diaphragmwall test

J.C.W.M. de Wit, H.J. Lengkeek. Full scale test on environmental impact of diaphragm wall trench installation in Amsterdam, . Proc. Int. Sym. on Geotechnical

Aspects of Underground Construction in Soft Ground, Toulouse, 2002, France. -1-

1 INTRODUCTION

The new North/South Line metro in Amsterdam willconnect the northern and southern suburbs with thecity centre (de Wit, 1998). The underground stations,20-25 m wide and 200-260 m long, will be con-structed at very busy locations and close to historicalbuildings. The excavation will be carried out to adepth of over 30 m in soft soil conditions with highground water levels.

Figure 1. Typical cross section of underground station.

The underground stations will be constructed in abuilding pit of 40 m long braced diaphragm walls.The historical environment requires a careful ap-proach of all construction processes needed to buildthe underground station:− excavation of the diaphragm wall trench;− excavation of the building pit;− construction of permanent body.

The excavation of the building pit and the construc-tion of the permanent body can be analysed with a2D FE-model, in which the full building sequencewith time schedule is processed (figure 1). The ex-cavation of the diaphragm wall trench however, isfar more complicated because of the 3 dimensionalbehaviour. Therefore it was decided to carry out aresearch project on the diaphragm walls’ excavation.The project consists of the following parts:− prediction of the impact with a three dimensional

finite element model (3D FE-model);− full-scale test program at the Mondriaan Tower

construction site in Amsterdam near the Amstelriver;

− interpretation test results;− validation of the 3D FE-model based on the test

results that have become available.The first two parts of the test were described exten-sively in an earlier paper (de Wit et al, 1999) andwill only be summarised here. The interpretation ofthe test results and the validation of the 3D FEM-model are discussed in this paper.

2 DIAPHRAGM WALL INSTALLATION

Diaphragm wall installation is carried out incre-mentally by the construction of individual panels tosome planned sequence. The panels’ dimensions canvary considerably depending on the design and thelocal circumstances. The construction of a dia-phragm panel is carried out from surface level bymeans of a mechanical device such as a bucket grab(figure 2) or hydro fraise.

A progressive excavation of a trench in theground is allowed in such a way that stabilising fluid

Full scale test on environmental impact of diaphragm wall trenchinstallation in Amsterdam

J.C.W.M. de Wit i) , H.J. Lengkeek ii)

ABSTRACTThis year the construction of the deep underground stations for the new Amsterdam metro line will start. As apart of the stations’ design a full scale test on the installation of diaphragm wall trenches has been carried outto investigate the impact on the environment. This paper discusses test program, test results, the interpretationof the test results and the 3D FE modelling that was used for prediction and validation purposes.

Keywords: deep excavations, underground stations, diaphragm walls, FE modelling, full scale test

i) grade engineer deep underground stations, Design Office North/South Line Amsterdam, ROYAL HASKONING Rotterdam,Division Transport and Infrastructure, Civil engineering structures and geotechnics , The Netherlands

ii) geotechnical engineer deep underground stations, Design Office North/South Line Amsterdam, Witteveen + Bos Deventer, TheNetherlands



(bentonite) is introduced simultaneously as thetrenching operation proceeds. Once the excavation iscompleted the reinforcement cage is inserted into thebentonite-filled trench. Furthermore the trench willbe filled with concrete by way of tremie pipes,thereby displacing the bentonite from the bottom up.

During excavation the stability of the trench issatisfied by means of a combination of support pres-sure of the bentonite slurry and 3D stress distributionin the surrounding ground, referred to as arching.

Figure 2. Excavation with a bucket grab.

3 FULL SCALE TEST

The objects of the full-scale test program at theMondriaan Tower construction site is to monitor:− vertical and horizontal deformations of the

ground adjacent to the excavated trench;− settlement of loaded piles;− impact on bearing capacity of piles,due to sequential installation (excavation and con-creting) of adjoining diaphragm wall panels.

3.1 Stratification and geotechnical parameters

The surface level is at NAP + 2.0 m (ordnance date).The stratification and geotechnical parameters arelisted in Table 1.

In general the ground conditions are representa-tive of a non-uniform stratum with made ground,Holocene soft clay and peat layers overlaying Pleis-tocene medium dense sand layers and overconsoli-dated clay layers. The ground conditions at the testsite are comparable to those at the locations of thefuture stations of the North/South line. The ground-water level is NAP -0.4 m, the piezometric level ofthe deeper aquifers is NAP -3.0 m.

Table 1. Stratification and soil properties.Layer Z [m NAP] γsat [kN/m3] qc [MPa]Fill (sand) 2.0 20.0 10Holocene Clay -1.0 14.3 0.5Peat (“holland” peat) -3.5 10.3 0.5Clay -7.0 15.2 1.0Peat (“basis”peat) -11.0 11.7 1.5Pleistocene Sand(1st sand layer)

-13.5 20.0 20

Clay (Eem clay) -17.0 18,5 2.0Sand -28.0 19,0 20Clay -42.0 18,5 2.0

3.2 Diaphragm wall and test location

The test program was carried out at a constructionsite for the future 100 m high Mondriaan Tower(figure 3). Underneath the office building a 2 storeyunderground parking is constructed. Diaphragmwalls(length 35 m) are applied as building pit walland will act as structural wall to the undergroundparking in the final stage. Some of the panels, with alength of 55 m, will also serve as foundation ele-ment.

Figure 3. Impression of the Mondriaan Tower.

Figure 4 shows (approximately ¼ of the buildingpit) the diaphragm walls at the test location includ-ing the sequence of excavation.

The panel no’s 2 and 3 were excavated in onecourse. The panel no’s 1, 4 and 5 were excavated in2 or 3 courses as a consequence of different widthsor shapes.



Figure 4. Layout of panel and instrumentation.

3.3 Instrumentation and tests

It was decided to focus more on monitoring of de-formations of the ground rather than on monitoringof changes in stresses. Vertical ground deformationswere monitored by:− 48 precise levelling points at surface level were

placed in a regular grid;− underneath 11 precise levelling points electronic

extensometers were installed to monitor verticalground deformations on deeper levels (NAP –8m,-15, -31 and –51).

These deformations and also the pile settlements(par.3.4) were monitored continuously with a totalstation (figure 5). During the period of excavation ofthe panels 1 to 5, which lasted about 3 weeks, eachof these instruments was monitored at least once inevery 20 minutes.

To monitor horizontal deformations 14 tubes forinclinometers were installed, that were monitored byhand on pre specified moments related to the exca-vation and concreting of the panels.

In two panels 3 piezometers were installed at dif-ferent depths. This was realised by attaching pie-zometers to the reinforcement cages. The instru-ments allowed to monitor the changes in pressure ofthe bentonite and concrete in time. The wet concretepressure is hydrostatic only to a certain depth be-neath the concrete surface, the so-called “criticaldepth” ( Lings et al. 1994). The critical depth ismainly influenced by the type and temperature of theconcrete mix and by the pouring rate of the concrete.In this way the lateral pressure of the wet concrete,which appeared to be an important parameter in theFE-model, could be determined.

To investigate if relief of original stresses withinthe stratum has occurred CPT’s (Cone PressiometerTest) were carried out before and after the installa-tion of diaphragm walls.

Figure 5. Impression of test site, the total station in back-ground.

3.4 Test piles

In the test field 3 steel piles with a diameter of 110mm, representative for typical historical timber pilesin Amsterdam, were driven into the first sand layer.During excavation the settlements of the piles weremonitored. The pile settlements are to be comparedwith the vertical ground deformation monitored atthe nearest reference point at surface level and thenearest extensometer at pile toe level.

On each pile, installed at a distance of about 0.7times the panel width to the trench, bearing capacitytests were carried out before and after the excavationof the diaphragm walls (figure 6).

Figure 6. Pile testing.

4 TEST RESULTS

The measurements were carried during the period of21-09-1998 until 08-02-1999. The diaphragm wallsinstallation at the test field took place from 10-11-1998 until 03-12-1998. Both the diaphragm wall in-stallation and test program were carried out satis-



factory. Based on the raw data only it was concludedthat:1. No significant environmental impact occurred:

- minor vertical ground deformations on surfacelevel and on deeper levels;

- minor horizontal ground deformations duringexcavation of the trench;

- substantial horizontal ground deformations insoft top layers, however impact on piles wasnot observed;

- minor horizontal ground deformations inbearing sand layers and deep clay layers.

2. Pile bearing capacity was not affected;3. An influence of the depth of a panel on the test

results was not monitored, which was alreadynoted at the predictions;

4. The influence of the width of a panel was limited,which was not confirmed by the predictions. Pos-sibly the sequence of installation played a role.This will be illustrated later;

5. Environmental impact at panels with an irregularshape was limited but significant larger than atrectangular panels.

Vertical deform ation [m m] Horizontal deform ation [m m ]

-40

-30

-20

-10

0

-40

-30

-20

-10

0

-20 -10 0 10 20

Figure 7. Shaded plot of maximum measured vertical and hori-zontal deformations.

4.1 Interpretation of the test results

After the test program the results were evaluated andinterpreted. First of all this resulted in an increase ofunderstanding of the effects of diaphragm wall in-stallation and secondly it made the test results acces-sible to back analyses to validate the FE model.For that purpose adjustments of the raw data setwere carried out:− Data cleaning, which means that obvious spikes,

temporary external influences or temporary badfunctioning were corrected. This only concerns

obvious and temporary errors and was performedby the monitoring experts of Soldata.

− The test interpretation is primary focussed on therectangular shaped panels since these are mostlyapplied in the stations’ design. This is achievedby using a limited data set in which the immedi-ate impact of the irregular panel shapes wasomitted for the benefit of the FE validations. Thiswas possible since these panels were installedfirst.

− Correction of the vertical deformations on deeplevels with the effect of the horizontal deformedwire of the extensometer. This will be explainedlater.

4.2 Surface settlements

The targets on surface level show a minor settlementduring excavation of the trench, whereas the areathat is influenced is limited. At a distance of 1.2 m tothe trench (1st grid line) settlements of max. 4 mmdue to excavation were monitored. When the con-creting of the panel was almost completed an instantheave of max. 4 mm at the 1st grid line was ob-served at surface level (figure 8). This instant heavecan be explained by undrained deformation of thesoft clay layers in between the sand layers. The lat-eral concrete pressure pushed the Holocene layersaside. Because of undrained behaviour the soils werenot compressed but horizontally displaced. This isconfirmed by the inclinometer measurements.

-10

-8

-6

-4

-2

0

2

40 2 4 6 8 10

Vertical surface deformation at several distances [m] during D-wall installation

Ver

tical

def

orm

atio

n [m

m]

initial during exc. after exc.after concr. final D-wall

1

3

2

4

5

Figure 8. Settlements on surface level.

4.3 Horizontal ground deformations

During excavation the horizontal ground deforma-tions were less than 10 mm. Due to concreting minor



horizontal ground deformations were measured inthe bearing sand and deep clay layers. Just before thecompletion of the concreting an instant large hori-zontal deformation away form the trench was meas-ured in the top soft clay layers (figure 9). The maxi-mum horizontal ground deformation occurred atNAP-6 m to NAP-9 m. For a single panel this wasapprox. 100 mm away from the trench (at a distanceof 1.2 m to the trench); for the combined effect ofseveral panels this was approx.150 mm.

-35

-30

-25

-20

-15

-10

-5

0-150 -100 -50 0 50

Horizontal displacements (inclinometer) in time [mm]

Dep

th [m

]

26-11-1998 12:13 27-11-1998 7:0230-11-1998 8:58 01-12-1998 14:21

26-11-1998 7:07Reference date:

01-12-1998 14:21

30-11-1998 8:58

Figure 9. Measured horizontal deformations in time at 1.2 mfrom trench.

Figure 10 shows the measured maximum hori-zontal ground deformations perpendicular to thetrench with an increasing distance to the trench. Thedeformations differ from the vertical deformations inmagnitude and extent. Although the horizontalground deformations decrease with the distance tothe trench, still some 30 mm was monitored at 8 mdistance as a combined effect of several panels.

-30

-25

-20

-15

-10

-5

0-200 -150 -100 -50 0 50

Total horizontal (inclinometer) displacements [mm] at distances to D-wall

Dep

th [m

]

1.25 m 2.45 m 4.0 m 5.6 m 7.2 m

Figure 10. Measured total horizontal deformations wide panelat several distances from trench.

4.4 Subsurface settlements

On deeper levels (15 to 30 m below surface level)the extensometer results show maximum settlementsthat are within the same range as on surface level,however heave due to the concreting effect was notobserved. In general the vertical downward grounddeformations due to excavation do not exceed 2 mm.However, most extensometers, especially the onesclose to the trench, indicated a instant settlement (toabout 4 mm at the 1st grid line) at the end of the con-creting phase. At first it was not clear what couldhave caused this movement. However an explana-tion could be found in relation with the inclinometerresults. As was explained before the concreting ofthe panel caused substantial horizontal ground de-formations of over 10 cm in the soft top layers. As-suming that the extensometer rod has developed asimilar horizontal deformation, this means that therod has to extend, resulting in the registration of asettlement that in fact has not occurred. When cor-recting the raw data the settlements develop moreequally which can be considered as a confirmation ofthe correction. In figure 11 the corrected and noncorrected settlement lines perpendicular to the trenchimmediate after concreting are presented. Finally itwas decided to use both settlement lines bounding anarea in which the actual deformations will occur.



-8.0

-6.0

-4.0

-2.0

0.0

2.0

4.00.0 2.0 4.0 6.0 8.0 10.0

Total vertical displacements (extensometer) [mm] at NAP-15 m excl./incl. correction for

horizontal displacementV

ertic

a di

spla

cem

ent [

mm

]

excl.correction V11/08 03:12incl.correction V11/08 03:12D-wallfit excl.correctionfit incl.correction

Figure 11. Extensometer results (corrected and non corrected).

4.5 Pressure tests in trench

From the FE predictions it was learned that the lat-eral wet concrete pressure in trench surface has asignificant effect on the horizontal deformations ofthe soft top ground layers. From literature (Lings etal. 1994) it is learned that up to a certain depth, theso called critical depth, the lateral pressure increaseshydrostatic, based on the wet concrete density. Un-derneath this level the increase of lateral is related tothe bentonite slurry density. This bilinear relation iscaused by the dissipation of water out of the concreteand the beginning of the hardening process. Fromliterature critical depths of 5 to 15 m are known.

-30.0

-25.0

-20.0

-15.0

-10.0

-5.0

0.0

5.00 200 400 600

Piezometer measurement during D-wall installation [kPa]

Dep

th [m

..NA

P]

11:00 12:3014:10 15:30bilinear concrete fit bentonite

1

2

3

4

Figure 12. Piezometer measurements in the trench.

In the test program, on three levels in the trenchpiezometers were installed (attached to reinforce-ment cage) allowing for monitoring the developmentof the lateral pressure during concreting and the be-ginning of the hardening process (figure 12). Thetest results allowed for a determination of the criticaldepth. However the critical depth appeared to bevarying over the depth of the trench and dependingon the pouring rate (in meters height per hour). Thehigher pouring rate in the deep part of the trench re-sulted in a larger critical depth.

In table 2 the results are presented. For the pres-sure envelope still a bilinear diagram can be ob-tained. However the lower part of the diagramdoesn’t relate to the bentonite slurry density but to ahigher density with a value between bentonite andconcrete.

Table 2. Results of piezometer measurements in the trench.Measurement: At the top At the base UnitGradient bentonite slurry 12.0 12.0 kN/mGradient wet/hardened con-crete

23.0 19.0 kN/m

Critical depth 6 >15 mPouring rate 7 >15 m/hr

4.6 Test piles

4.6.1 Pile settlementsThe settlements of the test piles provide informationwhether there is a change of bearing capacity due tothe installation of diaphragm wall panels. Pile set-tlements larger than the vertical ground deformationsat the same location could be an indication of loss ofpile bearing capacity.

However, the pile settlements are in the samemagnitude as the subsoil settlements. Test piles no.1and 3 (about 2.4 m from narrow panel) settled 3 mmto 6 mm; pile no.2 at (about 5.6 m from wide panel)hardly settled (less than 1 mm). Because the pilesettlements and the vertical ground deformations atthe same level correspond very well it is concludedthat the bearing capacity is not reduced.

It appeared from the raw data set that the irregularpanels (Z and corner shape), which are close to thepiles no. 1 and 3, had a significant influence on thesettlements of these piles. For example, the total set-tlements of pile no.1 were primary caused by the in-stallation of the irregular Z-shaped panel and not bythe nearest rectangular panel. For pile no.3 no set-tlements were measured due to the installation of theadjacent wide panel 4, which had a three day bento-nite slurry stage (weekend panel).



4.6.2 Horizontal deformationsAttention was paid to the effect of the large hori-zontal deformations in the top soft soil layers as a re-sult of concreting the panel. It was investigated ifthese deformations could affect the piles. Thereforean upper bound approach was adopted in which thedeformations were considered as induced deforma-tions to the piles. It then was learned that these de-formations are not affective, if the stiffness of thepiles is limited. A limited stiffness means that thepile can follow these deformations easily. This issurely the case with 100-120 mm diameter timberpiles. Concrete piles are more sensible to this phe-nomenon since their stiffness is usually substantiallarger. However, concrete piles are hardly appliedalong the metros´ alignment, but it was decided toinvestigate these locations in more detail.

4.6.3 Bearing capacity testsTo detect a possible loss of pile bearing capacity thepiles were tested before and after the installation ofthe diaphragm wall. In advance calculations werecarried out, using CPT’s, to determine the bearingcapacity based on the Dutch codes and also an upperbound value of the bearing capacity. In the DutchCode calculations only the bearing sand layers aretaken into account. The upper bound value includedall layers. The pile tests that were carried out beforethe installation of the diaphragm wall indicated abearing capacity comparable to the Dutch Codebearing capacity. However, the pile tests that werecarried out afterwards indicated a substantial in-creased bearing capacity approaching the calculatedupper bound value

Apparently the soft layers paid a contribution aswell. Possibly caused by the increased effectivestresses due to the wet concrete pressure, which wasindicated in the CPT’s that were carried out afte r-wards. The fact that the loading steps in the tests af-terwards took significantly more time was an indica-tion that the soft soil layers played a role andabsorbed load. Eventually it was concluded thatthere was no loss of pile bearing capacity due to theinstallation of diaphragm walls.

Table 3. Data pile tests and working load.Pile tests Pile 1

[kN]Pile 2[kN]

Pile 3[kN]

Prediction by CPT(min. max)

245-380 175 –300 175-300

Ultimate load capacity,test before

220 185 150

Ultimate load capacity,test after

360 280 270

Working load,during excavation

150 120 100

-30

-25

-20

-15

-10

-5

0

0 10 20 30 40

Cone resistance qc [MPa] ->

Dep

th [m

..NA

P]

05101520

<- Friction ratio (Rf) [%]

qc before qc after Rf before Rf after

Figure 13. CPT test results before and after diaphragm wall in-stallation.

4.7 Cone pressiometer test

Assessment of the effect of diaphragm wall installa-tion on changes in stresses and bearing capacity ismade by means of cone pressiometer tests (CPT),which were performed before and after the dia-phragm wall installation in vicinity of and furtheraway from of the trench. A change of the cone re-sistance is an indication for a change in the bearingcapacity of the piles. Figure 13 shows that there is nochange in cone resistance in the bearing sand layersand that there is a slight increase of cone resistanceand friction resistance in the top soft layers. Thiscould even be an indication for an increase of thepile bearing capacity, which was confirmed by thepile load tests (chapter 4.6.3).

4.8 Influence panel width and panel shape

As was stated before the panel width hardly influ-enced the test result, which was not expected. How-ever this may be explained by the fact that in the testfield the wide panel was installed in between two al-ready installed panels. The presence of those twoadjacent panels has probably a stiffening effect re-sulting in a reduced impact by the wide panel. Com-paring the horizontal deformations of two adjacentpanels 3 (only adjacent panel 2 present) and 4 (adja-cent panel 3 and 5 present) indicates that there is asignificant influence caused by the presence of adja-cent panels and the sequence of installation (fig. 14).The FE analyses confirms this effect (chapter 5).



-6.0

-5.0

-4.0

-3.0

-2.0

-1.0

0.0-20.0-15.0-10.0-5.00.0

Maximum measured Y-displacement, scaled 10 times [m]

Max

imum

mea

sure

d X

-di

spla

cem

ent,

scal

ed 1

0 tim

es

[m]

panel 5a panel 1c panel 2 panel 3 panel 4

INC02 INC04 INC08 INC08

INC09 INC14 fit panel 3 fit panel 4

small panel 3 large panel 3

Figure 14. Top view of measured horizontal displacements.

At the irregular panel shapes (corner or Z shaped)the maximum settlements that have been monitoredare about double the impact of a rectangular panel.This is true for surface settlements, settlements ondeeper levels and pile settlements. This can be ex-plained by the fact that arching in the sub groundwill not develop that easy. However the impact isstill limited it was decided not to allow irregularpanel shapes close to historical buildings in the un-derground stations’ design .

5 NUMERICAL MODELLING OF DIAPHRAGMWALL INSTALLATION

Numerical modelling of the diaphragm wall installa-tion was performed by means of finite elementanalyses. The main goal was to develop a validated3D FE-model that can be used to predict the soil de-formations and environmental impact during the in-stallation of the diaphragm walls for the under-ground stations for North/South line.

Using a 3D FE model predictions were performedbefore the test program was carried out. Afterwardsthe FE model was validated based on the test resultsthat had become available. An axial symmetricmodel was used as a verification of the 3D analyses.

The calibration was carried out by using the offi-cial North/South Line geotechnical parameter set,changing the soil parameters within a certain range.

5.1 FE - model

For the 3D FE-analyses the 3D version of PLAXIShas been used. The mesh is build of 3D-wedge-elements with 15 nodes and 6 Gauss-points. For cal-culations of one panel a mesh with approximately5,000 nodes has been used. Only a quarter of thepanel is modelled, which means that there are twoplanes of symmetry (see figure 15).

5.2 Simulation of installation

The installation of the diaphragm wall is usuallycarried out in a relative short period of time. There-fore undrained behaviour of the clay and peat layersand drained behaviour of the sand layers is assumed.The installation involves the excavation under beto-nite slurry support, followed by pouring and subse-quent hardening of concrete. The construction of thediaphragm wall is modelled using the followingstages:1. Excavate a single trench by removing the soil

elements and simultaneously, applying the bento-nite pressure on the faces of the trench.

2. Fill the trench with concrete by increasing the lat-eral pressure. The lateral pressure of wet concretecan be described by the bilinear relation :

σh = γc * z z < hcrit σh = γc * hcrit + γb * (z – hcrit) z > hcrit

3. To model the hardening of the concrete, the ele-ments in the trench are switched on, and the stiff-ness and volumetric parameters are changed tothose of concrete.

4. Construct the adjacent panels, one at the time,using the same procedure, simulating the installa-tion of a wall.

In case of a constant pouring rate, γ1=γconcrete can beadopted above the critical depth and γ2=γbentonite be-low the critical depth. However, the test programshowed a different bilinear load distribution in stage2 with a imposed load below the critical depth that isrelated to a higher density than bentonite slurry,which means γc>γ2>γb. However, the calculation re-sults appeared not to be too sensitive to the actualvalue of γ2 if γc>γ2>γbγc>γ2>γb , since the soil layersbelow the critical depth are bearing sand layers andstiff clay layers.

5.3 3D FE predictions

A main objective of the predictions is to gain under-standing of the mechanisms that occur during the in-stallation of a diaphragm wall panel. That is also thereason why a relatively simple FE model wasadopted, where:− limited number of ground layers is used;− the relatively simple Mohr Coulomb based soil

model is applied;− the soil properties are taken as the mean values as

determined at the soil investigation program ofthe North/South Line.

However this resulted in a somehow simplifiedmodel, it appeared to be one where the mechanismsthat occur during installation of diaphragm wallscould be detected in a better way and it allowed for



parametric studies. Summarising the next calcula-tions have been carried out:1. calculations of a single panel in homogeneous

ground as a comparison to a model presented in aconference paper. The results were similar to theresults that were presented in the paper.

2. calculations of a single panel in the Amsterdamsoil, which means a heterogeneous soil profile

3. calculations considering succeeding panels.Figure 15 illustrates some typical results of the pre-dictions.

Figure 15. FE prediction, shaded plot of total displacements.

From the results of these FE predictions the nextconclusions could be drawn :1. 3D stress distribution is clearly observed during

both excavation and concreting (figure 16)2. during the excavation of the trench small settle-

ments and horizontal deformations in the direc-tion of the trench occur.

3. during concreting heave and horizontal deforma-tions away form the trench occur as a result of thehydrostatic wet concrete pressures.

4. there is a limited influence of the depth of thetrench. This can be explained by the heterogene-ous soil profile where vertical arching betweenthe stiff sand layers reduce the effect of the trenchdepth ( figure 17);

5. there is a significant influence of the width of thetrench; this was not observed during the test pro-gram and became subject of further research atthe FE validations.

0

100

200

300

0 1 2 3 4 5 6

distance to centre of panel [m]

hori

zont

al e

ffec

tive

stre

ss [k

Pa]

K0-situation

fase I (bentonite)

stage II (wet concrete)

figure 16. effective horizontal stresses in excavation and con-creting stage close to trench in horizontal plane.

-80

-70

-60

-50

-40

-30

-20

-10

0-40-30-20-100

Prediction of horizontal deformations, stage 1 [mm] (variation of geometry)

Dep

th [m

NA

P]

b=6,2 l=30 b=2,7 l=30 b=2,7 l=50

2,05m

L

B

Figure 17. Illustration vertical arching in heterogeneousground.

5.4 3D FE validations

5.4.1 GeneralIt is important to mention that the aim of the valida-tion was not a perfect fit of the test results, but rathera safe upper bound approach that can be used in thestations’ design. Since the deformations that havebeen measured are very small inaccuracies in thesystem could play a role. Based on the results of thepredictions, the test program and the interpretationsthe FE model was already corrected in advance (step1):− More ground layers were introduced.− Use of advanced non linear soil models that is

necessary to obtain a proper validation (Gourve-nec et al 1999). The constitutive model that wasselected is the Hardening-Soil model. In theHardening Soil model the limiting states of stress



are described by means of the friction angle, φ,the cohesion, c, and the dilatancy angle, ϕ. Thesoil stiffness is described much more accuratelyby using three different input stiffness’: the tri-axial loading stiffness, E50, the triaxial unloadingstiffness, Eur, and the oedometer loading stiffness,Eoed. In contrast to the Mohr-Coulomb model, theHardening-Soil model also accounts for stress-dependency of stiffness moduli. This means thatall stiffness’ increase with pressure (figure 18).

− Higher strength and stiffness for the especially thedeeper soil layers, since the FE predictions did re-sult in substantial larger deformations than weremeasured.

The second step was a more fine tuning calibrationwhich eventually resulted in the so called best fit ofthe test results(BF). Besides that the soil parameterset which has been developed for the North/SouthLine was introduced in the validation process (NZ).However to get results which are comparable to thetest results, upper bound values from theNorth/South Line parameter set for strength andstiffness had to be applied for the first sand layersand deeper layers. The best fit parameter set con-tained even higher values for these properties. How-ever the upper bound North/South Line approach re-sults in a safe upper bound of the test results.

Figure 5: Plaxis Hardening Soil model

cap withcompressionhardening

hardeningshear locus

Figure 18. PLAXIS Hardening soil model.

5.4.2 Calculation results for a single panelFigure 19 shows the maximum horizontal displace-ments at the centre of the panel, at approximately 2m from the trench. The maximum displacement oc-curs in the Holocene top-layers. The calculationswith NZ parameters and in particular the BF pa-rameters are in good agreement with the test-results.

-40

-30

-20

-10

0

0 20 40 60 80 100 120Horizontal displacements, stage 2 [mm]

Dep

th [m

NA

P]

Test result Best Fit (BF) North/Southline par.set (NZ)

2,05m

Figure 19. Horizontal displacements after diaphragm wall in-stallation at 1.2 m distance from trench: 3D FE analyses.

Figure 20 shows the maximum vertical deforma-tions of the 1st sand layer at the level NAP-15 m.Again the FE results are comparable to the test re-sults. These vertical displacement are especiallyrelevant because the sand layer is in most cases thefoundation layer of the timber piles.

-8.0

-6.0

-4.0

-2.0

0.0

2.0

4.00 5 10 15 20

Horizontal distance from panel [m]

Ver

tical

dis

plac

emen

ts [m

m]

Extensometer (no corr.) Extensometer (correction)D-wall Best-Fit (BF)North/Southline par.set (NZ) Prediction (MC-model)

ca. -15m

Figure 20. Vertical displacements after diaphragm wall instal-lation at NAP-15 m: 3D FE analyses.

5.4.3 Influence of panel size and installation se-quence

At the test site 5 panels were monitored with differ-ent length, width and shape. Like the FE predictionsalso the validations showed an increase of impactwhen the width of a panel increases. However, thiswas not found at the test program where also a wide



panel was part of the test field. The reason for thisapparent discrepancy probably was the installationsequence of the panels. The adjacent narrow panels(width 2.7 m) were installed before the wide panel ina relatively undisturbed subsoil whereas the widepanel (length is 6.4 m) was installed in between twoformer installed panels. Those adjacent panelscauses a reduction of the displacements because of:− load transfer to the adjacent stiff concrete panels;− hardening of the soil due to the installation of the

adjacent panels the soil is overconsolidated.To confirm this explanation the described situationis simulated with the 3D FE model. Figure 21 showsthe horizontal displacements of single panels with awidth of 3.2 m (narrow) and 6.4 m (wide). The hori-zontal displacement after stage 2 of the single widepanel is about two times the displacement of the nar-row panel. In the same figure also the results areshown of a wide panel in between two former in-stalled panels “b=6.2m enclosed (NZ)”. It can beconcluded that the horizontal deformations are sig-nificantly smaller than for a single wide panel, whichmeets the explanation made above.

-40

-30

-20

-10

0

0 50 100 150 200

Calculated horizontal displacements 2 [mm]

Dep

th [m

NA

P]

b=3,2m (NZ) b=6,2m (NZ) b=6,2m enclosed (NZ)

2,05m

Figure 21. Horizontal displacements of single panel (narrowand wide) and enclosed wide panel: 3D FE analyses.

5.5 Axial symmetric model

In combination with the 3D FE model a 2D FE axialsymmetric model was used to test the 3D model andto investigate all kinds of features like the behaviourof the (soft) top layers. The 2D axial symmetricanalyses confirms the conclusions of minor dis-placements during excavation (stage 1). It alsoshows that the soft clay layers were pushed asidewhen the concrete load is applied (stage 2). How-ever, the results of the 2D axial symmetric modelcan not be used for validation purposes it appeared

to be a very powerful tool, since it can be appliedvery easily and parametric studies can be carried outin a very short period of time. Despite of the differ-ences of a circular and rectangular trench, the 2Daxial symmetric model gives a reasonable impres-sion of mechanisms, the stress distribution and theextent of deformations, which can be very useful de-veloping the 3D model.

6 FINAL CONCLUSIONS

First of all it is important to conclude that the fullscale test did come up to its expectations. The goalsthat were defined in advanced were all more or lesssatisfied :− the installation of a diaphragm wall has a minor

environmental impact and no affect on the bear-ing capacity of the piles;

− panels with a irregular shape, like corner panelsor a Z shaped panel resulting in a substantialhigher impact. It is recommended not to applythese panels when the distance to adjacent build-ings is small

− modelling the excavation and concreting of a dia-phragm wall is reliable using the 3D Finite Ele-ment Method including advanced non linear con-stitutive soil models.

− The application of the upper bound North/SouthLine set for soil properties results in a safe upperbound of the test results and can be used in theunderground stations’ design.

REFERENCES

Gourvenec, S.M. & W. Powrie (1999). Three-dimensional fi-nite-element analysis of diaphragm wall installation. Geo-technique 49, No. 6, 801-823.

Lings, M.L. & C.W.W. Ng & D.F.T. Nash (1994). The lateralpressure of wet concrete in diaphragm wall panels cast un-der bentonite. Proc. Instn. Civ. Engrs Geotech. Engng 107,July, 163-172.

Cowland, J.W. & Thorley, C.B.B. 1984, Ground and buildingsettlement associated with adjacent slurry trench excava-tion, Proceedings third conference on ground movementsand structures, p. 723-728.

Lings, M.L. & Ng, C.W.W. & Nash, D.F.T. (1994), The lateralpressure of wet concrete in diaphragm wall panels cast un-der bentonite, Proc. Instn. Civ. Engrs. Geotech. Engng, 163-172.

Ng, C.W.W & Lings, M.L. & Simpson, B. & Nash, D.F.T.(1995). An approximate analysis of he three-dimensional ef-fects of diaphragm wall installation, Geotechnique 45, No 3,497-507.

Ng, C.W.W. & Yan, R.W.M.(1998). Prediction of Ground De-formations during a diaphragm wall panel construcion, 13 eSoutheast Asian Geotechncal Conference, Taiwan.

Ng, C.W.W. & Yan, R.W.M. (1998b), Stress Transfer and De-formation Mechanisms around a Diaphragm wall panel,



Journal of geotechnical and geoenvironmental engineering.Vol. 124, No. 7, July 1998.

Wit, de J.C.W.M. (1998). Design of underground stations onthe North/South line. Proc. of the World Tunnel Congres,Sao Paulo.

Wit, de J.C.W.M. & J.C.S Roelands & M. de Kant (1999). Fullscale test on environmental impact of diphragm wall trenchexcavation in Amsterdam. Proc. Int. Sym. on GeotechnicalAspects of Underground Construction in Soft Ground, To-kyo, Japan, 723-730.

Wit, de J.C.W.M. & Lengkeek (2002), H.J.Full scale test onenvironmental impact of diphragm wall trench excavation inAmsterdam – the final results. Proc. Int. Sym. on Geotech-nical Aspects of Underground Construction in Soft Ground,Toulouse, France.

diaphragmwall test

Documents