Você está na página 1de 6

1 INTRODUCTION

Excavation protected by diaphragm walls may be


considered as the default execution method for sta-
tions and other rail- and underground structures in
urban areas. Diaphragm walls of reinforced concrete
are built section by section. The principal construc-
tion phases of such a section, called panel, are as fol-
lows: (a) construction of guide wall (b) panel trench
excavation under bentonite slurry or other support-
ing fluid (c) bottom-up concrete filling of the trench
using a tube (d) reinforcement installation. Dia-
phragm wall panels may reach important dimensions
(up to 1.2 m width, 8 m length and 60 m depth are
not exceptional).
There are several reasons why the study of the ef-
fects due to panel installation is interesting. First (di-
rect effect) to assess the possible impact of the panel
installation on its environment, as sometimes panel
collapse occurs and, quite more frequently, signifi-
cant movements in the vicinity of the panel are regis-
tered during construction. Second (indirect effect)
because even if ground movements induced by panel
execution are small, they may lead to stress changes
in the environment relevant to later excavation stag-
es.
Direct effect of panel installation has more signifi-
cance in the case of soft soils. The indirect effect is
more important in hard overconsolidated soils. As a
first approximation, ignoring the indirect effect tends
to leave design on the side of safety, since the broad
consequence of panel excavation is horizontal stress
relaxation. Ignoring the direct effect instead, is not
safe, since the movements induced by the diaphragm
wall installation are added to those produced by the
subsequent excavation of the structure.
Given the increasing requirements regarding limi-
tation of ground movements in urban areas, devel-
opment of tools to evaluate the effect of the installa-
tion of diaphragm walls in soft soils is necessary. In
the paper, we briefly review some observations on
the problem available in the literature and present
modeling results relevant to a real case.
2 STATE OF THE ART
2.1 Field observations
Although well documented cases of instrumented
excavation of diaphragm wall panels are rare, there
are enough in the literature (Di Biagio and Myrvoll
1973, Poh and Wong, 1988, Tsai and others 2000,
De Wit and Lengkeek , 2003) to obtain some general
lessons. Table 1 lists the main geometric features of
these cases, i.e. the width (W), length (L) and depth
(D) of the studied panel. A description of soil type
and a measure of the characteristic resistance (CR)
are also included for each case. The correspondent
values are also given for the case study presented
later in this paper.
Where panel installation does not lead to failure
(a) final superficial settlements measured at the edge
of the panel vary between 10 and 20 mm (b) those
settlements attenuate rapidly with the distance from
the panel, being negligible for distances greater than
10W (Fig. 1) (c) horizontal displacements at a cer-
tain depth tend to reach values several times higher
than the surface settlements. Its attenuation with dis-
Coupled analysis of movements induced by diaphragm wall installation
B. Garitte, M. Arroyo & A. Gens
Department of Geotechnical Engineering, UPC, Barcelona, Spain








ABSTRACT: The construction process of diaphragm walls can lead to movements in the surrounding area
that are seldom taken into account. However, these movements may be important in situations where soft soils
dominate. In this paper, after briefly reviewing the state of the art on this issue, we present results of a hydro-
mechanical simulation of the problem in a case located in deltaic soils of Barcelona. The auscultation record
of the settlements of a nearby building is employed to validate the computational model. Panel length and
bentonite slurry level were found to be the most influent parameters on induced displacements.
Preprint: 7th European Conference on Numerical
Methods in Geotechnical Engineering, NUMGE
2010, Trondheim, Benz & Nordal, eds., Taylor &
Francis pp 547-552
tance from the panel follows a law similar to the sur-
face settlement.

Table 1. Cases of panel installation experiments in the littera-
ture.
Case Soil CR W(m) L(m) D(m)
Barcelona
(this study)
Sand
q
c
=5
10 MPa
1-1.2 2.8-6 25-35
Clayey
silt
Su=40
80 kPa
Oslo (Di
Biagio and
Myrvoll,
1973)
Soft
clay
Su=30-
40 kPa
1 1.8- 5 28
Singapur
(Poh y Wong,
1998)
Soft
clay
Su=14-
40 kPa
1.2 2 - 6 55.5
Taiwan
(Tsai et al.,
2000)
Silty
sand
q
c
=20
40 MPa
0.9 8 15
Amsterdam
(De Wit and
Lengkeek,
2002)
Clay
and turf
Su=30-
40 kPa
0.8 3 35
Sand
q
c
=20
MPa


Figure 1. Effect of the installation of a panel on the superficial
settlements (Poh and Wong, 1998).

Some characteristics of the construction process are
clearly reflected in the observations. For example,
settlements and movements toward the panel that
occur during the excavation phase decrease during
concreting. Occasionally, the movement direction is
even reversed (Fig. 2). Unloading (excavation) and
loading (concreting) during the installation of a pan-
el also produce pore water pressure changes in less
permeable materials. In that case, measurements dur-
ing the excavation correspond to a situation of par-
tial drainage and more movements may occur during
the subsequent consolidation phase.
Other interesting observations are those concern-
ing the effect of variations of the basic construction
process. In practice, such variations may be due to
work incidents (a) excavation induced movements
are very sensitive to the sustaining fluid level in the
panel trench. Apparently small decreases (e.g. 1m
for a 55m deep panel) can produce significant addi-
tional movements and even the collapse of the ex-
cavation (b) movements are not increased signifi-
cantly during important waiting times (e.g. one day),
as long as a constant fluid level is maintained inside
the trench. (c) Once the impervious cake has devel-
oped satisfactorily on the panel wall, excavation in-
duced movements are relatively insensitive to the
density of the sustaining fluid (in the range 1 to 1.3
t/m
3
).
One aspect that is not clear from the cases de-
scribed in the literature is the influence of the size of
the panel on induced movements. Greater displace-
ments are expected to be induced by panels of larger
length and depth, but there are few data to quantify
this phenomenon.

Figure 2. Settlement at 3m from the panel measured with an
hydraulic cell (Di Biagio y Myrvoll, 1973).


Figure 3. Measured fresh concrete pressure in a 45m deep pan-
el (Schad et al., 2007)

Finally, observations have repeatedly confirmed
(Uriel and Oteo, 1977, Lings et al., 1994; Schad et
Al., 2007) that the pressure applied by fresh concrete
on the panel walls is not always hydrostatic. In fact,
hydrostatic pressure seemed to be maintained only
until a certain depth, referred to as the "critical
depth" (Fig. 3). The observed critical depths vary be-
tween 1/3 and 1/5 of the panel depth. This character-
istic is still poorly understood, but seems to depend
on the interplay between concreting rate and harden-
ing.
2.2 Analytical solutions
There are a number of analytical solutions to the
problem of panel stability (e.g. Fox, 2004; Tsai,
2000). Most of them were derived using limit equi-
librium and the differences between them lie mainly
in the degree of complexity of the alleged failure
surface. A major drawback of these analytical solu-
tions is their limited applicability for layered pro-
files. Moreover, only stability is dealt with and they
are not useful for quantifying the settlement induced
by panel excavation.
However, it is interesting that these analytic solu-
tions also indicate the enormous importance of the
level of bentonite on the panel stability. For instance,
Fox (2004) predicts a security factor reduction from
2 to 1, by a decrease of bentonite level of 2m in a
15m deep trench of 8m length in sand ( = 34, =
20 kN/m
3
).
2.3 Numerical models
An interesting alternative to obtain quantitative an-
swers to the panel problem is to use numerical mod-
els. However this approach is relatively costly. For
practical reasons numerical modeling of excavations
still takes place mostly in 2D. In such models, dia-
phragm wall construction is simultaneous for the en-
tire wall length. Available examples (Ng and Yan,
1998; Gourvenec and Powrie, 1999, Schafer and
Triantafyllidis, 2006) make clear that it is very diffi-
cult to obtain approximate results if the three-
dimensionality of the problem is ignored. The cited
authors also emphasized the importance of the initial
earth pressure coefficient K
0
, because of its influ-
ence on stress redistribution.
Several common modeling features can be noticed
from the precedents (1) guide wall construction is
not considered; (2) excavation under bentonite is re-
produced by removing the elements included in the
volume of the panel and prescribing the hydrostatic
pressure of bentonite on the new contour; (3) fresh
concrete pouring is represented by changing the
boundary condition of total stresses from the hydro-
static bentonite profile to a bilinear profile; (4) final-
ly, to represent hardened concrete, the total stress
boundary condition is removed and elements in the
panel volume are re-activated, with material parame-
ters corresponding to those of reinforced concrete.
Note that, while this procedure is in accordance with
the above mentioned field observations, it does ne-
glect the tangential frictional stresses between soil
and fresh concrete that will be necessary for equilib-
rium. The critical depth is usually taken equal to one
third of panel depth.
3 CASE STUDY
3.1 Background
The case that inspired the studies described here
can be considered typical of the excavations in soft
soil of deltaic areas near Barcelona. The motivation
arose from the observation of significant movements
in a building near some excavation works during di-
aphragm wall installation.
Construction activities were complex because
most diaphragm wall installation was simultaneous
to other potentially disturbing activities, like jet-
grout treatment and micro-pile installa-
tion. However, the construction sequence began with
the execution of a diaphragm wall section near the
building (Fig. 4). During that period, which preceded
all other construction activities, significant building
movements were already registered (Fig. 5). Those
records made clear that diaphragm wall construction
had produced some movement, but they did leave
open the magnitude, because settlement occurring
after the diaphragm wall section was finished might
have been due to consolidation or to other, later ac-
tivities. Since similar diaphragm walls needed to be
constructed in the vicinity a detailed study of this
problem by means of numerical simulations seemed
necessary.


Figure 4. Plane view of the study area . The investigated dia-
phragm wall (R1-5) and reference measurement points (P1-7)
are indicated.
3.2 Geotechnical site characterization
The geotechnical profile at the site might be de-
scribed by five main levels. Made ground (2m thick),
clay (2m), sand (11m), silt (33m) and gravel (unde-
fined). The water table is found at the top of the sand
layer. The geotechnical site characterization proce-
dure cannot be described here other than it was heav-
ily reliant on in situ tests. The in situ measurement
campaign also provided information on the earth
pressure coefficient K
0
which was used to prescribe
initial stress state.

-7
-6
-5
-4
-3
-2
-1
0
1
1
5
/
1
1
/
2
0
0
6
0
5
/
1
2
/
2
0
0
6
2
5
/
1
2
/
2
0
0
6
1
4
/
0
1
/
2
0
0
7
0
3
/
0
2
/
2
0
0
7
2
3
/
0
2
/
2
0
0
7
S
e
t
t
l
e
m
e
n
t
s

[
m
m
]
P1
P2
P3
P4
P5
P6
P7

Figure 5. Settlements measured during the execution of the R-
diaphragm wall. Vertical dotted lines indicate the day in which
each panel was built (R5 to R1).

The silt package is slightly overconsolidated and has
a relatively high deformability. It is likely that much
of its deformation takes place under plastic re-
gime. Hence, the characterization of this regime has
some importance and a modified Cam-clay (MCC)
model was chosen. For the other materials a simpler
Mohr-Coulomb (MC) model was selected. The most
important geomechanical parameters employed in
the calculations are shown in Table 2.

Table 2a. Geomechanical parameters (MC)
E c
(m/d) (kPa) (kPa) ()
Made
ground
0.22 3000 0.3 30 25
Clays 0.005 15,000 0.3 50 29
Sand 31.3 20,000 0.3 1 33
Gravel 105 31,000 0.3 50 36

Table 2b. Geomechanical parameters (MCC)
OCR
(m/d) ()
Silt 0.003 27 1.15 0.012 0.06
3.3 Characteristics of the numerical models
Precedent published analyses of this problem mod-
eled clayey soils, for which undrained behaviour
might be safely assumed. The geotechnical profile in
this case includes layers of very different permeabil-
ity and any generic assumption about drainage was
not granted. For this reason fully coupled hydrome-
chanical computations were performed.
The program employed was Code_Bright, a finite
element code developed in the Department of Ge-
otechnical Engineering of UPC (Olivella, 1995). The
modeling of the construction process of a panel re-
quired some modifications to the program, the most
important being the implementation of a boundary
condition with (bi)linear stress variation with depth.
The new implementation was verified by bench-
marking against a case reported in the literature
(Gourvenec and Powrie, 1999).
Two types of analysis were performed: a paramet-
ric study of the excavation of an isolated panel and a
detailed modeling of a particular excavation se-
quence of five panels, namely the R-diaphragm wall
(Fig. 4). Figure 6 shows the mesh used for modeling
the R-diaphragm wall excavation sequence (the
mesh used to simulate the installation of one panel is
similar). Advantage was taken of the vertical longi-
tudinal symmetry plane of the panels.



Figure 6. 3D mesh used for the analysis of the R-diaphragm
wall excavation sequence.

Coupled hydro-mechanical computations require the
explicit specification of construction times. After
some consultation with the site managers and in-
spection of construction records, a site-
representative construction sequence was established
for the base case. Excavation under bentonite sup-
port is modeled by removing meter by meter the el-
ements of a panel during 3.2 hours in the base case
(25 m deep panel). Once a certain volume has been
removed in a panel, hydrostatic bentonite pressure is
applied as total stress on the new wall. After the ex-
cavation sequence finishes, a five hour waiting time
represents bottom cleansing and reinforcement
placement. The hydrostatic bentonite pressure profile
is then replaced by the bilinear profile representing
fresh concrete. Concrete hardening time was esti-
mated as 12 hours, after which solid elements with
concrete properties are placed in the panel. Concrete
hardening is thus modeled as an instantaneous pro-
cess, which is clearly unrealistic.
4 RESULTS
4.1 Base case
Geometric features of the base case (depth D, length
L and width W) are shown in Table 3. Bentonite lev-
el within the panel, n
b
, and the assumed critical
depth D
c
, are also indicated.
Surface settlement histories at different distances
from the panel edge are given in Figure 7. Times at
which excavation ends, fresh concrete is poured and
hardening is assumed are indicated by vertical dotted
lines. Excavation produces settlement, reaching
about 3.5mm at 2m from the panel edge. Most set-
tlements induced by the excavation occur simultane-
ously to it and later consolidation has a moderate in-
fluence. Fresh concrete deposition results in heave
and after concrete hardening, settlements are re-
sumed. Final settlement values are similar to those
registered after excavation. It is worth noting the
qualitative similarity with the measurements by Di
Biagio and Myrvoll (Fig. 2), especially during fresh
concrete injection.

-5
-4
-3
-2
-1
0
1
0 10 20 30 40
Time [hrs]
S
e
t
t
l
e
m
e
n
t
s

[
m
m
]
Settlements @ 2 m
Settlements @ 3 m
Settlements @ 4 m
Settlements @ 5 m
Settlements @ 10 m
Settlements @ 16 m
End of excavation
Concrete hardening
Fresh concrete pouring

Figure 7. Time evolution of settlements at different distances
from the panel edge (base case).

Table 3: Modeling of single panel. Parametric study cases.
Case D(m) L(m) W(m) n
b
(m) D
c
(m)
Base (V1) 25 3.6 1.2 0 8
V2 25 3.6 1.2 2 8
V3 25 6 1.2 0 8
V4 35 3.6 1.2 0 8
V5 25 3.6 1.2 0 5
V6 25 3.6 1 0 8
4.2 Parametric study of a single panel
The parametric study includes five variations on the
base case. As outlined in Table 3, a single parameter
was changed from the base case for each variant. In
variant 2 (V2) the level of bentonite, n
b
, was lowered
by 2m; in variant 3 (V3) a length of 6m is considered
for the panel; in variant 4 (V4) a deeper panel was
excavated up to 35m; in variant 5 (V5) the critical
depth was modified to 1/5 of the panel depth instead
of 1/3 and in variant 6 (V6) the panel was made
thinner.
A comparison between the different cases is given
in Figure 8, in terms of settlement evolution and hor-
izontal displacement of the panel wall. According to
the simulation results, the most damaging cases are
V3 and V2, i.e. a longer panel and a lower bentonite
level. Differences between the other cases are negli-
gible for superficial settlement measurements, alt-
hough, as expected, a deeper panel causes higher
movements at depth.

-6
-5
-4
-3
-2
-1
0
0 10 20 30
Time [hrs]
S
e
t
t
l
e
m
e
n
t
s

[
m
m
]
V1 V2
V3 V4
V5 V6
-25
-20
-15
-10
-5
0
-40 -30 -20 -10 0
Hor. Displ. [mm]
D
e
p
t
h

[
m
]
V1
V2
V3
V4
V5
V6

Figure 8. Left graph: evolution of settlements at 3m from the
panel edge for the 6 cases of the parametric study. Right graph:
horizontal displacements of the panel wall after the excavation
phase for the 6 cases of the parametric study.
4.3 R-diaphragm wall construction sequence
The R-diaphragm wall panels (Fig. 4) have similar
dimensions to the base case panel: all panels have a
length of 3.6m, excepted one (R1), which is 4.9m
long and the depth of the wall is 24m instead of 25m
in the base case. The construction sequence was
simulated according to work site infor-
mation. Bentonite level and critical depth were taken
as in the base case because no measurements were
available.
This computation confirmed some of the previous
results. For instance, the installation of one 6m long
panel produces more settlement than the consecutive
installation of two 3.6m long panels (Fig.9). The dif-
ference in settlement behaviour appears mostly dur-
ing the excavation phase. After concrete hardening,
quite similar final settlement levels are predicted for
the 6m long and the joint 7.2m long excavation.
Simulated and measured settlements at observa-
tion points P1 and P2 are compared in Figure 10.
The position of those points was illustrated in Figure
4. Quite a good agreement between measurements
and simulation is obtained. One remarkable differ-
ence is the smooth evolution of the measurements
when compared with simulation results. This is like-
ly due to the abrupt modeling approach adopted for
concrete hardening (instantaneous).

-6
-5
-4
-3
-2
-1
0
1
0 20 40 60 80 100 120 140
Time [hrs]
S
e
t
t
l
e
m
e
n
t
s

[
m
m
]
2 panels (L=3.6m)
1 panel (L=6m)

Figure 9. Superficial settlement simulated at a distance of 3m
from the panel wall for two panels of 3.6m length and one pan-
el of 6m length.

-7
-6
-5
-4
-3
-2
-1
0
1
5
/
1
1
/
2
0
0
6
0
5
/
1
2
/
2
0
0
6
2
5
/
1
2
/
2
0
0
6
1
4
/
0
1
/
2
0
0
7
0
3
/
0
2
/
2
0
0
7
2
3
/
0
2
/
2
0
0
7
S
e
t
t
l
e
m
e
n
t
s

[
m
m
]
P1 (simulation)
P1 (measurements)
P2 (simulation)
P2 (measurements)

Figure 10. Comparison of measured and simulated settlements
at observation points P1 and P2.

Measured and simulated settlement profiles after
construction are drawn in Figure 11. Two simulation
profiles have been plotted: one at the level of R4,
corresponding to observation point P1 and one at the
height of R1-R2, corresponding with the remaining
points. The fit is quite good for measurement points
close to the wall, but its quality decreases with dis-
tance to the wall. The observed discrepancies are
probably due to the fact that the building structure is
not taken into account in the computation. The struc-
ture may act as a stiffening element, hindering set-
tlement recovery during fresh concrete injection. The
overall good match between simulation results and
measurements can be considered satisfactory and
gives more credibility to the results of the parametric
study.
CONCLUSIONS
Diaphragm wall installation in soft soils may pro-
duce settlement in its neighborhood. Numerical
models may help to quantify and understand the
problem. The presented parametric study allow for
isolating two influent parameters: bentonite level
and panel length. Other parameters, like panel width,
depth and critical depth were found to be less im-
portant.
-14
-12
-10
-8
-6
-4
-2
0
0 20 40 60 80
Distance to panel edge [m]
S
e
t
t
l
e
m
e
n
t
s

[
m
m
]
Simulated (R1-R2) @ 27/11/2006
Simulated (R4) @ 27/11/2006
Measurements @ 12/02/2007 (p1)
Measurements @ 12/02/2007 (other plots)

Figure 11. Measured and simulated settlement profiles after
construction.
REFERENCES

De Wit, J . C. W. M. & Lengkeek, H. J . (2002). Full scale test
on environmental impact of diaphragm wall trench installa-
tion in Amsterdam the final results. Proceedings of the in-
ternational symposium on geotechnical aspects of under-
ground construction in soft ground, Toulouse, France (eds
R. Kastner, F. Emeriault, D. Dias and A. Guilloux), pp.
433440. Lyon
DiBiagio E, Myrvoll F. (1972) Full scale field test of a slurry
trench excavation in soft clay. Proceedings of the 15th Eu-
ropean Conference Soil Mechanics Foundation Engineer-
ing, Madrid 1972; 461471.
Fox, P.J . (2004) Analytical solutions for stability of slurry
trench, ASCE J ournal of Geotechnical and Geoenvironmen-
tal Engineering, Vol. 130, No. 7, 749-758
Gourvenec, S. M. & Powrie, W. (1999). Three-dimensional fi-
nite-element analysis of diaphragm wall installation.
Gotechnique 49, No. 6, 801- 823
Lings M, Ng CWW, Nash DFT. (1994) The lateral pressure of
wet concrete in diaphragm wall panels cast under bentonite.
Proceedings of the Institution of Civil Engineers: Geotech-
nical Engineering; 107:163172.
Ng CWW, Yan RWM. (1998) Stress transfer and deformation
mechanism around a diaphragm wall panel. J ournal of Ge-
otechnical and Geoenvironmental Engineering; 128(7):638
648.
Olivella, S., 1995. Nonisothermal multiphase flow of brine and
gas through saline media. Doctoral Thesis, Technical Uni-
versity of Catalonia (UPC), Barcelona, Spain.
Poh TY, Wong IH. (1998) Effects of construction of diaphragm
wall panels on adjacent ground: field trial. J ournal of Ge-
otechnical and Geoenvironmental Engineering; 124(8):749
756
Schad, H., Vermeer, P.A., Lchler, A. (2007) Fresh concrete
pressure in diaphragm wall panels and resulting defor-
mations. In: Grosse, Ch. U. (Ed.): Advances in Construction
Materials, Berlin: Springer Verlag, 2007, pp. 505-512.
Schafer R, Triantafyllidis T. (2006) The influence of the con-
struction process on the deformation behaviour of dia-
phragm walls in soft clayey ground. International J ournal
for Numerical and Analytical Methods in Geomechanics;
30:563576
Tsai, J .S., J ou, L.D., Hsieh, H.S. (2000) A full scale stability
experiment on a diaphragm wall trench, Canadian Geotech-
nical J ournal, 37, 379-392
Uriel S. y Oteo C. S. (1977) Stress and strain beside a circular
trench wall. Proc. 9th ICSMFE, Tokyo, 1,781-788.

Você também pode gostar