Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Groeneweg
T.W. Groeneweg
T.W. Groeneweg
January 2007
Shield driven tunnels in
ultra high strength concrete
Reduction of the tunnel lining thickness
T.W. Groeneweg
January 2007
Graduation committee
Prof.dr.ir. J.C. Walraven
Dr.ir. C.B.M. Blom
Dr.ir. C.R. Braam
Dr.ir. O.M. Heeres
Ir. L.J.M. Houben
Preface
The research presented in this report is the graduation thesis to obtain my masters degree in Civil
Engineering at Delft University of Technology. Most work for this thesis was performed at the
engineering office of Gemeentewerken Rotterdam (Public Works Rotterdam).
The objective of the study is to investigate the technical feasibility of reduced lining thicknesses for
shield driven tunnels by the application of ultra high strength concrete.
I would like to thank the engineering office of Gemeentewerken Rotterdam for giving me the
opportunity to do the research within their company. My gratitude goes to my colleagues at the
office. I am very thankful to Jos, Leon, Tom and Wouter for the nice daily lunch breaks we joined.
I would also like to thank my graduation committee for the useful conversations and enjoyable
atmosphere during the meetings.
Special thanks I would like to address to Kees Blom for his personal guidance and ever lasting
enthusiasm that encouraged me to keep on going and get the research to its current level.
Finally I thank my family and friends for their help and understanding.
Tom Groeneweg
January 2007
v
vi
Summary
From the history of shield driven tunnelling in soft soil is known that the lining thickness of such a
tunnel is of a fixed ratio compared to the tunnels diameter. Consequently the thickness equals 1/20
of the diameter. In case of very large tunnel diameters the weight of the tunnel segments increases
dramatically. As a result costs strongly increase due to problems in the logistics of construction (the
production process, transport to the building site and placing of the segments in the tunnel).
However, the wish to construct ever larger tunnel diameters remains. Because the possibility is then
created to construct multiple-lane motorways, as for instance motorway A13/16 in the north of
Rotterdam, in such a tunnel. This report investigated the feasibility of using new steel fibre
reinforced concretes, very high strength concrete C100/115 and ultra high strength concrete C180/210,
to reduce the lining thickness of shield driven tunnels with very large diameters.
Several mechanisms are known to cause damage in existing shield driven tunnels and therefore may
lead to failure. In this report the following four have been studied intensively:
1. Common ring behaviour of the tunnel embedded in soil (serviceability phase)
2. Ring behaviour after grout injection along the tunnel (construction phase)
3. Introduction of thrust jack forces from the tunnel boring machine into the segments
(construction phase)
4. Torsion in tunnel segments by deformations due to the grout injection, also known as the
trumpet effect (construction phase)
Each mechanism resulted in a boundary condition on the required lining thickness. It was shown
that the serviceability phase is never governing. However, the strength related conditions by the ring
behaviour due to grout injection and by the introduction of thrust jack forces dictate the required
lining thickness. In case of ordinary concrete these mechanisms result in the standard thickness of
1/20 D as well. Hence the construction phase should never be excluded in the design of a shield
driven tunnel.
Torsion by the trumpet effect quickly leads to the formation of cracks in tunnel segments. Very high
lining thicknesses are required to prevent this mechanism from happening. These thicknesses are
beyond the standard required thickness of 1/20 D. Consequently cracks during construction of a
tunnel with such a lining thickness are likely to occur. Indeed in practice these cracks have been
observed. Also for very and ultra high strength concrete cracks will occur if the lining thickness is
based on the mentioned strength related conditions. Temporary measures such as adjuster (trusses
placed in the ring to prevent in from deforming) or the application of tunnel boring machines with
longer shields help to reduce cracks by the trumpet effect.
Very thin tunnel linings can be used if conventional reinforcement bars are added to a tunnel lining
in ultra high strength concrete. The required amount of reinforcement strongly depends on the
tunnels depth projection. However at each considered depth a thickness of only 1/58 D is possible. A
reduction of governing behaviour for the grout injection can be realised by the use of additional mass
(for instance sand fill) in the tunnel tube during construction only. Such temporary measures during
the construction phase make thinner linings, even below 1/60 D, achievable.
vii
viii
Samenvatting
Uit de bouwpraktijk van geboorde tunnels in slappe grond is gebleken dat een vaste verhouding
tussen de tunnels diameter en vereiste liningdikte (wanddikte) bestaat. De dikte is hierdoor gelijk
aan 1/20-ste deel van de diameter. Bij zeer grote diameters worden de tunnelsegmenten zodoende
zeer zwaar. Dit levert problemen op in de logistiek (het productieproces, vervoer naar de bouwplaats
en de plaatsing van segmenten in de tunnel), die de totale kosten sterk opdrijven.
De wens blijft echter bestaan om boortunnels met zeer grote diameters te maken. Hierdoor zal het
mogelijk worden ook snelwegen, zoals de toekomstige snelweg A13/16 in het noorden van
Rotterdam, in zon type tunnel aan te leggen. In deze studie is onderzocht of de nieuwe staalvezel
versterkte betonsoorten, zeer hogesterkte beton C100/115 en ultra hogesterkte beton C180/210,
kunnen bijdragen aan een reductie van de liningdikte voor boortunnels met een zeer grote diameter.
Verschillende mechanismen kunnen bij een boortunnel tot schade en daardoor mogelijk tot
bezwijken, leiden. In de studie zijn de volgende vier uitvoerig onderzocht:
1. Algemene ringwerking van de tunnel ingebed in grond (gebruiksfase)
2. Ringwerking na injectie van grout rond de tunnel (bouwfase)
3. Introductie van vijzelkrachten vanuit de tunnelboormachine in de segmenten (bouwfase)
4. Torsie in segmenten door vervormingen ten gevolge van de groutinjectie, ook bekend als
het trompeteffect (bouwfase)
De genoemde mechanismen resulteerden elk in een grenswaarde van de vereiste liningdikte. Het is
gebleken dat de gebruiksfase nooit maatgevend wordt. De sterkte-eisen door de ringwerking bij de
groutinjectie en de introductie van vijzelkrachten dicteren de vereiste liningdikte. Deze zijn ook
verantwoordelijk gebleken voor de liningdikte uit de standaard vuistregel van 1/20 D voor
conventioneel beton. De bouwfase mag in het ontwerp van een boortunnel daarom nooit buiten
beschouwing worden gelaten.
Torsie door het trompeteffect leidt snel tot scheurvorming in de tunnelsegmenten. Zeer grote
liningdikten zijn vereist om dit mechanisme te voorkomen. Deze dikten liggen voor conventioneel
beton in ieder geval boven de standaarddikte van 1/20 D. Hierdoor zijn scheuren tijdens de bouw te
verwachten, wat in de praktijk ook inderdaad is waargenomen. Ook bij zeer en ultra hogesterkte
beton zijn scheuren te verwachten als de vereiste liningdikte wordt gebaseerd op de genoemde
sterkte-eisen. Tijdelijke maatregelen als het gebruik van een adjuster (vakwerk in de ring om
vervormen te voorkomen) of de toepassing van een tunnelboormachine met een langer schild
kunnen dit effect gedeeltelijk terugdringen.
Door toevoeging van conventionele wapening aan een tunnellining van ultra hoge sterkte beton is
het mogelijk gebleken zeer dunne liningdikten te verkrijgen. Het benodigde wapeningspercentage is
sterk afhankelijk van de diepteligging van de tunnel. Echter op alle onderzochte diepten is een dikte
van slechts 1/58 D mogelijk gebleken. Voor een vermindering van de maatgevendheid van de
groutinjectie kan een tijdelijke massa (bijvoorbeeld een zandlichaam) in de tunnel worden
aangebracht. Eventueel kan deze massa na de bouwfase probleemloos worden verwijderd. Door het
toepassen van zulke tijdelijke maatregelen tijdens de bouwfase kunnen zelfs liningdikten onder de
1/60 D worden gerealiseerd.
ix
x
Notations
xi
= Poisson ratio (lateral contraction) concrete
sd = Specific gravity dry soil
sw = Specific gravity wet soil
h = Horizontal soil pressure
r = Radial soil pressure
t = Tangential soil pressure
v = Vertical soil pressure
w = Water pressure
'h = Effective horizontal grain pressure soil
'v = Effective vertical grain pressure soil
t = Shear stress due to torsion trumpet effect
= Rotation in longitudinal Janen joint
= Internal friction angle soil
Abbreviations
TBM Tunnel boring machine
UHSC Ultra high strength concrete
ULS Ultimate limit state
xii
Table of contents
Preface v
Summary vii
Notations xi
Chapter 1 Introduction 1
1.1 Introduction to shield driven tunnels 1
1.2 Problem description 2
1.3 Problem definition 2
1.4 Objective 2
1.5 Solution approach and arrangement of this report 2
xiii
4.2.2 Complete and incomplete grouting 48
4.3 Relation with the lining thickness 48
4.3.1 Maximum bending moment 48
4.3.2 Bending moment capacity 51
4.3.3 Retrieving the required lining thickness 51
4.4 Conclusions 52
xiv
Chapter 1
Introduction
The first excavated tunnel in soft soil was finished in 1843. The tunnel connected both sides of the
river Thames in London and was designed by Marc Brunel and his son Isambard. Brunel used a
shield to keep the open area of the tunnel and the surrounding soil apart. Through the front of the
shield miners removed soil through small cavities. Behind the forward moving shield a brickwork
tunnel lining was erected.
Basically this method is still used in the construction of todays shield driven tunnels. However
miners have been replaced by a fully automated tunnel boring machine (TBM) and the brickwork has
been replaced by a prefabricated segmented concrete tunnel lining.
Foreign successes resulted in a pilot project in the Netherlands as well. In 1995 construction of the
first Dutch shield driven tunnel, the Second Heinenoord Tunnel, commenced. Subsequently nine
tunnels have been finished or are still under construction:
1. Second Heinenoord Tunnel
2. Westerschelde Tunnel
3. Sophia Rail Tunnel
4. Botlek Rail Tunnel
5. Tunnel Pannerdensch Canal
6. Green Heart Tunnel
7. North/South Metro Line Amsterdam
8. RandstadRail Staten Tunnel Rotterdam
9. Hubertus Tunnel The Hague
During and after construction of these tunnels several research projects studied the behaviour and
forces on the lining. As a result of the growing awareness of the mechanisms a shield driven tunnel is
subjected to, the diameter of the tunnel increased from 7,6 m for the Second Heinenoord Tunnel to
13,3 m for the Green Heart Tunnel.
1
1.2 Problem description
From the foreign history of shield driven tunnel construction in soft soil it is known that the required
thickness of concrete tunnel linings has a directly proportional relation with the tunnel diameter. The
thickness should be approximately 1/20 of the diameter (1/20 D) [4]. Obviously the circumference of
the tunnel also increases proportionally to the diameter. Hence their product, which represents the
amount of concrete needed to construct the lining, increases quadratic proportional to the tunnel
diameter (see Figure 1). For tunnels with large diameters this results in high costs as a consequence of
the material volume, heavy transport and the assembly process.
concrete volume V
The mechanism that leads to the required lining thickness has been
and still is an intensively discussed subject. In his dissertation [5] D D
Blom described several mechanisms which lead to observed cracks V ~ 2 - - = -D
2 20 20
(damage) in existing shield driven tunnels. By considering cracks as
an early warning system for failure caused by overloading, these
mechanisms might lead to failure at some point.
Despite the ongoing discussion the demand for larger tunnel
diameters remains. And preferably these larger diameters should be tunnel diameter D
achieved using equal or even smaller lining thicknesses than
1 | Volume of concrete increases
applied today to achieve an optimal economic design, hence save quadratic proportional to tunnel
costs. diameter
The application of new concrete materials such as very high strength concrete and ultra high strength
concrete might be a solution to this controversy. The use of these steel fibre reinforced concretes
already resulted in some very slender structures for bridges and roofs. Simultaneously in [16] it is
stated that by the addition of steel fibres to tunnel segments of ordinary strength concrete cost
savings can be achieved in two ways: either by reducing the thickness of the elements or by reducing
the amount of traditional reinforcement.
1.4 Objective
The objective of this study is to investigate the technical feasibility of reduced lining thicknesses for
shield driven tunnels by applying very high strength and ultra high strength steel fibre reinforced
concrete in stead of ordinary concrete with steel bar reinforcement.
The research will be applied on a shield driven tunnel for the future motorway A13/16 in the North
of Rotterdam.
Several mechanisms which are assumed to possibly result in the required lining thickness of 1/20 D
have been presented by Blom [5]. In other studies as well the behaviour of these mechanisms has
been investigated:
In this report the relations between those failure mechanisms and the required lining thickness will
be studied for three concrete strength classes:
1. Ordinary concrete with steel bar reinforcement C35/45
2. Very high strength steel fibre reinforced concrete C100/115
3. Ultra high strength steel fibre reinforced concrete C180/210
Each mechanism will return a boundary condition of the required lining thickness for the case that
particular mechanism would be governing. For that purpose Chapter 3 and Chapter 4 discuss the
calculation methods and resulting lining thickness boundary conditions for the tunnels ring
behaviour when embedded in soil and in the uplift loading case respectively.
From literature [3] its known that shear forces are of minor influence in a circular shield driven
tunnel and will not turn out to be governing. Therefore the focus of these calculations will be on the
generated bending moments and normal ring forces.
Chapter 5 discusses the influence of two additional mechanisms from the tunnels construction
process. One of them is the already mentioned introduction of thrust jack forces, the other is the so-
called trumpet effect. The latter has been described in [5] and deals with observed cracking due to the
distortion of tunnel segments in the uplift loading case. Cracks resulting from this imposed
deformation do not result in a failure mechanism to occur, but only cause damage. Nevertheless this
mechanism will be studied for the concrete strength classes presented in order to investigate any
possible variations in the sensitiveness.
All boundary conditions regarding the lining thickness will be combined in Chapter 6. Then it will
become clear which mechanisms actually dominate the required lining thickness for the case studys
tunnel. The possible reduction of the lining thickness by the use of very and ultra high strength
concrete will then become visible as well. Since the governing mechanism is then known, alterations
to the tunnel design might reduce the lining thickness even more. Some examples are given in
Chapter 6.
Conclusions about the research presented in this report and recommendations for additional research
are described in Chapter 7.
C100/115
C180/210
Layout
Concrete material Tunnel depth
tunnel design
Lining capacities
Chapter 6
Adaptation of design
Governing failure mechanism
to reduce effect of
Required lining thickness(es)
governing mechansim
Chapter 6
Economy
2.1 Introduction
From practice it is observed that despite all effort put in analyses and engineering of shield driven
tunnels, the lining thickness is still of a constant ratio to the tunnel diameter: 1/20 D. However the
wish remains to construct ever larger tunnel diameters with similar or even smaller lining
thicknesses compared to the ones used today. The newly developed ultra high strength concrete
might be a solution to this problem. Concrete strength classes in existing shield driven tunnels did
never exceed regular strengths like for instance C30/40. In section 2.3 the rise of the new type of
concrete will be clarified briefly.
The subject of shield driven tunnels will be discussed in section 2.2. The construction process of the
rings and the tunnel segments themselves are dealt with.
In the north of Rotterdam a new motorway is planned, the A13/16. This motorway has to reduce the
amount of traffic on the heavily loaded motorways in Rotterdams urban areas. Section 2.4 describes
the location and environment of the new motorway. The option to construct this motorway in a
shield driven tunnel is presented. This project is used as a case study in this report.
Table 1 | Lining thickness over diameter ratio for Dutch shield driven tunnels
5
bolt pocket
handle hole
width
longitudinal joint
thickness
bearing pad
ring joint
length
longitudinal joint
As soon as the TBM cleared enough soil, a tunnel ring is assembled by several tunnel segments
within the protection of the TBMs shield. These segments are temporarily bolted together to the
previous ring in order to prevent shifting and dropping during the complete construction process.
Because the tunnel rings are assembled within the TBMs shield, the external diameter of this
machine is slightly larger than the tunnels diameter. Hence a tail void is created between the tunnel
lining and the surrounding soil due to the forward movement of the TBM. To prevent settlements of
the soil and to embed the tunnel the viscous material grout is injected. That material stiffens over
time.
In the German configuration thrust jacks at the edges of the segment are realised by one thrust jack
plate over the longitudinal joint of two adjacent segments. The total number of thrust jack plates per
ring is therefore similar for both configurations.
A significant disadvantage of the German configuration is the appearance of large tensile forces by
the introduction of thrust jack forces at the edge of the segment (see Chapter 5). Because of the
presence of three contact areas in the ring joint, this configuration is vulnerable to placing errors in
already assembled rings as well. If one out of three contact areas is not able to fully interact with the
other ring, the huge thrust jack force at the other front face of the segment has to be diverted to the
other remaining supports. The result is the creation of large internal tensile forces. In concrete this
implies the formation of cracks, hence damage (see Figure 5).
If the axial contact pressure in the ring joints is present, the joints are able to interact in radial and
tangential direction as well. The distribution of contact areas along the segmental length of the
German configuration, especially the area in the middle, then introduces significant additional peak
bending moments in the tunnel lining.
All mentioned problems are significantly reduced or diminished by reducing the number of contact
areas per tunnel segment to two and by locating these areas as far from the longitudinal joints in
adjacent rings as possible. These properties have been realised in the French thrust jack
configuration. From the finished Dutch shield driven tunnels only the Green Heart Tunnel made use
of this configuration.
a b
5 | Uneven support of tunnel segments for: a) German configuration; b) French configuration
By now a number of projects have been constructed in the new steel fibre reinforced concretes. These
are mainly bridges and roofs in both France and Japan. All were accomplished with very slender
structures. Figure 7 shows the Bridge of Peace in South Korea.
The creation of ultra high strength concrete is possible by making some changes in the design of the
concrete mixture [10, 6]:
Reduce the water-cement ratio
Increase the packing density by optimising the grading curve (see Figure 6)
Improve the homogeneity by using small sized particles only
Add steel fibres to the mixture to resist minor internal tensile bursting stresses and to bridge
minor cracks
Both types of ultra high strength concrete, with steel bar or steel fibre reinforcement, are extremely
dense. This results in outstanding durability properties. Ductal for instance claims a carbonation
depth of only 2 mm in 500 years [9].
For the construction of road traffic bridges in ultra high strength concrete in the French town of
Bourg-Ls-Valence design recommendations for the material used were required. As a result the
French organisations AFGC (Association Franaise Gnie Civil) and Setra (Service dtudes
techniques des routes et autoroutes) published a document with some interim recommendations in
2002. At the moment of publication of this report the French recommendations were the only
recommendations present; hence calculations in this study were based on their findings.
Prices of ultra high strength concrete are yet very high. A cubic metre of BSI-Cracem (C180/210)
costs approximately 400-800 euro. The same amount of ordinary concrete costs only 150-200 euro. If
the production of the material increases because more engineers and contractors dare to apply it in
their structures, the costs are supposed to show a significant decrease.
The route Rijkswaterstaat suggests runs close to (future) housing an existing and future green zones,
among which the picturesque river Rotte and park Lage Bergsche Bos. Construction of the motorway
in a shield driven tunnel appears to be a sympathetic solution to diminish hindrance during its
construction and use in those areas.
Prins Clausplein
Zoetermeer
Rijswijk A12
Nootdorp
Ypenburg
Pijnacker
Delft Bleiswijk
Delfgauw
Berkel en N209
Rodenrijs
A4
Bergschenhoek
N470
A13
Hoge Bergsche
Bos
Lage
Bergsche Bos
A13/16
Rotterdam
Airport Rotte
Terbregseplein
A20
Overschie
Kleinpolderplein
Kethelplein
Schiedam
Rotterdam
Vlaardingen
A16
Nieuwe Maas
A4
9 | Fitting two directions in one tunnel tube. a) Two tunnels with large useless free areas; b) One tunnel with less free space
Alignement
Rijkswaterstaat already suggested two routes for the new motorway. Because no hindrance of a
motorway in a shield driven tunnel is recorded on the ground surface, the shortest route can be
applied. Every possible route forces the motorway to cross the high speed railway line HSL-Zuid. At
the crossing location the railway is situated at a slab founded on a significant number of poles. These
piles require an underground crossing at a depth of at least -24,00 m + NAP (soil overburden of 19,26
m).
The research project of this report does not deal with the complete design of the tunnel. The effects of
the application of new concrete materials on the reduction of the lining thickness of this tunnel are
investigated. A vertical alignment of the motorway will be assumed only. For now its supposed to
be horizontal at a depth of -24,00 m + NAP. Only at the ends the tunnel will reach smaller
overburdens.
At the engineering office of Gemeentewerken Rotterdam soil data of the tunnels route was available.
It shows that the tunnel is fully embedded in packed sand, the so-called Layer of Kedichem. Figure 11
shows both the alignment and soil conditions of the tunnel.
escape route
BD 35 JR
safety strip RW 13 16
T
ranspo
rte
r
IG WR 19 VT UD 06
safety strip
10 | Cross-section of tunnel for motorway A13/16 with personal cars only (scale 1:150)
NAP 0 m
-10 m
-20 m
shaft
shaft
Peat -30 m
shield driven tunnel
Klay
Silt -40 m
Sand
-50 m
Unknown
11 | Schematic representation alignment of the tunnel for motorway A13/16
3.1 Introduction
Reaction forces and deflections in the serviceability stage of a shield driven tunnel are described by
calculations on ring behaviour for the embedded lining. Calculations are made on the cross-sectional
face of the tunnel perpendicular to its axis.
Soil surrounding the tunnel has certain stiffness, just like the concrete lining itself. Stiff parts attract
bending moments. Consequently the tunnel and soil will cooperate to bear all loads. These loads
result from the soils mass and ground water pressure surrounding the shield driven tunnel.
In the introduction of this report it was stated that shear forces in circular tunnels are not governing
[3]. On the other hand the bending moments, in combination with a normal force in ring direction,
are governing. Hence the scope of calculations in this chapter will be on those reaction forces.
In 1964 Schulze and Duddeck described ring behaviour of shield driven tunnels by a collection of
graphs. By means of those graphs bending moments and normal forces could be retrieved for various
depth projections of the tunnel and various ratios between the tunnel stiffness and soil stiffness.
When computers developed and the time needed for more comprehensive calculations decreased,
the creation of models specifically designed for one tunnelling project grew popular. The main
difference in the models created by now is the modelling of the soil. In finite element models soil is
normally introduced as a continuum around the tunnel lining. In more uncomplicated framework
analyses the soil has been reduced to springs and loads representing the supporting and loading
effects of the soil on the tunnel lining. This model focuses on the tunnel structure only; the
developments of deformations and stresses in the surrounding soil are omitted. Finite element
models however are able to return these soil results as well.
Modelling of the tunnel lining itself can be realized by reducing the ring to a homogenous ring beam,
a segmented single ring beam or a segmented double ring beam. The homogeneous ring beam is
most simplified, but ignores peak moments which develop in the lining due to the presence of
longitudinal joints and ring joints. The segmented single ring beam model takes care of the
longitudinal joints as well. This model is valid if no axial normal forces are present; hence no
interaction between rings occurs via the contact areas in the ring joints. The segmented double ring
beam model introduces the effects of both longitudinal and interacting ring joints in the calculation.
All models can be created in a 2-dimensional or 3-dimensional environment. A 3-dimensional model
however is a waste if no serious changes in loading, support or geometry occur in the third
dimension, the tunnels axial direction. The only significant defect in this direction is the ring joint.
Such a joint however can be modelled in a 2-dimensional environment as well, which is then referred
to as 2-dimensional as well. Hence a 2-dimensional environment is sufficient.
In section 3.2 the creation and validation of a segmented double ring beam model with soil
interaction represented by springs and loads will be described. This type of model is a reliable
representation of reality [5] focussing on the tunnel structure only with relatively short calculation
13
times. Therefore this model is widely used in practise and will be used in this research project as
well.
Section 3.3 abstracts parameters from the case studys tunnel for use in the yet created model.
In order to reduce the lining thickness of a shield driven tunnel, knowledge about the relation
between the thickness and safety level of the tunnel is required. Section 3.4 concludes this relation for
the case study tunnel of this research project, motorway A13/16. It will turn out that the depth
projection of the tunnel has a serious influence on the safety level. Therefore this chapter will not
focus on the depth projection as suggested in Chapter 2 only, but includes a wider range.
Comparison of the actual safety level with a required safety level generates the possibility to abstract
a boundary condition on the required minimum lining thickness. In section 3.4 the requirement by
the ring behaviour of the embedded tunnel lining will be concluded for the considered concrete
materials, namely C35/45, C100/115 and C180/210.
Conclusions concerning this chapter will be described in section 3.5.
Modelling of the interaction forces in one ring joint requires that the influence zone of this particular
joint is included in the model only. This implies that only half the width (dimension in axial
direction) of each tunnel segment is used. If this fact is considered in all calculations and checks of
the embedded ring behaviour, no difficulties occur.
Calculations in this study will not include the key stone. The narrow width of this segment results in
a small decrease of the length (tangential dimension) of adjoining segments. By excluding the stone
their width is normal again and their common longitudinal joint is positioned at the centre of the
actual key stone. The validation of the model will prove that no significant changes occur due to the
key stone.
Section 3.2.1 describes the reduction of the concrete tunnel segments to elements for the framework
analysis. Modelling of the longitudinal joints and ring joints is described in sections 3.2.2 and 3.2.3
respectively. In section 3.2.4 the modelling of soil is taken into consideration, which will be
represented by springs and loads as described before. The final model will be validated by
comparing its resulting bending moments to the ones of two other studies in section 3.2.5.
The software application ANSYS 6.1 is used to process the framework analysis from this study.
A certain number of nodes were needed any way. Connections between adjoining rings, representing
the ring interaction, can be attached to nodes only. For the French thrust jack configuration now at
least two nodes are required in the segments span. By adding the nodes at the ends of the segment
(in the longitudinal joints) to this number, at least four nodes (hence three beam elements) have to be
used. To keep the length of each beam element equal (for output handling reasons), another node is
added to the middle of the segment. So at the end at least five nodes and four beam elements make a
tunnel segment with the French thrust jack configuration. More detailed output data is achieved by
multiplying this required number of elements by any integer. In this study 12 beam elements will be
used for each tunnel segment.
In ANSYS five properties have to be defined for each elastic beam element:
1. Cross-sectional surface in axial-radial plane A = bh
2. Moment of inertia I = 121 bh 3
3. Height h
4. Youngs modulus E
5. Poisson ratio
One probably noticed that no property has been assigned to the segmental mass. It will not be added
as a loading of the structure either. Because the weight of the tunnel lining is very small compared to
the mass of the excavated soil, the high pressure on the tunnels external surface will hardly be
influenced by the tunnels own mass. Consequently this load will not be introduced to the
calculation.
The circular shape of shield driven tunnels generates an ideal opportunity to use a cylindrical system
of coordinates. In this system each node of the beam elements is defined by a fixed radius and an
angle. Of course this assumes a fully circular tunnel ring prior to loading.
Janen developed an often applied method to include the non-linear rotational behaviour of the
longitudinal joint in a rotational spring. This non-linear spring connects the end nodes of two
adjacent tunnel segments in the same ring.
The so-called Janen joint assumes that the contact area can be represented by a concrete beam with a
depth equal to the joint contact areas width (segmental width) and its height and width equalling
the joints contact height. Opening of the joint is included by the fact that the concrete beam is unable
to bear any tensile stresses at all. Now it is possible to find a relation between the rotation in, bending
moment in and normal force on the beam. Next this relation is translated into a spring stiffness of the
rotational spring.
At an increasing rotation in the joint or beam, three stages are distinguished:
The enumeration also shows the boundary conditions for the rotation or maximum strain in the joint
to be valid for that particular stage. The parameters represent:
b = Width contact area longitudinal joint (segmental width)
E = Youngs modulus
lt = Height contact area longitudinal joint
N = Normal ring force
= Rotation
'c = Compressive yield strain of concrete
'u = Ultimate compressive strain of concrete
In Appendix A.1 the equations for the Janen rotational spring stiffness kr have been derived. Here
the final equations are given for all three stages:
bl 2 E
1. Closed joint: k r = t
12
2
2M
9 bl t EM 1
Nl
2. Opened joint: k r = t
8N
6( 2Mf 'c b + l t Nf'c b + N 2 )ME
3. Opened joint plastic concrete behaviour: k r 1, 2 =
(6N )
21N 2 + 30Mf 'c b 15Nl t f'c b l t f'c2 b
Where:
M = Bending moment in the joint
f'c = Design value compressive strength concrete
In most studies only the first two stages (opened joint and non-plastic closed joint) are being used.
Plastic behaviour in the joints implies very large rotations and very large deflections, which is
undesirable in normal day use and in general will not and may not take place. Consequently only the
first and second stages will be applied in the model of this study as well.
The equation for the spring stiffness of an opened joint shows that the stiffness is related to the
rotation , bending moment M and normal force N. The software application ANSYS however only
knows a non-linear spring with a custom relation between the rotation and bending moment,
without the normal force. However no problems arise, for the normal ring force varies only little
along the rings circumference and is relatively easy to predict in advance.
For a ring under uniform radial pressure p (see Figure 12) the normal ring force is retrieved as
follows.
The vertical component of the pressure at an angle is given by:
pv = p cos
The force due to this vertical pressure at a section with an angular span of d holds:
fv = pv Rd
Where:
R = External radius
d
R
N N
12 | Uniform pressure leading to normal ring force tunnel
If the average radial pressure on the tunnel is used for pressure p in equation (1), a rather accurate
approximation of the normal ring force in the Janen joint returns. The resulting non-linear relation
between the rotation and bending moment is assigned to all non-linear rotational springs at the
locations of the longitudinal joints in the model. An example of this relation is shown in Figure 13.
moment M [kNm]
700 Mmax
600
500
Obviously the Janen method only focuses on the transfer of bending moments in the joint. Normal
forces and shear forces are transmitted in a far more straightforward way, namely by linear
translation springs. Provided that the joint is subjected to a pure compressive normal force the joint
will not be noticed at all, the normal force is simply transferred from one tunnel segment to the other.
The joint itself does not elongate by the compressive force. A spring in tangential direction should
therefore prevent such elongation. In other words: a spring stiffness of infinity should be assigned to
the spring. The software application however requires an actual value. A parameter study showed
that a value of 1010 kN/m results in minor insignificant values only; hence this value was selected.
So why not use a beam element when an infinite stiff behaviour in the axial direction is required? The
beam element has a property to transmit bending moments, shear forces and normal forces by
definition from one end to the other. For this joint however only normal forces and shear forces
(frictional forces) should be transferred, bending moments are taken care of by the Janen joint.
Hence a beam element is unsuitable.
At the end three springs are necessary to model a longitudinal joint between two tunnel segments: a
rotation spring with a non-linear spring stiffness according to Janens method and linear translation
springs in radial and tangential direction with high spring stiffnesses.
In most Dutch shield driven tunnels so-called packing materials have been used at the contact area of
the ring joint. For instance plywood planks are frequently used. The idea behind this phenomenon is
that if the ring joint rotates the plywood plank will deform unevenly, but remains in contact with
both concrete tunnel segments. Now the complete contact surface would still be available for the
transfer of frictional forces in the ring joint. The thickness of plywood planks automatically defines
the location of the ring contact areas.
The Green Heart Tunnel showed that packing materials could easily be left out of the design as well.
Its joints made use of pure concrete to concrete contact only. The bearing pads were defined by a few
millimetres of additional concrete and behaved just as well.
The force responsible for interaction between two adjoining rings is a friction force in radial
direction. Just like the frictional shear forces in longitudinal joints, this transfer is modelled by use of
a linear translation spring in radial direction again. The stiffness of this spring however does not
equal infinity.
Springs in ring joints represent both the stiffness of the concrete to concrete contact behaviour and
the stiffness of the tunnel segments bordering the joint. If one returns to the original modelling of the
concrete tunnel segments themselves the reason becomes clear: the beam elements representing the
tunnel segments are located at the centre lines of the segments. Even though only half the segmental
width is taken into account, the elements remain at the original centre lines. Hence at the cutting
faces of the narrowed segments now (see Figure 14). Recollection of the skipped third dimension
(tunnel axis) shows that there actually is a full segmental width in between the beam elements of
both adjacent rings. Over that width deflections might occur by interaction forces in the ring joint.
This will additionally define the total ring joint spring stiffness.
By considering the cross-sectional surface of the half tunnel segment in a vertical plane along the
tunnels axis, a spring stiffness may be assigned to that particular part of the structure. It is assumed
that the magnitude of the axial thrust jack force (still present in the ring joint contact areas) is of such
extend that no rotation occurs in the joint. By forcing one of the sides upward a deflection u occurs
due to a vertical force F. A linear bending moment distribution results, with its maximum values at
both ends. Figure 15 displays this simple mechanism.
The symmetric bending moment distribution allows reducing the problem to a beam clamped at one
end and loaded by a force F at the other free end. The maximum bending moment simply holds:
b
M=F
2
The rotation of the loaded endpoint (in the ring joint in the original problem) is then given by:
M( b / 2) Fb 2
= =
2EI 8EI
Where:
I = Moment of inertia in the radial-tangential plane
u
u/2
b/2
One ring interaction location is handled. Therefore the moment of inertia for the French thrust jack
configuration is defined by half the segmental length in tangential direction. Hence:
I = 121 l segm / 2 h 3
Where:
b = Full segmental width
h = Lining thickness
Di = Internal tunnel diameter
( Di + h)
segm = Segmental length =
n segm
nsegm = Number of segments per ring
During the description of the case study tunnel for motorway A13/16 it came forward that this shield
driven tunnel is primarily embedded in compacted sand. Since this study does not focus on a
detailed design of the tunnel, but tries to find an answer on the question whether the lining thickness
of shield driven tunnels may be reduced by applying ultra high strength concrete, a simplified soil
continuum is used now. Subsequently the complete soil continuum is assumed to consist of packed
sand only. This only affects the loads on the tunnel, not the supporting function as will become clear
from the following sections.
sw
xtop
h
w()
v()
'v()
w v
The total soil pressure at an angle of the tunnels external surface is:
( )
v () = v , wt + x w x top + h sw (3)
Where:
xtop = Top of tunnel relative to mark NAP
sw = Specific gravity of saturated soil
Parameter h is the vertical distance between the tunnel top and the point at an angle along the
tunnels external circumference. Its defined by:
h = (1 cos )D / 2 (4)
Where:
D = External diameter of the tunnel
The hydrostatic water pressure and the effective vertical grain pressure combined equal the total
vertical pressure. Hence for the effective vertical grain pressure it holds:
' v () = v () w ( ) (5)
Where w() is the hydrostatic water pressure at the point with an angle , reading:
(
w () = x w x top + h w ) (6)
Substitution of equations (2), (3), (4) and (6) in (5) defines the effective vertical grain pressure as:
( )
'v = (x 0 x w )sd + x w x top + (1 cos )D / 2 (sw w )
All vertical pressures simply result from the soil continuum in Figure 16. Water pressure is omni
directional, so its horizontal pressure and vertical pressure are equal at any point. Soil pressures do
Just like the total vertical pressure, the total horizontal pressure combines effective horizontal grain
pressure and the hydrostatic water pressure:
h () = ' h ( ) + w () (8)
Using these equations its relatively easy to determine the total vertical soil pressure at the top of the
tunnel. This pressure results from the soil and water masses on top of the tunnel. Implying that the
vertical pressure at the bottom of the tunnel should equal the pressure at the top, for the tunnel hole
does not introduce any additional mass. However a point at the same depth just next to the tunnel is
loaded by the complete soil overburden and should therefore have a significant higher vertical soil
pressure. The vertical pressures at the points next to and below the tunnel are connected by the
horizontal pressure. This implies that an instable situation would occur. As a result equilibrium will
be generated resulting in a total pressure below the tunnel somewhere in between the original
pressures of both points.
A commonly adapted solution is given in [5]. Its assumed that the effective vertical grain pressure
along the full tunnel circumference equals the effective vertical soil pressure by the original
overburden at the centre point of the tunnel. The hydrostatic water pressure remains unchanged,
resulting in a pressure difference between the top and bottom of the tunnel. This generates a floating
pressure component, which has been observed in the construction of actual shield driven tunnel as
well.
Tunnel rings are defined in a cylindrical coordinate system because of their circular shape. Therefore
the vertical and horizontal soil pressures have to be translated into radial and tangential ones (Figure
17). For this purpose conversion equations from Appendix A.2 are used. They read:
r cos 2 sin 2 v
= (9)
t cos sin cos sin h
v()
r()
t()
h()
17 | Orientation of vertical and horizontal versus radial and tangential soil loads
Now the radial total ground pressure holds after substitution of equations (5), (7) and (8) in (9):
r () = ' v ()(cos 2 + K 0 sin 2 ) + w ()
This also can be written as:
1+ K0 1 K0
r () = ' v () + cos 2 + w ()
2 2
Loads from the soil are introduced by forces at nodes in the framework model. The (radial) pressure
is therefore converted to the (radial) force:
bl
Fr ( ) = r () n
2
Where l n is the influence length along the linings circumference for one node. At an arbitrary node
in the span of the tunnel segment this length is defined as the full length of a beam element, at end
nodes (in the longitudinal joints) its defined as half the beam element length.
Excavation of soil by the TBM will induce deformations in the soil. In sand these deformations will
occur at once. The adhesive property of clayish materials or sand with clay included will spread the
deformations over a longer period of time and affect the tunnel in its serviceability stage. Hence these
induced deformations will result in loading of the tunnel in tangential direction. The complete loss of
loads in tangential direction (in case of sand) or the presence of a part of the tangential loads is only
introduced by multiplying the full tangential load by a reduction ratio t. In case of sand this ratio is
set to zero. If deformations remain it holds that t > 0. For the Botlek Rail Tunnel, which has been
constructed in soil with more-or-less similar conditions to the soil of this reports case study, a value
of t = 0,25 has been applied [17]. Hence that particular value will be used here as well.
Reduction of the ratio to t = 0 causes a bending moment saving of no less than 12 %. This shows one
of the advantages of shield driven tunnel construction in uncontaminated sand.
The Oedometer stiffness depends on the soil properties only. Similar to the soil loads the soil springs
are attached to the tunnel segments nodes. The stiffness in equation (10) was given per unit of
surface, hence it has to be converted as follows:
The soil loading was represented by stresses on top, sides and bottom without any floating
component. The soil properties were:
Soil stresses top and bottom = 0,448 MPa
Soil stresses sides = 0,411 MPa
Oedometer stiffness = 38 MPa
In Slenders report another equation was used to translate the soil Oedometer stiffness into a spring
stiffness. That equation was commonly used before, but appears to be valid only for uniformly
loaded (hence without ovalisation) tunnel rings [5]. The equation was:
c soil ,r E oed
=
A R
Hence this equation results in a double spring stiffness compared to equation (10). In the model
about to be validated a fictitious double Oedometer stiffness of 76 MPa has been used to prevent
uneven results.
Resulting bending moments from both models corresponded very well. Figure 18 shows both
distributions of the bending moment and their local maximum values. At the sides and bottom of the
tunnel only small deviations up to 2,5 % occur. At the top however obvious changes of the bending
moment distribution take place due to the missing key stone. The deviation is still reduced to
approximately 4,2 % only.
50
Angle
0 []
0
0 30
33 -50
-76,5
-79,9 -100
0
60
30
BRT-model 87,2
270
90
Model for this study 84,9
0
24
12
0
-77,4 -76,5
21 0
0 15
180
The absolute maximum values of both distributions are located at the right-hand side of the model
and vary just 2,5 %. Its therefore concluded that the generated model shows good results and that
variations between models with and without the presence of a key stone are only very small.
Therefore no key stone is included in this study.
Both validations in this section demonstrate the model corresponds well to the results of existing
studies with variations of the maximum bending moments up to 2,5 % only. It is therefore assumed
that the models results are correct.
50
Angle
0 []
0
0 30
33 -50
-76,5
-78,2 -100
0
60
30
LDesign 86,8
270
90
Model for this study 84,9
0
24
12
0
-78,2 -76,5
21 0
0 15
180
19 | Moment distribution in validation model LDesign and model from this study
From the concrete material properties only the Youngs modulus and Poisson ratio are required for
the model. Lateral contraction, or the Poisson ratio, equals 0,2 for all considered concrete strength
classes C35/45, C100/115 and C180/210. The Youngs moduli vary and hold for each material:
EC35/45 = 33.500 MPa (NEN 6720)
EC100/115 = 45.000 MPa [6]
EC180/210 = 65.000 MPa (French interim recommendations [1])
These values hold for short-term uncracked concrete and therefore appear relatively conservative for
calculations on the ultimate limit state for the tunnels serviceability phase (more long-term than
short-term). Reduction of these values might be considered if calculations on the linings ring
behaviour appear to result in the governing situation. Therefore at the end of this report the value of
the Youngs modulus will be reflected once more.
Figure 20 shows the complete framework model for this studys shield driven tunnel. Both rings
were assigned a different diameter in order to clarify the picture. Note that the diameters in the
actual model are equal.
L2 R L1
R R
L2
L1
R R
L2
L1
R
R
L1
L2
R
R
L2
L1
Lx Longitudinal joint in ring x
R Ring interaction
R
R
L2
L1
R
R
L1
L2
R R
L1
L2
R R
L2 R L1
20 | Complete framework model for the shield driven tunnel of this study
From the Youngs moduli for the considered concrete strength classes in section 3.3.3 it follows that a
lining of ultra high strength concrete is 94 % stiffer at similar thicknesses compared to ordinary
concrete (65.000 MPa versus 33.500 MPa for uncracked moduli). At very thin lining thicknesses no
reduction by the mentioned asymptote is observed yet. At vary low thicknesses (approaching to 0
mm) resulting in an additional bending moment for C180/210 compared to ordinary concrete C35/45
of 94 % as well. However when the lining thickness increases it becomes clear that stiffer concrete
will reach the asymptote at an earlier level. The relative surplus for concrete materials with a higher
Youngs modulus descends if the lining thickness increases. Figure 21b visualises this trend by means
of a diagram with this relative surplus for C180/210 and C100/115 compared to C35/45.
20
200
The normal ring force is not influenced by the tunnels lining thickness. Or actually: it shouldnt be.
In the model however the depth projection is defined by a certain soil cover on top of the tunnel and
the internal diameter is fixed. So at an increasing lining thickness, the external diameter grows,
forcing the heart of the tunnel (where the loads are defined and hence the normal ring force is related
to) down by the additional thickness. The resulting slight increase of the normal ring force is
presented in Figure 22.
0 200 400 600 800 1000
-2200
Thickness h [mm]
-2300 C35/45
C100/115
-2400
C180/210
Normal ring force [kN]
-2500
C180/210
-2600
C35/45
-2700
Similar to the normal ring force (see Figure 23a), the maximum bending moment in the lining is more
or less directly proportional to the soil overburden on top of the tunnels heart. The size of the soil
loads (at average defined at that point) is responsible once more. For a standard lining thickness of
675 mm the maximum values have been given in Figure 23b, apparently the ratios between all three
concrete strength classes dont change. Therefore no additional deviations of the maximum bending
moments occur for the three concrete strength classes an increasing depth. Except for the increasing
absolute values of course, as Figure 24 demonstrates.
C100/115
54,8 (2,9D) 54,8 (2,9D)
C35/45
a b
23 | Effect of tunnel depth projection with a lining thickness of 675 mm. a) normal ring force; b) tangential bending moment
Maximum bending moment [kNm]
3000 3000
C180/210
4000
C180/210
3000
C100/115
2000
C35/45
1000
Thickness h [mm]
0
0 200 400 600 800 1000
c
24 | Maximum bending moment by ring behaviour for soil overburden of: a) 1,2D; b) 2,9D; c) 4,6D
This section will demonstrate that the acting normal force has a positive effect (up to a certain depth.)
on the linings bending moment capacity Therefore the representative normal force Nrep has been
introduced, that does not include any safety factors. Minimum values resulting from calculations on
the embedded ring behaviour have been applied.
0,389 xu
xu
Nc
h
Mu
Nrep
Ns
c+/2
The equation is based on the mechanism shown in Figure 25. Where the tensile force in the steel
rebars is defined as:
N s = A s fs
Where:
As = Steel area of rebars
fs = Representative strength of reinforcement steel
Scalars 0,389 in the basic equation for Mu and 0,75 in the determination of xu hide the fact that this
calculation procedure assumes that the concrete is not able to cope with any tensile forces and that its
ultimate compressive strain is twice the compressive yield strain. Subsequently the procedure
assumes the strain in the concrete cross-section at the position of the rebars is of such tensile extent
that the maximum tensile strength from the stress-strain diagram (see Figure 26) is reached.
For calculations on the ultimate resisting moment of tunnel segments in this study a more
comprehensive calculation procedure is required. Several properties of the described simplified
procedure disagree with the following requirements:
Tensile stresses from the concrete should be introduced for the steel fibre reinforced
concrete materials very and ultra high strength concrete.
The ratio between the ultimate compressive strain and compressive yield strain of concrete
should be free, since this ratio does not equal 1:2 for very and ultra high strength concrete.
450
300
150
Strain []
-30 -20 -10 0 10 20 30
-150
-300
-450
Additionally to the enumeration the tensile behaviour of ultra high strength steel fibre reinforced
concretes turns out to be rather complicated. Consequently another procedure is required to include
all four requisites in the calculation of the ultimate resisting moment. The calculation will be altered
to make the stress-strain relations of concrete and steel the base of the entire determination.
Therefore the same principles from the simplified determination of the building code are applied
again, meaning the equilibrium of horizontal forces and the equilibrium of bending moments:
ns
Fh = 0 Nc + Fsj = Nrep
j=1
ns
M = 0 M u = M c + M sj M N
j =1
Where:
Fsj = Tensile steel force for reinforcement layer j
Mc,sj or N = Contribution to the ultimate resisting moment by the concrete (c), steel
reinforcement layer j (sj) or the normal ring force (N)
ns = Number of steel reinforcement layers
Obtaining the ultimate resisting moment Mu from these equilibrium equations demands a vast
number of identical calculations, which makes working out by hand a very uninteresting activity. For
that reason a Visual Basic .NET2 application has been implemented to do the job. In Appendix B the
calculation process of this application has been reproduced.
Stress-strain diagram for very and ultra high strength steel fibre reinforced concrete
The French interim recommendations on structures in ultra high strength concrete [1] also contained
a stress-strain relation or diagram for this special material. Two different effects by the steel fibres
were presented: strain hardening (Figure 27a) and strain softening (27b). Their use will be discussed
at a later stage.
E E
Calculations on the ultimate limit state of the structure demand an additional safety factor on strain
el and the tensile stresses fc and f1% according to the French recommendations. This partial safety
factor for fibres holds: f = 1,3
The value of stress fc defines whether the concrete behaves with strain hardening or strain softening:
only if fc > fel strain hardening will occur. The phenomenon of strain hardening is induced by the
steel fibres. The effect on the total ultimate resisting moment however is only extremely small and
may therefore be neglected in this calculation.
The product BSI-Cracem by Eiffage is applied for ultra high strength steel fibre reinforced concrete
C180/210 in this study. Its stress strain relation is determined by use of several parameters in the
interim recommendations:
f'c = 126 MPa
fc = fel= 9,1 MPa [1]
f1% = 7 MPa (value has been scaled from values used in [9])
E = 65.000 MPa
lf = 20 mm (amount of fibres 200 kg/m or 2,5 vol-%) [19]
For very high strength steel fibre reinforced concrete C100/115 a mixture described in [6] is used,
which principal values read:
f'c = 69 MPa
fc = fel= 5 MPa
f1% = 4 MPa (scaled from C180/210)
E = 45.000 MPa
'c = 1,75 (this value from the building code is larger than the one resulting from
equation (13) and therefore assumed critical)
lf = 40 mm (amount of fibres 125 kg/m or 1,6 vol-%)
Stress [MPa]
Stress [MPa]
-80 -120
-100
-60
-80
-40 -60
-40
-20
-20
Strain [] Strain []
30 20 10 0 -10 15 10 5 0 -5
b a
28 | Stress-strain diagrams steel fibre reinforced concretes a) C100/115; b) C180.210
Enlargement of the lining thickness results in an increase of the lever arms for reinforcing steel (h c
/ 2) and the contribution by the normal ring force (h / 2). This obviously results in a higher
Provided that no reinforcement bars are used, as is most desirable from the point of cost saving for
very and ultra high strength steel fibre reinforced concrete, the contribution by the reinforcing steel
to the total bending moment capacity is obviously omitted. Nevertheless these concrete materials are
able to resist some tensile stresses in their own cross-section. But as a result of the stress-strain
relation (see Figure 28) the value of these tensile stresses decreases if the distance from the concrete
compression zone increases. Hence if the lever arm for the contribution to the ultimate resisting
moment increases, the tensile stress actually decreases. On top of this the height of the tensile zone
decreases if the lining thickness increases (because of equation (12) on page 34), causing a smaller
positive contribution in stead of a larger one as in case of deep beams or linings. This negative
property undermines the contribution of steel fibres to the bending moment capacity, which is
thereby relatively inadequate.
An advantage of very and ultra high strength concrete should now exist of the capability to cope
with compressive normal forces in a narrower concrete compression zone compared to ordinary
concrete. By this means the negative contribution to the ultimate resisting moment is restricted. At
equal acting normal forces the height of the concrete compression zone remains almost unchanged
(some distortion occurs by the length of the tension zone). The contribution to the bending moment
capacity is thereby independent of the lining thickness. The most significant advantage of a small
compression zone is subsequently reached if the contribution by rebars in ordinary concrete is
relatively small because of a small lever arm, hence if the lining thickness is low. In that range the
length of the tension zone is larger and therefore more effective as well.
As soon as the lengths of both the concrete compression and tension zone are very small compared to
the lining thickness, the lever arm of the normal ring force only causes the bending moment capacity
to grow. Implying that at relatively low normal ring forces and high lining thicknesses the resulting
ultimate resisting moment for ultra high strength concrete might be smaller than the one for ordinary
concrete with reinforcement bars. For these rebars a steel quality of FeB435/500 and reinforcement
percentage of 0,22 % (minimum requirement for concrete C35/45 according NEN 6720) on top and
bottom of the cross-section are used, with a concrete cover of 35 mm and bars of 12 mm.
In Figure 29 the described phenomena are indeed observed for the most shallow depth projection.
Remarkably the ultimate resisting moment for C100/115 surpasses the one of C180/210 from
approximately 300 mm on. The higher compressive strength of the latter copes with the acting
normal force in a smaller concrete compressive zone. The absolute height of the compression area
defines the absolute height of the tensile region since they are interconnected in the stress-strain
diagram. Consequently the contribution by the tensile force is inadequate at smaller lining
thicknesses for C180/210, causing the bending moment capacity to approach the contribution by the
normal ring force only at an earlier stage (dashed line in Figure 29).
2000
C35/45
1500
1000 C180/210
C100/115
500 C180/210
Contribution normal force
Thickness h [mm]
0
0 200 400 600 800 1000
29 | Ultimate resisting moment at a depth of 0,7D
If the acting normal force Nrep (the minimum normal ring force) in the cross-section increases, the
height of the concrete compression zone is primarily manipulated. By its enlargement the variations
of the negative contributions for the considered concrete strength classes will rise. This is in favour of
very and ultra high strength concrete. As soon as this favour exceeds the contribution ordinary
concrete gets from its reinforcement bars, the bending moment capacity will be higher for ultra high
strength concrete even at large lining thicknesses.
A larger concrete compressive zone of course implies a larger height of the tensile zone in fibre
reinforced concrete as well. Subsequently higher tensile stresses occur at relatively large distances
from the compressive zone. These higher tensile stresses have a positive contribution to the
comparison of higher strength concrete and ordinary concrete as well.
The indisputable positive effect of the normal ring force or tunnel depth projection on the bending
moment capacity of ultra high strength concrete is shown in Figure 30. The capacities of all
considered concrete strength classes and depth projections are presented for a standard lining
thickness of 675 mm. Concrete C180/210 commences with the smallest capacity of all at a depth of
11,1 m (0,7D) (normal ring force approximately 2.550 kN) because its tensile zone is inadequate
already; at a depth projection of 74,2 m (4,6D) (normal ring force 11.100 kN) however it grew to own
the largest capacity, exceeding the one of reinforced concrete C35/45 by no less than 120 %.
Figure 31 displays the development of the bending moment capacity at increasing lining thicknesses
for the three remaining depth projections. The fact that the increment of the compressive zone at high
normal forces (depths) also causes the tensile zone to grow, is visible by the fact that the capacity of
C100/115 exceeds the one of C180/210 at ever higher lining thicknesses.
19,2 (0,7D)
27,4 (1,2D) C100/115
C180/210
82,2 (4,6D)
30 | Development ultimate resisting moment at various depth projections for lining thickness of 675 mm
4000 4000
C100/115
3000 3000
C35/45
2000 2000 C180/210 C35/45
C100/115
1000 C180/210 1000
5000
4000 C180/210
3000 C100/115
C35/45
2000
1000
Thickness h [mm]
0
0 200 400 600 800 1000
c
31 | Ultimate resising moment at depth projections of: a) 1,2D; b) 2,9D; c) 4,6D
The preceding discussion obviously demonstrates the main advantage of ultra high strength
concrete. Not the materials tensile strength, but its capability to cope with enormous normal forces
in a very small cross-section is positive. Therefore all present bridges constructed from ultra high
strength concrete have been prestressed to a very high extent. So actually its not the concrete itself,
but the large prestress force that generates a high bending moment capacity for thin beams.
5 5
C100/115
4 C100/115 4
C180/210
3 3
C35/45 C180/210
C35/45
2 2
= 1,5 = 1,5
1 1
Thickness h [mm] Thickness h [mm]
0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
a b
Safety factor Mu / Mmax
6 6
5 5
4 4
C180/210
3 3
C100/115
C100/115
2 2
= 1,5 = 1,5
1 C35/45 C180/210 1
C35/45
Thickness h [mm] Thickness h [mm]
0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
c d
32 | Development of safety factor Mu/Mmax at depth projections of: a) 0,7D; b) 1,2D; c) 2,9D; d) 4,6D
At shallow depths its no surprise the safety factor of ultra high strength concrete C180/210 is lower
than the one of ordinary concrete C35/45, particularly in thick linings. The reason of course is the
small value of its bending moment capacity due to the relatively low acting normal force. Figure 32a
indeed demonstrates this phenomenon. The other graphs in Figure 32 also show that the
performance of high strength concretes improves compared to ordinary concrete at higher depths. At
a soil cover of 74,2 m (4,6D) ordinary concrete C35/45 is unable to reach the required safety level. Up
to a certain level this problem is relatively easy tackled by applying more steel bar reinforcement.
As mentioned the trend of the safety development for the bending moment in embedded ring
behaviour is downward at an increasing lining thickness. Hence the lining thickness where the
required and occurring safety levels meet is an upper boundary condition for the lining thickness. In
other words: a required lining thickness that apparently follows from other calculations may not
exceed the requirement from this mechanism. For the considered amount of reinforcement in
ordinary concrete, a lower boundary condition is present at high depths as well. The lining is unable
to reach the required safety level.
Values of the upper boundary conditions have been shown in Table 4 for the considered concrete
strength classes and soil overburdens.
Of course these upper boundary conditions do not imply that a solid tunnel (or a tunnel with an
extremely large lining thickness) leads to failure by the embedded ring behaviour. As mentioned
before the resulting bending moment in the lining will reach a maximum value at some point.
Table 4 | Upper boundary conditions on the allowable lining thicknesses in case of embedded ring behaviour
3.5 Conclusions
Resulting normal ring forces and bending moments in the tunnel ring enlarge linear proportional to
the depth of the tunnels centre line. The bending moment significantly increases if the lining
thickness ascends.
Alterations of the tunnel rings design details, such as extension of the number of tunnel segments
per ring, result in a very small reduction of the bending moment only.
Steel fibres in very and ultra high strength concrete hardly contribute to the total bending moment
capacity (or ultimate resisting moment). Thereby the bending moment capacity of very and ultra
high strength steel fibre reinforced concrete strongly depends on the acting normal ring force.
The thicker a lining of very or ultra high strength concrete, the smaller the contribution of steel fibres
to the bending moment capacity becomes. As a result thin linings with a high acting normal ring
force (hence at large depths) provide the best use of these materials capacities concerning normal
forces and bending moment capacity when compared to reinforced ordinary concrete.
In very shallow tunnels (overburden of 11,1 m (0,7D)) with relatively thick linings (over 400 mm) the
bending moment capacity of reinforced ordinary concrete C35/45 grows bigger than the one of ultra
high strength concrete C180/210. Lacking of steel reinforcement bars in the latter is responsible for
this.
The linings bending moment safety level the safety margin to cope with the resulting bending
moments for ring behaviour of an embedded tunnel decreases if the lining thickness ascends.
Consequently the thickness where the required safety level is hit is the maximum value that can be
applied; hence an upper boundary condition of the required lining thickness. Reinforced ordinary
concrete C35/45 has no upper boundary condition in the considered range of lining thicknesses.
In very deep tunnels the high normal ring force reduces the bending moment capacity of reinforced
ordinary concrete that is now unable to reach the required safety level. Therefore a lower boundary
condition for embedded ring behaviour appears for this material at great depths.
All boundary conditions on the required lining thickness by embedded ring behaviour have been
displayed in Table 4 and Figure 33.
19,2 0,7D
27,4 1,2D
C180/210 C100/115
54,8 2,9D
C35/45
82,2 4,6D
33 | Upper boundary conditions on the allowable lining thicknesses in case of embedded ring behaviour (lower boundary
for C35/45)
4.1 Introduction
Tunnel rings are assembled within the shield of the tunnel boring machine (TBM). Therefore its
diameter is larger than the tunnels external one. Behind the machine a tail void between the soil and
tunnel is created during the excavation process. In order to prevent loss of soil support and possible
settlements at ground level, the semi-liquid material grout is injected in the void.
Grout that consists of sand, water, potentially cement and several additives like bentonite or
plasticizers has some special properties that make it suitable for its specific task. The material is not
actually liquid, but has a very low plastic yielding shear strength (viscosity) in the beginning. In
other words: the material has a low resistance against flowing. If it flows (visco-plastic behaviour) it
can easily be transported and injected into the void. Nevertheless if the yielding shear strength has
not been exceeded yet, the material is able to resist some shear stresses, which is not the case for a
liquid.
If grout is subjected to a compressive pressure (ground and water pressure in this case) for some
time, water will be squeezed out of the material and therefore the viscosity increases, thus the
material stiffens. Over time it will turn into a stiff shell making interaction of the tunnel rings and the
surrounding soil possible.
As long as the grout didnt harden yet the tunnel would like to float up in the semi-liquid material.
The support of the tunnel then completely changes and is no longer comparable to the principal
applied in Chapter 3 for a tunnel embedded in soil. Interaction between soil and lining is not acting
in fresh grout.
In Bloms dissertation [5] came forth that the bending moment in the grouting phase, or uplift
loading case, might play a key role in the determination of the required lining thickness. The
resulting bending moment of this loading case appeared to barely depend on the linings thickness.
However, because the bending moment capacity remains unchanged compared to the calculations in
the previous chapter, the safety factor for this loading case will obviously ascend. In Bloms study the
required safety level for the Botlek Rail Tunnel was reached at a lining thickness of approximately
1/20 D.
So if the lining thickness 1/20 D of a tunnel with ordinary reinforced concrete is determined by the
fact that its ultimate resisting moment equals the design value of the bending moment of the
grouting phase, this might help to retrieve the required lining thicknesses for another concrete
strength class. Implying the other way round it should also hold that the bending moment capacity
of another type of concrete with a certain lining thickness h equals the bending moment capacity of
ordinary reinforced concrete with a lining thickness of 1/20 D. Now the required thickness h can be
resolved.
The analysis on the Botlek Rail Tunnel in Bloms study has been performed for an overburden of
approximately 1,4D. The question should be raised whether this hypothesis applies to deeper tunnels
as well. Subsequently the effects on the ultimate resisting moment for the tunnel lining of motorway
43
A13/16 have been inquired at an increasing normal ring force, hence at an increasing depth.
Although the tunnel may no longer be in full contact with the soil, the normal ring force should still
make equilibrium with the ground pressure and therefore develops identical to the normal ring force
in the embedded situation.
For an unreinforced cross-section with concrete unable to cope with any tensile forces, the
equilibrium of horizontal forces holds according to the simplified calculation method for the ultimate
resisting moment on page 31:
Fh = 0 Nc = Nrep
At an increasing acting normal force on the cross-section the height of the compressive zone will
grow directly proportional to that force:
Nc Nrep
xu = =
0 ,75f' b b 0 ,75f' b b
Its negative contribution to the bending moment capacity therefore grows with a parabolic
proportional rate:
0 ,389 N rep 2
M c = N c 0 ,389 x u =
0 ,75f' b b
The positive contribution by the normal force itself obviously ascends with a directly proportional
rate only:
h
M N = N rep
2
Hence the total ultimate resisting moment is a combination of these parabolic and linear equations:
0 ,389 N rep 2 h
Mu = Mc + M N = + Nrep
0 ,75f' b b 2
At relatively low normal forces the linear part of this equation is predominant and the capacity will
rise if the normal force increases. However at some point a turning point is reached and the capacity
will drop due to the parabolic part. For this simple case that point is located at 8.785 kN for concrete
C35/45. This is clearly visible in Figure 34 as well, where line (a) represents unreinforced concrete.
A reinforcement layer at the bottom of the cross-section will elevate the ultimate resisting moment
(as described in last chapter). If however the normal force reaches such high values that the tensile
stress in the reinforcement layer drops or even turns into compression, the capacity will consequently
decrease. At line (b) in Figure 34 this provokes ordinary concrete with a reinforcement percentage of
0,22 % to descend from approximately 7.900 kN.
0 250 500 750 1000 1250 1500 2000
0
Ultimate resisting moment [kNm]
0D
2000 0,7D
1,2D 1D
4000
Estimated tunnel depth from top [m/m]
(a)
(b)
2D
6000
2,9D
3D
8000
Normal ring force [kN]
14000 6D
34 | Development ultimate resisting moment over the depth for 675 mm thick reinforced concrete C35/45
In the introduction to this chapter it was mentioned that grout has some capability to cope with shear
stresses. Only if this capacity is exceeded grout will flow. At that moment the so-called shear yield
strength is reached. If the shield driven tunnel moves upward due to the floating component the
grout shear stress will grow till it reaches its maximum value. By addition of its downward
component to the tunnels dead weight it follows:
DW + grout ,tan gential ,vertical = Dhb concrete + Db yield
Where:
DW = Dead weight tunnel
grout ,tan gential ,vertical = Vertical component of tangential frictional shear stress between tunnel and
grout
D = External diameter tunnel
h = Thickness tunnel lining
b = Width tunnel segment
concrete = Specific gravity concrete
yield = Shear yield strength grout (0,0015 MPa [5])
Plastic shear
yield stress grout
Grout surrounds the tunnel; therefore its this material that transmits the water pressure to the
tunnel. In addition the mentioned contact between tunnel lining and grout influences the pressure in
the grout as well. The product of both, the actual grout pressure, now also includes the floating
component and should consequently equal the downward directed forces. In Figure 35 this principal
has been visualised. The definition now holds:
grout ,radial , vertical = DW + grout , tan gential , vertical
Where:
grout ,radial , vertical = Vertical component of the radial grout pressure, consisting of:
Water pressure
Lining-grout contact pressure
Hence the upward component of the total grout pressure equals the downward components by the
tunnels dead weight and the tangential loading of the tunnel by the grouts shear stresses. As a
result for the vertical component of the grout pressure it holds:
grout ,radial ,vertical = Dhb concrete + Db yield
Provided that the grout pressure develops as a hydrostatic pressure along the lining, it may be
written as:
2
grout ,radial , vertical = 4 D beq
Where eq is an equivalent specific gravity assigned to the grout. Accordingly it holds:
Dhb concrete + Db yield h concrete + yield
eq = =4
2 D
D b
4
Assuming ordinary concrete C35/45 (concrete = 24 10-6 N/mm), a lining thickness of 675 mm and
grout with a shear yield stress of 0,0015 MPa the equivalent specific gravity holds for the tunnel of
motorway A13/16:
675 24 10 6 + 0 ,0015
eq = 4 = 4 ,1 10 6 N / mm 3 = 4 ,1 kN / m 3
(14.900 + 2 675)
This implies that vertical equilibrium occurs if the specific gravity of the floating component in grout
is lowered from the specific gravity of water w = 10 kN/m to the equivalent specific gravity eq = 4,1
kN/m. Now the tunnels dead weight and the grout shear stresses are able to compensate the
remaining upward component of eq. According to Bloms dissertation this reduction of the upward
pressure has been observed in measures on site as well. Besides, finite element models where a
tunnel embedded in a uniform soil continuum is loaded by the hydrostatic water pressure return the
Having a closer look on the grout pressures resulting from comprehensive finite element analyses
shows that the grout pressure does not develop perfectly hydrostatic. A pressure peak arises on top
of the tunnel. This behaviour will be explained by analysing the development and creation of the
grout pressure:
If grout has just been injected in the tail void along a certain tunnel ring, this ring is not yet moving
upward. The considered ring has only recently left the tunnel boring machine and the next ring, that
is still within the machine, prevents any vertical movement. Though as soon as the ring does move,
the grout shear stress builds up. At the top of the tunnel grout is compressed between the moving
tunnel and fixed soil. The shear yield strength has not been reached yet; hence grout is not able to
flow away from the top. As a result the grout pressure above keeps increasing due to interlocking.
Only if the shear yield strength is exceeded somewhere, grout at that particular location is able to
flow from the top to the sides and bottom of the tail void. Because of the tunnels upward movement
a relatively low grout pressure was generated in the void at the bottom of the tunnel. By flowing
grout this phenomenon is partly compensated.
Provided that the grout-lining contact pressure behaves like a uniform bedding, a local pressure
increase is simulated by a local enlargement of the beddings stiffness, an additional fictitious
bedding. In his study Blom presumed that this additional stiffness diminishes from the tunnels top
to its sides in accordance with a quadratic cosine distribution. His study also shows that the
magnitude of the pressure increment is related to the soil overburden, the soils specific gravity and
the tunnels external radius. Consequently the stiffness of the fictitious top support is defined as:
k fictitious = cos 2 () k uniform
Where:
h s s
= 1
R w
With:
hs = Overburden soil
kuniform = Stiffness uniform bedding
R = External radius tunnel
s = Specific gravity of soil
w = Specific gravity of water
= Angle along tunnel circumference (0 at the top)
At the standard depth with an overburden of 19,3 m (1,2D) the -ratio for the tunnel from this study
holds:
19 ,3 20
= 1 = 3,75
14 ,9 / 2 + 0 ,675 10
If the stiffness of the uniform bedding for the grout-lining contact pressure is set to a very low value
of for instance 1 MPa, the pressure is able to develop smoothly without significant bending moment
reduction due to ring behaviour as described in the previous chapter. Now a value of 3,75 cos(0) + 1
= 4,75 MPa holds for the total bedding stiffness of the top support , which is the sum of the fictitious
and uniform bedding (see Figure 36).
1
Angle []
0
0 45 90 135 180 225 270 315 360
This calculation procedure (implemented in LDesign) has been applied on the Green Heart Tunnel in
Bloms study. The resulting total grout pressure, consisting of the water pressure and grout-lining
contact pressure, was practically similar to the pressure resulting from finite element models.
Bending moments in the tunnel lining result from this grout pressure in LDesign.
Also in case of complete grouting, subtraction of the water content from the grout is of main
importance. Hardened grout will generate a stiff skeleton that supports the tunnel and hence makes
cooperation of the soil and tunnel to bear the ground and water pressures possible. From that
moment on ring behaviour of the embedded tunnel lining takes place. The porosity of sand subtracts
the water from the grout relatively easy. In more dense or saturated soils like clay and peat a long
period of time is required to squeeze out all water and therefore the grouting phase will stay active
for a longer period as well.
The combination of both described mechanisms on the grout bending moment is displayed in Figure
37 for a tunnel depth of 11,1 m (0,7D). The figure shows the resulting moments for complete and
incomplete grouting. Up to a lining thickness of approximately 300 mm the stiffness ratio between
the concrete tunnel lining and fictitious top support for the grout-lining contact pressure obviously
influences the maximum bending moment. The type of concrete with the highest Youngs modulus,
C180/210, returns the largest moments now. However from a thickness of approximately 600 mm its
the dead weight of the tunnel lining that causes the bending moment to descend now with an almost
linear pace. As a result the concrete materials with similar specific gravities, C35/45 and C100/115,
return the same maximum bending moment, independent of their unequal Youngs moduli.
Grout bending moment [kNm]
C100/115
Incomplete grouting
800
C35/45 C180/210
600
200 C180/210
C35/45
Thickness h [mm]
0
0 200 400 600 800 1000
37 | Bending moment by complete and incomplete grouting at a depth of 0,7D
Variations between resulting bending moments for complete and incomplete grouting are a result of
the difference in bedding stiffnesses at the top and bottom for both situations. The differences are
induced by the -factor, which reads:
h
= s s 1
r w
The stiffness ratio between the top and bottom support holds ( + 1) : 1 for complete grouting and
holds : 0 for incomplete grouting. If the soil cover on top of the tunnel increases, the -factor
increases as well. As a result the stiffness ratio for complete grouting approaches the one of
incomplete grouting more and more. Hence the variations in the bending moments for both
situations will decrease at an increasing depth.
At the same time the stiffness ratio for incomplete grouting between the fictitious top support and
the tunnel lining changes as well. The top bedding stiffness increases compared to the lining stiffness.
The resulting bending moment for incomplete grouting therefore reduces at larger depths.
The result is shown in Figure 38. The bending moment by complete grouting (line a) does approach
the descending bending moment by incomplete grouting (line b) more and more as predicted.
10000 4D
4,2D
12000 (a) Complete grouting 5D
(b) Incomplete grouting
14000 6D
(c) Mu,1/20D /
38 | Development bending moment by complete and incomplete grouting over the depth for 675 mm C35/45
The introduction to this chapter discussed the hypothesis that the bending moment capacity of a
lining in reinforced ordinary concrete with thickness 1/20 D might represent the bending moment by
grouting. The exceptional development of the capacity at increasing depths however resulted in the
conclusion that this might not apply to all tunnel depths. Therefore the bending moment capacity has
been shown in Figure 38 once more. Now to verify the assumption (note that the capacity has been
divided by the applied safety factor of 1,5 on resulting bending moments). It turns out that the
bending moment capacity indeed doesnt represent the grout moment at each depth. Especially at
Grouting bending moment [kNm]
C100/115
Incomplete grouting Incomplete grouting
C100/115
800 800
C35/45 C180/210
C180/210 C35/45
600 600
C100/115
400 400
C180/210 C35/45 Complete grouting
Incomplete grouting
800 C100/115
C35/45
600 C180/210
C100/115
200
C180/210
Thickness h [mm]
0
0 200 400 600 800 1000
c
39 | Bending moment by complete and incomplete grouting at depths: a) 1,2D; b) 2,9D; c) 4,6D
For the tunnel of motorway A13/16 the resulting bending moments by complete and incomplete
grouting have been given in Figure 39. Only the remaining depth projections have been shown, refer
to Figure 37 for the results of the most shallow projection.
C180/210
6 Complete grouting 6
5 5 Complete
grouting
4 C100/115 4
C180/210 Incomplete
3 grouting 3 Incomplete
grouting
2 2
= 1,5 = 1,5
5
11
1 1 C35/45
0/
C35/45
10
0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
c d
40 | Bending moment safety factor for complete and incomplete grouting at depths: a) 0,7D; b) 1,2D; c) 2,9D; d) 4,6D
The resulting lower boundary conditions by complete and incomplete grouting for the required
lining thickness have been listed in Table 5.
Table 5 | Lower boundary conditions on the allowable lining thicknesses in case of complete and incomplete grouting
4.4 Conclusions
The bending moment capacity of a standard lining thickness of 1/20 D is not representative for the
resulting bending moment by grouting at all considered depth projections in a tunnel lining of
reinforced ordinary concrete C35/45. Especially at relatively large depths the bending moment is
overestimated. Consequently slightly more slender lining thicknesses might be applied at these
depths if grouting appears to be governing.
The grout pressure in the tail void between tunnel and soil knows a local pressure increase on top of
the tunnel. The bending moment by grouting is defined by the extent in which the tunnel is pushed
into this pressure increment (described as fictitious top support) by the upward floating components
Incomplete grouting results in significantly higher bending moments in shallow tunnels compared to
complete grouting. However if the tunnel depth increases, the bending moment by complete
grouting approaches the value of incomplete grouting more and more. The absolute value of the
latter will somewhat descend at greater depths.
The resulting bending moment is only slightly dependent on the linings thickness. The stiffness of
the tunnel rings, such as the concrete Youngs modulus and the number of tunnel segments per ring,
do not considerably influence the resulting bending moments.
The weight of the tunnel ring however decreases the net upward floating force and therefore reduces
the extent in which the lining is pushed in the fictitious top support. Especially at high lining
thicknesses this results in a considerable reduction of the resulting bending moment for the heaviest
of all considered concretes, ultra high strength concrete C180/210.
Very and ultra high strength concrete require only very small lining thicknesses at high depths to
cope with the bending moment by complete and incomplete grouting.
At ascending lining thicknesses the bending moment safety level for both complete and incomplete
grouting increases. The lining thickness at the required safety level is therefore a lower boundary
condition of the final required thickness. Table 5 and Figure 41 show all boundary conditions.
19,2 0,7D
27,4 1,2D
82,2 4,6D
41 | Lower boundary conditions on the allowable lining thicknesses in case of complete and incomplete grouting
5.1 Introduction
Shield driven tunnels are constructed within the protection of the shield of a tunnel boring machine
(TBM). After assembly of a complete ring, the TBM moves forward by pushing its thrust jacks on the
bearing pads of the newest assembled ring. The high thrust jack forces that are hereby introduced in
the lining result in tensile bursting forces deep in the material and in high compression stresses
under the thrust jack plate. In section 5.2 the relation between the linings thickness and the
introduction of thrust jack forces is investigated.
Moving forward of the tunnel boring machine also causes the one-but-last ring of the lining to leave
the protection of the shield. Now the ring is loaded by the surrounding grout pressure and will start
to deform. The latest ring however is still covered by the shield; hence no loading is present to
initially deform it. As a result tunnel rings close to the TBM are subjected to torque in order to change
the initial circular shape into an ovalised one by the loading with semi-liquid grout. This torsion
phenomenon is also known as the trumpet effect and will be discussed in section 5.3.
In case of the French thrust jack configuration this system will occur just like described. By the
applied positioning of thrust jacks on the segment, each jack plate is located at the exact centre line of
its own half of a segment (two thrust jack plates per segment). Now the thrust jack force of each jack
plate is able to spread over the width of its own segmental half and the tensile bursting stresses will
appear as described. The high force introduced and the lateral contraction property of concrete result
in deformations of the segmental shape close to the thrust jacks (see Figure 44a). As a result tensile
stresses will appear in the deformed sections and cracks might occur. However if cracking does occur
the deformation is no longer restricted and the tensile stresses will fade away.
55
disturbance height
compression
schematisation
tension
bursting force
compression
trajectories
a b
42 | Development of bursting stresses. a) Compression trajectories; b) Latice analogy
The German thrust jack configuration has no such decent spreading of compression trajectories over
the full segment. The thrust jack plates in the middle of the segments do not result in any additional
tensile stresses (hence no problems for concrete). The jack force introduced on the edge of the
segment however, over the longitudinal joint, is unable to spread symmetrically to two sides and no
compensating tensile bursting force can be created. Still the force will spread over the only side thats
present. As a consequence of the changed direction of the
trajectories a completely different mechanism to ensure
horizontal equilibrium of forces is introduced. Deep in the
segment a compression strut links the forces of the jacks on the
tension
edge and in the middle. In between the actual introduction spalling force
locations, so just under the segments front face, a tensile
spalling force is created (see Figures 43 and 44b). This tensile
force is a structural force needed for equilibrium, so cracking compression
will not decrease its magnitude. Serious cracking with large 43 | Tensile force in latice analogy for
crack depths may therefore result in the German thrust jack German thrust jack configuration
configuration.
Obviously the French configuration is more effective in spreading the introduced thrust jack forces
from the TBM. This fact contributed to the decision made in Chapter 3 to apply this particular
configuration in the tunnel of motorway A13/16.
This section will discuss the resulting high compressive stresses under the thrust jack plates and the
tensile bursting stresses deeper in the segments by the introduction of thrust jack forces for the
French configuration. First of all the magnitude of this thrust jack force should be determined.
spalling stresses
spalling stresses bursting stresses
bursting stresses compression
French German
Tension Compression
a b
44 | Stress distribution in tunnel segments due to introduction of thrust jack forces in the: a) French and b) German thrust
jack configurations [5, 18]
Most properties for the soil have been discussed in previous chapters already, such as for the
calculations on ring behaviour of an embedded lining. Additionally a friction coefficient between the
soil and TBM shield is required (value of 0,45 for sand follows from tables in [12]), as well as so-
called Ky,front and Ky,side coefficients, which both equal the average of the active and neutral soil
support coefficients Ka and K0:
1 sin
Ka = Ka + K0
1 + sin K y ,front = K y ,side =
K 0 = 1 sin 2
The internal friction angle = 30 was already mentioned in earlier calculations in this report. Hence
it holds:
0 ,33 + 0 ,50
K y ,front = K y ,side = = 0 ,42
2
Dimensions of the tunnel boring machine for the shield driven tunnel of motorway A13/16 are not
known yet. Therefore values from the machine of the Botlek Rail Tunnel from De Rijkes study are
used to make an estimation. Just like in the mentioned machine the external radius of the TBM is 250
mm larger than the one of the shield driven tunnel. The length of the excavation chamber is 2 m and
the length of the shield is 9,9 m. De Rijke also proves that a reduction of the shield length results in a
significant decrease of the actual thrust jack forces. For now however it is not guaranteed that such
shortening can be achieved; hence for this study the conventional length of 9,9 m is assumed.
For a description of the TBM Shield Equilibrium Models principal is referred to the study by De
Rijke [12]. This report will show the final part of the calculation only. Here the principal horizontal
force Hjacks,total and the bending moment Mjacks,total on the machines face for all thrust jack forces
together, which is a result of the mentioned model, are converted to individual thrust jack forces. The
resulting principal forces for the considered tunnel with a lining thickness of 675 mm and a depth
projection of 19,3 m (1,2D) are:
H jacks , total = 1,34 10 5 kN; M jacks, total = 2 ,02 10 4 kNm
These reaction forces can be converted to a distributed load along the tunnels circumference. The
values of this load on the tunnels top and bottom are given by:
M jacks , total H jacks , total
q bottom =
R jacks 2 2 R jacks
H jacks , total
q top = q bottom
R jacks
Where Rjacks is the radius of the jacks centre line, which equals the radius of the centre line for the
tunnel segments. Consequently Rjacks holds for the applied lining thickness:
R jacks = 12 (Di + h ) = 21 (14 ,9 + 0 ,675 ) = 7 ,79 m
qtop
I
A
B
qmean
y,q
G
D
E
qbottom
45 | TBM Shield Equilibrium Model with line load to determine thrust jack forces
The mean value of the distributed load along the tunnels circumference (qmean) is easy to determine,
just like the variation between the top and bottom values (q):
( )
q mean = 12 q top + q bottom = 2,73 10 3 kN/m
q = q bottom q top = 212 kN/m
At each arbitrary angle in the linings circumference a cumulative force based on the distributed
load can be defined. This cumulative force is given by:
F( ) = R jacks q mean 21 R jacks q cos
If such a force is determined at each longitudinal joint of one individual tunnel segment, the resulting
force by both thrust jack plates on that particular segment is:
Fsegment = F2 F1
Where:
F1 = R jacks q mean 1 12 R jacks q cos 1
F2 = R jacks q mean 2 12 R jacks q cos 2
The axial force on one group of thrust jacks (one thrust jack plate) in this segment is now simply
defined as:
Fgroup = 12 Fsegment
Letters have been appointed to all tunnel segments in Figure 45. Table 6 displays the determined
values and calculation principal of the actual thrust jack forces for the tunnel segments based on the
described method.
According to the TBM Shield Equilibrium Model the maximum thrust jack force is 7.700 kN for this
tunnel configuration at a depth of 19,3 m (1,2D). Including a safety factor jack = 2 it now holds [4]:
Fjack = jack Fgroup = 2 7.700 = 15.400 kN
The safety factor used is relatively high compared to the factor which has been applied on the
bending moment in prior calculations. The main reason is the uncertainty about original thrust jack
force estimates. However since the model by De Rijke is no longer an estimation, but a calculation
based on soil and TBM properties, a lower safety factor may be suggested. Nevertheless a safety
factor of 2 is still common now and will therefore be used in this study.
If the tunnel lining thickness increases, the external diameter of the TBM increases as well.
Consequently the maximum thrust jack force grows (see Figure 46a). For the tunnel embedded in soil
(Chapter 3) the normal force developed directly proportional to the depth of its centre. For the thrust
jack force there is nothing different, as Figure 46b shows.
-14000
Tunnel depth from centre line [m]
19,2 (0,7D)
-14500 27,4 (1,2D)
-15000
Thrust jack force [kN]
54,8 (2,9D)
-15500
-16000
82,2 (4,6D)
a b
46 | Development thrust jack forces over: a) Lining thicknesses for depth of 1,2D; b) Tunnel depth
The maximum allowed compressive stress under the jack plates is known in the Dutch code NEN
6720 as fco. Hence the following requirement should be met at any time:
c , jack f'co
According to the code the maximum allowed compressive stress is given by:
lb
f' co = f' c (15)
ala b
Where fc is the design value of the compressive strength for concrete and l and b are the length and
width of the segmental cross-section over which the thrust jack force will spread in the end.
The latter dimensions are defined by the minimum values of:
a l + 2sl a b + 2s b
a +d a +d
l b
l = min and b = min
5a l 5a b
5b a l 5l a b
Dimension d is the height of the so-called disturbance area and sl and s b are the differences
between the length and width respectively of the thrust jack plate and half the area available for
spreading (see Figure 47).
sl al sb ab
a l + 2s l a b + 2s b
A-A' A-A'
47 | Definitions schematised stress distribution under a thrust jack plate
For the tunnel with a lining thickness of 675 mm the dimension a l + 2sl , that equals half the
segmental length, and a b + 2s b is given by:
(Di + h ) (14.900 + 675 )
a l + 2sl = l segm / 2 = = = 2.718 mm
2 n segm 18
a b + 2s b = h
Tunnel segments transmit the axial thrust jack force to older rings via bearing pads in the ring joint.
These pads are situated right behind the thrust jack plates. So only half of the segmental dimension
in axial direction is available for spreading the force. The other half is used to narrow the spread
force again in order to lead it to the bearing pads (see Figure 47). Therefore the disturbance length d
is half the segmental width:
2.000
d= = 1.000 mm
2
Determination of all remaining parameters is possible now:
a b + 2s b = h
a + d = 2 h + 1.000 mm if h 3.000 mm h 3.000 mm
b h
b = min = 2 b = h
3
5 a b = 3 1
3 h 3 h + 1 .000 mm if h > 3.000 mm
5l( a b ) = 12.500 mm
Accordingly parameter b equals the lining thickness h for all thicknesses considered. The governing
value for parameter l however varies within the considered range. The following definition applies:
a l + 2s l
a + d a l if h 300 mm
l
l = min = 5h if 300 < h < 500 mm
5a l a + d if h 500 mm
5 b( a l ) l
The maximum value of the permitted compressive stress is now defined by substitution of all
parameters in equation (15):
f'c 1,5 if h 300 mm
lh h
f'co = f'c == f'c if 300 < h < 500 mm (16)
1.500 23 h 10 2
f'c 2 ,5 if h 500 mm
For a lining thickness of 675 mm the permitted values for the three concrete strength classes are:
f'co , C 35 / 45 = f'c ,C 35 / 45 2 ,5 = 0 ,6 45 1,58 = 43 MPa
f'co , C100 / 115 = 0 ,6 115 1,58 = 109 MPa
f'co , C180 / 210 = 0 ,6 210 1,58 = 199 MPa
In Figure 48 the development of these values along an ascending lining thickness has been included.
From a comparison of the permitted stresses and the actual compressive stress by the introduction of
thrust jack forces in equation (14) (22,8 MPa), it appears that all concrete strength classes are able to
cope with the compressive stresses under the thrust jack plate in case of a lining thickness of 675 mm
and a depth projection of 19,3 m (1,2D).
Equation (14) also shows that the resulting compressive stress for constant thrust jack forces is
related to the inverse lining thickness and will therefore decrease with an asymptotic extent at
increasing thicknesses. However the real thrust jack force of thick linings is slightly larger than the
one of slender linings (Figure 46a). Nevertheless the global development of the resulting concrete
compressive stress is hardly assaulted by this variation, as Figure 48 proves.
The locations where the lines of the actual and permitted values meet demonstrate that the minimum
required lining thicknesses for the concrete strength classes are:
h 390 mm for C35/45
h 175 mm for C100/115
h 100 mm for C180/210
-120
-140
-160
-180
f'co,C180/210
If the depth projection of the tunnel increases, the thrust jack force grows as well (see Figure 46b).
Consequently the concrete compressive stress under the thrust jack plate ascends accordingly. No
changes occur in the permitted compressive strength, irrespective of the depth. The occurring
stresses for all four considered tunnel depths and the permitted stresses for all three materials have
been combined in Figure 49a. Again minimum required lining thicknesses can be isolated, which
have been included in Figure 49b.
Obviously concretes with a higher compressive strength require significantly thinner tunnel linings
to cope with the compressive stresses introduced by thrust jack forces. Especially in very deep
tunnels the advantage of C100/115 and C180/210 is clearly visible. At the end this may turn out to be
a very positive property of very and ultra high strength concrete, since these materials manifested
splendid behaviour for deep tunnels in calculations on the uplift loading case in the previous chapter
as well.
0 200 400 600 800 1000 0 200 400 600 800 1000
0
f'co,C35/45 h [mm] thickness h [mm]
-50 0,7D
Tunnel depth from centre line [m]
C180
-250 54,8
4,2D
/210
-300 C100/115
-350
82,2 4,6D
a b
49 | Introduction of thrust jack forces over the depth. a) Compressive stress under thrust jack plate; b) Required lining
thicknesses
at1 = 1.000
ar = h
ar = h
A-A' A-A'
tangential radial
50 | Spreading of compression trajectories in tangential and radial direction in tunnel segments
In the radial direction of the tunnel segments, the introduction width is restricted to the height of the
thrust jack plates, the spreading width is the thickness of the concrete segments. Hence:
r = 23 h r 2
=3
ar = h ar
In Iyengars diagram a ratio between the maximum radial tensile bursting stress cr and the spread
compressive stress cm is found, reading:
cr
= 0 ,15
cm
cx
cm 0,5
Fjack
y cm =
a =0 ab
cx
Fjack a x
0,4 0,1
cy
0,2
a b
a =0,3
0,3
0,4
0,5
0,2
0,17 0,6
0,15 0,67
0,7
0,1 0,8
0,9
x
a
0
0 0,25 0,5 0,75 1,0
51 | Diagram of Iyengar for tensile bursting stresses due to introduction of thrust jack forces
In order to determine the final spread compressive stress cm both the spreading height and width
are required. The spreading height (in radial direction) has been retrieved already and equals the
lining thickness, or segmental thickness. Determination of the spreading length (in tangential
Maximum spreading in Iyengars diagram occurs if = 0. Over the full spreading depth a, the
trajectories should spread over a distance of a/2 to either side. So maximum spreading of the
introduced thrust jack force on the sides of the jack plates holds a/2 as well. In this expression a
equals at1 from Figure 50 in this case. Consequently the maximum spreading length in tangential
direction of the thrust jack force is:
a t = t + a t 1 = 1.500 + 1.000 = 2.500 mm
This dimension is smaller than half the segmental length ((Di + h)/9 = 2.720 mm for h = 675 mm)
and is therefore thought to be critial for the spreading length.
The complete spreading area is known now, so the compressive stress after full spreading may be
retrieved by use of the definition from Figure 51:
Fjack Fjack Fjack
cm = = =
ab a ra t 2.500h
Subsequently the ratio between the tangential tensile bursting stresses and the spread compressive
stress is found as well. According to the diagram it is given by:
t = 1.500 t
= 0 ,6
a t = 2.500 at
Tensile bursting stresses in radial and tangential direction are now determined by:
0 ,15Fjack Fjack
cr = cr cm = =
cm 2.500 h 16 ,7 10 3 h
0 ,17 Fjack Fjack
ct = ct cm = =
cm 2.500h 14 ,7 10 3 h
The thrust jack force Fjack has a positive value under compression, for it points in the direction
defined by Figure 51.
Similar to the development of the compressive stress under the thrust jack plates, the tensile bursting
forces due the jack force introduction are related to the inverse lining thickness as well. Figure 52
shows both resulting stresses for an overburden of 19,3 m (1,2D).
Bursting stresses cr/ct [MPa]
fc,C180/210
8
ct
6 cr
fc,C100/115
2 fc,C35/45
h [mm]
0
0 200 400 600 800 1000
52 | Radial and tangential tensile bursting stresses due to the introduction of thrust jack forces
For the applied relation between the height and width of the thrust jack plate on one side and the
dimensions of the tunnel segment on the other side, tensile bursting stresses in the tangential
direction are governing. Figure 52 thereby returns the following minimum required lining
thicknesses for the tunnel in order to resist tensile stresses by the introduction of thrust jack forces at
a depth of 19,3 m (1,2D):
h 635 mm for C35/45
h 195 mm for C100/115
h 105 mm for C180/210
Apparently the lining thicknesses required by the tensile stresses are higher than the ones dictated by
compression under the thrust jack plates (also see Figure 48). In case of reinforced ordinary concrete
C35/45 however the possibility exists to include rebars to deal with the tensile stresses. Hence for this
material the lining thickness required by the resulting compression stresses is governing.
Again the actual thrust jack force is the only input parameter of the calculations that depends on the
tunnels depth projection. In Figure 53a the resulting tangential tensile bursting stresses are
presented. The required lining thicknesses of the considered concrete materials are shown in Table 7
and Figure 53b.
thickness h [mm]
25
4,2D
Tunnel depth from centre line [m]
19,2
20
2,9D 27,4
C35/45
15
10 1,2D fc,C180/210
C1 8
54,8
0,7D
0/21
fc,C100/115
5 C100/115
0
fc,C35/45
h [mm]
0 82,2 4,6D
200 400 600 800 1000
a b
53 | Introduction of thrust jack forces over the depth. a) Maximum tangential tensile bursting stresses; b) Required lining
thickness per concrete strength class by bursting stresses
Table 7 | Lower boundary conditions on the allowed lining thicknesses by the introduction of thrust jack forces
B
Initial circular shape
Segments
Grout loading subjected to
TBM shield
torsion
B
Protected Deformed segments
The torsional distance in axial direction determines the amount of damage to a big extent. The exact
distance is not known however. Depending on the interaction of rings in this direction possibly one
or perhaps multiple segmental widths are required to generate the deformation. The torsion width is
therefore indicated as Bt for now.
't
u2
Bt
55 | Rotation of tunnel segment in vertical plane
The torsional moment creates a linear shear stress in the concrete tunnel segments. Its maximum
value at the top and bottom of the segments is given by:
T
t = (17)
Wt
Where Wt is the elastic torsional modulus:
1
B h2
Wt = 3 t again with h < Bt
h
1 + 0 ,6
Bt
If its assumed that the shear stress t is the only stress present (hence stresses by the thrust jack
forces are omitted for now), the shear stress is a principal tensile stress and therefore should not
exceed the tensile strength fshr:
t < fshr
Substitution of the parameters in equation (17) gives the torsional shear stress:
E h 4u 2
3 B t h 1 0 ,6
1 3
GI t t 2(1 + ) Bt b(r + 21 h) 2Ehu2 h2 2Ehu 2
t = = = 1 0 ,36 2 =
Wt 1
B h 2
B ( r + 1
h )(1 + ) B B (r + 1 h)(1 + )
3 t t 2 144244 t
3 t 2
h 1
1 + 0 ,6
Bt
According to the Dutch building code NEN 6720 the mean tensile strength fcm is the boundary
condition for tensile stresses for calculations on cracks. Hence:
fshr = fcm = 1,4fcrep = 1,4(1,4fc ) = 2fc (19)
The maximum permissible deformation to prevent cracking by torsion appears to depend on several
concrete properties is presented in equations (18) and (19) (note that the Poisson ratios are similar for
all considered concrete strength classes):
80
60
C180/210
40
C100/115
20
C35/45
h [mm]
0
0 200 400 600 800 1000
56 | Maximum allowed deformation u2 due to torsion only in tunnel segments
direction
T principal stress 2 (yy,yx)
direction
principal stress 1
y 2 1 xx,yy
yy
yx
xy
x
(xx,xy)
xx
xy
a b
57 | Multi-axial stresses in twisted tunnel segments. a) Definition of stresses; b) Mohrs cirle
The principal tensile stress by the multi-axial stresses is restricted by boundary fshr again:
1 = 1
2 ( xx + yy + ) 1
4 ( xx yy ) 2
+ tj 2 fshr
If fshr equals fcm, the maximum allowed value for shear force tj can by isolated by:
( ( ))
fcm 12 xx + yy 14 xx yy 2 + tj2
2
( )
fcm 2 + 1
4 ( xx
2
) (
+ yy 2 + 2 xx yy xx + yy fcm ) 1
2 ( xx
2
)
2
+ yy 2 2 xx yy + tj 2
(
tj fcm 2 fcm xx + yy + xx yy )
Values for stresses xx and yy result from the calculations on tensile bursting stresses in the previous
section. At a depth projection of 19,3 m (1,2D) the admissible value for shear force by grouting tj now
develops as presented in Figure 58. Obviously the allowed values for very and ultra high strength
concrete benefit from the introduced compressive forces in thin linings in particular. As a
consequence even higher deformations are allowed. If the initial concrete tensile strength fshr is
replaced by tj these deformations are described by equation (18) once more:
Bt tj (r + 21 h )(1 + )
u2 <
2Eh
If concrete is unable to cope with the combined compressive and tensional stresses by the
introduction of thrust jack forces, the permissible value of tj reduces to zero. Thats what occurs at a
lining thickness of approximately 300 mm for ordinary concrete C35/45 in Figure 58. Evidently the
tunnel segments are unable to resist any deformation u2 now. At deeper tunnel projections this
phenomenon will appear for the other concrete strength classes as well. In Figure 59 the permitted
displacements have been visualised for all considered depths. Higher compressive and tensile
bursting stresses at higher depths result in a decrease of the local peaks at slender lining thicknesses.
30
25 tj,C180/210
20 fcm,C180/210
15 tj,C100/115
fcm,C100/115
10
5 tj,C35/45
fcm,C35/45
h [mm]
0
0 200 400 600 800 1000
58 | Allowed shear stress tj at a depth of 1,2D
100 100
C180/210
80 80
C180/210
60 60
40 40
C100/115 C100/115
20 20
C35/45 C35/45
h [mm] h [mm]
0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
a b
Permitted deformation u2 [mm]
Permitted deformation u 2 [mm]
100 100
80 80
C180/210
60 60 C180/210
40 40
20 C100/115 20
C100/115
C35/45 h [mm] C35/45 h [mm]
0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
c d
59 | Development of allowed deflection u2 at depths: a) 0,7D; b) 1,2D; c) 2,9D; d) 4,6D
19,2 0,7D
27,4 1,2D C35/45 C35/45
Incomplete grouting
54,8 2,9D
Complete
C180/210
grouting
C100/115
82,2 4,6D
60 | Development of actual deflections due to ovalisation of tunnel in the uplift loading case for a lining thickness of 675 mm
A reduction of the lining thickness significantly reduces the linings total stiffness (EI) against
deformations. Consequently the deformation for slender lining thicknesses increases dramatically.
The small reduction of the bending moment by incomplete grouting that has been observed at very
slender linings is of hardly any help. Figure 61 shows the principal.
Actual deformation u2 [mm]
C100/115
250
C180/210
200 C35/45
150 C100/115
C35/45
100
C180/210
50
Thickness h [mm]
0
0 200 400 600 800 1000
61 | Deflections by complete and incomplete grouting at overburden of 19,3 m (1,2D)
The dimension available for torsion in the tunnels axial direction (Bt) is set to one segmental width
again. A safety factor for the resistance against cracking by torsion in the trumpet effect is defined as
the permissible value of deformation u2 divided by the actual occurring deformations due to the
uplift loading case. Figure 62 contains the development of this safety factor for incomplete grouting
at a depth of 19,3 m (1,2D). Apparently the more flexible behaviour of thin linings (resulting in high
permissible deformations as presented in Figure 59b) is unable to compensate the larger increment of
the occurring deformations at these thicknesses (Figure 61). Consequently very large lining
thicknesses are required to prevent cracking by torsion. This does not hold for ordinary concrete
C35/45 only, which is unable to prevent cracking without applying additional rebars in the
considered range of thicknesses anyway. However it does hold for the steel fibre reinforced concretes
very and ultra high strength concrete. At greater tunnel depths the occurring ovalisation
deformations by incomplete grouting slightly descend. At thick linings however the permitted value
hardly changes (except for C35/45 at the deepest location). Consequently the minimum required
lining thickness for torsion cracks by incomplete grouting descends at an increasing depth projection.
Table 8 and Figure 64 show that development.
3
C180/210
2
C100/115
=1
1
C35/45
Thickness h [mm]
0
0 200 400 600 800 1000
62 | Safety factor for incomplete grouting against cracking by torsion for an overburden of 1,2D
Table 8 | Lower boundary conditions on the allowed lining thicknesses to prevent cracks by the trumpet effect
5.4 Conclusions
High compressive concrete stresses occur under the thrust jack plates by the introduction of the jack
forces. In case of the French thrust jack configuration tensile bursting stresses deeper in the segment
below the jack plates are the only structural tensile stresses resulting from the introduction.
Provided that the height of the thrust jack plates is linked to the lining thickness, the thrust jack force
only slightly increases if the jack plates height grows. The resulting stresses however descend at a
more extensive rate; hence the safety level increases at an ascending lining thickness.
The required lining thickness in order to cope with the tensile bursting stresses in tangential direction
is governing above the required thicknesses by the compressive stresses. In reinforced ordinary
concrete it is assumed that reinforcement bars will be utilised to cope with the tensile forces, as a
result the compressive stresses are governing for this material.
Required thicknesses for very and ultra high strength concrete are significantly lower than for
reinforced ordinary concrete. From a depth projection of approximately 50 m (3D) a larger lining
thickness is required for C35/45 than the thickness of 1/20 D from the commonly applied rule of
thumb.
All required lining thicknesses by the introduction of thrust jack forces have been shown in Figure
63.
19,2
27,4 +
C35/45
_ unreinforced
54,8 2,9D +
_ + C35/45
_
82,2 4,6D
C180/210 C100/115
63 | Lower boundary conditions on the allowed lining thicknesses by the introduction of thrust jack forces
The ovalisation deformations are defined by the tunnels diameter alteration during complete or
incomplete grouting. These deformations decrease at a growing tunnel depth. If thinner linings are
applied the resulting deformation will increase dramatically. Therefore the safety against cracking by
torsion increases at an ascending lining thickness.
Required lining thicknesses to prevent cracking are very high for all considered concrete strength
classes. If the standard lining thickness of 1/20 D is applied for ordinary concrete cracking will occur
if no additional reinforcement is used to prevent this.
Figure 64 gives all boundary conditions for the required lining thickness in order to prevent cracking
by the torsional trumpet effect in case of both complete and incomplete grouting.
19,2 0,7D
27,4 1,2D
(c) (ic)
54,8 2,9D
(c) (ic)
(c) (ic)
82,2 4,6D
C180/210 C100/115 C35/45
64 | Lower boundary conditions on the allowed lining thicknesses to prevent cracks by the trumpet effect
6.1 Introduction
Relations between several failure mechanisms and the required lining thickness for shield driven
tunnels were investigated in the previous chapters. Each individual mechanism resulted in a
boundary condition on the required lining thickness in order to cope with the forces or bending
moments from that mechanism. Four different depth projections for the potential tunnel of future
motorway A13/16 in Rotterdam were considered.
This chapter will bring all boundary conditions on the lining thickness together, in order to find a
minimum required thickness that meets the requirements by all mechanisms. Section 6.2 will deal
with this. Consequently will be visible what, considered a certain depth projection, the governing
mechanism for the tunnel is. For long a discussion is going on about this question of governing
behaviour in the design of shield driven tunnels. Perhaps this report may give some help to come to
a conclusion in that discussion. More interesting however is to find a way to reduce the influence of
the governing mechanism(s) and so to reduce the lining thickness even more. Section 6.3 will shortly
discuss some alterations of the considered tunnel which might lead to this goal, focussed on the high
strength steel fibre reinforced concretes.
75
0 200 400 600 800 1000 1200
Thickness h [mm]
ja jac
19,2 0,7D ck ks
ten
pl.
s s.
co
om
27,4 1,2D m
Tunnel depth from centre line [m]
pr
l.
grout
inc
.
mp
ion
nco
comp
tors
ut i
g ro
tor
l.
54,8 2,9D
sio
nc
ring
beha
om
v iour
pl.
82,2 4,6D
19,2 0,7D
27,4 1,2D
Tunnel depth from centre line [m]
mpl.
jac
torsion co
ks
.
pl
inco
j ac
m
ten
co
ks
s.
in
n
t
co
torsio
ou
mpl.
mp
gr
54,8 2,9D
r.
grou
io ur
ehav
t com
b
ring
pl.
82,2 4,6D
to
19,2 rsi
o nc
jack cks com
om
27,4
Tunnel depth from centre line [m]
p l.
s te
ja
pl.
ns.
m
co
t in
rou
g
pr.
l.
54,8 2,9D
incomp
ring behaviour
mpl.
grout co
torsion
82,2 4,6D
An economic design of a certain structure is commonly described as that the structures construction
phase may not be governing in stead of its serviceability phase. The serviceability phase is mainly
included by the bending moment in ring behaviour of the embedded lining. This bending moment
however results in an upper boundary condition of the required lining thickness. In other words:
only in case of large lining thicknesses this element of the serviceability phase will be governing. A
better description of the most economic design of a shield driven tunnel should therefore focus on
the smallest possible lining thickness in the construction phase.
Some relatively simple adjustments of the tunnels construction process might lead to a reduction of
crack formation by torsion. For instance if more rings are located within the TBM shield the
ovalisation deformations are spread over multiple rings and consequently cracks are reduced.
Nevertheless this implies an enlargement of the shields length that results in higher thrust jack
forces according to section 5.2.1. For ultra high strength concrete an increase of this force is no
problem at all, since only at very deep projections the thrust jack force is governing. A large grow of
the thrust jack force might lead to governing behaviour of this force for ordinary concrete at
shallower depths.
In practice so-called adjusters are occasionally used as well. These steel trusses are placed within a
new ring to prevent it from deforming up to a certain level. Consequently the ovalisation
deformations are gradually introduced to the tunnel segments in multiple rings. If the full
deformation is present the adjusters are removed.
Finally an optimised grouting regime can be used to prevent ovalisation of the tunnel. For that
purpose grout pressures have to be varied along the tunnels perimeter. If the tunnel intents to
deform as a lying egg higher grout pressures are applied on the sides in order to force it into its
original circular shape again.
19,2 0,7D
27,4 1,2D
C35/45
54,8 2,9D
C100/115
C180/210
82,2 4,6D
Figure 68 shows the combined ULS areas and minimum required lining thicknesses for all concrete
strength classes. From the diagram it obviously follows that it is now impossible to agree on one
lining thickness only that defines the required governing thickness for all arbitrary depth projections.
For ordinary concrete C35/45 however the range is relatively small up to an overburden of 50 m (3D).
Here the required thickness varies from 1/21 D to 1/26 D, all very close to the value from the rule of
thumb 1/20 D.
At large depths very and ultra high strength steel fibre reinforced concretes C100/115 and C180/210
only require very slender linings. According to incomplete grouting and the introduction of thrust
jack forces for instance, concrete C180/210 only requires a lining of 255 mm (1/58 D) at a depth of 61
m (4D). At the most shallow projection however no less than an additional 215 % of concrete lining is
required, resulting in a thickness of 805 mm, or 1/19 D. The application of shield driven tunnels in
ultra high strength concrete at great depths only seems essential. However in order to reach these
depths the tunnel should lead through more shallow ground as well. So how should a lining
thickness of approximately 255 mm fit in there? Or is the creation of very slender tunnel structures
from ultra high strength steel fibre reinforced concrete virtually impossible? These questions will be
dealt with later on in this chapter.
In Chapter 3 it has been stated that a relatively high concrete Youngs modulus would be applied in
the calculations on embedded ring behaviour, the modulus for short-term uncracked concrete. If this
In order to test this proposition an example will be discussed where additional reinforcement is
applied to the original steel fibre reinforced concrete C180/210. Regardless of the lining thickness 14
rebars per full segmental width, with a diameter of 20 mm (14 20), are used at the bottom as well as
at the top of the cross-section. Due to the dense concrete matrix a concrete cover of only 10 mm is
sufficient [8].
1,2D reinforced
2000 80
0,7D reinforced
0,7D
1500 60
1000 40
1,2D
1,2D unreinforced
500 0,7D unreinforced 20
Even though Figure 69b shows that the effect of the reinforcement bars is not at its optimum at very
thin lining thicknesses, the actual increment is sufficient to result in a significant lower required
lining thickness by the bending moments in the grouting phase. The lining thickness required by
embedded ring behaviour is influenced by the increased bending moment capacity as well. The
required safety level is now met at each considered lining thickness and depth in case ultra high
strength concrete with reinforcement bars is applied. Hence the original upper boundary condition is
no longer visible and the applicable ULS area is extended significantly.
In Figure 70 both the original unreinforced and reinforced situations are presented. Higher amounts
of reinforcement have been included in the figure as well. Obviously huge savings of the required
lining thickness can be realised by the addition of steel rebars. In case 56 bars per full segmental
width are used, a lining thickness of only 265 mm or 1/56 D is sufficient. At larger depth projections
less reinforcement is needed in order to obtain the same lining thickness. As a consequence similar
tunnel segments can be produced with a reinforcement percentage depending on the projected depth
of each individual segment.
f.
42 20
28 0
20
20
reinforcement:
in
2
re
14
56
un
19,2 jack
27,4
Tunnel depth from centre line [m]
s te
ns.
54,8 2,9D
ring behaviour
82,2 4,6D unreinforced
70 | Influence of reinforcement bars on the required lining thicknesses for C180/210
The improvement of the linings bending moment capacity by the application of reinforcement bars
nearly reduced the required lining thickness to the original minimum value of 255 mm from the
previous section. At the most shallow projection however very high amounts of reinforcement are
required by the bending moment for incomplete grouting. A reduction of the actual resulting forces
by this mechanism might obtain an even better and more considerable result. The following section
gives an example of that solution.
To restrict floating of the tunnel during the grouting phase mass of a temporary kind is thus
required. Straightforward ballast in the tunnels interior, for instance sand fill, seems most suitable
and most simple to use. At the same time as the grout is injected sand is now dumped at the bottom
of the tunnel. If the grout has hardened to such degree that the additional mass is no longer needed,
sand can be moved from the back of the fill to the front of the tunnel boring machine to serve as
ballast for the newest rings once more.
During construction of shield driven tunnels some fill is created by sand or a part of the future
internal structure already (see Figure 71). The resulting levelled surface provides a carriage way for
dump trucks to transport the excavated soil from the TBM out of the tunnel. By not only defining the
The software application LDesign, that was used to perform calculations on the grout uplift loading
case, has no function to add additional mass to the tunnel. The final net uplift force however is
reduced by the weight of the tunnel lining. Therefore the specific gravity of the concrete lining (the
specific gravity for C180/210 holds C180/210 = 28 kN/m) is increased in new calculations to imitate a
sand fill (specific gravity of sand = 18 kN/m) at similar lining thicknesses.
By iteration a specific gravity for the concrete lining is searched for that meets the required bending
moment safety level of 1,5 at a specific lining thickness. In other words: for each lining thickness the
linings specific mass is altered up to the moment that the resulting maximum bending moment
equals the bending moment capacity divided by the required safety level of 1,5. The additional
weight of the tunnel lining is now easily converted to a required height of the sand fill from the top
of the circular tunnel tube.
For the shield driven tunnel of motorway A13/16 several required sand heights have been
determined for ultra high strength concrete with and without additional steel bar reinforcement (14
20). The most shallow tunnel depth projections only have been considered. The required lining
thickness should be reduced to the minimum value from the original calculations as much as
possible. Therefore only values close to 255 mm have been applied, hence 200, 300 and 400 mm. The
maximum required sand fill has a height of 5,8 m for these cases (most shallow depth, 200 mm
without rebars). Figure 72 displays the required sand fills for all situations. In case a lining thickness
of 300 mm steel fibre reinforced concrete C180/210 with the additional reinforcement bars from the
previous section is used, temporary sand heights of 4,0 and 2,7 m are necessary at the most shallow
11,1 m (0,7D) and standard 19,3 m (1,2D) tunnel depth projections respectively.
200 mm 300 mm 400 mm
Sand fill [m]
0,7D 1,2D 0,7D 1,2D 0,7D 1,2D 0,7D 1,2D 0,7D 1,2D 0,7D 1,2D
Tunnel depth Tunnel depth Tunnel depth
72 | Required sandfill to resist bending moment by incomplete grouting for C180/210
6.4 Conclusions
Torsion by the trumpet effect during the construction phase leads to damage by cracks. If the rule of
thumb for a lining thickness of 1/20 D is used, cracks will appear in reinforced ordinary concrete
C35/45.
Even if damage by torsion is neglected the construction phase of the shield driven tunnel is very
important for the determination of the required lining thickness. Incomplete grouting in shallow
tunnels and the introduction of thrust jack forces in deep tunnels now dictate the required thickness.
The rule of thumb 1/20 D leads to a safe tunnel lining in reinforced ordinary concrete up to a depth of
approximately 50 m (3D).
No simple rule of thumb is present for the required lining thickness of steel fibre reinforced very or
ultra high strength concrete at each arbitrary tunnel depth projection. The required thickness of
tunnel linings in plane ultra high strength steel fibre reinforced concrete is lowest in deep tunnels.
Reinforced ordinary concrete behaves best at a depth projection of 2D. The best application of plane
very high strength concrete is in between both previous materials.
In order to reach high depths for tunnels in very and ultra high strength concrete shallow depths
should be passed as well. Reinforcement bars should be added to these steel fibre reinforced
materials to make construction of shallow tunnels possible as well. Tunnel linings with a thickness
from only 1/56 D are possible now.
Conclusions
The construction phase of shield driven tunnels is of major importance for the determination of a
required lining thickness. For shallow tunnels the bending moment by incomplete grouting dictates
the tunnels strength and therefore the required thickness. In deep tunnels concrete stresses by the
introduction of thrust jack forces are governing. In all considered concrete strength classes cracks are
likely to occur due to torsion in tunnel segments leaving the tunnel boring machine (known as the
trumpet effect) if the lining thicknesses dictated by the tunnels strength are applied. Also in case the
thickness of a lining from reinforced ordinary concrete is determined by the common rule of thumb
1/20 D, torsion cracks occur.
Therefore the tunnels construction phase can not be neglected in the design process.
Ring behaviour of a tunnel embedded in soil during the serviceability phase does not result in a
governing required lining thickness. The safety level for the bending moment in this failure
mechanism decreases if the lining thickness increases. As a result the requirement found for the
lining thickness is an upper boundary condition. Hence the applied thickness may not exceed this
relatively high value. Consequently alterations of the tunnels design in order to reduce the bending
moment in embedded ring behaviour only do not result in a lower required lining thickness.
Measures to decrease the governing behaviour of the construction phase however, do result in
thinner linings.
Steel fibres in very and ultra high strength concrete hardly contribute to the linings bending moment
capacity. If the lining thickness increases this minor contribution decreases even more. Consequently
slender tunnel linings in ultra high strength concrete with steel fibre reinforcement only are most
favourable at very high tunnel depths, where high normal ring forces are present. At a depth of
approximately 62 m (5D) concrete C180/210 only requires a lining thickness of 255 mm (1/58 D). At
very shallow depths however thicknesses of over triple this minimum value are necessary. The same
happens to very high strength concrete C100/115 that requires a lining thickness of only 375 mm at a
depth of 46 m (2,9D).
As a result no simple rule of thumb can be set for the required lining thickness of a shield driven
tunnel in steel fibre reinforced very or ultra high strength concrete at each arbitrary depth. The rule
of thumb 1/20 D for the lining thickness of reinforced ordinary concrete proved to be on the safe side
up to a tunnel depth projection of approximately 3D for the considered tunnel.
Tunnel linings in very or ultra high strength concrete can be constructed only if reinforcement bars
are added. The wide scattering of required lining thicknesses over the depth of a tunnel in plain steel
fibre reinforced concrete, makes it impossible to construct tunnels of these materials in both more
shallow and deep grounds. Reinforcement bars increase the linings bending moment capacity to a
high extent and therefore reduce the governing behaviour of incomplete grouting. In the considered
tunnel with an internal diameter of 14,9 m a lining thickness of only 265 mm (1/56 D) has been
85
presented for a heavily reinforced tunnel segment in ultra high strength concrete at each depth
projection.
Temporary measures can be taken to reduce the governing behaviour of incomplete grouting during
the construction phase of a shield driven tunnel. Reduction of the net uplift floating force by use of
additional mass in the tunnel has a positive effect on the required lining thickness. Consequently
tunnel linings in ultra high strength steel fibre reinforced concrete with additional reinforcement bars
are able to reach thicknesses below 250 mm or 1/60 D.
Recommendations
More (temporary) measures might be possible to reduce governing behaviour and damage due to
mechanisms in the construction phase of shield driven tunnels. This holds especially for torsion in
the tunnel segments that leads to cracking in all considered concrete materials and most lining
thicknesses. Research on how to prevent this topic can result in a significant decrease of damage in
tunnel linings during the construction phase.
Studies on modelling of the behaviour of shield driven tunnels in fresh grout should focus on
solutions to prevent governing behaviour of this mechanism. Especially in shallow tunnel depths
prevention of high resulting bending moments by grouting is needed in order to reduce the required
lining thickness.
This study showed that the minimum required lining thickness for very and ultra high strength
concrete very much depends on the tunnels depth projection. Only a very large tunnel diameter has
been investigated up to now. Likewise research on the applicability of these concrete materials in
shield driven tunnels with smaller diameters is needed. Comparison of both studies might show that
the application of shield driven tunnels in ultra high strength concrete is perhaps more favourable for
large or small tunnels.
Only strength related mechanisms have been studied in this report. Requirements on the deflections
have been omitted. Calculations on torsion in the grouting phase however showed that very large
deflections occur during incomplete grouting of very slender tunnel linings. Adaptations of the
linings cross-section for slender ultra high strength concrete in order to generate stiffer behaviour
might therefore be needed to meet the requirements. Hence more detailed research on quality
constraints, like deflections, durability and fire safety, should be performed before a shield driven
tunnel in ultra high strength concrete is constructed.
1. Acker, P., C. Baloche et al., Ultra High Performance Fibre-Reinforced Concretes, Interim
Recommendations. Bagneux Cedex (France): Setra and AFGC, January 2002
2. Bambridge, C., Focusing on fibres: CTRL experience. In: Tunnels & Tunnelling International, March
2006
3. Blom, C.B.M., A.P.M. Plagmeijer, Boren van tunnels met niet-ronde vormen. In: Cement 6 2003,
pages 74-80 (Dutch)
4. Blom, C.B.M., Concrete linings fors hield driven tunnels, lecture note. Delft: Delft University of
Technology, March, 2006
5. Blom, C.B.M., Design philosophy of concrete linings for tunnels in soft soil, dissertation. Delft: Delft
University Press, December 2002
6. Bruijn, H.J. de, HS, Literatuur- en voorstudie: hogere sterkte beton nader belicht. Utrecht: Delft
University of Technology and Holland Railconsult, December 2005 (Dutch)
7. Burgers, R., Non-linear FEM modelling of steel fibre reinforced concrete for the analysis of tunnel
elements in the thrust jack phase, thesis. Delft: Delft University of Technology, September 2006
8. CRC description. Hjallerup (Denmark): CRC Technology Aps (download at www.crc-tech.com)
9. Hollander, J. den, Technical feasiblity study of a UHPC tied arch bridge, thesis. Delft: Delft
University of Technology and Ingenieursbureau Gemeentewerken Rotterdam, May 2006
10. Kaptijn, N., Zeerhogesterktebeton, Toepassingen, handout. Utrecht: Rijkswaterstaat Bouwdienst
DIO, 2002 (Dutch)
11. Pruijssers, A.F. et al., Toetsingsrichtlijn voor het ontwerp van boortunnels voor weg- en
railinfrastructuur L500. Gouda: Centrum Ondergronds Bouwen, September 2000 (Dutch)
12. Rijke, Q.C. de, Innovation of stress and damage reduction in bored tunnels during construction based on
a shield equilibrium model, thesis. Utrecht: Delft University of Technology and Holland
Railconsult, February 2006
13. Rijkswaterstaat, SATO Deel 5 Tunneldetails. Utrecht: Bouwdienst Rijkswaterstaat (Dutch)
14. Rijkswaterstaat, Startnotitie Rijksweg A13/16 Rotterdam. Rotterdam: Ministerie van Verkeer en
Waterstaat, Directoraat-Generaal Rijkswaterstaat Zuid-Holland, November 2005 (Dutch)
15. Rijkswaterstaat, Technische haalbaarheidsstudie tunnelverbinding A6/A9. Utrecht: Ministerie van
Verkeer en Waterstaat, Directoraat-Generaal Rijkswaterstaat, September 2002 (Dutch)
16. Schumacher, P., Rotation Capacity of Self-Compacting Steel Fiber Reinforced Concrete, dissertation.
Delft: Delft University of Technology, November 2006
17. Slenders, B.M.A., Modellering van boortunnels, thesis. Utrecht: Delft University of Technology and
Projectorganisatie HSL-Zuid, January 2002 (Dutch)
18. Waal, R.G.A. de, Steel fibre reinforced tunnel segments. Delft: Delft University Press, January 2000
19. Walraven, J.C., Ultra-hogesterktebeton: een material in ontwikkeling. In: Cement 5 2006, pages 57-
61 (Dutch)
87
88 Shield driven tunnels in ultra high strength concrete
Table of figures
Figures
1 | Volume of concrete increases quadratic proportional to tunnel diameter 2
2 | Visualisation of this reports structure 4
3 | Dimensions and elements of the tunnel lining 6
4 | Thrust jack configurations: a) German method; b) French method 7
5 | Uneven support of tunnel segments for: a) German configuration; b) French configuration 7
6 | Concrete materials: significant increase of density and homogeneity for BSI 8
7 | Pedestrian Bridge of Peace in Seonyu, South Korea 9
8 | Surroundings of motorway A13/16 in the north of Rotterdam 10
9 | Fitting two directions in one tunnel tube. a) Two tunnels with large useless free areas; b) One
tunnel with less free space 11
10 | Cross-section of tunnel for motorway A13/16 with personal cars only (scale 1:150) 12
11 | Schematic representation alignment of the tunnel for motorway A13/16 12
12 | Uniform pressure leading to normal ring force tunnel 17
13 | Relation between rotations and bending moments in a Janen joint 17
14 | Beam elements (and nodes) in the centre line of a tunnel segment 19
15 | Determination of spring stiffness for ring joints 19
16 | Vertical stresses in uniform soil continuum along a tunnel 21
17 | Orientation of vertical and horizontal versus radial and tangential soil loads 22
18 | Moment distribution in validation BRT-model and model from this study 25
19 | Moment distribution in validation model LDesign and model from this study 26
20 | Complete framework model for the shield driven tunnel of this study 28
21 | Maximum bending moment by ring behaviour for depth of 0,7D. a) absolute values; b) surplus
relative to maximum moments for ordinary concrete C35/45 30
22 | Normal ring force by ring behaviour for depth of 0,7D 30
23 | Effect of tunnel depth projection with a lining thickness of 675 mm. a) normal ring force; b)
tangential bending moment 31
24 | Maximum bending moment by ring behaviour for soil overburden of: a) 1,2D; b) 2,9D; c) 4,6D 31
25 | Calculation definition ultimate resisting moment 32
26 | Stress-strain diagram of reinforcement steel 33
27 | Stress-strain relations of ultra high strength concrete: a) strain hardening; b) and strain softening
34
28 | Stress-strain diagrams steel fibre reinforced concretes a) C100/115; b) C180.210 35
29 | Ultimate resisting moment at a depth of 0,7D 37
30 | Development ultimate resisting moment at various depth projections for lining thickness of 675
mm 37
31 | Ultimate resising moment at depth projections of: a) 1,2D; b) 2,9D; c) 4,6D 38
32 | Development of safety factor Mu/Mmax at depth projections of: a) 0,7D; b) 1,2D; c) 2,9D; d) 4,6D 39
33 | Upper boundary conditions on the allowable lining thicknesses in case of embedded ring
behaviour (lower boundary for C35/45) 41
34 | Development ultimate resisting moment over the depth for 675 mm thick reinforced concrete
C35/45 44
35 | Forces and definitions in the uplift loading case [5] 46
89
36 | Fictitious top support 48
37 | Bending moment by complete and incomplete grouting at a depth of 0,7D 49
38 | Development bending moment by complete and incomplete grouting over the depth for 675 mm
C35/45 50
39 | Bending moment by complete and incomplete grouting at depths: a) 1,2D; b) 2,9D; c) 4,6D 50
40 | Bending moment safety factor for complete and incomplete grouting at depths: a) 0,7D; b) 1,2D;
c) 2,9D; d) 4,6D 52
41 | Lower boundary conditions on the allowable lining thicknesses in case of complete and
incomplete grouting 53
42 | Development of bursting stresses. a) Compression trajectories; b) Latice analogy 56
43 | Tensile force in latice analogy for German thrust jack configuration 56
44 | Stress distribution in tunnel segments due to introduction of thrust jack forces in the: a) French
and b) German thrust jack configurations [5, 18] 56
45 | TBM Shield Equilibrium Model with line load to determine thrust jack forces 58
46 | Development thrust jack forces over: a) Lining thicknesses for depth of 1,2D; b) Tunnel depth 59
47 | Definitions schematised stress distribution under a thrust jack plate 60
48 | Compressive stress under thrust jack plates for a depth of 1,2D 62
49 | Introduction of thrust jack forces over the depth. a) Compressive stress under thrust jack plate;
b) Required lining thicknesses 62
50 | Spreading of compression trajectories in tangential and radial direction in tunnel segments 63
51 | Diagram of Iyengar for tensile bursting stresses due to introduction of thrust jack forces 63
52 | Radial and tangential tensile bursting stresses due to the introduction of thrust jack forces 64
53 | Introduction of thrust jack forces over the depth. a) Maximum tangential tensile bursting
stresses; b) Required lining thickness per concrete strength class by bursting stresses 65
54 | Torsion in tunnel segments by deformation due to grout loading 66
55 | Rotation of tunnel segment in vertical plane 66
56 | Maximum allowed deformation u2 due to torsion only in tunnel segments 68
57 | Multi-axial stresses in twisted tunnel segments. a) Definition of stresses; b) Mohrs cirle 69
58 | Allowed shear stress tj at a depth of 1,2D 70
59 | Development of allowed deflection u2 at depths: a) 0,7D; b) 1,2D; c) 2,9D; d) 4,6D 70
60 | Development of actual deflections due to ovalisation of tunnel in the uplift loading case for a
lining thickness of 675 mm 71
61 | Deflections by complete and incomplete grouting at overburden of 19,3 m (1,2D) 71
62 | Safety factor for incomplete grouting against cracking by torsion for an overburden of 1,2D 72
63 | Lower boundary conditions on the allowed lining thicknesses by the introduction of thrust jack
forces 73
64 | Lower boundary conditions on the allowed lining thicknesses to prevent cracks by the trumpet
effect 73
65 | Boundary conditions on lining thickness for C35/45 76
66 | Boundary conditions on lining thickness for C100/115 76
67 | Boundary conditions on required lining thickness for C180/210 76
68 | Effect of the combined boundary conditions on the required lining thicknesses 78
69 | Utimate resisting moment with and without additional rebars: a) absolute values; b) surplus by
rebars 80
70 | Influence of reinforcement bars on the required lining thicknesses for C180/210 81
71 | Levelled tunnel interior during construction Green Heart Tunnel 82
72 | Required sandfill to resist bending moment by incomplete grouting for C180/210 82
73 | Tangential cross-section of the longitudinal joint (left) and an overview of the stresses in the
reduced beam problem of Janen 98
74 | Stresses in the Janen joint in case the compressive yield strain has been exceeded (left) and the
stress-strain relation for rotation in the joint (right) 100
Tables
Table 1 | Lining thickness over diameter ratio for Dutch shield driven tunnels 5
Table 2 | Classification of concrete strength classes [19] 8
Table 3 | Notation and values depth projection tunnel 29
Table 4 | Upper boundary conditions on the allowable lining thicknesses in case of embedded ring
behaviour 40
Table 5 | Lower boundary conditions on the allowable lining thicknesses in case of complete and
incomplete grouting 52
Table 6 | Calculation of axial thrust jack forces 58
Table 7 | Lower boundary conditions on the allowed lining thicknesses by the introduction of thrust
jack forces 65
Table 8 | Lower boundary conditions on the allowed lining thicknesses to prevent cracks by the
trumpet effect 72
Table 9 | Equations for the longitudinal Janen joint 103
Table 10 | Partial safety factors for strength (resistance) according NEN 6720 and DIN 1045 neu 119
Table 11 | Partial loading factors NEN 6720 (safety class 3) and DIN 1045 neu 120
Table of contents
Appendix A Derivations 95
A.1 Janen joint 97
A.2 Transformation of coordinate systems 105
A.3 Uplift of embedded tunnel 107
Appendix B Ultimate resisting moment for steel fibre reinforced concrete 109
Appendix C Safety factors in shield driven tunnels 117
93
94 Shield driven tunnels in ultra high strength concrete
Appendix A
Derivations
The longitudinal joint is an inhomogeneous part of the tunnel structure and therefore needs extra
attention in the modelling process of ring behaviour for a shield driven tunnel. The joint transfers a
bending moment and a normal force by contact; its unable to transfer tensile forces and opens in
case of a relatively high bending moment. Modelling of a contact area usually means longer
calculation times. If a solution is present to simplify this problem, it would be favourable for the
framework analysis.
A simplified solution has been presented by Janen. The contact problem was reduced to the
problem of a beam, which is unable to cope with any tensile stresses. The width and height in the
tangential cross-section of the beam equals the contact height of the longitudinal joint. Opening of the
joint is symbolised by formation of a crack in the complete tensile zone of this beam. The formation
of this crack depends on the applied normal force and bending moment at that particular joint. The
stiffness of the beam, also called Janen joint, now represents the stiffness of the longitudinal joint. By
applying this stiffness (being a function of the normal force and bending moment) to a non-linear
rotational joint its possible to simplify the contact problem in a longitudinal joint in a framework
analysis.
This appendix discusses the derivation of Janens method and computes some useful attributes of
the Janen joint in advance.
lt
lt
uM
lt
+
M
xu R
73 | Tangential cross-section of the longitudinal joint (left) and an overview of the stresses in the reduced beam problem of
Janen
Depending on the ratio between the bending moment and the normal force the joint will be opened
or closed. Both situations are discussed separately. This will also be done for the opened situation
when concrete behaves plastic.
From this equation the bending moment M is known. However the rotation still needs to be
derived. By having a look at figure 73 this rotation is given by:
2u M
= (2)
lt
In order to determine the displacement uM, first of all the stress by the bending moment shown is
needed:
6M
M = 2 (3)
bl t
Where: b = Width of the tunnel segment
For the strain resulting from this stress it holds:
M = M (4)
Ec
Where: Ec = Youngs modulus of concrete
Next the deflection uM is easy to determine:
uM = Ml t (5)
Substitution of equations (3), (4) and (5) into (2) gives the rotation by M:
12M
= (6)
E c bl 2t
Now substitution of equation (6) into equation (1) results in the constant rotational stiffness of a
closed Janen joint depending on the stiffness and lay-out of the contact area:
E bl 2
cr = c t (7)
12
lt
lt
f'c
u
lt
R1 R2 Ec
x1 x2
'c 'cu
74 | Stresses in the Janen joint in case the compressive yield strain has been exceeded (left) and the stress-strain relation
for rotation in the joint (right)
Now the reaction force R1 will be computed by (after substitution of equation (20)):
f' c x 1 b
R1 = (21)
2
The other way round holds for x1:
2R 1
x1 = (22)
f' c b
1,2 =
(6N )
21N 2 + 30Mf ' c b 15Nl t f' c b l t f' c2 b
(27)
6( 2Mf ' c b + l t Nf' c b + N )E c
2
By using the final value of the non-plastic opened joint rotation as a boundary condition, the correct
equation of both can be determined.
Hence the Janen rotational stiffness (equation (1)) for the plastic situation of the joint is known:
M 6( 2Mf' c b + l t Nf' c b + N 2 )ME c
c r1,2 =
1,2
=
(
6 N 21N 2 + 30Mf' b 15Nl f' b l f' 2 b
c t c ) t c
Upsetting strain
The upsetting strain, or ultimate compressive strain, is a material related property. This study
therefore investigates the effects of this property on the maximum strain actually occurring in the
joint. The upsetting strain max will be given as a function of the rotation .
In case of a closed joint equation (2) will be substituted into equation (5):
l t
= Ml t
2
Where M represents the strain by the bending moment in the joint. The strain due to the normal force
is given by:
N
N = N =
E' c E' c bl t
Now the total strain max in a closed joint holds:
N
max = M + N = +
2 E' c bl t
For the calculation of the maximum strain in case of an opened joint firstly equation (18) will be
substituted into equation 17:
x
x u = l t = u
lt
Substitution of equation (13) in this equation gives:
Next if concrete plasticity occurs in the opened Janen joint, the maximum strain is retrieved from
figure 74 and holds:
' (x + x 2 ) f' c (x 1 + x 2 )
max = c 1 = (29)
x1 Ec x1
Substitution of equations (20), (24) and (21) in (29) results in the following equation for the maximum
strain if plasticity occurs:
f ' l b + 2 NE c
max = c t
2E c f ' c l t b
In the non-linear equation for the rotation of the Janen joint in case of concrete plasticity (equation
(27)) too, a possibility for an invalid division by zero exists. Hence this expression knows a maximum
bending moment as well. This maximum is derived by the equilibrium of:
2Mf' c b + l t Nf' c b + N 2 = 0
Reordering and again replacement of M by Mmax now results in the maximum bending moment for a
plastic Janen joint:
l N2
M max = N t +
2 2f ' c b
The equations for the rotational stiffness and reactions in the Janen joint for all three situations are
summarised in Table 9.
A cylindrical coordinate system (r--z) is best used to model the tube of a shield driven tunnel. The
loads on the tunnel however are given by a Cartesian coordinate system (x-y-z). Equations to
transform one system into the other will be described in this appendix. Besides the conversion of
Cartesian soil stresses into cylindrical soil forces on the nodes of the tunnel will be discussed as well.
z z
Both the Cartesian and cylindrical coordinate systems share the same axis z. Therefore this appendix
only focuses on the transformation of the 2-dimensional Cartesian x-y and circular r- systems.
Stresses and forces act on a surface. The angle of this surface in the Cartesian system x-y is fixed.
Implying forces perpendicular to the axis r might occur as well. Therefore an axis t will be added to
the circular r- system. Now a new coordinate system r-t is created, showing the Cartesian x-y system
rotated over an angle of .
x
r
ur
ux (ux, uy)
ur2
ur1
y
uy
ut1
ut
ut2
t
76 | Distances in a Cartesian x-y and r-t-() coordinate system for a point (ux, uy)
cos
y
v
t
r h
t
Fh
sin
From figure 77 the following relation holds for the forces Fv and Fh in the Cartesian system:
Fv cos 0 v
=
h
(2)
F
h 0 sin
By substituting equation (2) into equation (1) (ur, ut, ux and uy have been replaced by r, t, v and h
respectively) the relation between the horizontal and vertical distributed loads and the radial and
tangential distributed loads is known (although r and t have been visualised as concentrated forces
in figure 77, their influence width of 1 makes them equal to a distributed load):
r cos 2 sin 2 v
=
t cos sin cos sin h
A tunnel embedded in soil should be in vertical equilibrium at all time. Hence the upward floating
force by the hydrostatic water pressure should be compensated by the mass of the tunnel and mass
of the soil on top of this tunnel. For safety reasons a safety factor of mass = 0,9 is applied on the masses
of both tunnel and soil.
By use of Figure 16 on page 21 of the actual report the vertical distance between the water table and
an arbitrary point at the tunnels circumference with an angle holds:
h , w = x w x t + R(1 cos )
Where:
xw = Water table relative to mark NAP
xt = Top tunnel relative to mark NAP
R = External radius tunnel
The vertical water force on the tunnel lining at an angle is then given by:
F = R w () cos
So the total bottom force by the hydrostatic water pressure is then defined by:
3
2
Fw , up = Fd = R w ( 2 x w 2 x t ) + R 2w ( 2 + 21 )
1
2
For the cross-sectional surface of the soil on top of the tunnel it holds:
A soil = 2R (x 0 x t + R ) R 2
2
Where:
x0 = Ground level relative to mark NAP
Ftunnel = concrete R 2 i
2
Where:
concrete = Specific gravity concrete
The software application Maple 6 has been utilised to solve equations (1) and (2). As a result the
values of xt as presented in Figure 78 were retrieved for the soil and tunnel properties from this
study.
0 200 400 600 800 1000
(NAP) 0
Thickness h [mm]
-2,5
Top of tunnel relative to mark NAP xt [m]
-7,5
-10
nnel
ht tu
ad weig
-12,5 de
ding
Inclu
-15
Excluding dead
-17,5 weight tunnel
Each individual considered lining thickness should be prevented against floating. Subsequently the
value of a lining thickness of 100 mm has been applied. This one requires the deepest depth
projection (if the tunnels dead weight is included). Obviously thicker linings are heavier and
therefore require a more shallow minimum depth projection.
The minimum depth projection is set to: -15,85 m + NAP, hence a soil cover of 11,1 m (0,7D).
Introduction
This appendix presents the calculation principal for the ultimate resisting moment of a reinforced
rectangular cross-section of concrete with any arbitrary stress-strain relation. The calculation has
been implemented in a Visual Basic .NET2 application.
stress strain
Ns1
xu
Nc,compr
Nd
Nc,tens
Ns2
Both equilibrium of horizontal forces and equilibrium of bending moments should exist at all time:
ns
Fh = 0 N c + Fsj = N d (1)
j=1
ns
M = 0 M u = M c + M sj M N (2)
j =1
Where:
Nc = Normal force in concrete
Fsj = Steel force in reinforcement layer j
ns = Number of reinforcement layers
Nd = Acting normal force
Mu = Ultimate resisting moment
M = Contribution to Mu by element defined by its subscript
The normal force in concrete is principally in compression (negative), hence the value of Mc is
negative as well. Consequently the maximum value of Mu is reached if the contribution by concrete is
minimised. The concrete strain in the upper fibre of the cross-section is therefore set to the ultimate
1 (1) 0
Stress
(2)
6 Strain
2
(3)
(6)
5 (5)
4 (4) 3
f ( )
i 1
av ,i = , then it holds for the normal force: N i = bh i av ,i
i i 1
The integral, that is the surface under the line segment, is simply determined for this linear
development:
i
fi + fi 1
f() = 2 ( i i1 )
i 1
Of course this principal can be applied on all individual line segments in the stress-strain diagram.
The total integral, from now on assigned as , follows from the sum of the preceding individual
integrals. Hence the integral in the upper fibre with strain 0 up to point i with strain i is:
i i
f() = i = f() + i1
0 i 1
If this integral is determined for each node of the stress-strain diagram, the normal force in the
concrete is known if the strain at arbitrary node i is the strain in the cross-sections bottom fibre. This
normal force then reads (by use of prior equations for Ni and av,i):
i
N ci = hb
i 0
The total normal force at the cross-section is known by performing this calculation on each node in
the concrete stress-strain diagram and by adding the steel forces due to the concrete strain at their
individual location to it. The strain in reinforcement layer j simply reads:
x sj
sj = (i 0 )
h
The strain that is connected to this strain follow from the multiple linear stress-strain relation for
reinforcement steel (see Figure 81):
(2)
2 Strain
(3)
4 (4) 3
81 | Stress-strain diagram reinforcement steel
A complete set of resulting total normal forces in the cross-section on each critical point is gained if
this calculation procedure is repeated at every node of concretes stress-strain relation and on the
nodes of the steel relation for each individual reinforcement layer. Comparison of the forces in this
set to the actual acting normal force Nd makes it possible to find the line segment in the concrete
diagram where the actual strain in the bottom fibre is at. If, for instance, the steel force disappears (if
su is exceeded) a non linearity is found. Now the concrete strain closest to neutral (= 0) is used. Hence
for concrete and each individual reinforcement layer a small linear zone of strains (the line segment
in its stress-strain diagram) is known. These zones are zc for concrete and zsj for reinforcement layer j.
First of all: a straight line between two arbitrary nodes is defined by the following equation (see
Figure 82):
f() = ( i ) + fi
fi+1 fi
Where: =
i+1 i
i+1
fi+1
(i)
i
fi
i i+1
82 | Determination equation by nodes of a linear element
( ) (
z c + zc + zc ns )
= 2 hb + j j zsj + fzsj A sj
0 j=1
Where: = zc + fzc
2
hb 2 2 j=1 hb
ns j zc fzsj A sj
Where: A =
j =1 hb
This can be written as:
a 2 b c = 0
ns j N ns
N
With: a = + , b = zc A d
j 0
, c = zc zc + A 0 + d 0
2 j=1 hb 2 hb j=1 hb hb
Consequently the solution for the strain at the bottom fibre is:
b b 2 4ac
1, 2 =
2a
Only one of these two strains is positioned within all boundary conditions (boundary nodes of the
zones in the stress-strain diagrams of concrete and steel). That value is therefore the maximum strain
in the cross-section.
If the values of the scalars and j equal zero (0), parameter a will turn out to be zero as well.
Consequently no valid solution is found by the equation for 1,2. A simpler calculation principal can
be used now. Namely the strain that is looked for is positioned at horizontal areas in all stress-strain
relations. The zones (line segments) in these diagrams are still known. This implies that the forces in
the reinforcement steel are known. Hence:
ns
N c = N d fzsj A sj
j=1
For that purpose the stress-strain relation between strains 0 and is spread over the entire height of
the cross-section.
The strain in each node is now converted to a lever arm relative to the upper fibre by:
0
xi = h i
0
The bending moment Mc is constructed of the bending moment of all individual segments in the
stress-strain relation combined:
nc
M c = M ci
i =1
These individual bending moments are determined by multiplying the surface under the stress-strain
section by the lever arm of that particular section. Hence: the integral of the stress-strain diagram
multiplied by the vertical distance x:
x
f xi x i 2 i+1
xi + 1 xi + 1
M ci = b f( x)x dx = b xi x 2 + (fi - xi x i )x dx = b xi x 3 + i x
xi xi 3 2 xi
f xi x i
M ci = xi (x i+13 x i 3 ) + i (xi+12 xi 2 )b
3 2
fi+1 fi
Where: xi =
x i+1 x i
Where the determination of steel strain sj has been mentioned before and reads:
x sj
sj = ( 0 )
h
3 2 j=1 2
Hence follows:
M u = 113 ,2 + 209 ,6 + 500 = 593 ,7 kNm
The application, wherein the described calculation principal has been implemented, returns the
following value:
M u = 593 ,7 kNm
The individual contributions are:
M c = 115 ,9 kNm; M s = 209 ,6 kNm; M N = 500 kNm
Consequently no variations are found between the values resulting from the custom calculation
prinipal and the Dutch code.
For this validation an Excel-sheet from the graduation thesis of Den Hollander [9] is used. That sheet
searches for an equilibrium of horizontal forces by iteration. In order to retrieve a valid result from
that method prestressing steel has to be added to Den Hollanders calculation. A steel area of 1 mm
has been positioned at a distance from the upper fibre of 0 mm in order to exclude the steel force
from the ultimate resisting moment. The model generated a bending moment capacity of:
M u = 801 kNm
Both in case of reinforced concrete and steel fibre reinforced concrete the application implemented
returns very accurate ultimate resisting moments. Consequently it is assumed that this method can
be applied in this study.
Building codes prescribe calculation methods and safety ratios for most standard structures. Several
additional regulations deal with special requirements for various specific structures like bridges, or
sheet piles. The structure of a shield driven tunnel shows several differences with normal structures
like bridges or buildings, for instance its vital interaction with the surrounding soil. However no
additional regulations exist for this type of structure. This chapter therefore discusses the policy from
Dutch and German building codes from the bottom to present a safety philosophy for shield driven
tunnels in the end.
Safety factors in building codes exist to include the probability of several risk aspects in the
calculation. In modern building codes these safety factors are split up in partial safety factors for the
strength or resistance of the structure and for the stress on the structure by loading. Via a simple
check its possible to check whether a structure is able to resist unforeseen risks as well. The check is
given by the following basic equation:
Rd
1 (30)
Sd
Where Rd represents the resistance of the structure, including partial safety factors for the strength,
and Sd represents the stress to the structure, including partial safety factors for loading.
The partial safety factors will be discussed in separate sections in this appendix. This appendix
describes the policy in the Dutch and German codes for partial safety factors on the resistance side
for concrete. The partial safety factors on the stress or loading side of the structure, including the
influence of soil, will be dealt with as well.
Next a combination of those partial safety factors and a safety philosophy that is applicable for the
unique structure of shield driven tunnels are presented.
The use of material factors from the building codes for concrete is limited by a maximum strength
class of concrete. The NEN is valid for strength classes up to C53/65 (ordinary concrete) and
additional recommendations exist for values up to C90/105 (high strength concrete); the DIN can be
used up to C90/105 as well. For ultra high strength concrete (C135/150 to C180/210) only French
Ordinary concrete
Building codes know three levels of compressive strengths of concrete:
1. Characteristic compressive strength (f'ck)
2. Representative value compressive strength (f'crep)
3. Design value compressive strength (f'c)
The final value, the design value, is used for strength calculations. In the Dutch NEN 6720 its value is
defined by:
f' crep
f' c = (31)
mc
Parameter mc represents the material factor for concrete, with a value of 1,2. In case of accidental
loading situation (gas explosion, collision and impact load) the material factor may be reduced to 1,0.
For the representative compressive strength holds:
f' crep = 0 ,72f' ck (32)
The representative compressive strength is the long-term uniaxial compressive strength of concrete.
The short-term uniaxial strength is retrieved by multiplying the 28-days cube strength of concrete by
0,85, from now on referred to as u. The long-term strength follows from the product of the short-
term strength and another 0,85, referred to as t. The product of both scalars is 0,72 and has been
included in equation (32).
Please note that the Eurocode no longer uses the 28-days cube strength for f'ck, it has been replaced by
the 28-days uniaxial compressive strength from cylindrical test specimens.
The original German code DIN 1045 used one overall safety factor to include all risks of a structure at
once. Code DIN 1045 neu was adopted to implement the partial safety factors from the future
Eurocode in a German code before the actual introduction of the Eurocode. Therefore the DIN 1045
neu makes use of the uniaxial compressive strength f'ck as well:
f' crep = term f ' ck = 0 ,85f' ck (33)
The value of material factor mc from the DIN neu has a different value from the NEN: for in situ
structures the value holds 1,5; for prefabricated elements from a constantly supervised production
process the value is reduced to 1,35. In case of extraordinary loading situations a value of 1,3 is
sufficient.
For reinforced ordinary concrete the tensile strength is negligible compared to tensile forces in steel
reinforcement bars, therefore no tensile strength will be taken into account.
Material factors for steel reinforcement bars are equal in both the Dutch and German codes. The design
value for the steel strength fs is given by the equation:
f
fs = sk (34)
ms
Where fsk represents the characteristic steel strength. The value for the material factor for steel ms is
given by 1,15 in both codes. It may be reduced to 1,0 in case of extraordinary loading situation in the
German code and in case of gas explosions, collisions and impact loads in the Dutch code.
Table 10 gives an overview of all partial safety factors related to the strength of the material for both
concrete and steel. For a comparison of partial safety factors for concrete the ratio between the
Compressive stresses
In CUR Recommendation 97 a higher level of safety is required for concrete strength classes from
C53/65. The additional safety is requested because of the more brittle behaviour of high strength
concrete and has been included in an increment of the combined ratio (u t). This ratio should be
multiplied by the following scalar 'c:
785 f' ck
' c = (35)
720
In the German DIN 1045 neu the safety should be increased from C55/67 as well. The Germans
multiply the material factor of concrete by the scalar 'c:
1
' c = 1 (36)
f' ck
1,1
500
Although both methods appear very different, this difference is actually relatively small. Inversion of
the German increment and conversion of the included characteristic compressive strength from
multiaxial to uniaxial in the Dutch equation, gives:
1 f' 1 f'
NEN: = 1,1 ck and DIN: = 1,1 ck 1 (37) and (38)
' c 612 ' c 500
The CUR and the DIN both deal with reinforced concrete without the addition of steel fibres. Steel
fibres however ensure a more ductile behaviour of concrete, resulting in less brittleness. According to
[6] the recommended extra safety is superfluous for high strength, very high strength and ultra high
strength steel fibre reinforced concretes. Implying the same safety strategy from as before can be
applied for these ranges of concrete strength classes if steel fibres have been added.
Tensile stresses
The values of the partial safety factor for the influence of steel fibres f hold [1]:
f = 1,3 for standard loading situations
f = 1,5 for accidental loading situations
The new equations for the influenced parameters from the stress-strain diagram for ultra high
strength concrete in figure 27 are:
w 0 ,3 f ( w 0 , 3 )
0 ,3 = + el and fc = (39) and (40)
lc f E el fK
w 1% f ( w 1% )
1% = + el and f1% = (41) and (42)
lc f E el fK
Loading on a structure can be favourable and unfavourable relative to the overall loading situation.
Table 11 shows all loading factors for the Dutch and German codes. In the Dutch code all loading
factors can be reduced to the value of 1,0 in case of a gas explosion, collision or an impact load.
Table 11 | Partial loading factors NEN 6720 (safety class 3) and DIN 1045 neu
1)
NEN 6720; 2) DIN 1045 neu; 3) Reduced factors for gas explosion, collision or impact load
1 Markovic, I., High-Performance Hybrid-Fibre Concrete: Development and Utilisation, dissertation. Delft:
Forces and moments in the lining resulting from calculations on the tunnels ring behaviour, very
much depend on loads and support from the surrounding soil, which depend on its density and
stiffness. Introduction of loading factors to the loads by the soil does not necessarily imply a more
severe loading situation of the tunnel structure. Increasing loads normally result in an increasing ring
force. A higher ring force itself can impose a higher moment capacity of the segments and can limit
the rotations in longitudinal joints. Up to a certain degree of loading both given results influence the
overall ring behaviour of the lining in a beneficial way.
Risks related to the soil and therefore related to the loading and support of the tunnel should be
introduced in a different way than codes prescribe for ordinary tunnels. The next section therefore
discusses a overall safety philosophy for shield driven tunnels.
Both approaches know a combined loading factor that should contain several risk aspects.
approach A approach B
f'crep f'crep
f'c f'c
original original
f' f'
arctan mc c
arctan c
'c compr. strain (') 'c compr. strain (')
'
'c 'u 'c1 = c 'u
mc
83 | Adapted stress-strain diagrams in case material factors are excluded from the strength calculations
This study uses approach B to adapt the stress-strain diagram in case an overall safety factor is used
(approach 2). Preservation of the linear elastic behaviour is thought to be more important than
retaining the ratio between the compressive yield stress and the ultimate compressive stress.