro,
Tunnelling and Underground Space Technology, 23, (2008), 674-681.
ORESTE P.P.; PEILA D.; POMA A. (1999). Numerical study of low depth tunnel behaviour, Challenge for the 21th
Century, Balkema, Rotterdam, Oslo 1999, 155-162
ORESTE P.P.; PEILA D. (1998). A new theory for steel pipe umbrella design in tunneling, Tunnel and Metropoles,
Balkema, Rotterdam, Sau Paolo 1998, 1033-1040
PEILA D. (1994). A theoretical study of reinforcement influence on the stability of tunnel face, Geotechnical and
Geological Engineering, 12, (1994), 145-168.
PEILA D.; PELIZZA S. (2003). Ground reinforcing and steel pipe umbrella system in tunnelling, Rational Tunnelling,
Logos Verlag, Berlin, (2003) 93-132.
PELIZZA S.; PEILA D. (1993). Soil and rock reinforcements in tunnelling, Tunnelling and Underground Space
Technology, 8, (1993), 357-372.
SHIRAKAWA K.; AOKI T.; FUJII Y.; NAKAO T. (1999). Excavation through semicircular-shaped shell formed by
umbrella method at fault fracture zone beneath densely residential area, Challenge for the 21th century, ITA-AITES,
Balkema, Oslo, 1999, 441-452.
SHIN J.; CHOI Y.; KWON O.; LEE S. (2008). Model testing for pipe-reinforced tunnel heading in a granular soil,
Tunnel and Underground Space Technology, 23, (2008), 241-250
TAKECHI H.; KAVAKAMI K.; ORIHASHI T.; NAKAGAWA K. (2000). Some considerations on effect of the long-
fore-piling-method, Tunnels under pressure, SAIMM, Durban 2000, 395-398
UHTSU H.; HAKOISHI Y.; NAGO M.; TAKI H. (1995). A prediction of ground bearing due to tunnel excavation
under shallow overburden with long-length forepilings, South East Asian Symposium of Tunneling and Underground
Space Development, JTA, Bangkok 1995, 157-165.
YOO C. S.; YANG K. H. (2001). Laboratory investigation of behaviour of tunnel face reinforced with longitudinal
pipes, Progress in Tunnelling after 2000, Patron, Bologna, Milano 2001, 757-764.
VOLKMANN, G.M.; BUTTON, E.; SCHUBERT, W. (2007). Pipe Umbrella Support Systems and Installation
Methods, 2nd Symposium on Underground Excavations for Transportation, Istanbul 2007, 395-402