Page 1



Experimental modelling on tunnel boring

Ten years of bored tunnels in The Netherlands

Engineering Highlights of RandstadRail in Rotterdam, The Netherlands The off-shore immersed tunnel in the Busan-Geoje Fixed Link project in South Korea 




This special edition of Geotechniek is powered by: P.O. Box 69 2600 AB Delft The Netherlands Tel. +31 15 269 3500 E-mail Rotterdam Public Works, Engineering Department P.O. Box 6633 3002 AP Rotterdam The Netherlands Tel. +31 10 489 6621 E-mail P.O. Box 33 6800 LE Arnhem The Netherlands Tel. +31 26 377 8911 E-mail

Contents 4

Experimental modelling on tunnel boring Adam Bezuijen Deltares / Delft University of Technology


Ten years of bored tunnels in The Netherlands K.J. Bakker Delft University of Technology A. Bezuijen Deltares / Delft University of Technology


Engineering highlights of RandstadRail in Rotterdam, The Netherlands D.C. van Zanten, V.M. Thumann Rotterdam Public Works, Engineering Department


The off-shore immersed tunnel in the Busan-Geoje Fixed Link project in South Korea ir. G. Meinhardt & ir. R.M.W.G. Heijmans ARCADIS Infrastructure

Preface The Netherlands, especially the densily


populated western part of the country, Geotechniek April 2008

is located several meters below sea-level. Generally speaking, the subsoil consists of

Special Issue on the occassion of the VI International Symposium Geotechnical Aspects of Underground Construction in Soft Ground Shanghai April 10th -12th 2008 Publication Uitgeverij Educom BV Mathenesserlaan 347 3023 GB Rotterdam The Neteherlands Tel. +31-10 - 425 6544 Fax +31-10 - 425 7225 E-mail Publisher Robert Diederiks Editorial Board Alboom, ir. G. van Barends, prof. dr. ir. F.B.J. Berg, dr. ir. P. van den Brinkgreve, dr. ir. R.B.J. Brok, ing. C.A.J.M. Brouwer, ir. J.W.R. Calster, ir. P. van Dalen, ir. J.H. van Deen, dr. J.K. van Diederiks, R.P.H. Eijgenraam, ir. A.A. Graaf, ing. H.C. van de Heeres, dr. ir. O.M. Jonker, ing. A.

Kant, ing. M. de Kooistra, mw. ir. A. Lange, drs. G. de Mathijssen, ir. F.A.J.M. Schippers, ing. R.J. Schouten, ir. C.P. Seters, ir. A.J. van Smienk, ing. E. Stam, ir. J.L. Thooft, dr. ir. K. Tigchelaar, ir. J. Veenstra, ing. R. Vos, mw. ir. M. de Wibbens, ir. H.G.P.

Editing Berg, dr. ir. P. van den Brouwer, ir. J.W.R. Diederiks, R.P.H.

Heeres, dr. ir. O.M. Kant, ing. M. de Thooft, dr. ir. K.

© Copyrights Uitgeverij Educom BV - April 2008 © ISSN 1386 - 2758

10-20 meters of very soft clay and peat layers. In addition there is a high groundwater level, almost up to the soil surface: a typical delta area. Delta areas all over the world are attractive places to live and to work. So, on the one hand, they have tremendous economic potential, on the other hand they live in the permanent threat of flooding and subsidence of the soft soil. Civil and hydraulic engineering are key factors in managing these challenges and providing sustainable solutions. Apart from the constant threat of water on the one hand and the soft soil on the other, the Netherlands face an additional problem as well. The claims on the available space in the Netherlands are becoming bigger and bigger. A solution to cope with this problem has been found into a new, additional direction: the underground. However, building underground in the soft soils of the Dutch delta is not without risks when the available knowledge and experience is limited. The last decade important steps have been made in order to be able to construct bored tunnels in the very soft soil conditions in the densily populated Western part of the Netherlands. Underground construction in a controlled way is an enormous challenge, especially the coming years, when a number of tunnels will be constructed underneath the city centers. An example: the new North-South metro line is being constructed underneath the old city center of Amsterdam. The sensitive structures surrounding the line are automatically monitored by a system of robotic total stations and prisms. A dedicated geographical information system automatically checks on an hourly basis whether individual measuring points have exceeded limit values. The limit values are based on the results of full scale tests and 3D finite element analyses. The large amount of monitoring data, every day several thousands of measuring points are checked for exceeding trigger values, requires special analyses. Once limit values are exceeded clear and comprehensive procedures have been defined to inform geotechnical engineers to take the most accurate action.

Many other examples can be worked out to indicate that there are still a lot of challenges for the geotechnical profession. Continuously new impulses are needed to comply with changing conditions: climate change, safety against flooding, settlement and failure of the soft subsoil, new concepts for infrastructure in a densily populated environment, increasing complexity in general in urban environment, asking for new technology and sophisticated risk control systems, and so on. This is a tough challenge for the geotechnical profession. Beside geotechnics, also mechanics, geophysics, geology, biology, chemistry, informatics, planology and law are involved to guarantee a multidisciplinary approach and to seek in close cooperation the best solution in situations of growing complexity. In our profession there is ample place for advanced technology and innovation such as new monitoring systems, on line data collection and interpretation, probabilistic methods to comply with weighted uncertainties, new materials, modified materials and ground mixing methods, further improvement of prediction models and methods of analysis directed to risk reduction and optimization of design, construction and maintenance. In this manner our profession remains valuable in the eyes of politicians, and policy makers, useful for the entire society and future generations. Dr. ir. O.M. Heeres Chairman of the Editorial Board of Geotechniek R.P.H. Diederiks Publisher

GEOtechniek – April 2008


Adam Bezuijen Deltares / Delft University of Technology


Experimental modelling on tunnel boring

Dutch experience in TBM tunneling is relatively recent. The knowledge development during the nearly ten bored tunnels up to now, was supported by experimental modeling on different scales. We will give two examples of the effectiveness of experimental modeling. The first concerns laboratory experiments

Introduction Dutch experience in TBM tunneling is relatively recent. In the early 1990s, Dutch engineers were uncertain whether the soft saturated soil in the western parts of their country was suitable for TBM tunneling. The first tunnelling projects were therefore accompanied by a research programme, with extensive monitoring during construction. The experience gained in every tunnel project was applied in the next one. The measurement results were analysed at a later date, and discrepancies with the predictions were explained where possible. Experimental research at a laboratory and model scale appeared to be crucial in understanding the observed phenomena.

It is important to consider that the thickness of the grout layer in a test should be identical to that in the field. This is to avoid scaling effects that occur because hardening of the grout is independent of the sample size. Some consolidation tests performed in a standard oedometer device with a sample thickness of 0.02 m over-predicted the maximum settlement with a factor of 4 compared to tests where a sample thickness of 0.2 m was applied, comparable to the thickness of the grout in the tail void. The volume loss found in these consolidation experiments was 3 to 8 %. Such a volume loss will lead to an unloading of the soil around the tunnel, possibly generating settlements of footings and pile foundations.

The abovementioned research programmes were aimed mainly at the processes at, in and around the TBM, like front stability, tail void grouting and flow of bentonite and grout around the TBM. A second important issue, however, in the soft saturated soil in the western parts of the Netherlands is the long term behaviour of the tunnel lining. Construction of the tunnel in the shallow clay layers gives inevitably rise to differential settlements of the soil surrounding the tunnel, and possibly also of the tunnel itself.

The composition of the grout is also important in compensation grouting. Experiments have shown that the fracturing behaviour in compensation grouting depends on the specification of the grout. If more cement is added, the permeability of the grout is higher and there will be more consolidation and leak-off during grout injection. At Delft University of Technology, the density of grout bodies made in two compensation grouting experiments was analysed in a CT-scan (X-ray Computerized

at a 1:1 scale on a small part of the system, particularly the bleeding behaviour of the grout from the tail void. The second example is a centrifuge experiment at scale 1:65 on a tunnel subjected to consolidation forces from the soft soil overburden.

Tomography). Such a CT-scan can be used to determine the density of the material tested. This was done with two grout mixtures, having a water-cement ratio of 1 and 10. The results of the CT-scans are shown in figures 1 and 2. The results of the first grout mixture show an increase in density from the injection pipe in the middle, to the outside boundary of the grout, from 1500 to more than 2100 kg/m3. Grout at the boundary of the sample is consolidated during the grout injection process. The grout body made with the second cementrich mixture has a more constant density across the fracture. The black lines in the figure are various contour lines showing the sharp increase in density at the boundary of the sample (figure 2). The more homogeneous density of the grout body in the second test is understandable

Grout bleeding or grout consolidation Consolidation of the tail void grout is an important process to understand the loading on the tunnel lining and the soil pressures in the direct vicinity of the tunnel. When the tail void grout is applied, the tunnel lining floats in a liquid with a density of up to 2100 kg/m3, since the average density of the tunnel itself (lining and air) is generally less than 1000 kg/m3. The flotation force has to be compensated by the yield stress in the grout, the weight of the TBM in front and by friction forces between the lining in the liquid grout and the part of the lining a bit further from the TBM in the solid grout. To minimize the forces and moments in the ’floating’ lining the liquid grout zone should be as short as possible The consolidation properties of the grout determine how long the grout will remain a liquid (chemical hardening due to the cement is a process with a longer time scale) and thus what will be the loading on the lining.


GEOtechniek – April 2008

0 3000

2500 50 2000

1500 100



150 0




Figure 1 Attenuation of x-rays over a grout sample with a WCR of 1. The surrounding air is blue. Red is the highest density (2100 kg/m3), green the lowest (1500 kg/m3).

Figure 2 Density of a fracture in sand with a WCR of 10. The blue is the fracture, the orange-yellow-green the sand.

Figure 3 Movement of consolidating clay around a tunnel. Result of centrifuge test.

gauges. After placement of the tunnel in the sand, a layer of Spesswhite clay slurry was applied and subjected to self weight consolidation. if the permeabilities of the grout are considered. The second sample has a lower permeability which results in much less grout consolidation within the limited injection time.

Consolidation pressure on the tunnel lining RandstadRail is a future light-rail link between Rotterdam, The Hague and Zoetermeer in the Netherlands. Building the Rotterdam section of RandstadRail involves the construction of two bored single-track tunnel tubes in the city area of Rotterdam. On several parts of the alignment the tunnel tubes are located at the boundary between the soft Holocene clay and the stiff Pleistocene sand, about 15 m below surface. It is expected that the top of the soft Holocene layers will settle 1.5 m due to consolidation and creep during the lifetime of the construction (100 y). Therefore the external loading on the tunnel lining will increase. The time dependent additional loading has been analyzed analytically as well as numerically. Since the mechanisms working on the tunnel wall and around the tunnel are known in principle but only partially in a quantitative sense, physical modeling using the Delft GeoCentrifuge, was performed in order to verify the design approach.

Scaling rules The basic purpose of centrifuge modelling in geotechnics is to increase the gravity artificially, such that the stress levels in the model are the

same as in the real situation. This is an important issue because material behaviour of soil (stiffness, strength) is strongly dependent on the stress level. For consolidation processes centrifuge modelling has an interesting side-effect. When the artificial gravity equals N times the natural gravity (Ng), the consolidation time is accelerated by a factor N2. This makes consolidation processes accessible for experimental study. Processes normally taking e.g. 10 years are at 100g accelerated by a factor of 104 implying a scaled process of about half a day. It should be kept in mind that it is not the time itself that is being scaled, but that geometrical effects cause an acceleration of processes. Settlement is not only created by consolidation, but also by creep: time-independent deformation under constant stress. Creep is basically a material property and only governed by the stress level, so the creep velocity is the same in model and in reality. Creep is therefore not sped up, and the settlement during the centrifuge experiment is caused by consolidation alone. The model was built at a scale of 1:65 in a strong box with a perspex window, through which a grid, applied on the clay, was observed with 2 video cameras. The model tunnel was made of an aluminum tube, for the lower part embedded in a dense sand layer. The tube was instrumented with pore pressure gauges and total pressure

After self weight consolidation had taken place in the GeoCentrifuge, a sand layer was applied in flight on top of the clay. To this purpose a special device has been designed that can be actuated hydraulically in flight. It is important to not stop the artificial gravity temporarily because stress release leads to deformations which are not realistic. This sand layer caused an overburden pressure of about 55 kPa. After consolidation of the clay a second layer of sand was applied leading to further consolidation of the clay layer. Figure 3 shows the deformation of the clay at the end of the consolidation of the first sand layer. The centrifuge tests allow seeing the processes in one day that will occur in the coming decades in the field. In these tests we checked the failure mechanism that was assumed in the design calculations. Although the assumed mechanism appeared to be right, the loading on the tube appeared to be smaller than expected, because the stiffness of the clay on top of the tunnel during consolidation was lower than the stiffness used in the calculations. With the results of the tests, the effects in the field can be predicted. When the field data corroborate these results, this implies that no steel lining is necessary when building a tunnel partly in soft Holocene soil layers. This would mean a significant reduction in lining and drilling costs. 

GEOtechniek – April 2008


K.J. Bakker Delft University of Technology A. Bezuijen Deltares / Delft University of Technology


Ten years of bored tunnels in the Netherlands

Ten years have passed since in 1997 for the first time construction of bored tunnels in the Netherlands soft soil was undertaken. Before that date essentially only immersed tunnels and cut-andcover tunnels were constructed in the Netherlands. The first two bored tunnels


initiative of the larger clients for underground

In 1992 a fact-finding mission was sent to Japan

infrastructure on the government side, was

by the Dutch government, which reported that

organised under supervision of the Netherlands

it should be possible to construct bored tunnels

Centre for Underground Construction; COB.

in the Dutch soft soil conditions. Up to that time

The research was organised in such a way that

essentially only immersed and cut-and-cover

results of a project would be beneficial for a

tunnels were constructed in the Netherlands,

next project starting a little later.

as boring of tunnels in soft soil conditions, at that time, was considered to be too risk full.

Unquestionably a lot has been learned from

After this conclusion things went quite fast; in

the performed monitoring and research. The

1993 the Dutch minister of Transport and Public

results of this process have been noticed abroad.

works ordered the undertaking of two pilot pro-

In 2005 the Netherlands hosted the fifth

jects, the 2nd Heinenoord Tunnel and the Botlek

International symposium of TC28 on Under-

Rail Tunnel. The projects were primarily aimed at

ground Construction in Soft Ground. Researchers

constructing new infrastructure and besides that

and experts from all over the world came to

for monitoring and research in order to advance

Amsterdam, to learn about the Dutch observations

the development of this new construction

on tunnelling and to visit the construction

method for the Netherlands. The projects started

works for the new North-South city metro

in 1997 and 10 years have passed since then.

system in Amsterdam.

After completion, the pilot projects have

The above event was also the occasion for the

triggered the start of a series of other bored

presentation of a book A decade of progress in

tunnelling projects.

tunnelling in the Netherlands by Bezuijen and van

were Pilot Projects, the 2nd Heinenoord tunnel and the Botlek Rail tunnel. Since then a series of other bored tunnels has been constructed and some are still under construction today. At the beginning of this period, amongst others Bakker (1997), gave an overview of the possible risks related to bored tunnels in soft ground and a plan for research related to the pilot projects was developed. After that in 1999 the 2nd Heinenoord tunnel opened for the public, the “Jointed platform for Bored tunnelling�, in short GPB, was organized, to coordinate further research and monitoring of bored tunnels. This platform is supervised by the Center for Underground Construction. In this paper a summary is given of some of the most characteristic observations on these 10 years of underground construction in the Netherlands.

Lottum (2006), where this research is described At the start of the pilot projects, the difficulties

in more detail. This paper gives some highlights

with respect to constructing bored tunnels in

of the main research result of the past decade.

soft soil conditions were evaluated and a plan

a major step forward, compared to past experience in the Netherlands; experience that was mainly based on constructing bored tunnels,

see Bakker (1997). Since then, the 2nd

Review of the 1997 situation and what came after

for monitoring and research was put forward,

pipes or conduits up to about 4.0 m diameter.

Heinenoord tunnel, see figure 1, and a series

In the design phase for the 2nd Heinenoord tunnel

This gave some concern with respect to the

of other bored tunnels were constructed.

a main concern were the soft soil conditions in

amount of extrapolation of empiric knowledge.

After the pilot projects a Joint Platform for

combination with high water pressures; in general

With respect to the soft-soil conditions, the low

Bored tunnels was established (GPB) that coordi-

in the Netherlands the water table is just below

stiffness of the Holocene clay and peat layers

nated the monitoring and research at the various

the soil surface. Furthermore the 8.3 m outward

and the high groundwater table; nearly up to the

other Dutch tunnelling projects. The GPB, an

diameter for this first large diameter tunnel was

soil surface, were considered a potential hazard and a challenge for bored tunnels. The soil profile at the 2nd Heinenoord tunnel, see figure 1, is indicative for the heterogeneous character and

Figure 1 Geological profile at the 2nd Heinenoord tunnel

on occasion sudden changes in underground soil layering, that one might encounter. In addition to the heterogeneity and the ground water, deformations due to tunnelling might influence the bearing capacity of any existing piled foundations in the vicinity. And as the common saying is that the Amsterdam Forest is underground, one might realize the potential risks involved for the North/South Metro works in Amsterdam. Characteristic for a high water table are buoyancy effects; the effect that the tunnel might be


GEOtechniek – April 2008

floating up into the soft upper layers above the

were involved with, without intent to minimize

tunnel due to the gradient in the groundwater

the importance of other research that is not

pressure. Besides the risk of breaking up of

discussed in this paper. Further issues related to

these soil layers, the rather flexible bedding of

groundwater effects and grouting are described

the tunnel and the deformations that this may

in more detail in a separate paper in this

cause need to be analysed.

symposium by Bezuijen & Talmon (2008).

Therefore research was aimed at clarifying the effects of the soft underground, groundwater effects, and the effect of tunnelling on piled

Experiences with bored tunnels in The Netherlands in the past decade


1 Structural damage After the successful construction of the two Pilot

An early experience with the difficulties for

projects, a number of other bored tunnelling

bored tunnels in soft ground was the damage

projects followed, see table 1. Mention worth is

to the lining that occurred during the first 150

that the Green Hart Tunnel holds until recently

metres of construction of the 2nd Heinenoord

the record as the largest diameter bored tunnel

Tunnel. On average the damage was too high

in the world.

compared to experiences from abroad and was

Still under construction are the tunnel for

considered to be unacceptable. Although, the

RandstadRail in Rotterdam, the Hubertus Tunnel

integrity of the tunnel was not at stake, there

for a road in The Hague and the North/South

was worry about the durability of the tunnel

metro works in Amsterdam.

and the level of future maintenance.

With respect to the construction of the North/

Characteristic to the damage was cracking and

South metro works in Amsterdam, the station

spalling of concrete near the dowel and notches,

works have made quite some progress and the

see figure 2. Quite often the damage was combi-

bored tunnel is in a preparation phase. The

ned with differential displacements between

elements of the immersed tunnel; the extension

subsequent rings and with leakage. The evaluation

to Amsterdam North under the river IJ, are

report, see Bakker (2000), attributed the damage

waiting for the completion of the immersion

to irregularities in the construction of the lining

trench under the Amsterdam Central Station.

at the rear of the TBM and subsequent loading

For the bored tunnelling part, the TBM is expected

during TBM progress. Further a correlation of

to start excavation at the end of 2008.

the damage with high jack forces was observed;

Ten years after the pilot projects, the question

these appeared to be necessary to overcome

arises whether the observations and related

the friction in this part of the track, which

research have confirmed the above issues to be

prevented smooth progress.

the critical ones or that advancing insight may

With respect to the tunnel ring construction, it

have removed these issues from the ’stage’ and

is difficult to erect a stress free perfect circular

swapped these for other topics giving more concern.

ring. The ring needs to be built onto the end of

Figure 2 Damage to the Dowel and notch sockets

Figure 3 Trumpet effect in tunnel ring construction of the (gray) inner ring is 8.3 m.

a former ring that already has undergone some In this paper some of the characteristic events

loading and deformation from the tail void grou-

and results of this past decade will be described.

ting while it partially has left the tail of the TBM,

The choice for the topics being discussed is

see figure 3.The further deformation is characte-

influenced by the projects that both authors

rised by the trumpet shape of the tubing that it

Completion Year

Bored length m

2nd Heinenoord tunnel







Western Scheldt tunnel







Botlek Rail tunnel







Sophia Rail tunnel







Pannerdensch Canal Rail tunnel







Green Hart tunnel







Table 1 Bored tunnels completed after 1997 in the Netherland

Depth m

Outward Diameter m

Figure 4 Large-scale tunnel ring testing in the Stevin Laboratories at Delft University (the diameter of the (gray) inner ring is 8.3 m.

GEOtechniek – April 2008


causes, with the inevitable related stress devel-

testing, the details of assembling tunnel seg-

already had collapsed. This collapse created a

opment in the lining. The trumpet shape and the

ments into subsequent tunnel rings and these

shortcut between the excavation chamber and

high jacking forces lead to local stress concentra-

into a tube were investigated. Amongst others

the river. The action of pumping air was noticed

tions and irregular deformations in the lining

the main results of the project were reported

by shipmasters on the river, who reported air

and occasional to slipping between the different

by Blom (2002), and Uijl et al (2003).

bubbles rising to the water surface, which caused

tunnel elements. The slipping of elements was

This research was fundamental for the choice

the failure to be known as the “blow-out”.

blamed to the use of kaubit in the ring joint.

to omit the dowel and notches for the Green

In this case the pumping of air had not been

Hart tunnel; which led to a nearly damage free

beneficial to the restoration of stability because

tunnel lining.

pressure loss was not the cause but one of the

Originally kaubit strips had been used in the ring

results of the event.

joint. These kaubit strips, of flexible bituminous like material, were used to prevent the occurrence

A different issue, not settled yet, is the durability

of stress concentrations; so some slipping was

of plywood and the consequences of wood rot

This frontal stability at the 2nd Heinenoord

meant to occur, but the “dynamic” character of

on the long-term tunnel behaviour. An unwanted

tunnel has attracted some public attention.

the slipping that actually occurred that influenced

loss of the longitudinal pre-stress of a tunnel

Presumably it is less known that loss of frontal

the final geometry of the lining and had

might influence the tunnel flexibility and defor-

stability has also occurred since then with some

triggered cracking was unexpected. Especially

mations, possibly leading to leakages. On the

regularity at the other tunnels under construction

the cracking and overloading of the dowel

other hand, experience learns that compression

in the years after, e.g. during construction of the

and notch system was unforeseen.

largely increases the durability of wood. The ply

Sophia Rail Tunnel and the Green Hart Tunnel,

Failure of the dowel and notch system, see figure

wood material is compressed to a strain of more

however without much delaying the construction

2. led to spalling and in some cases to leakage.

than 50% during tunnel construction. At such a

process. At the 2nd Heinenoord Tunnel, con-

In the cases that leakage was observed this must

high level of straining the wood cells might have

struction work was delayed for several weeks

have been correlated to damage to the notch at


before the crew succeeded in restoring frontal stability, filling up the crater in the river bottom

the outer side of the lining, creating a shortcut to water penetrating behind the rubber sealing

2 An instability of the bore front

with clay and bringing in swelling clay particles


During the construction of the 2nd Heinenoord

in the excavation room.

After the main conclusions were drawn, it was

Tunnel, approximately in the middle underneath

From the evaluation of the monitored pressures

decided to exchange the kaubit strips for thin

the river Oude Maas an instability at the excava-

in the excavation room, it appeared that before

plywood plates. Due to the larger stiffness and

tion front developed, see figure 5; afterward

the development of the instability, the frontal

shearing resistance, shearing of the concrete

commonly referred to as “The Blow-out” (see

pressure was raised above the advised pressure

elements at large was further prevented and

also Bezuijen & Brassinga, 2001).

for frontal support; i.e. at about 470 kPa instead

the damage limited.

Backtracking the situation learned that after

of about 310 kPa. see figure 5 (pressure gauge

Besides this technical measure, the evaluation

that a pressure drop was observed, in his efforts

P15 is in the excavation chamber at tunnel axis

was the trigger for the undertaking of funda-

to restore frontal support, the machine driver


mental research into lining design, that included

first pumped bentonite to the excavation chamber;

large scale physical testing of tunnel tubing at

considering a deficiency in the bentonite system.

In retrospect it was understood that during

Delft University, see figure 4. In this project that

When this did not help, air was pumped to the

standstill, the pressures were raised to get

was a combined effort of physical and numerical

bore front; not realizing that the front itself

a larger gradient in the pipes in order to

Figure 5 Support pressures before, during and after the ’Blow out’ at the 2nd Heinenoord tunnel.


GEOtechniek – April 2008

Figure 6 Pore water pressure distribution in front of the TBM

Ten years of bored tunnels in the Netherlands

improve the transport of excavated material;

Bezuijen et al (2001), indicates that it normally

meaningful pressure measurements would be

i.e. Kedichem clay that was found in the lower

takes about 4 to 5 minutes to build up a new

short and to prevent bridging effects the size of

part of the excavation front and appeared to

cake sealing after the excavation wheel has

the pressure cells would have to be large and

be difficult to pump through the hydraulic

removed the sealing during excavation. The time

therefore costly.

muck transport system.

between passings of chisels, in the order of 20

The measurements indicate that excavation

seconds is too short for that. It must be emphasized

Still, against this advice, the measurement of

had started without releasing pressure to the

that this effect is not only important for the

grouting pressures was undertaken, and repeated

standard support level during excavation.

upper limit to face support pressures, but may

for a number of tunnel projects. It appeared that

In that condition instability developed within

also give a limitation to the lower limit of the

the interpretation was difficult when the grout

15 seconds after that the wheel started cutting.

support pressure. A method to discount for this

has hardened, but for the fresh grout until 17

At stand still, when sufficient time has passed

effect was given by Broere (2001), see also figure 7.

hour after injection it was possible to give an

for a proper cake sealing of bentonite to build

The situation of a low soil cover underneath the

accepted interpretation of the measurement

up at the front, a high support pressure is not

river bottom is not the only situation that might

results (Bezuijen et al, 2004), and a lot of expe-

much of a problem, as the pressures used are

be critical to the above effect, also if the soil

rience has been gained that has contributed to

way below those that might override the passive

cover itself is relatively light, such as in the case

a better understanding of the grouting process

resistance at the front.

of the thicker layers of peat overlaying the sand

and the pressures acting on the tunnel lining.

However, as the pressure itself is fluid pressure,

where the Green Hart Tunnel was excavated,

With these results it was possible to predict

when the cake-sealing is taken away during

this might lead to a critical situation. A local

grouting pressures and tunnel loading, see

excavation, and water can penetrate the front,

failure might be triggered where the generated

Talmon & Bezuijen (2005).

according to Pascal's law for a fluid without

excess pore pressure in front of the tunnel face

shear stresses, the pressure also works in the

can lift the soft soil layers.

Based on various evaluations of the force distribution in the tunnel lining, see amongst others,

vertical direction, and if this pressure exceeds the vertical soil pressure this will trigger an

The knowledge gained with the monitoring of

Bakker (2000), it came forward that the initial

uplift and possibly a breaking out of soil layers,

the 2nd Heinenoord tunnel was applied for the

in-situ soil stresses around the tunnel do not

and apparently that is what has happened here.

Green Hart tunnel, and may have prevented

have a dominant influence on the compressive

In their paper on face support Jancsecz and

instabilities at the bore front at larger scales; see

loading of the tunnel. Due to the tapering of the

Steiner (1994), for this reason gave a warning

Bezuijen et al. 2001 & Autuori & Minec (2005).

TBM and in spite of the tail void grouting there is a significant release of the radial stresses

about the limits to the face support pressure,

around the tunnel, see figure 8.

for situations with little overburden.

3 Tail void grouting and grouting pressures

The final loading on the lining relates more to

Research learns that for the fine sand that we

To measure the soil pressures on a tunnel lining

the effectiveness of the grouting process, the

have in the Pleistocene sands layers in the

is difficult. In the start-up phase for the monito-

distribution of the grout openings and the

Netherlands, penetration of bentonite in the

ring of the 2nd Heinenoord Tunnel, a number of

consolidation of the grout than to the initial

pores is negligible. Excavation therefore means

international experts on tunnel engineering

in-site soil stresses, see Bezuijen et al. 2004).

removal of the cake sealing; Research by

advised not to put too much effort on this topic,

Whether this reduction of the in-situ radial

as “the results would probably be disappointing”.

stresses is a lasting effect that will remain for

Due to the hardening of the grout, the period for

the full lifespan of the tunnel may depend on the creep sensitivity of the soil, see Brinkgreve and Bakker (2001).

4 Surface settlements Hoefsloot et al. (2005), have shown that the application of a stress boundary condition between tunnel and soil in 3D tunnel analysis has a better corroboration between measurement and calculation of soil deformations around the tunnel and subsequently of the loading on the tunnel, than the application of the so called “contraction method”. Although this effect was known in the literature, see for example Mair (1997), for the research team for the 2nd Heinenoord tunnel the obser-

Figure 7 The effect of removal of the cake sealing during excavation on pore-pressures in the front. The influence zone for excess pore-pressures may be larger that the zone normally considered in stability analysis.

vation that the numerical predictions of surface settlements lacked accuracy was disappointing. Figure 8 Surface settlements; measured and

At the start the expectations on numerical analy-

back-calculated with different material models.

sis had been quite high. Shortly after the first

GEOtechniek – April 2008


urban areas, such as for the Amsterdam North-

figure 9. This means that the pressures on the

South line metro works. However, since then a

TBM tail are higher than anticipated in the past

lot of effort was put in the improvement of the

and there might be no bedding reaction. This

numerical prediction of soil deformations.

could well explain the occurrence of buckling and the deformations of the TBM tail. A 1-D

Figure 9 Under circumstances the Grout material from the tail void might flow into the gap behind the tail of the TBM, giving cause to increased loads.

To begin with it was the project team for the

calculation model has been developed and is

Amsterdam Metro works, see Van Dijk &

verified with FEM simulations (Bezuijen &

Kaalberg (1998), that gave a first indication for

Bakker, 2008). This model shows that also the

an improvement, with the proposal to model the

high stiffness of soil during unloading, which

stresses at the tunnel soil interface instead of

led to the HS and the HSsmall material models,

the deformations. With the introduction of this

made it likely that the common tapering,

grout pressure model the results improved. Later

approximately equal to an equivalent volume

on, when the physics in the process became better

loss of 0.4 %, is sufficient to lose the larger

understood, i.e. the importance to account for

part of the effective radial stresses, which

the high stiffness of the soil in unloading, dou-

helps to develop a gap between the tail of

ble hardening was introduced with the develop-

TBM and the soil.

ment of Hardening Soil, as a material model;

The grout pressures exerted on the tail of TBM

with this development, the calculation results

might be much higher that the soil stresses,

largely improved compared to the measure-

and in absence of bedding, buckling could well

ments, see figure 8. The latest development is

explain for the deformations.

the introduction of small strain stiffness in the to now gives the best results, see MĂśller (2005).

6 The influence of tunnel boring to piled foundations

Theoretically the result might further be impro-

Large scale testing of pile foundations was

ved introducing anisotropy in the model; such

performed during construction of the 2nd

models are being developed nowadays, e.g. in

Heinenoord Tunnel. This was done in order

the framework of European Research; AMGISS,

of the Project Bureau of the Amsterdam

e.g. see

North/South metro works to get a better

Hardening Soil Model, see Benz (2006), that up

Figure 10 Test site for the Pile-tunnel interaction test .

understanding of the processes,

5 Deformations of the TBM machine during construction of the Westernscheldt tunnel

A trial field with loaded piles and pile configurations was installed in the area near and above

During construction of the first tube for the

the track of the TBM, see figure 10. One of the

Westernscheldt tunnel, unexpected deformations

main concerns was that due to an increase in

of the tail of the TBM were observed; i.e. the air

pore pressure the effective stresses around the

space between tubing and tail of the TBM narro-

pile tip might be affected and that a release in

wed at a certain stage in an unexpected way.

isotropic stresses might trigger a drop in pile

The shape of the observed deformations did not

bearing capacity.

coincide with the assumed soil loading and gave

However, against this reasoning there is also

the impression that it was a large deformations

numerical and analytic evidence, (assuming

effect; i.e. buckling. However, at first buckling

cylinder symmetric analysis), that indicates that

was not accepted as a cause because the tape-

the release in stresses due to tunnelling is limited

ring of the TBM was assumed to give sufficient

to a rather small plastic zone in the close vicinity

stress release to guarantee a sufficient decrease

of the tunnel lining, see also Verruijt (1993).

observations were evaluated it was realized within

in isotropic stress. Further a certain bedding

The analytical model reveals that strain as a

the team, that it were only the empirical predic-

effect was assumed to be always present and the

function of the distance drops as a function of,

tions by Peck (1969) for a volume loss of about

combination would make buckling unlikely.

which would indicate that the influence zone

1 % that gave a reasonable corroboration with

Buckling would only be plausible for a much higher

would be limited in size. This reasoning in

the measurements. The finite element calculations,

loading of the tail of the TBM in combination

combination with the fact that the strains due to

at that time mainly based on an application of

with the absence of any bedding reaction.

tunnelling in general are quite small; the largest

the elastic-plastic Mohr-Coulomb model in

However, the insights have changed since then.

strains often being less than 0.5 or 1.0 %, makes

combination with a contraction type of model-

In general there may be no overall contact

plastic zones further away than D/2, measured

ling for the tail void loss, predicted a too wide

between the soil and the tail of the TBM; when

from the tubing, unlikely. Only above the tunnel

and too shallow surface settlement.

grout is injected in the tail void, the increased

this zone can be larger.

This disappointing result created a problem for

pressure on the soil, compared to the original

However, reasoning and analysis is one thing;

the intentions to apply 3D numerical analysis in

stress will push the soil from the TBM and grout

measuring and validation is another; based on

deformation predictions for tunnel projects in

will flow between the TBM tail and the soil, see

the field measurements and physical model

Figure 11 Zones that indicate different effects on piles foundations.


GEOtechniek – April 2008

Ten years of bored tunnels in the Netherlands

research in Delft and Cambridge Kaalberg et al.

itself has turned out to be cumbersome. Within

remain small, so the soil reactions will most

(2005), proposed a zoning as shown in figure 11,

the monitoring scheme for the 2nd Heinenoord

probably have mainly been elastic.

with the following indicators; a zone ’A’ above

a trial measurement was undertaken. In addition

The main conclusion with respect to this effect

the tunnel where the settlement of a pile is

to that measurements from the Sophia Rail

was that this issue can be properly analysed with

expected to be larger than the soil deformations.

Tunnel were back-analysed with 4D finite

a relatively simple model based on the concept

A zone ’B’ adjacent to the tunnel, with an incli-

element analysis, and after that the longitudinal

of a beam with an elastic bedding and a series

ned influence line, where the pile will follow the

stresses were also measured during the con-

summation, such as developed by Boogaards

soil deformation at the tip of the pile, and furt-

struction of the Green Hart Tunnel.

(1999), and later on applied by Hoefsloot (2002).

her a zone ’C', outside Zone B, where at soil sur-

To begin with the latter situation: measurements

For the model concept see figure 12 and figure 13

face level the settlement of the pile will be less

were taken with a tubular liquid level devise of

for a comparison between model outcome and

than that of the soil surface. This zoning propo-

the longitudinal deformations of the tunnel

measurements. However, using generally accep-

sal more or less coincides with the main results

during the grouting process. From these measu-

ted parameters, the measured deformations are

as published by Selemetas (2005) that were

rements the observation came forward that the

much higher than according to these models.

mainly based of physical testing in a geotechnical

tubing exhibited large vertical movements,

Recently, Talmon et al. (2008) have presented


up and down, between 20 to 30 mm during

results that may explain the lower stiffnesss that

The results published by Kaalberg et al. and

excavation and tail void grouting was measured,

are found in the measurements (the lining stiff-

others are valid for the average volume loss that

and a total vertical shift of the tubing vertical

ness can be lower due to only local contact

can be expected during tunnelling (0.5 to 1%)

of about 60 mm at one location (See also Talmon

between the elements and the soil stiffness

Earlier centrifuge testing by GeoDelft indicated

& Bezuijen, 2008).

reduces due to unloading of the soil around the

that larger deformation effects are possible for

This amplitude was surely unexpected and is

tunnel), but these are not yet generally accep-

higher volume losses (up to 7% was tested).

not fully accepted yet. Nevertheless it is clear


Such volume losses are well above nowadays

that vertical deformations do occur in the zone

practice, but it means that during a calamity,

where the grout material is still fluid, and during

Cross passages

piles over a larger area may be affected

excavation and may lead to an alternating defor-

The design for the Westernscheldt tunnel in the

mation; upwards when the TBM is excavating

Netherlands did trigger a debate on tunnel safe-

7 Longitudinal deformations of the tunnel tube

and during grouting and downwards if the TBM

ty. Some major accidents with tunnel fires, such

In a paper by Bakker (1997), the development of

is at stand still.

as occurred in the Channel tunnel and at the

longitudinal stresses in a tunnel lining due to

With respect to the 3D staged construction

Mont Blanc tunnel in the Alps did reveal the

irregular bedding in soft soil was mentioned

analysis of tunnel construction for the Sophia

vulnerability and relative unsafe situations in

as an item for research. Irregular bedding that

Rail Tunnel, that was undertaken for the COB

tunnels with oncoming traffic or in a single

could be the result of zones with different elasti-

F220 committee, a combined DIANA and PLAXIS

tube in general.

city or else due to the stiff foundation of a shaft

3D analysis was performed, see Hoefsloot et al,

For the Westernscheldt tunnel, a twin tunnel

or bedding in the deeper Pleistocene layers;

(2005). The outcome of these various analyses

with one way traffic per tube, the discussion

especially near the transition between Holocene

more or less coincided; which might have been

focussed on what distance between cross

and Pleistocene layers.

expected as the mathematical base of both

passages would be acceptable to guarantee that

The measurement of longitudinal stresses in

models is quite similar, and in general deformations

escaping people would be able to find a safe

Figure 12 Conceptual model for the analysis of beam effects in the tube of a bored tunnel by Boogaards (1999) Thorbeckerveld Gouda.

Figure 13 Langetermijngedrag uit centrifugetest, Stewart, Jewel & Randolph, 1996 [11,12].

GEOtechniek – April 2008


haven by entering the other tube; assuming that

may think of a loading that may be on the level

deformation behaviour of sand, see Benz (2006),

the traffic is stopped, by an automatic control

of the initial soil stresses before tunnel construc-

Further it is recommended, and planned for, to

system. The outcome of these safety studies was

tion; the Ko stress situation or even higher.

integrate the Delft Cluster Grout pressure model

a cross connection at least every 250 m, which is

Such a situation was accounted for in the design

in the Plaxis 3D Tunnel software. The latter

nowadays more or less the reference situation in

for RandstadRail in Rotterdam, where a full steel

would contribute to the applicability of the

the Netherlands.

lining was chosen for a part of the track where

numerical models as a more general tool for

The task to construct these cross passages is a

the tubing mainly rests in the upper much softer

underground construction. This would enable

further technical effort. During the construction

Holocene clay and peat layers, that foreseeable

a better analysis for the loading on the tail of

of the Botlek Rail Tunnel a vertical shaft and

would have an extra loading on the lining due to

the TBM and of the tunnel lining.

freezing were the main construction techniques

consolidation and creep (Pachen et al. 2005).

Within certain limits some cost saving structural

as the cross passages could be positioned outside

Nowadays it's not the soil deformation during

improvements are expected to be possible and,

the area under the Oude Maas River. The positive

“normal� excavation process that makes us

even more important, insight is obtained in the

experience with freezing for the Botlek Rail

worry. With an average tail void loss of about 0.5

mechanism involved.

Tunnel was helpful in the decision making for the

% of the diameter or less, the deformation might

Westernscheldt Tunnel, but there the freezing

only be a problem for situations of under-exca-

Concluding remarks

was done from the tunnel tube as the track

vation of buildings or if the structures are located

Ten years have passed since the first large

underneath the estuary is too long and too deep

very close to the excavation track. For tunnels in

diameter bored tunnelling project in the

with respect of the water table to enable the

urban area, there is more concern with respect to

Netherlands in Soft soil was undertaken. Since

shaft type method.

bore-front stability; especially when the upper

then some world records with respect to tunnel-

Although the method in itself is costly, its

stratum of the soil above the Pleistocene layers,

ling have been broken in the Netherlands; i.e.

reliability is an important advantage and there-

where the tunnels are usually positioned in,-

the largest diameter (for the Green Hart Tunnel),

fore it is also used for the cross passages of the

consists of soil with a relative low density, as in

the highest outside pressure on a segmental tun-

Hubertus Tunnel and is expected to be used in

the Netherlands. For the situation of a relatively

nel (for the Westernscheldt Tunnel), the applica-

future projects. For the single tube Green Hart

light upper stratum with peat or clays with orga-

tion of an Earth Pressure Balance shield in coar-

Tunnel tunnel safety is achieved by construction

nic parts, one has to be very careful in control of

se sand, and the largest length of constructed

of a separation wall with doors.

the support pressures during excavation, as on

tube in one day, (Pannerdensch Canal Tunnel).

the one hand there is a lower bound value of the

Before the underground construction works

Evaluation of ther learning isues

support to prevent cave in, but on the other

were started, and the tunnelling projects were

The research on grout pressures, in combination

hand, the upper limit triggered by an uplift of

in a pre-design stage, the softness of the

with the structural research on lining design has

light upper layers may also be not far. This will

Netherlands underground attracted a large part

gained us the insight that the lining thickness

limit the pressure window to work in.

of the attention, see Bakker (1997). In retrospect the influence of a low stiffness as a source of

and the necessary reinforcement are mainly determined by the loading in the construction

Front instability has occurred at various tunnel-

risk and influence on underground construction

phase and to a lesser degree to the soil pressures.

ling projects in the Netherlands. If the tunnel

was confirmed, but sometimes in a different

In engineering practice the thickness and rein-

is outside any urban area this might not give

perspective, or related to other physical proces-

forcement of the tubing is mainly determined by

too much problems; however if the tunnel is

ses than foreseen.

the most unfavourable jack-forces during TBM

underneath a city road system, or close to pile

With respect these new insights the following

excavation in combination with an unfavourable

foundations this may cause severe problems,

conclusions were drawn:

tail void grouting scenario. Difficulty with these

as instability might cause a sinkhole in the

is, that it's the contractor's prerogative to decide

pavement and foundation settlements.

1. The low stiffness of the ground support may

on the necessary jack-forces that will enable him

With respect to the accuracy in the prediction of

give rise to increased vulnerability of the lining

to construct the tunnel and also what scenario he

soil deformations: Apart from the well known

for jacking forces by the TBM during excavation.

will use for the tail void grouting. This might

empiric model of Peck (1969) that predicts the

Care must be taken to precise shape of the ele-

lead to conservative assumptions in the design

shape of the settlement trough but not the

ments and joints to prevent too high stresses

office in order to avoid liabilities if a problem

volume loss, the numerical models have become

during assembly.

would occur during construction.

quite reliable in predicting surface and subsurface

With respect to the generality of this conclusion

deformations, both vertical and horizontal.

2. The low stiffness of the soil may also lead to

it has to be considered that the main observations

The improvement, mainly achieved in 2D analysis

increased flexibility of the tunnel tube. The

that were discussed relate to tunnels that are

has opened up the possibility for a reliable

deformation of the tube during hardening of

safely located in stiff Pleistocene sand layers.

deformation analysis in 3D of tunnelling in

the grout, and the additional Eigen stresses that

We must however consider the possibility of

urban areas. For an adequate prediction of

this may cause is still a research topic.

tunnels in softer soil layers that are more suscep-

deformations it is important to model the

3. For a proper prediction of surface settlements

tible to consolidation and creep. The consolida-

grouting pressures as a boundary condition

and soil deformations, it is important to model

tion and creep can counteract the general ten-

to the excavation, in combination with the

the grouting pressures at the interface between

dency of stress release and arching in the soil

application of a higher order material model,

soil and tunnel (or grouting zone).

and lead to a much higher radial loading. One

that takes into account the small strain

Further to improve the prediction of the width of


GEOtechniek – April 2008

Ten years of bored tunnels in the Netherlands

the settlement trough, the use of small strain


– Broere, W, 2001, Tunnel face stability &

analysis is advised.

– Autuori, P. & S. Minec, 2005, Large diameter

new CPT applications, Delft University Press.

4. During excavation in fine sand, such as the

tunnelling under polders, Proceedings 5th Int.

– Hoefsloot, F.J.M. & K.J. Bakker, 2002,

Pleistocene sand layers in the Netherlands,

Symposium on Underground Construction

Longitudinal effects bored Hubertus tunnel

during excavation the supporting cake fluid will

in soft Ground, IS-Amsterdam 2005.

in The Hague, Proceeding 4th Int. Symposium

be removed. In case of limited overburden the

– Bakker, K.J., P. v.d. Berg & J. Rots, 1997,

on Underground Construction in soft Ground,

upperbound to the support pressure must be

Monitoring soft soil tunnelling in the


carefully determined to prevent instability of

Netherlands; an inventory of design as-pects,

– Hoefsloot F.J.M. & A. Verweij, 2005,

the overlaying soil.

Proc. ISSMFE, Hamburg.

4D grouting pressure model PLAXIS, Proceeding

– Bakker, K.J. 2000, Soil Retaining structures,

5th Int. Symposium on Underground Construction

5. In addition; for the determination of the

development of models for structural analysis,

in soft Ground, IS-Amsterdam 2005

lower limit to the support pressure, the increa-

Balkema, 2000, Rotterdam.

– Kaalberg F.J., Teunissen E.A.H., van Tol A.F.

sed pore pressures in the front also needs to be

– Benz, T. 2006, Small strain stiffness of soils

and J.W. Bosch, 2005, Dutch research on the

taken into account.

and its consequences, Doctor Thesis,

impact of shield tunnelling on pile foundations.

IGS Universität Stuttgart.

Proceedings of 16th ICSMGE, Osaka.

6. The stiffer Pleistocene sand layers might not

– Bezuijen, A. & Brassinga H.E.,2001.

– Jancsecz, S. and W. Steiner, 1994, Face Support

always be able to follow the tapering of the

Blow-out prsessure measured in a centrifuge model

for a large Mix-Shield in heterogeneous ground

TBM. It is expected that this may give rise to

and in the field. Proc. Int. Symp. on Modern

conditions, Tunnelling 94, British Tunnelling

gapping behind the tail of the TBM. If grout

Tunnelling Science and Techn. Kyoto.

Association, 5-7 July 1994

penetrates this gap, this may cause higher loads

– Bezuijen, A., A.M. Talmon, F.J. Kaalberg and R.

– Möller S.C. & P.A. Vermeer Prediction of

on the TBM than is normally assumed.

Plugge, 2004, Field measurements of grout pressu-

settlements and structural forces in linings due

res during tunneling of the Sophia Rail Tunnel.

to tunneling, Proceeding 5th Int. Symposium

7. No proof was found that tunnel driving in

Soils and Foundations vol, 44, No1, 41-50, Feb.

on Underground Construction in soft Ground,

normal operation might give cause to loss of

– Bezuijen, A. & A.M. Talmon, 2006, Grout pro-

IS-Amsterdam 2005

bearing capacity of piles. Settlements in general

perties and their influence on back fill grouting.

– Pachen, H.M.A., H. Brassinga & A. Bezuijen,

are related to the settlement of the ground and

Proceeding 5th Int. Symposium on Underground

2005, Geotechnical centrifuge testst to verify

the position of the pile toe with respect to the

Construction in soft Ground, IS-Amsterdam 2005

the long-term behavior of a bored tunnel, Proc.

zones indicated in figure 11. 

– Bezuijen A., Pruiksma J.P., and H.H. van

5th Int. Symposium on Underground Construction

Meerten, 2001, Pore pressures in front of tunnel,

in soft Ground, IS-Amsterdam 2005.

With acknowledgement to the

measurements, calculations and consequences for

– Peck, R.B., 1969, Deep excavations and

Netherlands Centre for Underground

stability of tunnel face. Proc. Int. Symp. on

Tunnelling in soft Ground, Proceedings 7th

Construction for their consent to publish

Modern Tunnelling Science and Techn. Kyoto.

ICSMFE Mexico.

about the research they commissioned

– Bezuijen, A. & H. van Lottum (eds), 2005,

– Selemetas D., J.R. Standing and R.J. Mair,

and coordinated, and with thanks to

Tunnelling A Decade of Progress. GeoDelft

2005. The response of full-scale piles to tunnelling.

Cees Blom for the use of some of


Proceeding 5th Int. Symposium on Underground

the figures.

– Bezuijen & Bakker, 2008, The influence of

Construction in soft Ground, IS-Amsterdam 2005.

flow around a TBM machine, Proceeding 6th Int.

– Talmon, A.M. & A. Bezuijen, 2005, Grouting

Symposium on Underground Construction in soft

the tail void of bored tunnels: the role of hardening

Ground, Shanghai.

and consolidation of grouts. Proceeding 5th Int.

– Bezuijen A. & A.M. Talmon, 2008, Processes

Symposium on Underground Construction in soft

around a TBM, Proceeding 6th Int. Symposium

Ground, IS-Amsterdam.

on Underground Construction in soft Ground,

– Talmon, A.M. & A. Bezuijen, 2008, Backfill


grouting research at Groene Hart Tunnel.

– Blom C.B.M. 2002, Design philosophy of

Proceeding 6th Int. Symposium on Underground

concrete linings for tunnels in soft Soil,

Construction in soft Ground, Shanghai.

Delft Univ. Press, The Netherlands

– Uijl, J.A. den, A.H.J.M. Vervuurt, F.B.J.

– Bogaards P.J., Bakker K.J. 1999, Longitudinal

Gijsbers and C. van der Veen, 2003, Full scale

bending moments in the tube of a bored tunnel.

tests on a segmented tunnel lining. In Proc. ITA

Numerical Models in Geomechanics Proc.

World Tunnelling Congress 2003,

NUMOG VII: p. 317-321

Amsterdam, The Netherlands, 12-17 April 2003.

Brinkgreve, R.B.J. & K.J. Bakker, 2001, Time-

– Verruijt, A. 1993, Soil Dynamics, Delft

dependent behaviour of bore tunnels in soft soil

University of Technology.

conditions; a numerical study, Proceedings, ICSMGE Istanbul, Turkey.

GEOtechniek – April 2008


D.C. van Zanten, V.M. Thumann Rotterdam Public Works, Engineering Department

Abstract RandstadRail is a new lightrail connection between the cities of Rotterdam, The Hague and Zoetermeer (The Netherlands). A tunnel of three kilometres length is being built at present to realize RandstadRail in Rotterdam. The project contains building of one new station (Blijdorp) and re-building of an existing underground station (Rotterdam CS). Latter station will be transformed into a three-track configuration with two platforms while regular traffic is not to be interrupted by building activities. Within this project a number of special techniques have been used aiming at minimizing the effect of the building activities on every day life within the city of Rotterdam. This paper gives a quick-scan of the engineering highlights of the project. Special attention is paid to the shield-tunnelling in soft clay as well as the design of the building pit of Central Station in which ground freezing techniques are applied.

Engineering highlights of RandstadRail in Rotterdam, The Netherlands Introduction The four major cities of The Netherlands, Amsterdam, Rotterdam, The Hague and Utrecht, are situated in the western part of the country. Within this densely populated area, there is an increasing demand for public transport on a high service level. One of the keyprojects in this context is RandstadRail. RandstadRail is a lightrail line, by which it will be possible to travel from the centre of Rotterdam to the towns and cities in northern direction without transfer. In Rotterdam, RandstadRail will be linked to the existing metro line (Erasmus line) at its terminal station Rotterdam CS, which has to be enlarged to provide sufficient passenger transfer capacity. Re-building of the underground station Rotterdam CS as part of the RandstadRail

project is closely related to the overall project Rotterdam Centraal. This major project comprises the building of a large OV (public transport) terminal in the centre of Rotterdam, in the vicinity of the Rotterdam Centraal railway station. It is designed to facilitate passenger transfer between (inter)national trains including High Speed Line (HSL) and local public transport like trams, buses and metro/lightrail. Inside Rotterdam urban area, a new tunnel has been built since 2004 to create the connection between RandstadRail and the existing Erasmus line. Subject tunnel is approximately three kilometres in length, and has one underground station (called Blijdorp) halfway (figure 2). Due to the fact that several infrastructure (railway, highway as well as waterway) and the inner

Figure 2 Alignment of RandstadRail in Rotterdam.

Man made soil (sand fill)

Photo: D. Sellenraad; Aeroview

Figure 1 Bore tunnel

city of Rotterdam had to be crossed, 80% of the tunnel length is built by means of shield tunnelling technique. The remaining part is constructed through conventional cut and cover method. RandstadRail Rotterdam is the first tunnelling project in The Netherlands that has been executed in soft soils conditions and in densely populated urban area. This paper presents a quick-scan on some of the engineering aspects of the bored tunnel and the new underground station Rotterdam CS.

Soil Conditions The general subsurface conditions, which are typical for the Rotterdam region, are summarized in Table 1. The ground water level along the line is approximately 2.5 metres below reference level (NAP).

Bored tunnels main design considerations An extensive study regarding the options of one double track tunnel (1*Ø11.2m) versus two single











-0.3 (surface)

GEOtechniek – April 2008







Clay type a
















Clay type a








Clay type b








Pleistocene sand








Kedichem clay








Kedichem sand








Table 1 Average soil parameters.



track tunnels (2* Ø6.5m) led to the recommendation to use a configuration of two single track tunnels. This recommendation was based on (i) lower estimated building costs and (ii) lower risk profile of the project due to the greater influence on infrastructure when crossing with a larger TBM. It is noted that the sharp horizontal curves of the tunnel alignment (radius 240 metres) gave an additional technical reason for choosing two single track tunnels. Passenger safety is incorporated in the overall design of the tunnel through the safe haven concept. All underground stations are designed to be safe haven. In case of an emergency situation, standard procedure will be that trains run to one of these safe havens. In the unlikely event that this procedure fails and a train stops somewhere in the tunnel, additional cross connections (each 350 metres) are available for escape to the other (safe) tunnel. The cross connections are built using ground freezing technology (brine coolant), to allow for excavating from one tunnel to another. Shotcrete with thickness 300 mm is applied for stabilising the excavation. The 6.5 m external diameter bored tunnels have a concrete lining of 0.35 metres thickness. Concrete design strength is 55 N/mm2 (B55). Each tunnel ring is 1.5 metres in width. The tunnel rings have a conicity of 5 centimetres for realising the horizontal and vertical curves. Every ring consists of eight (including one key-stone) pre-cast concrete segments. The tunnel rings are placed in stretching bond and are connected by specially designed concrete dowels. The bored tunnels are placed mainly into the Pleistocene sand layers, which provide suitable soil conditions for this type of tunnel structure (figure 3). However, near the starting shaft in the Northern part of Rotterdam as well the receiving shaft near the existing underground station Rotterdam CS, the tunnel had to be placed in the soft Holocene clay. Reason for this are the fixed connections of RandstadRail to the already present rail infrastructure. The soft clay does not provide sufficient support to the concrete lining. Also, continuously developing settlement of the Holocene clay (~1 cm/year on surface elevation) will result in additional time dependent forces on the tunnel. These two problems have been solved by use of ground improvement techniques for the shallow parts of the tunnel alignment. Several ground improvement techniques have been used, considering the typical conditions and circumstances at each individual location. Soil deep mixing and soil replacement

Figure 3 Vertical alignment of bored tunnels; distance start tunnel – Blijdorp station – Rotterdam CS station ca. 2,0 & 1,2 km.

Figure 4 Jet grouting

near the receiving shaft.

have been executed near the starting shaft. Jet grouting has been applied near the receiving shaft (figure 4). It was observed that, when realising the ground improvement techniques of jet grouting and soil deep mixing, substantial horizontal deformations occurred (several centimetres), but this did not cause any damage. However, when using these techniques in densely built areas, these deformations should be anticipated for in the design. Ground improvement techniques could not be applied at all desired locations, for example near the receiving shaft which is located underneath the Rotterdam main railway station embankment consisting of 16 railway tracks, and near future station Blijdorp where the tunnels are located under a road. Using ground improvement techniques at these locations was not advisable as this would lead into influencing rail and road traffic. At these locations a steel tunnel lining is used, for a total of ca. 5% of the bored tunnel length. However, for this type of lining it is required to consider the possibility of stray currents originating from nearby railway tracks. Corrosion of the steel lining as induced by stray currents had to be avoided, which resulted in application of a steel lining with special coating and cathodic protection.

Tunnel boring process The boring process was done using a TBM with a slurry shield (figure 5, next page). The length of the TBM was 68 metres. The length of the shield was 9.8 metres. The machine had a diameter of 6.78 metres and contained a 5-arm cutting wheel. A pivoting joint between middle and tail shield was built-in to make passage of the tunnel curves having relatively small radii possible. The tunnel boring process started in December 2005. When crossing the sealing block, consisting of low strength concrete (expected strength ~15 MPa) and a diaphragm wall (expected strength ~45 MPa), substantial wear of the cutters was noticed. Here, the cutting wheels had to be changed several times. When boring in the original soil conditions (sand, clay) no significant wear of the cutters was noticed. The boring speed was ~40 mm/minute. When crossing the ground improvement areas (jet grouting, soil deep mixing) the boring speed was reduced (~20 mm/minute). Crossing the diaphragm wall the boring speed was even lower (~1 mm/minute), so passing the diaphragm walls (thickness of 1,5 metres) took several days. The influence of the boring process to the environment was relatively small. At surface level settlement of 10-15 mm was encountered when

GEOtechniek – April 2008


crossing the railway tracks. In the design phase of the project these settlement have been determined using the simple formula of Peck (1969), as well as a finite element calculations (Zanten, 2002). The actual surface settlement was small, but larger than expected. Post-diction of the settlement along the line resulted in a higher volume loss than initially expected (table 2).

retracting of the piles was very difficult (e.g. under the embankment for the railway tracks). At these locations the TBM had to pass through the piles. The RandstadRail project experience on boring through wooden piles is good. No influence on TBM face stability was encountered, just some minor effect on the bentonite cycle was noted. In February of 2008 the tunnel boring process was succesfully completed.

Underground station Rotterdam CS

δs;x δmax V x i k z

surface settlement at distance x from tunnel; maximum settlement; volume of settlement through; horizontal distance from tunnel axis; horizontal distance of tunnel axis to point of inflection; indicator for stiffness trough; depth tunnel axis.

Also, settlements in the vicinity of the tunnel have been measured. This is shown in figure 6 for one location. From this graph, it can be seen that settlement near the tunnel is twice as high as settlement at surface level. This effect has to be accounted for in projects containing tunnelling underneath pile foundation structures. Within the RandstadRail project, no such structures are located directly above the tunnel. Other obstacles have been encountered, such as remnants of wooden pile foundations. The design philosophy of the project is to retract these piles if possible. However, at some locations

The existing ’Erasmus’ underground tunnel, of which underground station Rotterdam CS is the end stop, was built in the period 1962-1967. The tunnel was assembled from prefabricated segments, which were built in dry docks. The tunnel segments were floated to their final destination through a canal (sunken tubes). Once arrived on the spot, the segments were sunk onto their permanent foundation which consists of pre-installed concrete ’oppers’-piles. The present underground station Rotterdam CS having a two-track lay-out and single platform will be transformed into a three-track configuration with two platforms. The building pit as required for the reconstruction works of underground station Rotterdam CS covers ca. 7500 m2 (figure 7a, b). Governing design condition is that regular underground traffic and passenger transfer at the existing underground station is not to be affected during the construction works, thus no damage (e.g. cracks, water leakage) to the tunnel due to the works is allowed. The existing underground tunnel reaches to depth 10 m below surface. Excavating is done to 14 m depth to allow for installation of new

foundation piles underneath the tunnel. Therefore, the surrounding soil around the existing tunnel had to be removed completely. Consequently, as the tunnel elements have not been designed to carry any horizontal loads under these circumstances, special supporting frames have been put in place (figure 8a, b). The excavating method itself is based on isolating the water carrying Pleistocene sand layers inside the building pit by means of a diaphragm wall to a depth of 38.0 m below reference level (NAP). This so-called “Kedichem”-method provides a water regime inside the building pit which can easily be maintained, as only a very limited amount of water is expected to pass through the diaphragm wall and the low permeability clay/peat/loam layers of the Kedichem formation below depths NAP -35 m. The existing underground tunnel enters the building pit at the east side of the construction site (figure 7). Closing the building pit at this location requires a watertight solution, especially at shallow depths. Final design for closing the gap in the cut-off wall at this location by means of a collar construction around the underground tunnel comprises application of ground freezing technology (figures 9,10). The collar construction consists of a frozen soil body, thus creating an arch-shape collar construction supported by long diaphragm wall sections to carry the hoop forces as induced by

Prediction Post diction

Average volume loss % 0,5 1,1

Table 2 Parameters for formula Peck.

Figure 5 RandstadRail Tunnel Boring Machine (TBM).


GEOtechniek – April 2008

Figure 6 Settlement in time above tunnel

k 0,4 0,35



Figure 7a Picture of

the building site. 7b Top view of Stations-

plein. contour lines of excavation (diapragm walls, yellow) and re-built underground station are shown. On the right: location of collar construction (green). 8a

Figure 8a Fixation

frame for stabilising horizontal position of station. 8b Piling equipment

working on installation of new pile foundation below existing tunnel.

excavating. The frozen soil body is generated through two parallel rows of vertical freeze pipes to 40 m depth. Freeze-up is done using combined brine (inner row) and liquid nitrogen (LN2, outer row) freezing. During maintenance, freezing is done using the brine pipes only. The LN2 facilities remain standby on site for back-up reasons. Some key-data are as follows:  Collar construction retaining height ~14 m (incl. hydrostatic water pressures).  Minimum thickness 2.5 m (i.e. frozen soil volume at least ~4000 m3).  Approximately 100 tubulars installed: 50% brine freeze pipes, 30 % LN2 freeze pipes and 20% temperature monitoring pipes.  Additional temperature monitoring of connection between tunnel structure and frozen soil.  Distance between freeze pipes ca. 0.9 m (also inside existing tunnel and in between rail tracks).  Freezing operation for at least one year.

Conclusion Building the tunnel for RandstadRail to the existing underground infrastructure has been done under very strict design conditions, aiming at minimizing the effect of the building activities on every day life within the city of Rotterdam. For this reason, some special techniques a.o. shield tunnelling in soft clays and combined brine and LN2 ground freezing, have been applied on unprecedented scale within urban area in The Netherlands. The connection between RandstadRail and the existing underground station Rotterdam CS is expected to be in service by end of 2009. 


Figure 9 Plan of designed collar construction around the tunnel at station platform elevation, showing frozen soil area (blue) and supporting diaphragm walls (green).

References – Peck. R.B.; Deep excavation and tunnelling in soft ground; Proceedings of 7th International Conference on Soil Mechanics and Foundation Engineering; Mexico, 1969. – Thumann, V.M., Hass, H.; Application of ground freezing technology for a retaining wall at a large excavation in the centre of Rotterdam, The Netherlands; Proceedings ECSMGE 2007 Madrid. – Zanten. D.C. van; Tunnelling for RandstadRail in Rotterdam; 28th ITA General Assembly and World Tunnel Congress; Australia, 2002.

Figure 10 Artist impression of collar construction.

GEOtechniek – April 2008


ir. G. Meinhardt & ir. R.M.W.G. Heijmans ARCADIS Infrastructure


The off-shore immersed tunnel in the Busan-Geoje Fixed Link project in South Korea The project The Busan-Geoje Fixed Link, with a length of 8.2 kilometres, forms a link between Busan Newport (the port expansion project of Busan) and Geoje Island (see figure 1). The link is necessary to improve the accessibility of the archipelago in the south of Korea. The islands of the archipelago accommodate many shipyards and industries; the population, however, lives in the area of influence of Busan. The main parts of the link are an immersed tunnel of 3,240 metres long, a 300 metre long rock tunnel, an artificial island and two cablestayed bridges, one having a main span of 475 metres and one having two spans of 230 metres each. The entire link has two lanes in each direction for car traffic. The immersed tunnel will be constructed with 18 concrete elements of 180 metres long and


GEOtechniek – April 2008

26 metres wide. About two thirds of the immersed tunnel (16 elements) lies in an approximately 15 metre deep trench below the seabed (see figure 2). The maximum depth of the seabed is about 40 metres below still water level. At the end on Daejuk island the tunnel lies above the seabed and the tunnel elements will be located in a dam there. Thus the tunnel features a deep and a shallow part.

Figure 2 Position of the tunnel in the sea bottom and bottom profile.

Figure 1 Overview of the Busan-Geoje Fixed Link.

At the present moment, the large infrastructure project Busan-Geoje Fixed Link is being implemented in the south of Korea. Apart from two cable-stayed bridges, an off-shore immersed tube tunnel will be constructed. During the design of the immersed tunnel it was found that the loads on the tunnel due to 9 metre high ocean waves, in combination with its foundation on a layer of weak marine clay, are determinative factors for the design. Furthermore, analyses were carried out of the integrity of the tunnel during an earthquake.

The project is being developed by the ’Special Purpose Company’ GK Fixed Link. Participants in this company are several banks and the contractor, Daewoo Engineering & Construction, along with seven local contractors. The design bureau is the combination Cowi Daewoo Engineering. Commissioning authority is Busan Metropolitan City together with the Province of Gyeongnam, with which a 40-year concession contract has

been concluded for the construction and operation of the link.

Independent Design Checker ARCADIS, together with the French company Ingérop and Korean SeoYeong, acts as Independent Design Checker (IDC) in the project, charged with making an independent analysis of the design and supplying so-called ’design certificates'. The IDC should prevent design errors and guarantee the required technical quality to the commissioning authority. The IDC has an operational relationship with the designing parties and a contractual relationship with the permitting authorities or the concession holder.

construction dock, towed to the desired location by tugs during a suitable period of the year, and immersed into an approx. 15 m deep trench in the seabed. The tunnel is located in an area that is prone to typhoons and steep waves due to the proximity of the coast. The significant wave height at the tunnel location may be as high as 9.2 m. The largest significant wave height documented so far for an off-shore immersed tunnel is 5.3 m in the Naha Immersed Tunnel project [1]. The following two critical scenarios have been identified for the stability of the immersed tunnel in Busan:  Flotation of the tunnel: Flotation of the

For the Busan Geoje Fixed Link project, all designs of the immersed tunnel and the two bridges, but also the designs of the construction dock (see photo on the left), the temporary harbour and the auxiliary constructions, have been checked independently by the IDC. For this purpose only the design drawings and the raw data – such as geotechnical and hydraulic data, functional requirements and design criteria – were made available to the design checker by the commissioning authority. On the basis of the data and the drawings, independent assessments and check calculations were made of all relevant components of the design.

construction is checked in this project for the lower limit of the tunnel weight including ballast concrete, with a safety factor of 1.060 to 1.075. In general, water level variations are not significant for the check on flotation of an immersed tunnel. Model tests and calculations have proven, however, that the tunnel is sensitive to flotation due to differences in permeability of the backfill material, for instance, due to silting up of the backfill on one side of the tunnel and water level differences across the tunnel. During the passage of a wave trough, insufficient weight may be available to guarantee a minimum pressure on the foundation plane.

Wave loads and their effects on the tunnel The immersed tunnel in Busan will be built up from 18 concrete elements, which are provided with temporary steel bulkheads at the ends. The elements are prefabricated in a dry

 Horizontal loads on the tunnel tube: a high water pressure gradient across the tunnel due to steep waves, combined with the possible difference in permeability between the inflow and outflow areas, will result in a horizontal

force on the tunnel (see figure 3). The horizontal force on the tunnel proved to be linked to a corresponding upward vertical force. Heavy backfill material beside and on top of the tunnel is required to guarantee sufficient stability of the tunnel (see figure 4). The relevant analyses will be explained below. The hydraulic boundary conditions were calculated with the numerical model MIKE 21. Using this program the significant wave height for the normative cross-section was calculated at 9.2 m with a frequency of exceedance of once in 10,000 years. The maximum wave height is limited by the water depth and was calculated at approx. 15 m. The shape of the waves at the water surface depends on the water depth in relation to the wave height. In deep water, the shape of the wave can be described by a simple sine function. Furthermore, the pressure on the sea bottom can also be determined with a simple function. Relatively shallow water makes the wave steeper to the point where it wants to break. Description of the wave shape on the basis of a sine function is no longer possible. In that case a description based on the ’stream function theory’ can be used [2]. At the seabed, the waves result in pressure variations and an orbital movement. This is an elliptical or circular movement of the water particles during the passage of a wave. For the specific project situation, the effect of orbital movement on the pressure variations was negligible. The pressure variations of the waves on the sea bottom result in a groundwater flow and different water pressures around the tunnel and in the backfill material.

Figure 3 Effect of the difference in permeability on the water pressure distribution around the tunnel.

Figure 4 Cross-section of immersed tunnel, deep position.

GEOtechniek – April 2008


To determine the effect on the tunnel of the pressure variations on the seabed and around the tunnel, extensive PlaxFlow and Plaxis calculations have been carried out. The advantage of combining Plaxis with PlaxFlow is that it allows the correct geotechnical and geohydraulic boundary conditions to be implemented in a model. Firstly, PlaxFlow was used to determine the water pressures in the backfill material and the bottom, and the pressures around the tunnel. The PlaxFlow calculations were based on the pressure distribution of the waves on the seabed (see figure 5). In the flow calculation a steady state groundwater flow was assumed. This is a realistic scenario, because the pressure waves run at a much higher velocity than the waves themselves. Inertia effects, in particular of the tunnel, have been ignored, because of the fact that the waves are present above the tunnel long enough to accelerate the entire mass system of soil, tunnel and water. The pressures around the tunnel were calculated with PlaxFlow while varying the location of the waves above the tunnel. The effects of different permeabilities on the calculated pressure distribution, as well as the resulting loads on the tunnel, were also studied in this way. The permeability was varied within the possible limits. The result for the 10,000 year wave and

for a difference in permeability of a factor 10 (permeability of outflow area is 10 times lower than that of the inflow area) is presented in figure 6. The calculated maximum vertical force upward per running metre is approx. 380 kN. The maximum horizontal force is approx. 450 kN. This makes it clear that the horizontal and vertical reaction forces are not in phase. The largest upward force occurs when a wave trough moves over the tunnel. The largest horizontal force occurs when the largest pressure difference across the tunnel is present. On the basis of this, the worst case situation was calculated, in which the tunnel is subjected to the largest horizontal deformation. Furthermore, the minimum cover was determined to safeguard a minimum support pressure of 5 kPa. To determine the deformations of the tunnel and to assess the stability of the tunnel as whole, numerical calculations were carried out using Plaxis. The water pressure situation was calculated first with PlaxFlow and implemented in Plaxis as a boundary condition. The bottom soil parameters for Plaxis were determined on the basis of oedemeter tests, which were available in sufficient numbers to perform a statistical analysis for this project. Triaxial tests were available to a lesser extent and were used for verification purposes. In view of the relief and

Figure 5 Distribution of water pressure on sea bottom in PlaxFlow.

Figure 8 Stress and deformation behaviour of clay and structured cla.

GEOtechniek – April 2008

Foundation The foundation of the tunnel is formed by a thick layer of marine clay down to about 30 metres below the seabed (see figure 2), followed by a layer of sand and gravel, underlain by weathered rock. The properties of the marine clay require special attention. Normally, clay deposits exhibit some degree of overconsolidation due to ageing. On the basis of the available research data, such an effect could not be found for this marine clay. A remarkable feature was that the clay exhibited fairly stiff behaviour at low stress levels, whereas the deformations increased sharply at higher stresses (see table 1). Furthermore, the specific gravity was low in relation to the stiffness upon reloading.

Figure 6 Resulting forces on the tunnel due to water pressures.

Figure 7 Possible failure mechanisms around the tunnel.


reload situation, the Hardening Soil model was used for the marine clay. The backfill material, the sand and the weathered rock were modelled with Mohr-Coulomb. In view of the short duration of the wave load on the tunnel, the clay was assumed to be undrained and to have increased dynamic stiffness. The calculated maximum horizontal deformation due to wave passage was in the order of magnitude of 40 mm. Finally, insight into possible failure mechanisms was obtained on the basis of the Plaxis model (see figure 7).

The off-shore immersed tunnel in the Busan-Geoje Fixed Link project in South Korea

Soil type


γ nat


Compressibilty parameters

Strenght parameters








C rc

kN/m3 Marine clay ’stuctured clay’










Band width









n.d. = not determined

* depends on OCR and vertical effective stress, indication: Cu = 20 à 50 kPa

Table 1 Soil parameters of marine clay.

Eventually the clay was classified as ’structured clay'. This means that the clay contains calcium compounds, which cause a relatively high stiffness at relatively low stresses [3], [4]. In figure 8 this is shown schematically for ’normal’ clay and for ’structured’ clay. If the calcium compounds are broken up by an increase of the stress level beyond the so-called ’limit stress', then they collapse like a house of cards and the clay behaves like a very weak material (see figure 9). Furthermore, creep effects may be reactivated. Usually the weight of the tunnel under water and the backfill beside the tunnel is lower than the weight of the excavated soil. Then, by definition, the stress level under the tunnel cannot become higher than the stress originally present. With this tunnel, however, due to the off-shore conditions, as explained above, a heavy backfill on and beside the tunnel is necessary to offer sufficient resistance to wave pressures. Consequently, the soil stress below the tunnel increases beyond the limit stress, thus significantly increasing the settlements of the tunnel. This can be expected to be associated with large

settlement differences. These in turn may lead to opening of joints, possibly beyond the capacity of the waterstops. To remove the uncertainties described above with regard to the settlements, different soil improvement techniques have been used. At the ends of the tunnel, the clay layer under the tunnel is relatively thin. Here the clay is replaced with rock fill. Along the major part of the tunnel route the tunnel is founded on a grid of columns formed in the soil, using the Cement Deep Mixing (CDM) process. This process involves a pontoon from which a vane on a drilling shaft is turned into the soil, which mixes the clay with cement. The columns each have a diameter of 90 centimetres, and are placed as continuous walls under the tunnel (see figure 5). The stiffness and strength of the clay are sufficiently increased by the CDM columns to avoid undesired settlements. At the western end, the tunnel will be resting in a dam construction built on the sea floor. The CDM method is considered less suitable here, as it may cause disks to be created, which would form a potential slip surface during an earthquake. The solution chosen here is the use of sand columns according to the Sand Compaction Pile (SCP) process. These sand columns, with diameters of 1.2 to 1.8 metres, are inserted to a record depth of approx. 60 metres (see photo this page). The dam is built onto the column grid with some excess height, so that after placement of the tunnel elements the remaining settlement will be very small. For the design of the foundation with soil improvement, use has also been made of a large number of Plaxis models for both the deep position of the tunnel in the trench and the high position in the dam. All possible load situations were taken into account, such as a sunken ship on the tunnel. Subsequently the length and location of the columns were optimised.

Soil improvement with SCP process at the western end of the tunnel.

In the context of the CDM and SCP methods it

Figure 9 Failure mechanism of structured clay.

should be noted that, apart from the design, the independent design checker has also assessed implementation aspects. Technical specifications have been prepared for these two techniques. For the CDM method, for instance, the correct strength is an important aspect. On the one hand the columns must be sufficiently strong and stiff. On the other hand, the columns must not be too strong in order to avoid damage to the columns during cutting off and excavation to the correct depth.

Earthquake analysis of the dam At the western end, the last two elements of the tunnel run through a dam in an elevated position. In the preliminary design, the dam was provided with 1:1.5 slopes without shoulders. However, the preliminary design was based on the scale model tests carried out for dimensioning of the tetrapods of the dam. No account was taken of the total stability of the overall soil massif below the dam. Moreover, earthquakes and the steepness of the waves above the slopes may have a negative effect on slope stability, which has not been included in the model tests either. An earthquake will lead to an increase in the gradient of the wave above the slope of the dam (see figure 10). Furthermore, excess pore pressure will be generated in the soil below the tunnel during an earthquake.

GEOtechniek – April 2008






Acceleration Safety factor Effect of earthquake kh


Santa Barbara

1925 6.3

Sheffield dam



Complete failure

Santa Fernando

1971 6.6

Santa Fernando dam downstream



Waterside slope collapsed



Caving in on land side


Table 2 Observed failure modes of earth dams during earthquakes (Source: Seed 1979). Figure 10 Schematic representation of hydrodynamic effects on the slip plane of a dam during an earthquake.

Figure 11 The Plaxis model for earthquake analysis.

The fact that these effects can lead to instability of earth dams is demonstrated by Table 2, which has been derived from article [5]. In 1925, the ’Sheffield dam', with a calculated safety factor SF = 1.2, collapsed completely due to an earthquake of M = 6.3 on the Richter scale with an effective horizontal earthquake acceleration of 0.1*g (acceleration of gravity). An analysis of the stability of the dam with the embedded tunnel was necessary. The earthquake analyses for the immersed tunnel were made using accelerograms suitable for the seismic zone. However, these provide data on the accelerations of the deep rock. To obtain representative acceleration levels of the earth dams, dynamic Plaxis calculations were performed. The NERA software [6] was used to determine the output signal on the rock, the so-called ’rock outcrop signal'. This has been entered into Plaxis as a boundary condition (see figure 11). Possible accelerations of the dam with the tunnel were calculated on the basis of the methodology described in the Plaxis manual [7]. At the location of the tunnel, earthquakes may occur that generate accelerations of up to


GEOtechniek – April 2008

Figure 12 Schematisation with MStab.

approx. 0.15g. These earthquakes are in the order of 6 to 6.5 on the Richter scale. The current version of Plaxis does not accurately calculate the generation of excess pore pressures due to an earthquake. Therefore it was decided to perform the stability calculations using a slip circle model. The MStab program of DelftGeoSystems [8] was selected for this purpose. To ensure correct modelling of the stress situation with MStab, a number of schematisation steps were necessary. For instance, due to arching effects increased stresses occur on the columns as a result of the load from the dam (see figure 12). Here the stress distribution on the columns was based on the results of the static Plaxis calculation. On the basis of the MStab model, the sensitivity of possible excess pore pressures in the columns during an earthquake and the influence of the strength parameter (internal friction angle) of the columns could be analysed with a larger number of calculations than with more elaborate numerical calculations. At first, the classic approach to stability calculations was chosen, using a pseudo-static method

as described in the MStab manual, in which a horizontal earthquake acceleration is added to the vertical acceleration due to gravity [8]. The results showed, however, that even with relatively gentle slope gradients (more than 1:2) and relatively large shoulders of more than 20 m wide, the level of safety was insufficient (SF < 1.0). This led to an approach in which plastic deformations in the dam beside the tunnel are regarded as acceptable. In literature this approach is known as the Newmark method [9]. On the basis of this method, Makdisi & Seed presented a graph in 1978 (see figure 13), which allows estimation of the possible plastic deformations of a dam on the basis of the ratio between the maximum acceptable horizontal acceleration to prevent failure (yield acceleration, kv) and the maximum possible local acceleration (kmax) [9]. The yield acceleration (kv) is calculated with the slip circle model. The expected maximum acceleration (kmax) is determined with Plaxis and on the basis of NERA, as described above. On the basis of the basic assumptions with respect to the soil parameters, the hydraulic boundary conditions,

The off-shore immersed tunnel in the Busan-Geoje Fixed Link project in South Korea

a slope of 1:2 and a shoulder of 20 m wide, ky/kmax ratios between 0.2 and 0.3 were calculated for the present project. According to the graph of Makdisi & Seed this will result in plastic deformations of 10 to 20 cm, which is considered acceptable for the soil massif beside the tunnel.

Summary During the final design phase of the immersed tunnel in the Busan-Geoje Fixed Link, the following three issues were found to be critical from the hydraulic and geotechnical point of view due to the off-shore location of the tunnel. These issues have led to the following adjustments to the reference design:  Vertical and horizontal pressure differences

across the tunnel due to waves have resulted in the necessity of heavy backfill material on and beside the tunnel.  The heavy backfill material, in combination

with the mechanical properties of the marine clay, a so-called ’structured clay', makes soil improvement for the foundation of the immersed tunnel necessary to limit deformations and guarantee stability. On the basis of earthquake and implementation considerations, it has been decided to use sand columns as soil improvement in areas where the tunnel is embedded in a dam. In stretches where the tunnel runs below

the sea bed, use has been made of Cement Deep Mixing columns.  Earthquake analyses of the dam have resulted

in the use of a more gentle slope gradient and a shoulder. Plastic deformations of the dam in accordance with the theory of Makdisi & Seed are considered acceptable during an earthquake.

Conclusion The stability of an immersed tunnel under offshore conditions calls for a multidisciplinary approach between hydraulic, geotechnical and structural specialists in a design team. This project is a good example of how the Dutch expertise in the fields of immersed tunnels and off-shore constructions can make a vital contribution to an international project. The following Dutch parties are involved in implementation of the project: the tunnel trench was dredged last year by Van Oord, TEC is the engineering consultant of the concessionaire, Trelleborg-Bakker is supplier of the sealing strips, and MARIN in Wageningen carried out a series of scale model tests of the immersion process. At the moment Strukton is busy to immerse the tunnel elements into their final position. During the last audit of the IDC in Busan, implementation was in full swing, and completion of the Busan-Geoje Fixed Link is expected in 2010. 

Literature [1] Aono T., Sumida K., Fujiwara R., Ukai A., Yamamura K. and Nakaya Y.; 2003; Rapid stabilization of the immersed tunnel element; Proceedings of the Coastal Structures 2003 Conference Portland, Oregon, August 26-30, 2003, 394-404; American Society of Civil Engineers. [2] Heijmans R., Jackson P., Kasper T., Meinhardt G., Schmitt J., Voortman H.; 2007; The effect of wave passage on immersed tube tunnels - Busan Geoje Fixed Link in South Korea; IABSE 2007 Weimar. [3] D. Masín, S. E. Stallebrass and J. H. Atkinson; 2003; Laboratory modelling of natural structured clays; Int. Workshop on Geotechnics of Soft Soils - Theory and Practice. [4] J.-C. Chai, N. Miura, H.-H. Zhu, and Yudhbir; 2004; Compression and consolidation characteristics of structured natural clay; Can. Geotech. J. 41: 1250-1258. [5] G. Biondi & M. Maugeri; year unknown; A modified Newmark type-analysis according to EC-8 requirements for seismic stability analysis of natural slopes; University of Catania, Italy. [6] University of Southern California, Department of Civil Engineering, NERA, A computer program for nonlinear earthquake site response analyses of layered soil deposits by J. P. BARDET and T. TOBITA, April 2001. [7] PLAXIS, 2D - Version 8, Manual [8] Geo Delft, MStab User Manual, Release 9.8, February 2004. [9] Steven L. Kramer, Geotechnical earthquake engineering, Prentice Hall, 1996, ISBN 0-13-374943-6.

Figure 13 Estimation of plastic deformation of soil, Makdisi & Seed (1978).

GEOtechniek – April 2008


Educom Publishers

Journals and magazines with substance Besides publishing the technical journal Geotechniek, a leader in its field for 10 years now, Educom Publishers also specialise in developing and realising publications focusing on (city) promotion, tourism, culture and policy topics.

All our publications are distinguished by the high quality of both technical implementation and content. As Educom Publishers co-invest in promising projects, it is almost always possible to realise such a publication, even with a limited budget. Feel free to contact us for a no-obligation meeting to discuss your ideas.

EDUCOM PUBLISHERS P.O. Box 25296, 3001 HG Rotterdam,The Netherlands

You want to reach the Dutch and Belgian GEO-market? Choose for the scientific journal Geotechniek Ask for information about: - Sponsorship - Publicity Package - Advertising

EDUCOM PUBLISHERS P.O. Box 25296, 3001 HG Rotterdam,The Netherlands

Profile for Uitgeverij Educom

Geotechniek april 2008 Special congres Shanghai  

Twaalfde jaargang Special ICSMGE Shanghai 2008 Special Issue on the occassion of the VI International Symposium Geotechnical Aspects of Un...

Geotechniek april 2008 Special congres Shanghai  

Twaalfde jaargang Special ICSMGE Shanghai 2008 Special Issue on the occassion of the VI International Symposium Geotechnical Aspects of Un...

Recommendations could not be loaded

Recommendations could not be loaded

Recommendations could not be loaded

Recommendations could not be loaded