ALTERNATIVE TUNNELLING METHOD FOR THE SUBWAY IN

MADRID.
A. Vazquez B.
Rover Grupo, Madrid, Spain
N. Bravo V.
Subterra Ingeniería, Santiago de Chile, Chile

ABSTRACT: The Belgian method or Madrid Traditional Method is a tunnelling method that has been
used for centuries to build tunnels in soils, which considers the integral and progressive shoring of the
advanced section, appropriate for soils that cannot be completely unconfined.
This system in Madrid has proven to be reliable for many years as there have been no significant
accidents or settlements. However, the working conditions are “tough” for the workers and in other
countries, the use of this method would be inconceivable due to the health and safety standards of
public administrations. For this reason, the need to improve health and safety conditions in the
execution of tunnels by conventional methods in Madrid has been chosen in this publication.
It is considered that the umbrella arch method with the composite piled roof (CPR) using hollow bars
or IBO anchors as a temporary support measure to build the permanent roof is a more convenient
alternative as it presents improvements in efficiency, safety, time, and quality. In addition, the use of
micropiles to underpin the crown increases safety and reduces settlement in the bench phase.
In this way, there is a safe working tunnel face without the need for a hand pneumatic hammer or
shoring by manual methods in constrained hard conditions.
This publication describes the Belgian method or MTM and provides the feasibility study of the
alternative with micropiles with finite elements (FLAC3D) in the soils of Madrid.

1. INTRODUCTION
1.1 BACKGROUND AND PURPOSE
For the subway of the Madrid Metro, is still common practice to build tunnels with the Belgian method
or Madrid´s Traditional Method (MTM). This paper suggests an alternative tunnelling method that
improves safety and quality, as well as an optimum performance during execution.
The Belgian method is a constructive system of excavation of tunnels in soils, which considers the
integral and progressive support of the advance section, appropriate for soils that cannot be completely
unconfined.
The MTM is an adaption of the Belgian method chosen since the nineteenth century for the Metro in
Madrid. This construction method means dozens of workers confined in small compartments, heat,
humidity, high levels of noise, dust, vibration and poor visibility. It is mainly due to these poor health
and safety conditions why it is suggested to modify the current MTM.
When executing tunnels with conventional methods in soils, reinforcement and / or improvement of the
terrain is necessary. The main problem in complex geotechnical conditions is the control of deformation;
Without support or treatment the soil remains "loose" and tends to sink into the cavern (falling material
from the crown), which is known as deconfinement or decompression [1].
To eliminate this risk of deconfinement, a pre-confinement or pre-support action is required in the
cavern, in other words, any action that enables the formation of an arching effect on the ground over
tunnel face. This can be done through reinforcement and/or protective actions in front of the tunnel face.
In the case of the Belgian method, this effect is achieved by manually inserting the support of stamps and
wooden supports.
Including an umbrella of micropiles of long length [13] intervenes in the tensional state around the
tunnel therefore performing a protective function ensuring that the resistance and deformation
properties of the interior of the tunnel are maintained (the main effect is to contain the reduction of the
main stresses σ3). See Error! Reference source not found. below as an illustration of the umbrella and
arching effect.

Figure 1: Illustration of the micropile umbrella with self-drilling micropiles and the arch and longitudinal effect. Source: TITAN
Ischebeck.

Considering that the NATM is prohibited in the Madrid Metro [4], it is not proposed to combine this
method with shotcrete, so an umbrella solution and permanent lining is proposed as performed in the
Belgian method (or current MTM).
The reason why using this type of self-drilling micropiles with dynamic injection is on the one hand due
to the versatility of installation since they can be drilled (and injected) with light machinery in limited
headroom. And, on the other hand, because they work as tension cables which differs from conventional
methods of micropile umbrellas in which there is dependence on the rigidity of the micropile when
dimensioned as supported beams.
In this case, as seen in the figure below, the micropile is fixed between the last casted section and the
long embedment length of micropile injected through the front of the tunnel (>11 m). The micropile
therefore works in tension which is ideal for this type of element, and it is not necessary to use large
diameters of micropiles.
Results with numerical methods of Mämpel & Faber (2005), Brandl (2005), Ischebeck (2005) and Eckle
(2012), Lopez & Severi (2018) show the behavior of these micropile arc umbrellas as shown in the
figures below (Figure 2 and Figure 3).

Figure 2: Simplified static model from Ischebeck (2005)

 Figure 3: Axial efforts on the micropiles arc umbrella, after two excavation steps (modified from Eckl (2012)).

The proposed design considered detailed analysis with numerical methods in FLAC3D to define the
dimensions of the micropiles, analyze the ground deformation and verify the stability of the front for the
different excavation phases. The results are compared with traditional method and discussed below in
Chapter 3.

2. ALTERNATIVE MADRID TUNNELLING METHOD

In this method micropile arc umbrellas are installed each 11 m in advance as shown in Figure 4. An
excavator with conveyor belts e.g., ITC, allows the transport of material 12-15 m distance behind from
the excavation phase A. The micropiles allow up to 3 m span without support. Then, the small drill rig
(e.g., Tescar) installs the underpinning micropiles due to the limited headroom (<3m). In order to
connect the head of the micropiles with the final lining, lattice girders are installed with c/c 3 m. Metal
ribbed formwork is moved to the face to cast each 3 m pass, optionally over rails. While concrete is
hardening (2 days) the bench phase B is performed 12-15m behind the face. The bench phase C1 is
performed with regular excavators and trucks until the wall is casted. Then, the bench phase C2 can be
performed. The last element casted is the invert. Below is a list of the main features of this method and
illustrated in Figure 4:
- Two rows of TITAN 40/16 micropile umbrellas of 15 m in length and inclination of 5 degrees
with c / c 0.35 m to be installed every 11 meters in plan with overlaps of 4 meters between each
umbrella and 3 meters embedded in the vault. Approximately 46 micropiles / 12 m tunnel.
- Row of TITAN 40/16 micropiles of 6 m in length in vault supports, are installed every 3 m
systematically and with inclination between 20-30 degrees with the vertical.
- Metal ribbed formwork and hydraulic carriage on rails.
- Excavator loader with conveyor belt, e.g., ITC for heading and bench phase B.
- Micropile drill rig for limited headroom e.g., Tescar, less than 3 m only for underpinning
micropiles.
- Conventional micropile drill rig for the arch umbrella.
- Light lattice girders of 3 bars for vault reinforcement and micropile connection.
 Figure 4: Proposed section for the Madrid method.

3. GEOTECHNICAL AND STRUCTURAL ANALYSES

3.1 NUMERICAL MODELLING
In order to evaluate the construction sequence proposed, three geomechanical models were performed
in the software of finite differences FLAC3D (V 5.01, by ITASCA). Figure 5 shows the 3D model considered
for this paper. As shown, the box dimensions are 60m long in the tunnel axis (y-axis), 160m wide (x-
axis) and 74.68m high (z-axis), the coverage is about 28m.

≈28 m.

Figure 5: Tree dimensional model mesh. Dimension in m.

3.2 GEO - MECHANICAL MODELS

As mentioned, three geo-mechanical models were performed in order to study the construction
sequences and its effects on the surrounding ground and surface. The models are described below:
 - Model 1 (M1): This is the baseline comparison model. This constitute the current excavation
method MTM. In this model the vault phase is excavated sequentially with 2.5m span. Once it is
finished, the bench phase is excavated.
- Model 2 (M2): The proposal section and sequence are modelled. In this case the vault phase is
excavated sequentially with 3m span. Every 11 m, a micropile umbrella arc is installed before the
excavation. Once the vault phase is finished, the bench phase is excavated.
- Model 3 (M3): A third model is performed in order to analyze effect of the vault-bench phase gap.
The vault and bench phase are excavated with a gap of 15m. The central bench is excavated 15m
behind the vault, the left sidewall is excavated 3m behind the central bench, and right sidewall is
3m behind the left one. The invert is finished 3 meters behind the sidewalls.

3.3 CONSTITUTIVE MODELS AND GROUND PROPERTIES

The constitutive model used for each soil layer is Mohr Coulomb. To represent correctly the water effect,
all analyses performed are coupled, i.e., the water table is explicitly represented and the flow of the water
due the excavation is considered. The parameters used in all the numerical analyses are shown in Table
1:
Table 1: Geomechanical parameters of typical Madrid soils from [2].

Geological unit Thickness Density Young Cohesion Friction Poisson Permeability 𝒌𝟎

Moduli angle Coefficient
Units m kN/m3 MPa kPa ° - cm/s -
Manmade Fill 2.75 18 10 0 28 0.35 10-3 0.53
Alluvial terrace 1.9 19 15 10 28 0.32 10-2 0.53
(Silt/sand/gravel)
Clay I (Peñuelas) 43 19 200 60 28 0.28 10-5 >19m 0.53
10-6 ≤19m
Clay II (Arcillas 27 19.3 250 50 28 0.28 10-5 0.53
negras)

4. RESULTS AND ANALYSIS

In each model performed, 3 points of control were monitored in the cross section from the beginning
until the excavations are finished, in different points along the tunnel axis (y - axis). These points and
the monitored parameter are summarized on Table 2:
Table 2: Monitoring points of the geomechanical models.

Point Location Longitudinal Parameter monitored

Position
P1 Over the crown of the y=15; y=30; y=45 Vertical displacement.
section. Within the soil.
P2 Left Sidewall. y=15; y=30; y=45 Vertical displacement,
Horizontal displacement.
P3 Right Sidewall. y=15; y=30; y=45 Vertical displacement,
Horizontal displacement.

The Figure 6 shows the maximum vertical displacement calculated on the crown for each model.
First, it is compared the data from different observation points in the same model. For M2 and M3 the
maximum value is the same regardless of the longitudinal point analyzed, but not in the case of M1 where
in point y=15 the greatest vertical deformation was obtained. In this point, the displacement reaches a
value of 28,3 mm, meanwhile in the point y=45 the displacement is about 23,7 mm. This means that the
excavation front has an important effect on the deformations of the crown for the first model, whereas
in M2 and M3 this effect is not appreciated, since the deformation is the same regardless of the
observation point. This also translates into more stable deformations along the tunnel. On the other
hand, comparing the results between the models, it can be seen that in any point analyzed, the
deformations are greater in M1. The M3 presents the least deformations, with a value of 21.2 mm, and
in the second model, the displacements reach 21.5m. That means that the distance between the vault
phase and the bench phase has no important effect on the vertical displacement, due the differences in
the vertical displacement are less than 2%.

Figure 6: Maximum vertical displacement in the crown calculated for each model in every longitudinal position. Units in
mm.

With the same methodology, all parameters monitored are compared between models and longitudinal
position. The maximum values of every point are summarized in Table 3. Comparing the data between
the models, same conclusions can be inferred. M1 presents greater values than the other models
regardless the point of observation and the parameter observed. The vertical displacements of the left
sidewall in the M2 decrease values by 44.4% on average relative to M1, while M3 decreases by 48.7%.
Similarly, for the vertical displacement of the right sidewall, the decrease corresponds to 36.2% for M2
and 40.4% for M3 on average. In the cases of the horizontal movements of the sidewalls, for M2 and M3,
the right sidewalls present more deformations that the left one, because this sidewall is excavated after
the left one, which leaves a bigger section without support, due the invert is casted behind the
excavation. This no happened in the model 1, because the ground tends to compress the excavation
vertically instead horizontally due the low value of 𝑘0 , so the vertical movements are more important
than the horizontal movements, as discussed before. Despite this, the horizontal movements of the
sidewall, in the M1 are greater than those presented in the other models.
Table 3: Maximum values of the monitoring points obtained for each model.

Position "y" Z disp. P1 Z disp. P2 Z disp. P3 X disp. P2 X disp. P3

Model
Units mm mm mm mm mm
15 -28.28 -22.25 -20.95 7.68 6.92
M1 30 -24.77 -21.99 -20.66 7.73 6.96
45 -23.71 -22.11 -20.73 8.02 7.26
15 -21.10 -11.87 -12.87 5.52 6.11
M2 30 -21.24 -11.01 -12.04 5.51 6.06
45 -21.20 -11.16 -12.23 5.84 6.44
15 -21.51 -12.55 -13.51 5.81 6.39
M3 30 -21.56 -12.20 -13.17 5.84 6.36
45 -21.39 -12.12 -13.10 5.99 6.53
The lower deformations observed in the new method proposed are due to two important factors, (1) the
presence of the micropiles in the crown and (2) then in the smaller excavation section. This section
reduction implies directly in lower deformations even the span excavated is bigger. As shown in Table
4, the new proposal section is up to 16.7% smaller than the current section, and use up to almost 49%
less concrete. This can be translated in a material saving, less ground excavated, and so, less waste
material.
Table 4: Cubage of excavation and concrete volumes for both sections.

Current New Variation

Item Units
Section Section [%]
Total Excavation m3/ml 67.28 56.02 -16.7
Vault Excavation m3/ml 35.22 30.66 -12.9
Bench Excavation m3/ml 32.06 25.36 -20.9
Concrete m3/ml 23.07 11.68 -49.4
Vault concrete m3/ml
12.42 7.68 -38.2
support
Bench concrete m3/ml
10.64 4 -62.4
support

Next, an evaluation of the damage thresholds is made in M2 and M3 according to the methodology
proposed by Boscardin & Cording for a central section of the model on the x-axis, i.e., in a plane that cross
y=30, and in the y-axis, i.e., in a plane that cross in x=0. The evaluation is shown in the Figure 7. As can
be seen, both M2 and M3 are under the “very slight” damage limit, this validates that the proposed
section does not present major drawbacks for surface structures, which are sensitive to high
deformations or differential settlements.

Figure 7: Estimated damage thresholds at surface level in the middle of the model (in a cross section) according to the diagram
of Boscarding and Cording.
Finally, the stability of the excavation front is analyzed. So the Safety Factor (S.F.) of the front is
calculated in the M3, for a intermediate stage of the excavation. The thrid model is chosen because the
excavated section is bigger than the M2, due the vault and bench phases are excavated simultaniously,
so both phases are evaluated. The safety factor is calcullated by mean a reduction of the resistant
parameters of the front excavation until the failure is achieved. The results are showed in Figure 8. As
shown , the S.F. calculated is 2.1, so the the excavation is considered stable.
 Figure 8: Safety Factor of the front for an intermediate stage of excavations.

5. CONCLUSION
Regarding the Belgian method (or MTM), it is a system that in Madrid has proven reliable for many years
since there have been no accidents or significant settlements. However, working conditions are poor for
tunnel workers and in other European countries it would be inconceivable to use this method due to the
health and safety standards of public administrations. That is why it has been looked into an alternative
method to improve the conditions health and safety in the execution of tunnels by conventional methods
in Madrid.
It is considered that the umbrella arch method as a temporary support measure to execute the vault is
possible and presents improvements in efficiency, safety, time and quality. In addition, the use of
underpinning micropiles before executing the bench phase increases safety and reduces settlement.
As can be seen in the results shown, the settlements and deformations around the tunnel are less in the
new proposed section. This is mainly due to the presence of the micropiles umbrella and the section that
has smaller dimensions than the traditional method (16% smaller). Due to the fact that the cross section
is smaller than the current one, it translates into less use of concrete to reinforce the section, smaller
excavation areas, which also reduces waste material, in addition to reducing earth moving costs,
materials and transportation. The distance between the vault and bench phases has no important effect
on the deformation. This can be seen due the variation in M2 and M3 results are similar, with less than
2% of difference between the two models. So, in the execution phase, the vault-bench phases gap, can be
more flexible and can be adjusted to the work conditions.
According to the evaluation, the proposed construction sequences are below the very light damage
threshold, allowing this sequence to be used in areas sensitive to superficial deformations. Finally, the
Safety Factor of the excavation is analyzed. As shown in Figure 8, the S.F. obtained has a value of 2.1,
which means the excavation front is stable.

6. ACKNOWLEDGEMENTS
The authors thank Rover and Subterra for the support in the research of an alternative method for
conventional tunnel construction in the Madrid Metro based on one project currently in execution [14].
Moreover, to AETOS (Spanish tunnelling association) for encouraging the author to publish a paper
about this alternative method.
REFERENCES
[1] • D. U. DEERE, R. B. PECK, and H. W. PARKER. Design of Tunnel Support Systems
[2] • OTEO, C.; RODRÍGUEZ ORTIZ, J.M and MENDAÑA, F. (2003) “Sobre los sistemas y parámetros geotécnicos de diseño
en la Ampliación del Metro de Madrid”. Revista de Obras Públicas, nº 3429, January pp. 49-67
[3] • OTEO, C. (2000) “Movimientos controlados durante la Ampliación del Metro de Madrid”. Revista de Obras Públicas, nº
3405 Dic. Pp. 91-102.
[4] • Melis M, Madrid, (2003) Madrid Metro and Railway Infrastructure 1995-2003
[5] • Brandl, J 2005 (Geoconsult Holding). Metro de Santiago Tunel de Acceso Pique Estación. Design of Ischebeck Pipe Roof
system (unpubl)
[6] • Fierdr. Ischebeck GmbH 2018 TITAN for tunnels and minning (Technical brochure, unpubl)
[7] • Ischebeck, E. F 2005. New approaches in Tunneling with composite canopies TITAN
[8] • Eckl, M 2012.Tragverhalten von Rohrschirmdecken beim Tunnelbau im Lockergestein. Doctoral Technische
Universität Munchen.
[9] • Lopez, F & Severi G, G 2018 Micropiling in Urban Infrastructure Advantages, Conference on Deep foundation and
Ground Improvement, Rome 2018.
[10] • Lopez, F Havranek &Severi G, Alternativ Umbrella Arches: The use of composite Pile Roofs
[11] • Mämpel, H&Faber, U. 2005 Ischebeck, GmbH, Verbundhaus Injeketionsbohrankern Typ TITAN 40/16.
[12] • Volkmann G.M 2004. A contibution to the effect and Behavior of Pipe Roof Supports.
[13] • Oke J, Vlachopoulos, N, Marinos. V 2017 Umbrella Arch Nomenclature and Selection Methodology for temporary
support systems for the design and construction of tunnels. Geotechncial and Geological Engineeging. An international
Journal (2014)
[14] • Proyecto constructivo ampliación de la línea 11 Metro Madrid PLAZA ELÍPTICA – CONDE DE CASAL.
[15] • Marco D. Boscardin and Edward J. Cording, (1989) “BUILDING RESPONSE TO EXCAVATION-INDUCED
SETTLEMENT”

(Provide contacts for author/s at the end of the paper, separated from References by one empty row.)
MSc, Alejandro Vazquez Borragan: Author1
Place of work: WSP Spain (Recently changed from Rover)
E-mail address: [email protected]

MSc, Nicolás, Bravo Vásquez: Author2

Place of work: Subterra Ingeniería Chile
E-mail address: [email protected]