CONTRIBUTION TO THE DESIGN OF TUNNELS WITH PIPE ROOF

SUPPORT

G.M. VOLKMANN1 and W. SCHUBERT1

1
Graz University of Technology, Institute for Rock Mechanics and Tunneling; Graz, Austria
(e-mail of corresponding author: volkmann@tugraz.at)

The working area of tunnels is often supported with a pipe roof support system in weak ground. The experience
gained from former projects led to the conclusion that this support system decreases the subsidence during
tunneling. The lack of knowledge about the geotechnical system behavior of this support system disables the
designer from determining the basic design parameters depending on analytical and/or empirical solutions. This
fact often leads to conservative and uneconomical designs. In order to overcome this lack an extensive
monitoring program was executed on site. Using these data sets and the results of laboratory investigations the
geotechnical model for this support system was determined in numerical simulations. Starting with this back
calculated model the variation of basic design parameters was investigated. In this publication the focus will be
set on the influence of the dimension and number of pipes on the displacement magnitudes at the tunnel level.

Keywords: shallow tunnel; weak ground; pipe roof support system; design parameters.

1. Introduction
In the last decades shallow tunnels were increasingly constructed in weak ground. These tunnels
are often situated in urban areas, where project requirements, such as limited settlement
requirements, constitute the necessary support. The experience gained from former tunnel projects
indicate that the pipe roof support system not only increases the stability of the working face but
also decreases the subsidence induced by the excavation. Due to these experiences a number of
tunnels were additionally supported by this system without clear design rules for the
determination of the design parameters.
Before such necessary rules can be established the ground support interaction has to be
monitored during construction and the system behavior identified. The result of these
investigations was used for calibrating the geotechnical model in numerical simulations. The
authenticity of the numerical investigation was increased by using laboratory results for the rock
mass and support parameters. In contrast to the experience gained during construction the same
tunnel section can be excavated again and again with different support parameters in numerical
simulations.
The differences of these simulations demonstrate the influence of different design parameters
on the ground support interaction. Based on these evaluations relevant parameters can be derived
and design rules can be determined leading to a transparent design for pipe roof support systems.

2. Definition of Pipe Roof System

Some pre-support systems are not separated from each other by clear definitions. For this reason
the pipe roof system should be shortly described to inhibit confusion with other systems.
In the literature the pipe roof umbrella support system is also mentioned with the terms “pipe
forepole umbrella” (Hoek, 2003), “umbrella arch method” (Kim et al., 2004), “long-span steel
pipe fore-piling method” (Miura, 2003) or “steel pipe canopy” (Gibbs et al., 2002). These terms
all contain the words for describing this system. Normally steel pipes but also fiber glass pipes are
installed from the actual face to the front (forepoling systems) arranged like an umbrella or
canopy around the later excavated area. The diameter of the steel pipes is usually between 60 mm
and 200 mm with a wall thickness of 4 mm to 8 mm. The length of one umbrella is commonly
12 m or 15 m. The excavated length underneath (pipe roof field length) ranges from 6 m to 12 m.
The pipes can be installed with both special machines and conventional drill jumbos. From
the geotechnical point of view there are basically two different methods for the installation: the
pre-drilling system and the cased-drilling system. The significant difference is, when using the
cased-drilling system the pipe follows directly behind the drilling bit immediately supporting the
installation hole. When using a pre-drilling system the hole for the installation is drilled first and
in a 2nd step the pipe is installed in the unsupported hole. When the ground weakens a cased-
drilling system is therefore less susceptible to settlements than a pre-drilling system (Volkmann,
2004).

3. Projects and their Geological Conditions

The in situ measurement program was performed at two projects. 130 m of excavation were
overall observed in more detail for the investigations on the pipe roof support system.

3.1. Birgl tunnel

The 950 m long Birgl tunnel (AUSTRIA) is a double track railroad tunnel. The excavated area is
approximately 130 m2. The west portal and the following 80 m are situated in the so called
“Tauernnordrandstörung”, a major Alpine fault zone. This section of the tunnel was constructed
using the New Austrian Tunneling Method (NATM) with a pipe roof support as pre-support
system. In the section, where the additional measurements took place, the overburden increased
from 30 m to 50 m. The rock mass consisted of clayey, cataclastic fault zone material with shear
lenses composed of more competent blocks (3G & BGG, 2001). The design rock mass parameters
are shown in table 1.

3.2. Trojane tunnel

The Trojane tunnel (SLOVENIA) is a 2900 m long twin tunnel, located on the motorway section
AC A10 connecting Lubljana and Celje. The diameter of the tunnel is about 11 m. The tunnel was
driven using the principles of the NATM. In this section, where the measurement campaign was
performed, the overburden is 15 m. The geological conditions encountered during the construction
are dominated by a meta-sediment sequence including mudstone, claystone and sandstone. Alpine
thrusting resulted in heavily sheared zones that varied in thickness from a few centimeters to more
than 50 cm. The basic rock mass parameters are described by Zlender (2003) (table 2).

4. In situ data
Observations and measurements already have a long tradition in tunneling (Rabcewicz, 1944).
The measurement data are used to control the excavation induced movements (Rabcewicz 1963,
Steindorfer & Schubert, 1997). These movements reflect influences of the surrounding ground as
well as the construction method and the involved support methods. Altogether the measured data
represent the so called “system behavior”.

4.1. State-of-the-art measurements

Nowadays geodetic three dimensional observations are state-of-the-art for collecting displacement
data in tunneling. Using this system the positions of systematic located points in the tunnel and on
Table 1. Basic rock mass parameters for the Birgl Table 2. Basic rock mass parameters for the
tunnel. Trojane tunnel (Zlender, 2003).

Parameter Value Parameter Value

Matrix rock mass strength 0.3 – 0.8 MPa Young’s modulus 20 MPa
friction angle 20° Poisson’s ratio 0.25
cohesion up to 0.03 MPa friction angle 18°
Blocks rock mass strength up to 100 MPa Cohesion 0.016 MPa

the surface are determined in a local coordinate system during excavation. Such surveys are
normally executed daily.
The measured movements, induced by the construction process, are usually displayed in time
settlement diagrams, deflection curve diagrams and vector orientation plots. These display
methods enable to observe the ground support interaction. Special evaluation techniques allow
estimating parts of the pre-displacements by using the characteristics of the measured
displacements (Sellner, 2000). This estimation increases the quality for the geotechnical
evaluation of the survey data.
Surface settlements measured above shallow tunnels in weak ground are often larger than
those measured in the tunnel. This characteristic indicates that a significant part of the
displacements occur before the observation starts at the tunnel level (figure 1). To increase the
information gained by the geodetic survey an additional measurement system was applied at both
the Birgl and the Trojane tunnel during construction.
CS 130 m

Excavation Top Heading

CS 130 m

Excavation Top Heading

Excavation Bench

100

80

60
displacement [mm]

pre-disp. excavation top heading

40
top heading
20
zero reading
0
excavation bench
-20

-40
pre-disp. bench
-60
13.02.99 13 .03.99 10.04.99 08 .05.99 05.06 .99 03 .07.99 31.07 .99 28 .08.99 25.09 .99
Time

Figure 1. Time-settlement diagram: the grey shaded area in the diagram highlights the part of displacements that can be
measured by a geodetic survey at the tunnel level (Sellner, 2000)

4.2. Additional measurement system

The purpose of the additional measurement system is to obtain more information about the pre-
displacement characteristic at the tunnel level. Assuming homogeneous rock mass conditions only
settlements have to be measured in the crown ahead of the face without losing important
information.
 The instrumentation is installed in an extra steel pipe in the time of the pipe roof installation.
The 21 m long pipe is situated in the roof, parallel to the other pipes. The two executed
measurement campaigns used 2 m long inclinometer links. In each measurement section 10 links
were connected to each other to a horizontal, continuous, 20 m long, in-place inclinometer chain.
The instrumentation was connected to a data acquisition system that stored the measured data
every minute. This allows a very detailed observation of the settlements in the instrumented
section without interrupting the construction process (Volkmann & Schubert, 2005).

4.3. Benefits during construction

At the tunnel crown level the information gained from the inclinometer measurements supplement
the geodetical data ahead of that measurement cross section, which is nearest to the working face.
Due to the position of the inclinometer instrumentation (up to 20 m ahead of the working face) the
settlements are recorded directly above the working face and in the ground ahead of the face. The
additional data obtained from the inclinometer chain are leading to the total, measured settlement
path at the crown level. Behind the face the geodetical data additionally display the three
dimensional ground support interaction.
The evaluation of the geodetic data catches the combined influence of all construction
processes between two surveys on the ground support interaction. When constructing a shallow
tunnel in weak ground, the survey of the tunnel primarily presents the stability conditions in the
supported section behind the face. Around the heading the stress conditions change more rapidly
due to the ongoing construction process. The frequent change of the stress situation in this section
can be adequately observed by collecting the data in smaller time intervals, like the inclinometer
measurement system does. The high resolution of the data in time enables one to observe not only
the total settlement path but also the detailed development of settlements during single
construction processes. This includes the development of settlements during the excavation of
sequential parts of the cross section, the time dependent stabilization process afterwards and the
settlements induced by the installation of every support system (figure 2). This evaluation
identifies all settlement increasing construction phases (Volkmann & Schubert, 2005).

Figure 2. Time settlement diagram (left) and single deflection curve diagram (right) for one sequential excavation round

The short distance in between the measured points (2 m) enables the distribution of the
settlements in the longitudinal direction to be displayed as well as the position of the maximum
values. Changes in this characteristic behavior indicate changes in the ground support interaction:
e.g. changes in the effectiveness of the pre-support system or changes in the ground quality ahead
of the face.
The information gained from both geodetical and inclinometer data enable the understanding
of the mechanisms involved during construction. Both data sets have to be evaluated and
interpreted in a time span relevant for the tunnel advance (a few hours). Only in this way the
ground support interaction can be continuously controlled and uncertainties in the ground
properties can be followed by an adequate support adaptation leading to a safe and economical
construction process.

5. Laboratory Tests
Numerical calculations require a lot of input parameters, which can significantly influence the
results. For this reason multi stage triaxial tests and shear tests were performed on representative
samples from the rock mass. The different pipe dimensions were also tested with and without
grout to get input values for the simulations. With these results from the laboratory it is possible to
delimit the unknown parameters to a minimum leading to a more reliable simulation.

6. Numerical Investigations
The numerical studies were done with the program “Fast Lagrangian Analysis of Continua in
3 Dimensions” (FLAC-3D, Version 2.1). In order to decrease the boundary influence the length of
the model is 100 m. For the later discussed simulations the overburden is 15 m (Trojane tunnel).
The distance between the sidewall and the outer boundary of the model is 35 m. In order to catch
all mechanisms involved the geometry of the tunnel includes the saw-tooth shaped geometry in
the upper part of the top heading, which is typical for a pipe roof supported tunnel. With a
maximum finite element size of 0.5 m near the tunnel the memory limitation of the FLAC-3D
Version 2.1 only enables to simulate one half of the tunnel.
The strain-hardening/softening model was chosen for representing the ground behavior due to
the results of the laboratory tests. The shotcrete was simulated with a time dependent increase of
stiffness and strength based on Aldrian (1991) and Müller (2001). The smallest time increment
used for the aging process is 6 hours and the definition of age for the shotcrete is taken from the
advance rate at the Trojane tunnel. The heavy steel beams in the shotcrete are simulated with
beam elements. The face bolts and the pipe roof pipes are simulated with pile elements.
One meter of the top heading area is excavated at once in the simulation even though the
excavation of the top heading was done in 5 sequences at the Trojane tunnel with a design
excavation length of 0.8 m. Another 0.5 m is added to the excavated length as working area. After
the excavation the model is calculated until stability is reached. With stable conditions the support
consisting of shotcrete and heavy steel beams is installed behind the face and updated to its
current age values.

6.1. Comparison to the in situ data

The first exercise was to find the correct geotechnical model for the simulation. In contrast to the
publications of Bae et al. (2005), Hefny et al. (2004) and Kim et al. (2004) the grouted pipes were
not simulated as a homogenized area. Each pipe was simulated as a pile element and the grout was
neglected because the measuring of the grout volume indicated that only the pipes were filled with
grout. Other important points for modeling the construction process were the longer unsupported
span (0.5 m working area) and the calculation of stability before installing the support behind the
face. The adaptation of the bonding properties of the pile elements to the actual rock mass
properties also displayed an effect on the results.
 In figure 3 the upper deflection curve diagram shows the in situ settlement values. The
measured values after every second excavation round are displayed as black lines. Additionally
the trend line at the face is drawn in grey. The lower diagram shows the results of the numerical
calculation. The comparison of the diagrams shows a good correlation of the settlement values
with the little knowledge about the ground. The maximum settlement values as well as the
settlement distribution between pre- and total settlement amount can be simulated correctly.
Behind the face the increase of settlements is stopped in both cases after a few meters due to
the stiff support in this area. The geotechnical model for this part is therefore correct. The
settlement increase in the area of the unsupported span can also be observed in both diagrams. The
section ahead of the face also shows the same characteristic areas of faster and slower increasing
settlement values with respect to the uncertainties and continuous changes in real ground.
The implemented geotechnical model seems to be correct because the results of the
measurement campaign agree very well with the results of the numerical simulation, which is the
basis for further variations to evaluate the influence of the pipe number and dimension.

Figure 3. In situ deflection curve (upper) and simulated deflection curve diagram (lower) without pre-displacements.

6.2. Influence of design parameters

The influence of number and dimension of the installed pipes on the settlements is shown in
simulations from another Trojane tunnel section, where the pre-settlements increased up to
16.0 cm at the crown level.
For this comparison only the simulated pre-support is changed for one 8 m long pipe roof
field. Up to this position a 15 m long pipe roof support consisting of 20 pieces
114.3 mm x 6.3 mm pipes was used. The pre-settlement values at the crown level are evaluated
and compared. The reference calculation for this comparison is done without installing a pre-
support system. As can be seen in figure 4 and table 3 the pre-settlements at the face are 20.90 cm.
The maximum settlement value in the working area is 25.00 cm for the case without pre-support.
Another three calculations were performed with a pipe roof support consisting of 10, 20 or 30
pieces of 114.3 mm x 6.3 mm pipes. The installed pre-support decreases the pre-settlement values
depending on the number of pipes. The decrease of settlements increases with the quantity of
pipes (table 3).
The calculation with 20 pieces for the pipe roof support was also performed with the pipe
dimension 139.7 mm x 8.0 mm. The steel area per cross section is in this simulation comparable
to the simulation with the 30 pipes from the earlier mentioned case. During the construction on
site this case would be a little more time consuming than the case with the 20 smaller pipes but the
increase in stiffness decreases the settlement values again. Compared to the case with the 30
smaller pipes the reduction of settlements is nearly equal but one third more pipes usually need
more time for the installation.

Figure 4. Calculated settlement values without (left side) and with pre-support (30 pieces of 114.3 x 6.3) (right side)

Table 3. Calculated settlement values for the different cases

without 10 pieces 20 pieces 30 pieces 20 pieces

pre-support 114.3 x 6.3 114.3 x 6.3 114.3 x 6.3 139.7 x 8.0

at the face 20.90 cm 100.0 % 17.10 cm 81.8 % 16.15 cm 77.3 % 15.75 cm 75.4% 15.60 cm 74.6 %

maximum pre-
25.00 cm 100.0 % 19.70 cm 78.8 % 18.10 cm 72.4 % 17.50 cm 70.0 % 17.30 cm 69.2 %
settlement value

7. Conclusion
Pipe roof support system design is usually based on experience although their application has
increased in shallow, weak ground tunnel projects in the last decades. By performing additional
horizontal inclinometer measurements to supplement the state-of-the-art geodetic survey the first
step for understanding the system behavior was done. On site this additional data can be used to
optimize the construction process of pipe roof supported tunnels as well as to determine changes
in the ground quality ahead of the face. For this study the detailed data in combination with
laboratory data was used as input and control parameters for numerical investigations. This back
calculation clarified at first the geotechnical model for the pipe roof support system. Afterwards
the number and dimension of pipes in one pipe roof field was investigated.
Even though the grout was neglected due to the ground conditions of the investigated projects
the calculations clearly showed a decrease of the pre-settlement amounts up to 30 % at the tunnel
level when using a pipe roof pre-support system. The different cases displayed that the decrease of
pre-settlements increases with increasing number of pipes as well as with bigger dimensions.

8. Reference
3G and BGG (2001). “Gutachten zur Geologie, Geomechanik und Hydrologie.”, unpubl.
Aldrian, W. (1991). “Beitrag zum Materialverhalten von früh belasteten Spritzbeton.” Ph.D.
Thesis, Leoben: Montan University Leoben 1991.
Bae, G.J., H.S. Shin, C. Sicilia, Y.G. Choi and J.J. Lim. (2005). “Homogenization framework for
three-dimensional elastoplastic finite element analysis of a grouted pipe-roofing
reinforcement method for tunneling.” International Journal for Numerical and Analytical
Methods in Geomechanics. 2005. DOI: 10.1002/nag.402
Gibbs, P. W., Lowrie, J. Keiffer, S. and McQueen, L. (2002). “M5 East – Design of a shallow soft
ground shotcrete motorway tunnel.” In Proceedings of the 28th ITA-AITES World Tunneling
Congress, Sydney, Australia, March 2002
Hefny, A.M., W.L. Tan, P. Ranjith, J. Sharma and J. Zhao. (2004). “Numerical Analysis for
Umbrella Arch Method in Shallow Large Scale Excavation in Weak Rock.” In Proceedings of
30th ITA-AITES World Tunneling Congress, Singapore, 22 - 27 May 2004
Hoek, E. (2003). “Numerical modeling for shallow tunnels in weak rock.” PDF available at
www.rocscience.com/library/rocnews/Spring2003/ShallowTunnels.pdf
Kim, C.-Y., Kim, K.-Y., Hong, S.-W., Bae, G.-J. and Shin, H.-S. (2004). “Interpretation of Field
Measurements and Numerical Analyses on Pipe Umbrella Method in Weak Ground
Tunneling.” Proceedings of the 53rd Geomechanics Colloquium & EUROCK 2004. Copyright
VGE, October 2004, Salzburg, Austria
Miura, K. (2003). “Design and construction of mountain tunnels in Japan.” Tunneling and
Underground Space Technology, 18 (2003): 115-126
Müller, M. (2001). “Kriechversuche an jungen Spritzbeton zur Ermittlung der Parameter für
Materialgesetze.” Master Thesis, Leoben: Montan University Leoben 2001.
Rabcewicz, L. (1944). “Gebirgsdruck und Tunnelbau.” Vienna: Springer Verlag.
Rabcewicz, L. (1963). „Bemessung von Hohlraumbauten. Die “Neue österreichische Bauweise”
und ihr Einfluss auf Gebirgsdruckwirkungen und Dimensionierung.“ Felsmechanik und
Ingenieurgeologie, Special Issue Vol. I/3-4, 224-244. Vienna: Springer Verlag.
Sellner, P. J. (2000). “Prediction of displacements in tunneling.” Doctoral Thesis, Graz University
of Technology, Institute for Rock Mechanics and Tunneling. Graz, Austria
Steindorfer, A. and Schubert, W. (1997). “Application of new Methods of Monitoring Data
Analysis for Short Term Prediction in Tunneling.” In Proceedings of the 23rd General
Assembly of the International Tunneling Association, Vienna. Balkema, Rotterdam, 65-69
Volkmann, G. M. (2004). “A Contribution to the Effect and Behavior of Pipe Roof Supports.” In
Proceedings of the 53rd Geomechanics Colloquy and EUROCK 2004, Copyright VGE,
October 2004, Salzburg, Austria
Volkmann, G.M. and Schubert, W. (2005). “The Use of Horizontal Inclinometers for the
Optimization of the Rock Mass – Support Interaction.” In Proceedings of the 31st ITA-
AITES World Tunneling Congress, Istanbul, 7 – 12 May 2005, 967–972. London: A.A.
Balkema.
Zlender, B. (2003). “Triaxial Tests of Carboniferous Slates with Static and Dynamic Loading”. In
Proceedings of the 10th ISRM Congress, Johannesburg, South Africa, 1391-1394