You are on page 1of 14

International Journal of Civil Engineering (2019) 17:659–672

https://doi.org/10.1007/s40999-019-00391-9

RESEARCH PAPER

Numerical Modelling and Experimental Monitoring of a Full-Scale


Diaphragm Wall
Valentina Lentini1   · Francesco Castelli1 

Received: 3 April 2018 / Revised: 21 November 2018 / Accepted: 2 January 2019 / Published online: 28 January 2019
© Iran University of Science and Technology 2019

Abstract
Ground movements due to deep excavation may cause damage or loss of functionality to nearby building, especially in urban
areas: embedded diaphragm walls between the new and the existing buildings can be effective in reducing these movements.
The paper reports the results of an experimental and numerical study carried on a full-scale anchored piles diaphragm used
to supporting deep excavation in urban area devoted to the new Library of the University of Enna “Kore”, Enna (Italy). Two
piles, located in the central part of the diaphragm wall, have been instrumented with conventional inclinometer cases and
embedded piezoelectric accelerometers. The experimental measurements recorded during the construction and in operation
are reported and discussed in detail. Plane strain finite element analyses using the code PLAXIS 2D are presented in which
the diaphragm was modelled to evaluate the geotechnical system behavior. It has been observed that this model allows for a
satisfactory simulation of the displacement of the wall during the construction phases under a static loading scenario. The
results in terms of horizontal displacements obtained by numerical modelling are in good agreement with those derived by
measurements. Furthermore, on the basis of measured data, an empirical method has been used to evaluate the surface set-
tlements, whose values guarantee a good safety threshold to the nearby building.

Keywords  Numerical modelling · Excavation · Monitoring · Diaphragm wall

1 Introduction Deformation control is often as critical as assurance


against collapse [1–5]. Therefore, limiting values of dis-
In recent decades, the demand for underground space placements or strains should be specified to define service-
increased. Deep excavations are required to meet this ability in terms of just acceptable conditions and an accu-
demand, and, in many cases, excavation sites are in closing rate prediction of displacements under working conditions
proximity to existing structures. Consequently, advanced is required. There are two common techniques for estimating
techniques are needed in the excavations to mitigate lateral wall deflections and soil settlements involving either inter-
wall deflections and surface settlements for the purpose of polation from a published empirical database or numeri-
avoiding damage of the adjacent structures. cal analysis using finite elements. Finite element modelling
Large-scale excavations require retaining structures, such (FEM) is important in geotechnical design of excavations,
as diaphragm walls, to be installed before the soil excava- representing a powerful tool in the cases in which excava-
tion. Diaphragm walls have been widely used as primary and tions are located in urban areas. Although the influence of
permanent structural elements for supporting deep excava- excavation work on the surrounding ground and on existing
tions in urban area. structures has been commonly evaluated through numerical
simulation [6–10], the main interest focuses on the influ-
ence of the excavation process on these areas after retaining
* Valentina Lentini
valentina.lentini@unikore.it structures are installed in the soil. In these cases, monitor-
ing of the response of the wall (in terms of displacements)
Francesco Castelli
francesco.castelli@unikore.it plays a relevant role for a quality assessment of the design
and construction process. It is usually based on comparisons
1
Geotechnical Engineering, Faculty of Engineering between the predicted and the observed behavior.
and Architecture, University of Enna Kore, Via delle
Olimpiadi, 94100 Enna, Italy

13
Vol.:(0123456789)

660 International Journal of Civil Engineering (2019) 17:659–672

Furthermore, empirical methods for estimations of On the basis of measured data, an empirical method has been
ground deformation are often sufficient for practical applica- used to evaluate the surface settlements.
tions. These methods are subjected to important limitations,
because they have been developed, or have been validated,
for a limited number of cases. Nevertheless, their advantage 2 Metrology
is that only a few parameters for the soil and geometry are
required. In engineering practice, surface settlements are Measurements of displacements and pressures during and
usually estimated using empirical formula [11], where it is after the construction of geotechnical structures are com-
assumed that the ground settlement can be approximated by monly used to check design assumptions and they are even
a normal Gaussian distribution curve. prescribed by some codes [12, 13]. In order to record the
The paper provides a description of the monitoring sys- influence the deep excavation and, moreover the whole new
tem used to monitor the anchored diaphragm wall designed building, has on the neighboring structures, a monitor of the
for the construction of the new Library of the University displacements has been performed.
of Enna “Kore” (Italy), the interpretation of the obtained The diaphragm wall, located at Enna, in the Sicily
data and a discussion about assumptions and results of the Region, consists of two rows of reinforced concrete (r.c.)
numerical analyses carried out by the FEM code PLAXIS. piles 1000 mm in diameter and has a free height of about
10 m with an overall length of 22 m. An r.c. top-beam has
been constructed on the top of the structure and one row
of anchorage has been realized at depth of 4.50 m (Fig. 1).
An integrated structural and geotechnical monitoring
system for the diaphragm wall has been used. Two piles
(Fig. 2), located in the central part of the wall, have been
instrumented with conventional inclinometer cases (Fig. 3)
and embedded piezoelectric accelerometers. The two piles
were chosen in order to consider about the middle point of
the two parts making up the diaphragm. The response of
the two piles, chosen in order to avoid as much as possible
boundary effects, was monitored. Static displacements were
registered during the construction of the structures.
Frequent measurements have been carried out during the
excavation and the construction phases. The need to moni-
tor deflections along the pile axis has driven the choice of
embedding sensors in the piles. They have been equipped
Fig. 1  View of the building and anchored embedded wall

Fig. 2  Plan view of the dia-


phragm wall and the monitored
piles

13
International Journal of Civil Engineering (2019) 17:659–672 661

behavior. Two piles have been instrumented with no.5


embedded piezoelectric accelerometers placed at different
depths (Fig. 4) and conventional inclinometer casings.
Sensor module consists of two seismic accelerometers,
placed in two orthogonal directions and encapsulated in a
stainless steel enclosure (Fig. 4) to avoid any damage during
the casting operations and assure waterproofing. Specific
procedures for concrete casting have been adopted to ensure
the effectiveness of sensors after installation, which are bur-
ied in concrete and, therefore, not repairable. Furthermore,
the instrumented piles had to show similar characteristics
with respect to the adjacent ones, to assure reliability of
measurement.

Fig. 3  View of the diaphragm wall and instrumented pile 2.2 Sensor Installation and Data Acquisition

with two inclinometer casings and embedded piezoelectric In this section, the main features of a monitoring system
accelerometers. architecture, developed by the Laboratory of Experimental
The short distance between building and wall suggests Dynamics of the University of Enna “Kore”, is described.
that there should exist an interaction between the two struc- First, the relevant features of the used sensors are reported
tures. Thus, the knowledge about the building behavior can and then the monitoring system characteristics are detailed.
help in better understanding measurement results obtained The sensor package constitutes the main device of the moni-
from the sensors in the piles. Sensor characteristics and toring system and it is comprised of two parts, namely the
installation phases are going to be described in detail in the processor board and the sensor board, respectively.
following sections. The processor board main element is a microprocessor
with 16 k bytes flash (program) memory, 768 bytes SRAM
2.1 The Embedded Piezoelectric Accelerometers and 256 bytes of data memory, able to interpret commands
from a master node and to elaborate data at the sensor level.
Over the past 50  years engineers have used traditional, The board mounts a 18-MHz clock sensor in order to syn-
manual inclinometers to monitor the soil displacements. chronize data acquisition. Analogical data read from the
However, the frequency and accuracy of this labor-intensive sensor board are conditioned and converted by means of a
monitoring depend on a trained engineer in manually col- dedicated 16 bit sigma-delta ADC. Digital data are then sent
lecting and managing the data. to the master node (PC) by means of a transceiver RS485.
Recent advances in miniaturizing sensors and electronics The sensor board consists of an InvenSense, Inc. model
have enabled a paradigm shift in the monitoring of soil and MPU-6000 6-axis MEMS sensor. The sensor includes a
structures. In order to observe soil displacements, highly 3-axes MEMS accelerometer that uses separate proof masses
sensitive sensors are necessary. These sensors are typically for each axis. Acceleration along a particular axis induces
expensive due to their high sensitivity and precision. displacement on the corresponding proof mass and capaci-
Abdoun et al. [14] developed a microelectromechanical tive sensors detect the displacement differentially. When the
system (MEMS) accelerometer that is designed to monitor device is at rest, its three axes are able to sense the static
soil displacements and soil accelerations. The recent MEMS component of the gravitational acceleration. The full-scale
technology may offer some distinct advantages over other range of the digital output can be adjusted to ± 2 g, ± 4 g,
types of tilt sensors, and in fact many manufacturers are now ± 8 g, or ± 16 g.
offering MEMS-based inclinometer probes. The wireless The architecture of the monitoring system consists mainly
system can remotely collect both deformation and accelera- of: (1) a general purpose personal computer to manage the
tion readings anywhere where there is cell-phone coverage. flow of data coming from the network of sensor nodes; (2)
The sensor array is capable of measuring soil acceleration the master node which is connected to the personal computer
and permanent ground deformation to a depth of 100 m. by USB port and it has the main function of translating the
Each sensor array is connected to a wireless earth station to physical USB protocol to the RS485 slave nodes protocol;
enable real-time monitoring of a wide range of soil and soil- (3) the slave nodes or sensor nodes which represent the
structure systems as well as remote sensor configuration. peripheral units of the system.
In this study MEMS accelerometers and traditional Each slave node is connected to the master one by a
inclinometers have been used to monitor diaphragm wall serial bus with a communication protocol based on the

13

662 International Journal of Civil Engineering (2019) 17:659–672

Fig. 4  Scheme of the position of accelerometers (a); stainless steel enclosure (b, c)

physical layer RS485 and with a logical protocol developed of 0.006°. The table has been rotated from 0.0° to − 1.4° in
by Wisenet Engineering s.r.l. The RS485 standard ensures 35 steps so that, once the reference gravitational accelera-
an optimal electromagnetic noise isolation while the logical tion components are known, it was possible to determine the
protocol guarantees the optimization of the data flow on the decomposition of the gravitational vector along the sensor
bus, the coherence control of the data packs by means of a measurement axes and, consequently, the corresponding Euler
16-bit cyclic redundancy check (CRC-16) and the identifica- angles. For each sensor, this test has been repeated twice in
tion of the node ID information. Moreover, the logic proto- order to measure angle variations on two axes per time. Sensor
col allows the transmission of the data on a single RS485 bus data have been corrected, as explained before, and the com-
by means of a time-division multiplexing. Data acquisition puted angles have been compared to the actual angles meas-
is synchronized since the command is sent from the master ured on the reference table.
node to all the slave nodes (sensors) via broadcast digital The charts in Fig. 5 show these comparisons. Real angles
information. After the acquisition, each node sends the measured by the index table are plotted against those obtained
acquired data to the master node according to the sensor ID. by the sensor outputs, showing a very good agreement. By
An experimental campaign aimed at the calibration of means of classical statistical error propagation analysis [15],
the sensors has been performed. Sensors have been tested it has been found that the smallest measurable angles for each
by means of static tests in order to determine ground noise, sensor axis are approximately ± 0.002°, ± 0.002° and ± 0.003°
static sensitivity and their dependence on environmental for the x-, y- and z-axis, respectively. Although measured
temperature. All tests have been carried out at the Labora- angles present small deviations from the real table angles,
tory of Experimental Dynamics of the Faculty of Engineer- anomalies are smaller than the tolerances expected.
ing and Architecture of the University of Enna “Kore”. To detect soil movement, the static acceleration due to
In order to use the MEMS sensors in a monitoring sys- gravity sensed by the accelerometer has been measured. The
tem for civil structures, the basic idea is to discard their angle between the longitudinal axis and the gravity vector is
dynamic behavior and use them as inclinometers, exploiting computed as follows:
their capacity to read the static component of the gravity ( ) ( )
−1 accelerometer data 180
acceleration along their three axes. For such a reason, the 𝜗 = sin × . (1)
gravity 𝜋
angle sensitivity of the MEMS sensors has been determined
by means of multi position static tests. The mean values of the recorded accelerations at each
The sensors have been mounted on an index table capa- axis have been compared to the gravity acceleration
ble of axis rotation which can be measured with a resolution

13
International Journal of Civil Engineering (2019) 17:659–672 663

Fig. 5  Comparison between the real angle observed by the index table and those measured by the sensors

component in order to compute the scale factors (sf) that 2.3 The Inclinometers
have been used to convert the sensor output to the physical
variable of interest. The procedure adopted to find the scale Measurements of horizontal displacements of the diaphragm
factors for each axis consisted in a series of static acquisi- wall were carried out using inclinometer casings and were
tions of accelerations when, in turn, one of the axis of the performed with steps of 1.0 m, starting from the bottom of
accelerometer was matched with the vertical direction, once the inclinometer hole.
downward ­(sf−) and once upward ­(sf+). Experimental data have been acquired over a 4-month
The gain gi and the bias bi (i = x, y, z) of the raw data can period (from July 29th, 2014 to November 4th, 2014) dur-
be then obtained by assuming a linear variation from − 1 ing the construction and excavation phases of the embedded
to + 1 g: diaphragm wall. Measurements recorded on September 15th
and November 4th, 2014 are reported in Fig. 6 for the moni-
⎧ sf+i − sf−i tored piles in terms of lateral displacement.
g
⎪ i =
2 Displacements of the two piles are not very similar. A
⎨ i = x, y, z. (2)
+
⎪b = i sfi
sf + −
continuous increase in the horizontal displacements of the
⎩ i 2 wall has been observed, until it has reached a final value of
about 49 mm for inclinometer I1 and of about 23 mm for
inclinometer I2.
Gain and bias are then used to compute the static cor-
rected output as,i for each axis according to the following
expression:
3 Geological and Geotechnical
as,i = ms,i ⋅ ai + bs,i i = x, y, z, (3) Characterization
where ai is the raw output from accelerometer axis, and
ms,i and bs,i are the coefficients of the linear transformation To realize the new Library of the University of Enna
defined as follows: “Kore”, a detailed soil investigation of the site has been
performed. In particular, were carried out no.3 boreholes
⎧ms,i = 1 up to a depth of 30 m to determine the stratigraphic col-
⎪ gi umns, n.1 down-hole tests (DH) to evaluate compres-
⎨ b
i = x, y, z. (4)
⎪bs,i = − i sion (VP) and shear (VS) wave velocities, cross hole sonic
⎩ gi logging tests across several piles to check its integrity.
Undisturbed samples have been retrieved from boreholes
Accordingly, it is possible to state that the proposed sen- to determine, through standard classification test, the grain
sor can be also used as an inclinometer for application in size distribution and the physical properties in terms of
civil engineering structural monitoring [16]. unit weight γ, water content w and plasticity index I P.

13

664 International Journal of Civil Engineering (2019) 17:659–672

Fig. 6  Lateral pile displace- Disp lacement [mm] Disp lacement [mm]


ments versus depth 0 10 20 30 40 50 0 10 20 30 40 50
0 0

I nclinometer I 1 I nclinometer I 2
2 2

4 4

6 6

8 8

10 10
Dep th [m]

Dep th [m]
12 12

14 14

16 16

18 18

2014 Sep tember 15 th 2014 Sep tember 15 th


20 20
2014 N ovember 4 th 2014 N ovember 4 th

22 22

Furthermore, direct shear test and unconsolidated und- As an example, the stratigraphic column derived from
rained triaxial test have been carried out to evaluate the a borehole is reported in Fig. 7a. The stratigraphy shows
shear strength parameters in drained (effective cohesion c′ clayey silt up to 4 m below the ground surface, sandy silt
and effective friction angle 𝜙′ ) and undrained (undrained from 4 up to 6 m, gypsum clay from 6 up to 10 m and finally
cohesion cu) conditions, respectively. The results of labo- gypsum from 10 up to 30 m. The results in terms of VP and
ratory test, summarized in the Table 1, are reported in VS versus depth obtained by DH test are reported in Fig. 7b.
detail by Castelli et al. [17–19]. Values between 254 and 1936 m/s and 111 and 1159 m/s

Table 1  Geotechnical properties Sample Depth (m) Clay (%) Silt (%) Sand (%) γ (kN/m3) wn (%) IP (%) cu (kPa) c′ (kPa) 𝜙′ (°)
derived by laboratory tests
S1/C2 8.00 10.15 30.01 46.11 19,27 24.74 14.80 – 0 31
S1/C3 18.0 25.13 53.54 21.27 19,88 31.86 36,39 – 0 38
S2/C1 3.00 23.35 43.79 25.63 20.02 21.97 19.97 – 10 24
S2/C3 13.2 37.45 51.80 10.72 20.52 22.50 47.44 – – –
S3/C2 5.70 26.24 34.80 36.44 19.25 28.18 22.79 – – –

γ unit weight; w water content; IP plasticity index; c′ effective cohesion; 𝜙′ effective friction angle; cu und-
rained cohesion

13
International Journal of Civil Engineering (2019) 17:659–672 665

Waves velocity (m/s )


0 400 800 1200 1600 2000
0

VS
VP

10

Depth (m)
15

20

25

30

(a) (b)

Fig. 7  Stratigraphic column (a); waves velocity versus depth (b)

have been estimated for compression and shear waves,


respectively.
To adopt a high level of quality control and assurance,
the CSL testing procedure has been performed to detect
anomalies inside concrete. Cross-hole sonic logging (CSL)
is the most accurate and reliable technique for assessing the
integrity of deep foundation. One transmitting probe and one
receiving ultrasonic probe are lowered into parallel access
tubes, which have to be fixed to the reinforcement cage
before casting the concrete (Fig. 8). Measurements are taken
each 1–5 cm with both probes at the same depth. First arrival
time (which can be converted to apparent wave speed) and
amplitude are recorded as well as the raw times series. The
data are presented as log or waterfall diagrams, respectively
(Fig. 9). For both piles tested (PA42 and PA100) have been
obtained undisturbed signals from top to bottom with an Fig. 8  Cross-hole sonic logging
average value of ultrasonic waves velocity equal to 3500 m/s.
Based on the laboratory and in situ investigation results,
a simplified geotechnical model was adopted and used to
set a numerical model of the diaphragm wall for PLAXIS
code [20].

13

666 International Journal of Civil Engineering (2019) 17:659–672

Fig. 9  Cross-hole sonic logging plot of a sound pile for PA42 and PA100 piles

4 Empirical Solution for Maximum Surface monitoring and empirical solution coupled with numeri-
Settlement Estimation cal modelling are necessary to model the effect of surface
subsidence due to excavation.
In new buildings design causing minimum disturbance to According to previous researches, most commonly used
the routine daily lives of the city is very important, as infra- approach is based on the assumption that the surface set-
structure, buildings and services stretch through the densely tlement can be approximated to a Gaussian distribution.
populated and scarcely limited land space. In order to fully Martos [21] first observed that the shape of the surface
understand the extent of disturbance due to excavation in settlement trough above mining excavations can usually be
urban areas, a comprehensive knowledge of the deformation represented by a Gaussian curve. Later, Peck [22], Schmidt
caused by excavation will be useful. Hence construction of [23] and Rowe et al. [24] investigated surface settlement
new buildings requires meticulous considerations in terms data from a large number of tunnels and proposed that the
of geotechnical site investigations, construction methods, Gaussian function, as shown in Eq. (5), can be applied for
instrumentation and monitoring of surface subsidence and describing the surface settlement trough. This conclusion
the subsequent impact on nearby buildings and services. was based on statistical evaluation of field observed data
Among these considerations, the most important aspect is over 20 case histories.
the control of surface subsidence to minimize any distur- ( 2)
x
bance to nearby buildings and services. Sx = Smax exp − 2 , (5)
2i
The study of surface subsidence is limited to mainly
empirical solutions based on field studies, and very few where x is the horizontal distance from the diaphragm cen-
analytical studies have been carried out. Hence in situ terline and i is the horizontal distance from the diaphragm

13
International Journal of Civil Engineering (2019) 17:659–672 667

B Distance from diaphragm [m]


0 10 20 30 40 50
y x 0

1
Vw Vs
S max

Settlement [cm]
Sx 2
ymax
H 3
i
4

Fig. 11  Settlements versus distance from diaphragm

D
border disturbances. At the base, the boundary conditions
were assumed to be pinned in both vertical and horizontal
directions and laterally the model is free to move vertically.
Meshing plays one of the key aspects in obtaining a rea-
Fig. 10  Gaussian curve for the evaluation of the settlement profile
sonable result. A too coarse meshing fails to capture the
subtle changes in the stresses generated in different parts of
centerline to the point of inflection on the surface settlement the medium, especially at the points where stress concentra-
trough. tions are expected. On the other hand, a too fine mesh ren-
According to Fig. 10, the maximum surface settlement ders the time of obtaining a solution to be very long. Hence,
(Smax) is the one of effective parameters describing the a trade-off is required to obtain a nearly accurate solution
nature of surface settlement. This parameter (Smax) has been in reasonable less time. In this study, the global meshing is
studied by different researchers. Mair [25] proposed Eq. (6) achieved with a medium mesh, and the central part of the
to calculate the value of Smax: model is then refined twice. Figure 12 depicts the extent of
the model and the type of meshing used in the present study
Smax = √
Vs in which the colors green, yellow and light blue represent
, (6) silt with sandy clay (material 2), clay with chalk (material
2𝜋 × i
3) and chalk (material 4), respectively.
where Vs = Rv × Vw is the volume loss with Rv = 0.65 and Vw The numerical analysis includes the use of 15 node trian-
has the significant shown in Fig. 10 that is the soil volume gular elements to model the soil. The retaining wall structure
affected by settlement. Using Eqs. (5) and (6), the curve Sx was simulated with one-dimensional linear beam element
may be defined by any two of the parameters Vs, Smax or i. that can resist axial load and bending moments. The stiffness
For the diaphragm of the Library of University of Enna for the wall element is represented by means of the flexural
Kore, the surface settlements Sx have been evaluated by the rigidity (EI) and the normal stiffness (EA).
measured value ymax and the volume Vw affected by set- Figure 12 shows the various soil clusters used to define
tlement. The values obtained confirm that a good safety regions with same properties within the finite element mesh.
threshold is guaranteed to the nearby buildings, as shown The soil model was run with a medium mesh of triangu-
in Fig. 11. lar elements, leading to 1729 elements and 13,983 nodes.
The reticule average dimension is strongly refined near the
structure, for a better evaluation of the gradient of stress
5 Finite Element Analysis and strain.

5.1 Numerical Model
5.2 Calculation Stages
The modelling of the behavior of the pile diaphragm wall
has been performed using the FEM code PLAXIS 2D. The Many phases have been used in the numerical analysis to
numerical analysis was carried out in plane strain condi- establish the initial effective stress state condition existing
tions. The geometry of the model, the material properties prior to the construction of the diaphragm wall, the excava-
and the boundary conditions were specified. To analyze the tion to remove filling material, the excavation for positioning
full-scale diaphragm wall, the model extends about 100 m the diaphragm wall, the excavation for anchor placement, the
horizontally and 30 m vertically to be sufficient to avoid stage of pre-stressing and locking of the anchors.

13

668 International Journal of Civil Engineering (2019) 17:659–672

Fig. 12  Finite element mesh

Calculations have been organized in 8 stages, following deformation behavior. Based on the stratigraphic columns
the technological phases: obtained by boreholes performed in the site, four different
soil types have been considered. In particular, from ground
• phase 0: initialization of the stress state; level to bottom filling material (1), silts with sand clay (2),
• phase 1: excavation step 1 (to level − 4.00 m); clays with chalk (3) and chalk (4). Table 2 reports the geo-
• phase 2: activation of diaphragm wall; technical properties adopted in the numerical modelling for
• phase 3: activation of concrete retaining wall; the four categories of soil considered. Strength and deform-
• phase 4: excavation step 2 (to level − 8.00 m); ability parameters have been obtained by laboratory testing
• phase 5: activation of anchor 1 at level − 4.00 m; performed on the soil and by a literature database [27–29]
• phase 6: excavation step 3 (to level − 11.00 m); of model parameters used in PLAXIS for similar soil type.
• phase 7: excavation step 4 (to level − 14.00 m); The predicted ground movements are primarily influ-
• phase 8: activation of surcharge load. enced by soil stratigraphy and soil model parameters used.
To investigate the suitability of the HS model parameters for
5.3 Soil Properties subsoil, the analysis of diaphragm behavior was performed
for comparative study of observed values of lateral ground
Choosing an appropriate constitutive model for the soil is movements and predicted values using adjusted parameters
the most important factor in the numerical analysis. The of HS model in comparison of the MC model prediction. HS
constitutive model should reflect the main behavior of the soil parameters are iteratively adjusted until the calculated
soil, and it also should be simple and easy to implement. results show reasonable agreement with the observed data.
As constitutive law for the soil, the linear elastic–plastic
behavior is generally assumed in numerical analysis, such
as Mohr–Coulomb (MC) model. On the other hand, using a 6 Predicted Performance
complex constitutive model, the behavior of the soil would
be reflected wonderfully and the displacements of ground The analysis of diaphragm wall behavior was performed
could be predicted better, but the parameters of the model for comparative study of observed values of lateral ground
are difficult to obtain. Since the MC model is simple to be movements and predicted values and for the comparison of
understood and the parameters can be obtained easily, it is surface settlement deducted by analytical solution and those
widely used. derived by FEM analysis.
In this study, the MC model and hardening soil (HS) The results of the numerical analysis are presented in
model are used in order to compare the prediction results of terms of anchored diaphragm wall and soil displacements
diaphragm and ground movements so that assessment could
be made on the better constitutive model for use. HS model
[26] is based on the plasticity theory, evaluating the stiffness Table 2  Geotechnical parameters for numerical analyses
as a function of the stress level according to an exponential Soil layer γ (kN/m3) c′ (kPa) 𝜙′ (°) E (kPa) Eed (kPa)
law.
In the first step, the soil behavior has been modeled 1 19.6 0 24 6.0 × 104 8.9 × 104
adopting a linear elastic perfectly plastic constitutive law, 2 19.6 21.6 24 2.9 × 105 3.8 × 105
using the geotechnical parameters issued from laboratory 3 18.6 50.0 22 1.8 × 106 2.4 × 106
and in situ tests. The MC criterion in PLAXIS requires 4 19.6 114.0 29 7.5 × 106 10.1 × 106
five parameters: E and ν for linear elastic behavior and c γ unit weight, c′ effective cohesion, 𝜙′ effective friction angle, E
and 𝜙 for yield function and ψ as dilation angle for plastic Young modulus, Eed oedometer modulus

13
International Journal of Civil Engineering (2019) 17:659–672 669

computation. The predicted ground movements are influ- the calculations, especially when complex constitutive laws
enced by soil stratigraphy and soil model parameters used. are used for the ground; difficulty in estimation of the initial
In particular, Fig. 13 shows the deformed mesh of the stress state in the ground, taking into account its lithology,
numerical simulation corresponding to the state of end con- the presence of neighboring structures, the execution of the
struction and Fig. 14 reports the soil displacements predicted retaining structure itself, etc.; complexity of the numerical
at the final stage. According to the calculations, the maxi- model itself, considering all implied parameters; three-
mum horizontal displacement of the soil is about 49 mm at dimensional behavior of the retaining structure.
the top of the diaphragm. In order to obtain reliable results using numerical model-
The values of bending moment along the diaphragm wall ling it is important to calibrate and validate the model based
as function of calculation phases and at the end of construc- on experimental measurements. Taking into account the
tion (phase 8) are reported in Figs. 15 and 16, respectively. geometry and dimensions of the diaphragm wall, the loading
From phase 3 to phase 8, the bending moment increase until conditions and the soil model, it was found that the predicted
a value approximately equal to 650 kNm/m. data were close to the observed results.
As depicted in Fig. 17, the observed data correspond-
ing to last on-site measurements (November 4th, 2014) are
7 Comparison of Experimental compared to the results of the numerical analyses in terms
and Numerical Results of soil horizontal displacements. In particular, the values
of displacement versus depth of diaphragm computed by
Retaining structures imply complex soil–structure interac- PLAXIS and those measured by inclinometers are repre-
tion phenomena. A correct estimation of their behavior is sented with red (MC model), green (HS model) and blue
possible only by using numerical models, allowing a com- points, respectively. The plot shows a good agreement
plex modelling of the system formed by the retaining struc- between measured and computed results, with a maximum
ture, foundation ground and neighboring buildings. Even horizontal displacement of 49 mm. A difference of 5% can
when such methods are used, the results can present signifi- be observed between the values calculated with HS and MC
cant differences from the real behavior. model and those measured.
The reasons for these differences are many, including Furthermore, the surface settlements derived by the
incertitude regarding the geotechnical parameters used for empirical solution (blue line) described in Sect.  4 are

Fig. 13  Deformed mesh

Fig. 14  Total displacements of
the diaphragm wall

13

670 International Journal of Civil Engineering (2019) 17:659–672

Bending Moment [ kNm/ m] to the excavation depth is for both cases about 0.5. Obvi-
-800 -400 0 400 800 ously the location where the maximum settlement occurred
4 might have a significant effect on the performance of adja-
Phase 3 cent buildings.
6 There are several parameters that are effective on amount
Phase 4
of maximum surface settlement and recognizing effects of
Phase 5
these parameters can be very helpful for accurate meas-
8
Phase 6
urement of maximum surface settlement derived by FEM
Phase 7
analysis.
Phase 8
10 According to the numerical and experimental investi-
gations, cohesion, angle of internal friction, diaphragm
depth, Poisson’s ratio, Young’s modulus, unit weight, face
Depth [ m]

12 support pressure are the most effective parameters. On the


other hand, the empirical solution adopted is based on the
14 assumption that the surface settlement can be approximate
to a Gaussian distribution.

16

18 8 Concluding Remarks

The increasing use of underground spaces has driven many


20
researchers to update the knowledge on the behavior of
embedded structures in soil. The availability of FE codes,
22 advanced constitutive models for soil and the spread of suit-
able laboratory and in situ testing techniques are allowing
Fig. 15  Bending moment along the diaphragm wall function of calcu- more reliable evaluation of the safety conditions and the
lation phases prediction of the behavior of these structures during both
the construction and the serviceability phases.
In this paper, the results of an experimental and numeri-
compared with those predicted by FEM analysis (red line) cal study carried on a retaining structure are presented and
in Fig. 18 as function of distance from diaphragm wall. discussed. Static displacements were carried out starting
The comparison shows a reasonable agreement in terms of from the excavation phase and until the construction of an
maximum value of surface settlement (about 3 cm) but a anchored diaphragm wall for the new Library of the Univer-
discrepancy in the trend is clear. Figure 18 also displays sity of Enna “Kore” (Italy). Two piles of the diaphragm have
the comparison in terms of ground surface settlements with been instrumented with conventional inclinometer cases and
those (black line) found in literature [5] for different stages. embedded piezoelectric accelerometers. A continuous incre-
The distance where the maximum ground surface settlement ment in the horizontal displacement of the wall has been
occurred is about 11 m and 5 m for literature results and for observed, until it has reached a final and stable value of
FEM analysis, respectively, whereas the ratio of this distance about 49 mm.

Fig. 16  Bending moment along


the diaphragm wall

13
International Journal of Civil Engineering (2019) 17:659–672 671

Displacement [mm] The use of MEMS provides as advantages the low cost in
0 10 20 30 40 50 relation to the possibility to monitor the behavior of the
0 structures more easily and for long periods. The calibra-
Computed (M C)
tion procedure, which takes into account static sensitivity,
2 adopted to use the sensors as inclinometers for static moni-
Computed (H S)
toring applications, is also reported. The low-cost MEMS
M easured
sensors have been calibrated and tested to be used for sev-
4
eral applications of civil engineering structures monitoring.
Details of the installed monitoring system are reported and
6 the preliminary results are discussed. It has been proved that
the proposed system is reliable and the rotations measured
8
correspond to the actual deformations of the structures.
Based on careful interpretation of geotechnical data
derived by soil investigation, a numerical modelling by the
10 Finite Element code PLAXIS 2D has been performed and
Depth [m]

calibrated. The comparison between computed results and


12 experimental measurements in terms of horizontal displace-
ment has been carried out, highlighting a very good agree-
ment between experimental evidences and computed results.
14
Bending moment distributions have been assessed according
to computed lateral displacements along the diaphragm wall.
16 Finally, an empirical method to evaluate the surface set-
tlement based on the results of measured horizontal dis-
18
placements has been used and the results have been com-
pared with those obtained by FEM analysis, guaranteeing a
good safety threshold to the nearby building.
20

Compliance with Ethical Standards 


22
Conflict of interest  On behalf of all authors, the corresponding author
states that there is no conflict of interest.
Fig. 17  Comparison between computed and measured pile lateral dis-
placements

References
Distance from diaphragm [m]
0 10 20 30 40 50 1. Garrett C, Barnes SJ (1984) Design and performance of the Dun-
0 ton Green retaining wall. Geotechnique 34(4):533–548
2. Raison CA (1985) Discussion on performance of proposed and
2 cantilevered rigid walls. Geotechnique 35(4):540–544
3. Ikuta Y, Maruoka M, Aoki M, Sato E (1994) Application of the
observational method to a deep basement excavated using the top–
Settlement [cm]

4 This study - FEM Analysis


This study - Empirical Solution down observation method. Geotechnique 44(4):655–664
Stage 1 - Ou et al. 2000 4. Young DK, Ho EWL (1994) Observational approach to design of
Stage 3
6
Stage 5 a sheet-piled retaining wall. Geotechnique 44(4):637–654
Stage 7 5. Ou CY, Liao JT, Cheng WL (2000) Building response and
Stage 9
8 Stage 11
ground movements induced by a deep excavation. Geotechnique
50(3):209–220
6. Gunn MJ, Clayton CRI (1992) Installation effects and their impor-
10
tance in the design of earth-retaining structures. Geotechnique
42(1):137–141
Fig. 18  Predicted surface settlement by FEM analysis versus those by 7. De Moor EK (1994) An analysis of bored pile/diaphragm wall
empirical solution installation effects. Geotechnique 44(2):341–347
8. Ng CWW, Yan RWM (1999) Three-dimensional modeling
of a diaphragm wall construction sequence. Geotechnique
49(6):825–834
A detailed description of a monitoring system with piezo- 9. Conti R, de Sanctis L, Viggiani GMB (2012) Numerical mod-
electric accelerometers instead of conventional inclinometer elling of installation effects for diaphragm walls in sand. Acta
for measuring horizontal displacements has been explained. Geotech 7(3):219–237

13

672 International Journal of Civil Engineering (2019) 17:659–672

10. Chen J, Lei H, Wang J (2014) Numerical analysis of the installa- 20. Brinkgreve RBJ, Vermeer PA (2006) PLAXIS. In: Brinkgreve
tion effect of diaphragm walls in saturated soft clay. Acta Geotech RBJ, Broere W, Waterman D (eds). The Netherlands ISBN-10
9:981–991 90-76016-02-X, ISBN-13 978-90-76016-02-34
11. Peck RB (1969) Advantages and limitations of the observational 21. Martos F (1958) Concerning an approximate equation of the sub-
method in applies soil mechanics. Geotechnique 19(2):171–187 sidence trough and its time factors. In: International strata control
12. Rainieri C, Fabbrocino G, Santucci de Magistris F (2013) An congress, Leipzig, pp 191–205
integrated seismic monitoring system for a full-scale embed- 22. Peck RB (1969) Deep excavations and tunneling in soft ground.
ded retaining wall. Geotech Test J 36(1):40–53. https​://doi. In: 7th International conference on soil mechanics and foundation
org/10.1520/GTJ20​12006​7 engineering, state of the art report. Proceedings of 7th interna-
13. Rainieri C, Dey A, Fabbrocino G, Santucci de Magistris F tional conference on soil mechanics and foundation engineering,
(2017) Interpretation of the experimentally measured dynamic Mexico City, pp 225–290
response of an embedded retaining wall by finite element mod- 23. Schmidt B (1969) Settlements and ground movements associ-
els. Measurement 104:316–325. https​://doi.org/10.1016/j.measu​ ated with tunneling in soils. Ph.D. Thesis, University of Illinois,
re-ment.2016.05.100 Urbana
14. Abdoun T, Danisch L, Ha D (2005) Advanced sensing for real- 24. Rowe RK, Lo KY, Kack GJ (1983) A method of estimating surface
time monitoring of geotechnical systems. In: Proceedings of geo- settlement above tunnels constructed in soft ground. Can Geotech
Frontiers congress, 2005, Austin, Texas, USA, American Society J 20(1):11–22. https​://doi.org/10.1139/t83-002
of Civil Engineers 25. Mair RJ, Williamson MG (2014) The influence of tunneling and
15. Taylor JR (1997) An Introduction to error analysis: the study deep excavation on piled foundations. In: Geotechnical aspects
of uncertainties in physical measurements. University Science of underground construction in soft ground—proceedings of 8th
Books, Sausalito. ISBN 0-935702-75-X international symposium on geotechnical aspects of underground
16. Lo Iacono F, Navarra G, Oliva M (2017) Structural monitoring construction in soft ground, TC204 ISSMGE—IS-SEOUL 2014.
of “Himera” viaduct by low-cost MEMS sensors: characteriza- pp 21–30
tion and preliminary results. Meccanica 52:3221. https​://doi. 26. Schanz T, Vermeer PA, Bonnier PG (2000) The hardening soil
org/10.1007/s1101​2-017-0691-4 model: formulation and verification. In: Beyond 2000 in compu-
17. Castelli F, Freni G, Lentini V, Fichera A (2017) Modelling of tational geotechnics—10 years of PLAXIS. Balkema, Rotterdam.
a debris flow event in the Enna area for hazard assessment. In: ISBN 90 5809 040 X
Proceedings of 1st international conference on the material point 27. Likitlersuang S, Surarak C, Wanatowski D, Oh E, Balasubrama-
method, MPM 2017, Delft, January 10–13, 2017, Procedia Eng niam A (2013) Finite element analysis of a deep excavation: a case
175:287–292. https​://doi.org/10.1016/j.proen​g.2017.01.026 study from the Bangkok MRT. Soils Found 53(5):756–773. https​
18. Castelli F, Castellano E, Contino F, Lentini V (2016) A web-based ://doi.org/10.1016/j.sandf​.2013.08.013
GIS system for landslide risk zonation: the case of Enna area 28. Yasushi A, Kusakabe O, Murata O, Konishi S (2008) A numeri-
(Italy). In: Proceedings of 12th international symposium on land- cal study on ground displacement and stress during and after the
slides, Napoli, Italy. Aversa et al. (eds) Landslides and engineered installation of deep circular diaphragm walls and soil excavation.
slopes. Experience, theory and practice. Associazione Geotecnica Comput Geotech 35:791–807
Italiana, Roma, Italy. ISBN 9781138029880 29. Brinkgreve RBJ (2005) Selection of soil models and parameters
19. Castelli F, Lentini V (2013) Landsliding events triggered by rain- for geotechnical engineering application. In: Soil constitutive
falls in the Enna area (South Italy). In: Proceedings of second models: evaluation, selection, and calibration. ASCE, geotechni-
world landslide forum, WLF2 2011, Rome (Italy), vol 2: Early cal special publication no. 128. Evaluation, selection, and calibra-
warning, instrumentation and monitoring. Margottini C et al (eds), tion, 2005. Geo-Frontiers congress 2005, Austin, Texas, USA.
Springer, Heidelberg, pp 39–47. ISBN 978-3-642-31444-5, ISBN https​://doi.org/10.1061/40771​(169)4
978-3-642-31445-2 (eBook). https​://doi.org/10.1007/978-3-642-
31445​-2

13

You might also like