bài viết về cọc plaxis 3d
bài viết về cọc plaxis 3d
bài viết về cọc plaxis 3d
ABSTRACT
Due to advancements in computation, most deep excavation projects are now designed based on the results of numerical
analysis. However, designers are usually careless about structural modelling. This study presents the effect of structural modelling
on wall deformations of deep excavation. A 3D finite element method was employed to model the problem, and a residential
building project located in Bangkok was selected for the study. Structural models, including diaphragm wall, diagonal braces, and
bored piles, were the focus of study, which indicates that the types of elements used to model the structure system significantly
affect the finite element analysis result.
Key words: Finite element modelling, deep excavation, diaphragm wall, diagonal bracing, bored pile.
1. INTRODUCTION
Remarks: 1 is major principle stress (kN/m2); 3 is minor principle stress (kN/m2); pref is reference pressure (100 kN/m2)
complex condition, and irregular shape of excavation. Chheng and observing the deformation near boundaries and no boundary effect
Likitlersuang (2018) provided evidence that 3D FE modelling is was observed.
effective for deep excavation analysis in Bangkok. In the previous
study, two deep excavation projects in Bangkok were used to val-
idate the performance of 3D FE modelling. However, there is a
question about the effect of the structural models on the results of
FEA; therefore, this study extends the previous study (Chheng and
Likitlersuang 2018), focusing on the structural modelling of 3D
FEA in a deep excavation.
the non-volume elements (Hsiung et al. 2016a). Input parameters The second consideration in this study focused on diagonal
for plate elements are summarised in Table 4. Nevertheless, for the bracing at the end of each strut. Diagonal braces are used to reduce
volume element modelling option in this study, the D-wall was the spans between the struts, and are usually installed at a 45o angle
modelled by tetrahedral volume elements. The volumes adopted to the struts. Two options were studied: modelling with and with-
linear elastic and non-porous parameters for drainage type as tab- out diagonal bracing, as illustrated in Figs. 6(a) and 6(b), respec-
ulated in Table 5. tively. In PLAXIS 3D modelling, the diagonal bracing option can
be activated or deactivated for different cases – those with and
Table 4 Simplified elastic structural element (plate element) without diagonal bracing, respectively. Struts and wailings were
parameters for diaphragm wall system (Chheng and modelled by beam elements, and their input parameters are tabu-
Likitlersuang 2018) lated in Table 6. Noted that to model the diagonal bracing system
more realistic, the beam element was selected instead of a simpli-
Parameters d (m) (kN/m3) E (kN/m2) 12 fied anchor element in this study.
D-wall and barrette pile 0.8 16 24,800 0.15 Lastly, the model for existing bored piles was considered.
Slabs 1.0 23 24,800 0.15 Bored piles are structures which are used to support superstructure
loads. Piles are usually installed before the excavation and they
Table 5 Volume element parameters for diaphragm wall system can reduce both horizontal movements and settlement behind the
(Chheng and Likitlersuang 2018) retaining wall (McNamara 2011). In this study, two optional mod-
els with and without modelling of bore piles were looked at in this
d E Drainage study as presented in Figs. 7(a) and 7(b), respectively. Bored piles
Parameters
(m) (kN/m3) (kN/m2) 12 type
were modelled by embedded beam elements, which require the in-
D-wall and barrette pile Non-
0.8 23 24,800 0.15 put parameters found in Table 7. The embedded beam consists of
(Linear elastic) porous
beam elements with special interface elements providing the
(a) Model with diagonal braces (b) Model without diagonal braces
Fig. 6 Models of construction work
Table 6 Beam element parameters for bracing system (Chheng and Likitlersuang 2018)
A γ E I3 I2
Parameters
(m2) (kN/m3) (kN/m2) (m4) (m4)
Steel strut (400 400 172 kg/m) 2.19 × 102 77.15 210 000 2.24 × 104 6.66 × 104
Steel wailings (400 400 172 kg/m) 2.19 × 10-2 77.15 210 000 6.66 × 10-4 2.24 × 10-4
Skin End
E γ Diameter
Type resistance resistance
(kN/m2) (kN/m3) (m)
(kN/m) (kN)
Massive
24,800 23 1.2 260 2,443
circular pile
Massive
24,800 23 0.8 173 1,086
circular pile
(a) Model with bored piles (b) Model without bored piles Remarks: The average skin friction (fs) = 69 kN/m2 and the end bearing
Fig. 7 Models of construction work (qb) = 2,160 kN/m2
126 Journal of GeoEngineering, Vol. 14, No. 3, September 2019
interaction between pile and the surrounding soil. The interaction 4.2 Diagonal Bracing Modelling
may involve a skin resistance (fs) and an end resistance (qend).
Noted that the skin friction and end resistance were approximated Figure 9 presents the numerical results of the wall defor-
from fs = su,aveD and qend = 9su,end D2/4, where su,ave and su,end are mation for each construction stage for both optional models of di-
the average undrained shear strength of soil through the plie length agonal bracing (i.e., with and without modelling the diagonal
and the undrained shear strength of soil at pile tip location, respec- braces), in which the symbol numbers in circle are referred to the
tively. construction stages as described in Table 3 and Fig. 3(b). The sim-
ulations are also compared with the data measured from inclinom-
eters at IN1 and IN2. The results indicate that the differences are
4. ANALYSIS RESULTS very insignificant. In this study, it was found that the diagonal
braces can be ignored in order to reduce the effort required and the
4.1 D-wall Modelling calculation time. However, in a complicated excavation site, such
as one with an irregular shape and heavier diagonal bracing mod-
This section aims to show the results of wall deformation elling may be required.
from the different methods used for modelling the D-walls. The
simulations were performed following the construction sequence 4.3 Bored Pile Modelling
as described in Table 3 and Fig. 3(b). For the plate elements and
volume elements modelling of D-walls, the wall movements are The bored pile modelling results are extracted and plotted
plotted and compared with monitoring data as shown in Fig. 8, in in Fig. 10 for both options (i.e., with and without piles
which the symbol numbers in circle are referred to the construction modelling), in which the symbol numbers in circle are referred
stages as described in Table 3 and Fig. 3(b). The stages 2 and 3 to the construction stages as described in Table 3 and Fig. 3(b).
data, measured by inclinometers (IN1 and IN2) are included in Fig. The simulations are also compared with the data measured from
8, while the locations of the inclinometers are referred to in Fig. inclinometers at IN1 and IN2 as shown in Figs. 10(a) and 10(b),
3(a). The results reveal that the estimated D-wall deformation respectively. Noted that the IN1 is closed to the group of bored
modelled by plate elements tends to be a little larger than that mod- piles, whereas the IN2 is located far from the bored piles. In
elled by volume elements in all stages of construction. Addition- general, the model including bored piles was found to reduce
ally, the modelling D-wall by plate elements is more convenient the horizontal movements of wall, which also agree with
for the designer. Since bending moment and shear force can be McNamara (2001). In fact, the wall deformation was influenced
obtained directly from the plate elements, whereas the volume el- by the presence of the bored piles only near the tip of the D-wall
ements present the internal stresses. To calculate the bending mo- (Fig. 10(a)). In the models, most of the pile heads were at levels
ment and shear force from the internal stresses requires more ef- lower than the excavation depth. However, the presence of piles
forts on post-processing calculation. A separated numerical code under the excavation site could affect the heaving stability as
may be required for the calculation. Therefore, the selection of pointed out by McNamara (2001). This may need further study
which type of model to use for predicting D-wall deformation is a in the future.
decision for the engineers on a case-by-case basis.
Wall movement, h (mm) Wall movement, h (mm) Wall movement, h (mm) Wall movement, h (mm)
0 5 10 15 20 25 0 5 10 15 20 0 5 10 15 20 25 0 5 10 15 20 25
0 0 0 0
2 2 2 2
1
4 4 1 4 4
6 6 6 6
1 1
8 8 8 8
2
2 2
Depth (m)
2
Depth (m)
Depth (m)
Depth (m)
3
10 10 10 10
3 3
3
12 12 12 12
14 14 14 14
16 16 16 16
18 18 18 18
20 20 20 20
Case 1: 3D FEA - Plate element Case 1: 3D FEA - Plate Element
Case 1: 3D FEA - With diagonal bracing Case 1: 3D FEA - With diagonal bracing
Case 2: 3D FEA - Volume element Case 2: 3D FEA - Volume element
Case 3: 3D FEA - Without diagonal bracing Case 3: 3D FEA - Without diagonal bracing
IN1 - Stage 2 IN2 - Stage 2
IN1 - Stage 2 IN2 - Stage 2
IN1 - Stage 3 IN2 - Stage 3
IN1 - Stage 3 IN2 - Stage 3
Depth (m)
2 tion times for the cases with and without diagonal bracing are
10 10
3 not much different (see Table 8), the modelling of diagonal
3
12 12 braces may be required for more complex sites, such as one
14
with an irregular shape and heavier diagonal bracing. This
14
should be elaborated in the future.
16 16
(3) In general, the small differences of wall movements were ob-
18 18 served when including bored piles in the model. However, the
bored piles can be significantly influenced in the modelling
20 20
especially at the deeper excavation zone, where the D-walls
Case 1: 3D FEA - With piles Case 1: 3D FEA - With piles
Case 4: 3D FEA - Without piles Case 4: 3D FEA - Without piles
are near the pile head. Moreover, when bored piles are acti-
IN1 - Stage 2 IN2 - Stage 2 vated in the model, the calculation time can be significantly
IN2 - Stage 3
IN1 - Stage 3
increased compared to cases in which bored piles are deac-
(a) IN1 (b) IN2 tivated. Based on this study, the simulations took 70 minutes
for the model with bored poles, while the simulations took
Fig. 10 Comparison between models with and without bored piles only 54 minutes for the model without bored pile (16 minutes
difference) as shown in Table 8. It is recommended that the
study of bored piles modelling should be elaborated in the fu-
5. DISCUSSIONS AND CONCLUSIONS ture such as modelling of heaving problem and the effect of
pile head location.
3D Finite Element Analysis (3D FEA) has become an im-
(4) It is aware from Figs. 8 to 10 that it has significant differences
portant numerical tool in the geotechnical engineering field due to
between field measurements and numerical results for vari-
the power of today’s computer resources. A great deal of calcula-
ous stages of different sections, no matter which model used
tion effort is required in 3D analysis; hence, an optimised model
for the diaphragm wall, with or without modelling diagonal
is needed in order to reduce the calculation time. Table 8 summar-
bracing and simulating the piling works inside the excavation
ies the calculation times influenced by the structural modelling op-
or not. Thus, these factors are recognised not to be main rea-
tions. In this study, which looked at the underground construction
sons for the difference. Therefore, it is suggested that explo-
of a residential building, a project in downtown Bangkok was se-
rations on impacts from small strain characteristics of soils
lected as the focus of a case study to investigate the effects of struc-
(Likitlersuang et al. 2013c) and axial stiffness of the strut
tural modelling, including D-wall modelling, diagonal struts and
shall be carried out in the future. Moreover, the difference of
existing bored piles. Some highlights from this study can be con-
locations of selected cross section for analyses and monitor-
cluded and discussed as follow:
ing is also a possible reason leading to the difference.
(1) D-walls can be modelled by plate elements or volume ele-
ments. It was found that the wall deflection is larger when the
wall is modelled by plate elements. This difference of wall ACKNOWLEDGEMENTS
movements between modelling D-wall by plate elements and
volume elements agrees well with previous studies such as This research was supported by the Thailand Research Fund
Zdravkovic et al. (2005) and Dong et al. (2014). However, Grant No. DBG-6180004 and the Ratchadapisek Sompoch En-
the use of volume elements requires more effort for model- dowment Fund (2019), Chulalongkorn University (762003-CC).
ling the D-wall than plate elements. Such effort includes the The authors would like to thank Lt. Col. Dr. Chanaton Surarak and
modelling time, complexity, calculation time, and results ex- Dr. Boonlert Siribumrungwong for providing useful information
traction (Chheng and Likitlersuang 2018). In this study, the throughout the research. The second author wishes to thank the
calculations took 124 minutes for the volume element model, AUN/SEED-Net (JICA) for a scholarship during his study.
while the calculations took 70 minutes for the plate element
Case D-wall model Diagonal bracing Bored piles Calculation time (minutes)
1 Plate element Yes Yes 70
2 Volume element Yes Yes 124
3 Plate element No Yes 65
4 Plate element Yes No 54
Remarks: 3D FE calculations were conducted using Intel® Core™ i7-6500U CPU@2.50GHz
128 Journal of GeoEngineering, Vol. 14, No. 3, September 2019