Design of Deep Underground stations in soft soil ... - Royal Haskoning

Design of Deep Underground stations in soft soil conditions 

J. Herbschleb 1 , J.C.W.M. de Wit 2 

ABSTRACT 

The municipality of Amsterdam wish to 

reduce the level of car traffic within the 

City Centre. As a consequence the public 

transport is to be extended by a new 

North/South Metro line. This paper will 

focus on the design philosophy of the deep 

station boxes. The excavations depths for 

the stations will exceed 30m and will be 

constructed in difficult soft soil conditions. 

A further significant aspect is that the 

building pits are very near (3 to 5 m) to 

buildings of historical importance. This is 

illustrated by the following topics that are 

investigated. 

• Geology/Geotechnics 

• Structural design station box 

• Requirements/Impact surroundings 

• FE-Modelling 

• Additional test project diaphragm walls 

• Monitoring during construction 

INTRODUCTION 

Central 

Station 

Rokin 

Station 

Vijzelgracht 

Station 

Figure 1: Aerial photograph of Amsterdam with projected 

North/South Metro line 

The Amsterdam North/South metro line will connect several important public transport systems (train, tram, 

bus, and metro) with the north, centre and south districts of Amsterdam. The total length of the metro is 9 

km, of which 4 km will be constructed underground. During the construction (early 70’s) of the first metro 

line in Amsterdam it was necessary to demolish buildings in city centre. The client wishes to avoid any 

demolition during the construction of the new metro line. Hence the following two major restrictions were 

imposed for this project: 

1) No significant damage to buildings may occur. 

2) Disruption of city life should be limited. 

To minimise the effects of the construction to adjacent historical buildings a bored tunnel will be adopted for 

the underground sections in the city centre. The tunnel follows the street pattern as closely as possible, see 

Figure 1, and descends to a great depth. Consequently the metro stations are located at significant depths 

(over 30 meters below surface level). A building pit will be constructed for the stations, having braced 

diaphragm walls to a level of over 40m below street level. 

The influence of the construction of the North/South line on historic and vulnerable buildings was a 

significant design issue. 

Much design effort was required and special test procedures were carried out to investigate the realistic 

effects of underground construction in the City. 

1 De Weger, Architects and Engineers, Heer Bokelweg 145, 3032 AD Rotterdam, the Netherlands, email: jh@deweger.nl 

2 De Weger, Architects and Engineers, Heer Bokelweg 145, 3032 AD Rotterdam, the Netherlands, email:hdw@deweger.nl 

De Weger participates together with Witteveen & Bos bv and IBA in the North/South line Design Office, which is responsible for the 

design of the North/South line metro.

GEOLOGY 

The subsurface of Amsterdam is 

composed of sediments, up to depths 

varying between 800 to 1000 meters 

below ground level. The sediments 

(sands, silts, clays and peat) have 

originated from marine-, glacial-, 

eolian- and river deposits. In the 

upper 350 meters of sediments two 

main geological important units are 

distinguished, namely Holocene 

(10.000 years - present) and 

Pleistocene (10.000 - 2.5 million 

before present) deposits, see Figure 2 

for a general lithological profile of 

Amsterdam. 

The oldest and deepest Pleistocene 

deposits are marine clays and fine- Figure 2: Geological cross section of Amsterdam 

grained sands, which extend to a level 

of 250- 350 meter below groundlevel 

With the second last (Saalien) ice age sediments were deposited on top of these river deposits. The ice cap 

came near to the South of Amsterdam and created the Amsterdam Basin. During this period mainly glacial 

and melt water deposits were formed within the basin. After melting of the ice cap the Amsterdam Basin was 

flooded with the sea and partially filled with marine sands and clays (Eemclay). During the last ice age 

(100.000 - 10.000 years ago), the ice cap did not extend to the south of Amsterdam. In that period the 

Netherlands experienced a tundra climate. The Amsterdam Basin was filled with mainly sand. These sand 

layers are very important to foundation practice in Amsterdam and marks also the end of the Pleistocene. 

Holocene deposits (mainly peats and clays) were mainly formed under the influence of the sea. The rising 

sea (0.5m per century) reached the present location of Amsterdam 7000 year ago. A transgression started due 

to lowering of the sea level, 5000 years ago. During the Holocene a river connection existed to the sea. This 

river produced erosion channels in the Pleistocene sand deposits. In these channels, sometimes with widths 

over 100m, thin sand and (soft) clay layers were deposited. 

SITE INVESTIGATION 

Along the route of the North/South line a comprehensive soil investigation has been made. This soil 

investigation consisted of a total of 125 boreholes and about 400 CPT’s each up to depths of 70 meters. As 

well as these borings and CPT’s, special tests like Cone Pressure Meters (CPM) were also carried out. 

In the laboratory the samples of the borings were classified and tested. The laboratory tests were used mainly 

to determine the strength and stiffness parameters of the soils, specially the Pleistocene Marine Eemclay. The 

reason to concentrate the site investigation to the Eemclay is that very little geotechnical information was 

available for this layer prior to the North/South line project. Also this will be the first time a construction 

(both the bored tunnel and the deep building pits for the stations) is made in this clay. 

Table 1: Global overview of laboratory tests 

Soil layer Classification Sieve analysis Atterberg Triaxial Oedometer 

limits tests test 

Holocene 405 35 45 50 19 

Pleistocene Sand 260 80 38 

Pleistocene Eemclay 360 170 170 96 135 

Other layers 685 265 165 49 37 

Total 1710 550 380 233 191

GEOTECHNICAL DESIGN 

Geotechnical profile 

The geotechnical data (CPT’s, boreholes) are each 

interpreted to obtain a soil profile. Instead of drawing 

one geotechnical profile along the centreline of the 

alignement of the tunnel it was decided to use a GIS. 

With the aid of a GIS a 3D block model, see Figure 

3, is constructed. The advantages of the 3D GIS 

block model are: 

1) A geotechnical profile can be created in each 

direction and location. 

2) Updating the profile is fast and easy when 

additional data becomes available. 

3) The ability to calculate the volumes of excavated 

soils with a distinction in sands, clays, etc. 

The main disadvantage is that subjective 

information, such as possible filled-in channels, are 

more difficult to take into consideration. The final 

geotechnical profiles are therefore corrected and 

adapted by hand. 

Geotechnical parameters 

The determination of geotechnical parameters can 

not be seen apart from boundary conditions like: 

1) Calculations models 

2) Required safety 

Figure 3: Automatic Generation of profile 

The calculation models and design of the building pit 

determine the type of geotechnical parameters. 

Advanced finite element programs with second order material models require different geotechnical 

parameters than analytical models. The careful selection of calculation (soil) models, the level of safety (risk 

analysis) and site investigation is the start of the determination of the geotechnical parameters. In this project 

some of the important geotechnical parameters can be derived in several ways. In Table 2 a global overview 

is given of the applied investigation methods and the connection to some geotechnical parameters. 

Table 2: Global overview of Geotechnical parameters 

Investigation Type Phi E’50 Eoed E’ur 

Atterberg limits Laboratory C C 

Triaxial test (loading) Laboratory X X 

Traixial test (unloading) Laboratory X 

CPT In-situ C C C 

CPM In-situ X X 

Oedometer test Laboratory X X 

Void ratio Laboratory C C 

C Correlation 

X Direct measurement 

By careful selection of the samples and the characterisation it was possible to distinguish a set of geological 

layers. When studying these parameters further it was concluded that the variations found in the values of the 

stiffness parameters were not only caused by the different test methods (laboratory, in-situ), stress path, etc. 

But mainly due to differences in the strain levels of the tests. With an in-situ test (CPM) low (10 -4 ) strain 

levels are introduced in the soil, the resulting E’ur is therefore much higher than in an Oedometer test where 

high (10 -1 ) strain levels are introduced to the soil samples. This has led to the conclusion that when 

determining the design values of the geotechnical parameters it is necessary to derive the stiffness values in 

the range of the expected strain level. With careful interpretation of all tests eventually a geotechnical 

parameter set that meets the above mentioned requirements was derived.

DESIGN PHILOSOPHY 

The design philosophy of the station boxes was to determine an acceptable balance between the safety 

requirements and construction costs. The guidelines for the design, both for the building pits and the bored 

tunnel, are the predicted deformations of the adjacent building foundations. The deformations depend on the 

following: 

a. Geotechnical behaviour (parameters) 

b. Calculation models (Finite Element or Analytical) 

c. Construction design 

Knowledge of the optimal geotechnical parameters allows the soil behaviour to be determined. However 

because it is not possible to derive the geotechnical parameters exactly from laboratory-or in-situ tests, a 

certain safety margin is introduced. The larger the safety margin the more secure but also the more costly. To 

calculate the deformations mathematical soil models are used. By using the latest validated FE models and 

technology the most accurate predictions of the deformations can be assessed. Apart from the geotechnical 

behaviour, the deformations depend also on the design of the building pit. 

The main objective is not to cause any significant damage to the adjacent buildings, the criteria (allowable 

deformations of ground level and foundations) are assessed a conservative manner. In Figure 7 all the 

involved elements are illustrated including their interactions. 

UNDERGROUND STATIONS 

Four underground stations will be built in 

the city centre of Amsterdam as a part of 

the North/South line metro. For the 

construction of the underground stations, 

a building pit will be constructed having 

1.2m thick braced diaphragm walls. 

These walls reach down to approx. NAP - 

35 m/ -40 m, see figure 4. The minimum 

distance between the building pit and the 

historical buildings is about 3 m. The 

building pit will be excavated in the dry 

after the water level inside the pit has 

been lowered. This can be done without 

influencing the water level outside the pit 

because the building pit walls continue 

down into the sealing Eemclay at a depth 

of 20-25m. The excavation depth varies 

per station and will be between 26-32m 

below street level. With a dry excavation 

the vertical equilibrium of the building- 

pit base must be considered. For the 

deepest building pits safety against 

vertical equilibrium cannot be achieved 

by the weight of the sealing ground 

stratum (Eemclay) alone. Additional 

measurements are to be taken by 

excavating the deep part of the building 

pit under compressed air conditions. 

When the building pit is excavated the 

concrete structure can be built. The 

reinforced concrete diaphragm wall, which 

Depth [NAP + m] 

0 

-10 

-20 

-30 

3.0 to 9.0 m 24.5 m 

3.0 to 9.0 m 

Plaxis 2D 

predictions 

-40 

-40 -20 0 

Horizontal deformation [mm] 

Excavation NAP -10.0m 



Groutstrut 

Figure 4: Design Rokin Station 

-60 

0 10 20 

qc [MPa] 

30 

is originally the retaining wall, will be integrated in the permanent body. This was so decided because a 

composite structure of the diaphragm wall and a cast-in-situ inner wall results in the minimum depth of the wall 

structure and therefore a maximum distance between the building pit and the adjacent buildings. 

0 

-10 

-20 

-30 

-40 

-50 

Depth [NAP + m]

INFLUENCE ON THE SURROUNDINGS 

In considering the environmental impact of constructing these stations the following elements require closer 

examination: 

1. Excavation of diaphragm wall trench, which introduces a 3D stress strain behaviour in the soil layers. 

2. The self weight of the wall and the applied deck which can cause settlement of the wall. 

3. Excavation of the building pit which causes horizontal deformation of the diaphragm wall and a 

relaxation of the Pleistocene Eemclay which forms the bottom of the building pits. 

Some of the described mechanisms can cause settlements of nearby pile foundations, others may result in 

heave. Also the deformation behaviour of the soil layers (clay) is time dependent due to consolidation and 

creep effects. This means that the final situation is not necessarily the critical situation, and that the entire 

construction process should be taken into account. To obtain a complete picture of deformations ,it is 

necessary to use a FEM program, in which the full construction sequence is considered. 

Settlement risk assessment studies were carried out to predict the risk and degree of settlement induced 

damage to adjacent structures. The specific surrounding conditions of the Amsterdam city centre necessitated 

the application of analytical and numerical studies to investigated this risk. The study results were used to 

define boundary conditions for the historical masonry buildings and the requirements for the settlement 

inducing construction activities (Kaalberg et al). 

2D & 3D FE MODEL 

The vertical loading of the diaphragm wall, as well as the excavation and unloading of the Eemclay are 

analysed with a 2D FE model. The excavation of the diaphragm wall trench however is more complicated. 

This is due to 3D stress/strain effects, therefore a 3D FE model is used. The 2D FE model is furthermore 

used to estimate the deformations, at ground- and foundation level, during the several stages of the 

excavation of the deep building pits. Three FE software programs were tested in order to select the one most 

suitable for the construction of the station box. Eventually Plaxis was selected mainly because of the 

extensive soil mode library. After research, both internally and by the University of Stuttgart, the Hardening 

Soil model of Plaxis appeared to a most suitable model for very large (unloading ) excavations in sand and 

clay. 

Hardening Soil model 

The Hardening-Soil model represents a much more advanced model than the Mohr-Coulomb model. In the 

Hardening Soil model the limiting states of stress are described by means of the friction angle, φ, the 

cohesion, c, and the dilatancy angle, ψ. The soil stiffness is described much more accurately by using three 

different input stiffnesses: the triaxial loading stiffness, E50, the triaxial unloading stiffness, Eur, and the 

oedometer loading stiffness, Eoed. In contrast to the Mohr-Coulomb model, the Hardening-Soil model also 

accounts for stress-dependency of stiffness moduli. This means that all stiffnesses increase with pressure. 

Building sequence 

The building sequence as implemented in the FE model 

is more or less as follows: 

1) Installation of diaphragm wall (stresses only) 

2) Installation of deck 

3) 6 excavation phases up to NAP -32m. 

a) lower water table inside pit 

b) dry excavation 

c) placing struts 

d) consolidation (2-4 months) 

4) Installation of concrete and structure 

Sometimes special building steps are included to model 

excavation under air-pressure or the installation of groutstruts 

before excavation. Including steps for numerical 

stability more than 100 calculation steps are used to 

model a single excavation. 

hardening 

shear locus 

cap with 

compression 

hardening 

Figure 5: Plaxis Hardening Soil model

FE MODEL VALIDATION 

The main objective of the FE calculations is to create a frame of reference for the expected deformations. For 

this reason a large amount of sensitivity calculations were made after which a set of new calculations was 

created. In this new set of calculations the soil parameters (lower, average and upper limit value) were varied 

such that special features like maximum horizontal bending of diaphragm wall could be calculated. 

The 3D FE model is validated with the results of a special test project. This test project was conducted to 

gain more insight in the deformations that occur during installation of a diaphragm wall in Amsterdam soil 

conditions. 

After validation of the 3D FE model the total deformations, installation of diaphragm wall (3D FE model)+ 

excavation of building pit (2D FE model) are calculated. 

Diaphragm wall test 

The main objectives for the full-scale test on a construction site in Amsterdam are: 

• Expanding the knowledge on this subject, so that design if necessary can be made in advance. 

• Validation of 3D- FE model based on the test results. 

• Gained insight in the specific execution aspects of diaphragm wall installation in Amsterdam soil. 

The test project of the diaphragm wall installation consists of three parts: 

1. Prediction of the impact with a 3D FE model 

2. Full scale test at a construction site in Amsterdam 

a. Measurement of vertical and 

horizontal deformations of 

the soil adjacent to the 

excavated trench, see 

figure 6. 

b. Settlement of loaded piles 

c. Impact on bearing capacity 

of piles, due to excavation 

and concreting of 

succeeding diaphragm wall 

panels. 

3. Validation of 3D FE model based 

on the test results that have 

become available. 

At this moment the validation of the 

3D model is finalised. Nevertheless 

the following conclusions can be 

taken: 

• The test results show that the 

installation of a diaphragm wall 

on this project has a minor effect 

on the surroundings and no 

significant impact on the piles of 

nearby foundations (de Wit, et 

al, 1999). 

• The prediction by means of a 3D 

FE model proved, apart from 

consolidation effects, to be able 

to predict the deformations 

caused by the succeeding stages 

in the installation of the 

diaphragm wall. In general the 

FE predictions result in larger 

deformations than monitored at 

the test program. 

60 minutes 

Horizontal Deformation Inclinometer during 

concreting of diaphragm wall [mm] 




Figure: 6 Results of inclinometer measurements 

Depth [NAP +m]

MONITORING 

To anticipate ground movement and damage to buildings during construction and to enable timely and 

appropriate actions, an effective monitoring concept has been designed. Buildings will be monitored using a 

fully automatic system of total stations and sensors. This on-line system will provide direct data for 

interpretation in the design office. A system of partly automatic and partly manual monitoring will be carried 

out during the construction activities for the stations: 

• vertical ground deformations on surface level (outside the pit) with precise levelling points and on 

deeper levels (inside and outside the pit) with extensometers; 

• horizontal ground deformations with inclinometers; 

• horizontal and vertical deformations of the diaphragm wall (precise levelling points and inclinometer); 

• strut monitoring 

An area of about 70 m around the building pit will be monitored. The monitoring activities for the 

North/South Line are divided into three stages: 

a. Base monitoring: Base monitoring is carried out before construction works starts and will determine the 

natural settlement behaviour and seasonal (temperature) effects on the deformations of the buildings. 

b. Process monitoring: Process monitoring is carried out during construction and results will be provided 

on-line. The sample rate depends on the process (bored tunnel/building pit). 

c. Close out monitoring: Closeout monitoring continues after building activities have ended and register 

long-term settlement behaviour. 

The monitoring information is analysed and processed in such a way that during the construction period 

changes to the design/process can be made. This approach has been used to some success in several projects 

in the UK and is usually called the Observational method [Anon 1996]. 

Settlement Risk Management with GIS 

Settlement risk management is a important tool in controlling the effects of construction activities on the 

surroundings. It is implemented by a special developed Geographic Information System (GIS). The GIS has 

various modules that are used in the settlement risk management : 

• Design 

• data of defect surveys 

• settlement predictions by FE modelling of stations’ construction; 

• damage classification system by 3D numerical risk assessment studies on typical masonry structures 

using settlement predictions; 

• building classification based on damage classification system and settlement predictions 

• mitigating measures for locations with unacceptable damage risk 

• Construction 

• monitoring data 

Within the design phase GIS is used to combine settlement predictions and building classifications along the 

station in order to confirm or adjust the stations’ design or to design mitigating measures. During 

construction the GIS allows for rapid interpretation of the monitoring data using the information from the 

design modules. If necessary back analyses can be carried out and be implemented in the GIS. In this way a 

early decision can be made about active measures. 

CONCLUSIONS 

Extensive parametric studies were carried out taking into account the upper and lower bound values of both 

geotechnical parameters as well as material properties of the concrete structure. The result is a 

range/envelope of deformations in the subsoil that represents the impact of construction on the surroundings 

and that will have to meet the boundary conditions of the masonry buildings. This is achieved with both 

passive as well as active measures in the stations design. Passive measures are taken in advance e.g. six or 

seven strut levels, a deep jet grout strut etc. An active measure is taken during construction and based on 

relation between numerical analysis results and monitoring results during construction (observational 

method). Active measures can be the pre-stressing level of the struts or increasing the compressed air. This 

means that the design must be suitable to this possible active measures. In a complex project like the

excavation of a building pit (dimensions : 

lxhxw=120mx30mx20m) in a busy city very close 

(

Design of Deep Underground stations in soft soil ... - Royal Haskoning

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?