Professional Documents
Culture Documents
Transportation Geotechnics
journal homepage: www.elsevier.com/locate/trgeo
A R T I C L E I N F O A B S T R A C T
Keywords: This paper introduces the plastic stability solutions of braced circular excavations in anisotropic and non
Basal stability –homogeneous clays. Using the framework of Finite Element Limit Analysis (FELA) under axisymmetric condi
Circular excavation tions, the upper bound (UB) and lower bound (LB) solutions of the stability of excavations can be obtained. The
Anisotropy
clay is set to be anisotropic, where the Anisotropic Undrained Shear (AUS) model is used as a failure criterion of
MARS
FELA
the surrounding soil. The results of this study are the proposed stability number which is the normalized
parameter of the maximum unit weight and the anisotropic undrained shear strength of clay. Four dimensionless
parameters are considered in the study: the anisotropic strength ratio, the depth–radius ratio, the
depth–embedment ratio, and the strength gradient ratio. The impact of all considered dimensionless parameters
on the results of the FELA solutions is examined. A machine learning regression approach, Multivariate Adaptive
Regression Splines (MARS), is employed to develop an empirical design equation to predict the stability number
of braced circular excavations in anisotropic and non–homogeneous clays. The proposed MARS equation can be a
useful and reliable equation to estimate the basal stability of this excavation problem in practice.
Introduction heave failure. Numerical solutions of the basal heave stability of exca
vations under plane strain conditions were also proposed by Goh
To launch a tunnel boring machine, a shaft excavation with a circular [20,21], Faheem et al. [22], and Li et al. [23], where the dis
shape has been required to install a boring machine into the shaft before placement–based Finite Element Method (FEM) was employed in their
operating the machine. Currently, space is limited in urban areas for the studies. Recently, Yodsomjai et al. [24] presented the use of Upper
construction of shafts. As a result, the demand for circular shafts in Bound (UB) and Lower Bound (LB) Finite Element Limit Analysis (FELA)
tunneling construction has increased significantly compared to the use to derive the plastic solutions of the basal stability of braced plane strain
of rectangular shafts. In addition to the requested small area for con excavations. However, only a few studies on the basal heave stability of
struction, circular shafts require less steel reinforcement than rectan braced circular excavations have been carried out in the past by Cai et al.
gular shafts due to the impact of circumferential hoop forces on the [25] and Goh [26] using the displacement–based FEM and Keawsa
cylindrical shell of circular shafts. wasvong and Ukritchon [27] using the LB FELA.
Previous studies on excavation problems primarily focused on lateral It has been proven that the partial strength anisotropy of natural
earth pressures or wall movements (e.g., [1-17]). These previous studies clays results from depositional and sedimentation processes leading to a
are limited to the serviceability aspect of excavation works. Early studies specific particle orientation [28]. Thus, the shear strength anisotropy of
on the basal heave stability of braced plane strain excavations were clays is directional and can be evaluated based on the major principal
carried out by Terzaghi [18] and Bjerrum and Eide [19] using the Limit stress rotation relative to the depositional direction [29,30]. By
Equilibrium Method (LEM) with the assumption of the pattern of basal considering the anisotropic undrained strength of clays under different
* Corresponding author.
E-mail address: ksurapar@engr.tu.ac.th (S. Keawsawasvong).
https://doi.org/10.1016/j.trgeo.2023.100945
Received 15 November 2022; Received in revised form 19 January 2023; Accepted 20 January 2023
Available online 25 January 2023
2214-3912/© 2023 Elsevier Ltd. All rights reserved.
V.Q. Lai et al. Transportation Geotechnics 39 (2023) 100945
shearing modes, Ladd [28] found that the anisotropic strengths consist Keawsawasvong and Ukritchon [27], only fully-fixed supports were
of three undrained shear strengths, which can be acquired via triaxial applied to the vertical plane of the excavated side of the excavation. This
compression (Suc), triaxial extension (Sue), and direct simple shear (Sus) leads to unrealistic simulation in the field situation since the plate ele
tests. In 2019, Krabbenhøft et al. [31] proposed the Anisotropic Un ments were not used in the in-house LB FELA codes by Keawsawasvong
drained Shear (AUS) yield criterion to capture the failure characteristics and Ukritchon [27]. Moreover, previous studies of a similar problem
of anisotropic clays. The AUS has been successfully implemented in the [25-27] did not consider the impact of the anisotropy of clays. The
FELA and employed to solve several geotechnical problems such as the problem definition of a braced circular excavation is shown in Fig. 1(a).
bearing and pullout capacity of foundations [e.g., [32–38]], stability of The geometry parameters include the width of R, the depth of H, and the
contiguous pile walls [e.g., [39]], tunnel stability [e.g., [40–41]], slope wall embedment of D. The wall is assumed to be very rigid and is fully
stability [e.g., [42]], and stability of unsupported excavations [e.g., supported by lateral bracing over the excavated plane. The interface
[43–45]]. However, there is no published solution for the problem of the between the wall and the clay is set to be rough.
basal heave stability of braced circular excavations in anisotropic and As mentioned earlier, the input undrained shear strengths of the AUS
nonhomogeneous clays complying with the AUS failure criterion. failure criteria are three anisotropic undrained shear strength compo
In this paper, the LB and UB FELA software, OptumG2 [45], in nents, including Suc, Sue, and Sus. In addition, two anisotropic strength
conjunction with the AUS model, is carried out to solve the plastic so ratios are introduced: re = Sue/Suc and rs = Sus/Suc. Based on the finding
lutions of the undrained basal stability of circular excavations with fully by [31], the relationship of re and rs can be computed from rs = 2 re/(1 +
braced supports (no horizontal movement and rotation of wall). Even re). As a result, this study uses only re as the dimensionless anisotropic
though the present study and that of Keawsawasvong and Ukritchon strength parameter. The undrained shear strength profile of a non
[27] employed the FELA to study the undrained basal stability of cir –homogeneous clay is a linear function of Suc(z) = Suc0 + ρz, where Suc0
cular excavations, the main difference between both studies is that denotes a surface undrained shear strength and ρ represents a strength
Keawsawasvong and Ukritchon [27] considered the vertical distance gradient.
between the excavation bottom and rigid base without an embedment of Numerical analyses are performed following the plastic bound the
a wall. However, this study does not consider the rigid base but rather orems to achieve the most accurate stability solution for the problem of
the wall embedment depth. In addition, Keawsawasvong and Ukritchon fully braced circular excavations in anisotropic and non–homogenous
[27] did not model the wall by plate elements. To simulate the wall in clays. The completely plastic material and accompanying flow rule is a
Fig. 1. (a) Model of fully braced circular excavations; (b) Final adaptive meshes; (c) Incremental displacements and movement of soils; (d) AUS failure criterion.
2
V.Q. Lai et al. Transportation Geotechnics 39 (2023) 100945
fundamental concept of this limit analysis. According to this paradigm, AUS model, where the dimensionless strength parameter of re = Sue/Suc
the precise solutions to the stability problem may be determined by is considered. Fig. 1(d) demonstrates that a change in re can affect the
upper bound (UB) and lower bound (LB) computations [46]. Due to the development of the AUS failure criterion’s failure surface. Note that the
limit analysis, the traditional hand calculation approaches are compli valuables TC and TE in Fig. 1(d) represent the triaxial compression and
cated to complete for broad stability challenges such as heterogeneous triaxial extension, equal to 2Suc and 2Sue, respectively. The form of the
soil, unexpected shapes, and complex pressure. For these reasons, Finite yield function of the AUS model with the harmonic mean of three un
Element Limit Analysis (FELA) techniques are developed by combining drained shear strengths can be expressed by Eq. (1):
plastic bound theorems, the finite element concept, and mathematical
Fu = σ1 − σ3 + (re − 1)(σ 2 − σ3 ) − 2Suc = 0 (1)
optimization. This approach has rapidly gained recognition and has
been proposed by many scholars. However, none of those existing where σ1 ≥ σ2 ≥ σ3 are the principal stresses (positive in compression),
studies address the implications of the heterogeneity and anisotropy of and Fu is the yield function. Basically, re ranges from 0.5 to 1, where the
fully circular braced excavation. case of re = 1 (i.e., Suc = Sue = Sus) can be used to represent the case of
In this study, the FELA technique is employed to solve the UB and LB isotropic undrained shear strength similar to the Tresca failure criterion.
solutions of undrained basal stability of braced circular excavations in The upper and lower bound solutions were produced using an
anisotropic and non–homogeneous clays. The mesh adaptivity tech automatically adaptable mesh refinement. In addition, the upper and
nique [47] is also used to improve the rigor of LB and UB solutions. Note lower bound solutions predicted by the efficient method of automated
that this mesh adaptivity technique has been successfully used by many mesh adaptivity with shear dissipation optimization are more accurate.
works in the past [e.g., [48–54]]. After obtaining the results, a machine This feature allows the modification of mesh density in sensitive areas
learning technique, multivariate adaptive regression spline (MARS), is with high plastic shearing strain. This study employs five adaptive
adopted to provide the correlation between the investigated and output meshing processes with a minimum mesh size of 5,000 elements and a
variables. Additionally, a sensitivity analysis is conducted to determine maximum mesh size of 10,000 elements. The failure behavior of fully
the relative impact of different input variables on the output. The MARS braced circular excavation in anisotropic and non–homogeneous clays
results can assist practicing engineers in predicting the undrained basal can be reported using the final adaptive mesh, as shown in Fig. 1(b). It is
stability of braced circular excavations in anisotropic and non essential to remember that adding more elements to the delicate areas
–homogeneous clays. might increase the accuracy of the solutions. The meshing processes
represent the evolution of the initial meshes toward the desired value.
Method of analysis Since there are sufficient numbers of elements and processes, the solu
tions are unaffected.
In the FELA program OptumG2, numerical analyses of circular Throughout this study, the influence of the following examined pa
braced excavations in anisotropic and non–homogenous clay are per rameters is related to the numerical analysis of braced excavation
formed, and an effective simulation model is established to monitor the anisotropic and non–homogenous clay under axisymmetric conditions,
plastic behavior of the clay. The computed circular excavation model which includes γ, Suc0, re, ρ, R, H, and D. The definitions of these pa
with specific dimensions is shown in Fig. 1(a), where the line of axial rameters are noted previously. In this study, the output from FELA is the
symmetry is applied at the left boundary, which acts as the problem maximum unit weight (γ) that causes a circular excavation to be a basal
centerline. The circular excavation has a diameter of R, a depth of H, and heave failure. The dimensionless technique by Butterfield [54] is applied
a wall embedment of D. The wall is considered to be a fully rigid material to reduce the examined dimensional parameters to dimensionless pa
with no failure, and the unit weights of supports and walls are neglected. rameters and to derive the stability solutions in terms of the stability
Based on Terzaghi [18] and Bjerrum and Eide [19], the assumption of number (N = γH/Suc0). Therefore, four dimensionless parameters are
the basal heave stability problem requires fully braced support at the utilized to investigate the influences of these variables on the computed
wall. As a result, the circular excavation in this study is set to have a fully stability number of circular braced excavation in anisotropic and non
braced support so that horizontal movement and rotation of the wall are –homogenous clay, which can be set as a function as follows:
not allowed. To ensure the fully braced support condition, the feature of
the plate boundary condition is activated at the top of the wall, as shown R D
N = ( , , re ,
ρH
) (2)
in Fig. 1(b), where only vertical movements are permitted. As demon H H Suc0
strated in Fig. 1(c), there are only vertical moments of soil masses
where R/H is the radius–depth ratio;
without the horizontal and rotation movements of the wall, so the basal
heave failure based on the assumptions by Terzaghi [18] and Bjerrum
D/H is the depth ratio of the embedded wall;
and Eide [19] can be simulated. Furthermore, the boundaries of all the
ρH/Suc0 is the strength gradient factor;
models are identical and considered roller supports at the left and right
re is the anisotropic strength ratio.
boundaries (with no horizontal movements), whereas that at the base
boundary acts as the fixed support (with no vertical and horizontal
To use Eq. (2), the input dimensional parameters (e.g., R, H, Suc0,
movements). The top surface is thought to be a limitless, unfettered
Sue0, and ρ) are first given. Then, these dimensional parameters are
surface. The domain sizes of the models were carefully determined to
normalized to dimensionless parameters, including R/H, D/H, re = Sue0/
prevent any overlapping of the developed plastic yield zone among the
Suc0, and ρH/Suc0. After obtaining these dimensionless variables on the
excavation boundaries. As shown in Fig. 1(b), the size of the horizontal
right-hand side of Eq. (2), the dimensionless output variable or the
domain is set to 7R, and the size of the bottom domain is D + 2R.
stability number (N = γH/Suc0) can be determined. Then, the maximum
It should be noted that, in all numerical models of this study, all parts
unit weight (γ) can be computed. If this maximum unit weight (γ) is
of the wall (the part along the excavated depth and the embedded part)
greater than the true value (γtrue), the excavation system is safe. In
are assumed to be rigid and have fully horizontal bracing. However, in
contrast, if the maximum unit weight is lower than the true value, basal
practice, the failure of the embedded part is a significant threat when the
heave collapse occurs due to the instability of the excavation system.
diameter of the excavation and the embedded depth of the wall are
Note that the design equations for obtaining the stability number N will
large. Hence, the models and the solutions of this study are limited to the
be developed later in the following sections. A thorough parametric
cases where the walls of circular excavations are designed to be rigid and
study of this issue is conducted utilizing the existing UB and LB FELA to
robust enough so that there is no collapse at the embedded part of the
determine the dependence relationship of the four dimensionless factors
wall.
(R/H, D/H, re, and ρH/Suc0) on the stability number as described earlier.
The plastic behavior of the surrounding clay is described using the
3
V.Q. Lai et al. Transportation Geotechnics 39 (2023) 100945
Table 1
Dimensionless analysis.
Cases R (m) H (m) D (m) Suc0 (kPa) Sue0 (kPa) ρ (kPa/m) R/H D/H re ρH/Suc0 γ (kN/m3) γH/Suc0
4
V.Q. Lai et al. Transportation Geotechnics 39 (2023) 100945
Fig. 3. Effect of D/H on N, where re = 0.5; R/H = 0.5, 1, 2, 4, and 8; ρH/Suc0 = 0, 1, 2, 3, and 4.
is observed. As the radius of the circular excavation increases, the failure observed only in the cases of D/H = 0. The effects of both re and ρH/Suc0
zone expands wider across the embedded wall to the ground surface. The are not demonstrated in the paper because no significant incremental
variation in the radius–depth ratio directly affects the incremental displacement information is observed with the variation in re and ρH/
displacement for every single value, as seen in Fig. 13(a–e), in contrast Suc0.
to the variation in the wall embedment ratio, where the effect can be
5
V.Q. Lai et al. Transportation Geotechnics 39 (2023) 100945
Fig. 4. Effect of D/H on N, where re = 0.6; R/H = 0.5, 1, 2, 4, and 8; ρH/Suc0 = 0, 1, 2, 3, and 4.
Proposed empirical equation from MARS (i.e., [55-59]). Furthermore, previous studies have employed MARS and
other machine learning techniques in braced excavation works (i.e., [60-
For more convenience in predicting the investigated stability num 63]). In short, the MARS algorithm consists of two main steps, as shown
ber, MARS is adopted to formulate the empirical model for generating in Fig. 14. In the first step, MARS partitioned the training data and
the complex relationship between input and output variables. It has generated the multi–linear regression model using splines. The multi–
been applied successfully in much geotechnical research for this purpose linear regressions are mathematically presented in terms of basic
6
V.Q. Lai et al. Transportation Geotechnics 39 (2023) 100945
Fig. 5. Effect of D/H on N, where re = 0.7; R/H = 0.5, 1, 2, 4, and 8; ρH/Suc0 = 0, 1, 2, 3, and 4.
functions. In the second step, MARS eliminated the insignificant term D/H, re, ρH/Suc0, and the output variable of N are considered input and
based on the CGV criterion to build a sufficient model. A more detailed target values. The optimal MARS model is first selected by varying the
MARS model can be found in previous studies [35,55,63-66]. initial number of splines. The performance of the examined MARS model
In this study, the numerical results from the plastic stability solutions is investigated through classical regression variables, including the Root
are used for the MARS model, where the investigated variables of R/H, Mean Square Error (RMSE), Mean Absolute Error (MAE), and Coefficient
7
V.Q. Lai et al. Transportation Geotechnics 39 (2023) 100945
Fig. 6. Effect of D/H on N, where re = 0.8; R/H = 0.5, 1, 2, 4, and 8; ρH/Suc0 = 0, 1, 2, 3, and 4.
of Determination (R2). values vary insignificantly when the number of BFs exceeds 30. Thus, in
Fig. 15 illustrates the performance of the investigated MARS models this study, the MARS model with 30 BFs is selected as the optimal MARS
when the number of basic functions is changed. Consequently, as the model to formulate the empirical equation and conduct the sensitivity
number of BFs changes from 10 to 30, the value of R2 rapidly increases, analysis.
and the values of RMSE and MAE explicitly decrease. In contrast, these The generated empirical equation from the optimal MARS model is
8
V.Q. Lai et al. Transportation Geotechnics 39 (2023) 100945
Fig. 7. Effect of D/H on N, where re = 0.9; R/H = 0.5, 1, 2, 4, and 8; ρH/Suc0 = 0, 1, 2, 3, and 4.
expressed by combining all basic functions as shown in Eq. (3). The the predicted and verified values, with an R2 of 99.95%. This indicates
details of the basic function are presented in Table 2. To validate this that the proposed empirical equation can be successfully applied in
empirical equation, a comparison is performed between the predicted practical design.
stability numbers from the empirical equation and those from the nu N = –1.63854 + 6.77849 × BF1 + 10.3635 × BF2 – 1.44788⋅BF3 +
merical results. As depicted in Fig. 16, there is good agreement between 5.69174 × BF4 + 20.5376 × BF5 – 2.80889 × BF6 + 25.9618 × BF7 +
9
V.Q. Lai et al. Transportation Geotechnics 39 (2023) 100945
5.38857 × BF8 – 8.33981 × BF9 + 0.0659546 × BF10 + 4.58593 × BF27 + 5.15467 × BF29 – 19.6357⋅BF30 (3).
BF11 + 17.6117 × BF12 – 2.12705 × BF13 + 2.42215 × BF14 – The sensitivity analysis results are presented in terms of the critical
8.01509 × BF16 + 12.4826 × BF17 + 8.96883 × BF18 – 2.61055 × relative index (RI, %), as shown in Fig. 17. The critical relative index is
BF19 + 4.15874 × BF20 – 8.01231 × BF21 + 1.17749 × BF22 + 100 percent, indicating that the corresponding input parameter is the
3.78956 × BF24 – 2.06993 × BF25 + 7.66372 × BF26 + 7.61055 × most crucial. As illustrated in Fig. 17, the strength gradient factor (ρH/
10
V.Q. Lai et al. Transportation Geotechnics 39 (2023) 100945
Fig. 9. Effect of B/H on N, where re = 0.7, ρH/Suco = 0 and 1, D/H = 0, 0.25, 0.50, 0.75, and 1.
Fig. 10. Effect of re on N, where R/H = 2, ρH/Suc0 = 1 and 4, D/H = 0, 0.25, 0.50, 0.75, and 1.
Fig. 11. Effect of ρH/Suc0 on N, where R/H = 2, re = 0.7 and 1, D/H = 0, 0.25, 0.50, 0.75, and 1.
11
V.Q. Lai et al. Transportation Geotechnics 39 (2023) 100945
Fig. 12. Effect of D/H on incremental displacements, where, re = 0.5, ρH/Suc0 = 1, R/H = 2.
Suc0) is the most significant parameter, whereas the size parameters (D/ with RIs being more significant than 30%.
H, R/H) of the excavation are less critical, with RI values of 75.12% and Notably, the proposed empirical equation and sensitivity analysis
64.23%, respectively. Finally, the anisotropic strength ratio (re) is the results from the MARS model are derived using the investigated range of
least important parameter, with RI values of 35.38%. It is noted, how input parameters. Consequently, the results from ANN may be less ac
ever, that these parameters should be considered in practical design, curate if the input value of the parameters is outside the investigated
12
V.Q. Lai et al. Transportation Geotechnics 39 (2023) 100945
In this section, the MARS solutions from the present study are vali
dated with the previous studies by Goh [26] and Keawsawasvong and
Ukritchon [27]. Goh [26] proposed a design equation based on the
displacement-based FEM, while Keawsawasvong and Ukritchon [27]
presented a design equation based on the LB FELA solutions of the basal
stability of supported circular excavations in isotropic and homogeneous
clays (re = 1 and ρ = 0). Note that the value of D is set to zero in this
comparison. The empirical equation by Goh [26] is presented below.
[ ( )2 ( ) ]
H H
N = − 0.4187 + 2.769 + 6.5443 (4)
2R 2R
Conclusions
Funding
R/H =
Fig. 13. Effect of R/H on incremental displacements, where, D/H = 1, re = 0.5,
This research was funded by the National Science, Research and
ρH/Suc0 = 1. Innovation Fund (NSRF), and King Mongkut’s University of Technology
North Bangkok with Contract no. KMUTNB–FF–66–12. This work was
supported by the Thailand Science Research and Innovation Funda
range. Another limitation of the MARS results is the length of the pro
mental Fund fiscal year 2023.
posed empirical equation (Eq.3). As presented in Fig. 15, the proposed
equation can be shortened by 10 or 20 BFs with R2 values of 98.6% and
CRediT authorship contribution statement
99.83%, respectively. However, the purpose of this paper is to propose
the best empirical equation based on the optimal MARS model, such that
Van Qui Lai: Methodology, Software, Investigation, Conceptuali
the proposed empirical equation has 30 BFs. Although it can be difficult
zation, Writing – original draft, Data curation. Khamnoy Kounlavong:
to calculate by hand, it can be programmed as the Marco function in
Methodology, Software, Investigation, Conceptualization. Suraparb
Microsoft Excel, such as the MAX or MIN function. Thus, the proposed
Keawsawasvong: Methodology, Software, Investigation, Conceptuali
empirical equation can be an easily applicable engineering tool for
zation, Writing – original draft, Data curation. Rungkhun Banyong:
predicting the stability number.
Software, Investigation, Writing – original draft. Warit Wipulanusat:
13
V.Q. Lai et al. Transportation Geotechnics 39 (2023) 100945
N
Fig. 14. Two main step of MARS algorithm.
Table 2
Basic function of the optimal MARS model.
BF Equation BF Equation
14
V.Q. Lai et al. Transportation Geotechnics 39 (2023) 100945
References
[1] Li MG, Zhang ZJ, Chen JJ, Wang JH, Xu AJ. Zoned and staged construction of an
underground complex in Shanghai soft clay. Tunn Undergr Space Technol 2017;67:
187–200.
[2] Zhang Z, Huang M, Wang W. Evaluation of deformation response for adjacent
tunnels due to soil unloading in excavation engineering. Tunn Undergr Space
Technol 2013;38:244–53.
[3] Tan Y, Wang D. Structural behaviours of large underground earth–retaining
systems in Shanghia. I: Unpropped circular diaphragm wall. J Perform Constr Facil
2015;29(2):04014059.
[4] Mu L, Huang M. Small strain based method for predicting three–dimensional soil
displacements induced by braced excavation. Tunn Undergr Space Technol 2016;
52:12–22.
[5] Zhang W, Zhang Y, Goh AT. Multivariate adaptive regression splines for inverse
analysis of soil and wall properties in braced excavation. Tunn Undergr Space
Technol 2017;64:24–33.
[6] Arai Y, Kusakabe O, Murata O, Konishi S. A numerical study on ground
displacement and stress during and after the installation of deep circular
diaphragm walls and soil excavation. Comp Geotech 2007;35(5):791–807.
[7] Kim KY, Lee DS, Cho J, Jeong SS, Lee S. The effect of arching pressure on a vertical
circular shaft. Tunn Undergr Space Technol 2013;37:10–21.
[8] Cho J, Lim H, Jeong S, Kim K. Analysis of lateral earth pressure on a vertical
circular shaft considering the 3D arching effect. Tunn Undergr Space Technol 2015;
48:11–9.
[9] Borges JL, Guerra GT. Cylindrical excavations in clayey soils retained by jet grout
walls: Numerical analysis and parametric study considering the influence of
consolidation. Comp Geotech 2014;55:42–56.
[10] Peng FL, Wang HL, Tan Y, Xu ZL, Li YL. Field measurements and finite–element
Fig. 16. The comparison between the MARS equation and FELA results. method simulation of a tunnel shaft constructed by pneumatic caisson method in
Shanghai soft ground. J Geotech Geoenviron 2010;137(5):516–24.
[11] Walton G, Delaloye D, Diederichs MS. Development of an elliptical fitting
algorithm to improve change detection capabilities with applications for
deformation monitoring in circular tunnels and shafts. Tunn Undergr Space
Technol 2014;43:336–49.
[12] Huynh QT, Lai VQ, Boonyatee T, Keawsawasvong S. Verification of soil parameters
of hardening soil model with small–strain stiffness for deep excavations in medium
dense sand in Ho Chi Minh City. Vietnam Innovative Infrastruct Solut 2022;7:15.
[13] Huynh QT, Lai VQ, Boonyatee T, Keawsawasvong S. Behavior of a deep excavation
and damages on adjacent buildings: A case study in Vietnam. Transport Infrastruct
Geotechnol 2021;8:361–89.
[14] Likitlersuang S, Chheng C, Keawsawasvong S. Structural modelling in finite
element analysis of deep excavation. J GeoEng 2019;14(3):121–8.
[15] Zhang W, Hou Z, Goh ATC, Zhang R. Estimation of strut forces for braced
excavation in granular soils from numerical analysis and case histories. Comput
Geotech 2019;106:286–95.
[16] Goh ATC, Zhang F, Zhang W, Chew OYS. Assessment of strut forces for braced
excavation in clays from numerical analysis and field measurements. Comput
Geotech 2017;86:141–9.
[17] Goh ATC, Zhang F, Zhang W, Zhang Y, Liu H. A simple estimation model for 3D
braced excavation wall deflection. Comput Geotech 2017;83:106–13.
[18] Terzaghi K. Theoretical soil mechanics. New York: Wiley; 1943.
[19] Bjerrum L, Eide O. Stability of strutted excavations in clay. Géotechnique 1956;6
(1):32–47.
[20] Goh ATC. Estimating basal heave stability for braced excavations in soft clay.
J Geotech Eng 1994;120(8):1430–6.
[21] Goh ATC. Deterministic and reliability assessment of basal heave stability for
braced excavations with jet grout base slab. Eng Geol 2017;218:63–9.
[22] Faheem H, Cai F, Ugai K, Hagiwara T. Two–dimensional base stability of
excavations in soft soils using FEM. Comput Geotech 2003;30(2):141–63.
Fig. 17. The impact of dimensionless parameter on the stability number.
[23] Li Y, Zhang W, Zhang R. Numerical investigation on performance of braced
excavation considering soil stress-induced anisotropy. Acta Geotech 2022;17(2):
563–75.
Table 3 [24] Yodsomjai W, Lai VQ, Banyong B, Chauhan VB, Thongchom C, Keawsawasvong S.
Comparison between the present MARS equation and previous equations by A machine learning regression approach for predicting basal heave stability of
braced excavation in non–homogeneous clay. Arab J Geosci 2022;15:873.
[26,27].
[25] Cai F, Ugai K, Hagiwara T. Base stability of circular excavations in soft clay.
No. R/H N [26] N [27] N (MARS) J Geotech Eng 2002;128(8):702–6.
[26] Goh ATC. Basal heave stability of supported circular excavations in clay. Tunn
1 0.5 3.860 8.750 13.801 Undergr Space Technol 2017;61:145–9.
2 1 4.044 7.804 13.032 [27] Keawsawasvong S, Ukritchon B. Undrained basal stability of braced circular
3 2 4.438 7.134 9.286 excavations in non–homogeneous clays with linear increase of strength with depth.
4 4 6.884 6.660 8.477 Comput Geotech 2019;115:103180.
[28] Ladd CC. Stability evaluations during stage construction. J Geotech Eng 1991;117
(4):540–615.
[29] Zhang R, Goh ATC, Li Y. A simple estimation model for basal heave stability of
Data availability
braced excavations in anisotropic clay. Acta Geotech 2022;17:5789–800.
[30] Zhang R, Goh ATC, Li Y. Assessment of apparent earth pressure for braced
No data was used for the research described in the article. excavations in anisotropic clay. Acta Geotech 2021;16:1615–26.
[31] Krabbenhøft K, Galindo-Torres SA, Zhang X, Krabbenhøft J. AUS: Anisotropic
undrained shear strength model for clays. Int J Numer Anal Meth Geomech 2019;
Acknowledgments 43(17):2652–66.
[32] Keawsawasvong S, Yoonirundorn K, Senjuntichai T. Pullout capacity factor for
We acknowledge Ho Chi Minh City University of Technology cylindrical suction caissons in anisotropic clays based on Anisotropic Uundrained
Shear failure criterion. Transport Infrastruct Geotechnol 2021.
(HCMUT), VNU–HCM for supporting this study.
15
V.Q. Lai et al. Transportation Geotechnics 39 (2023) 100945
[33] Keawsawasvong S. Bearing capacity of conical footings on clays considering [51] Shiau J, Chudal B, Mahalingasivam K, Keawsawasvong S. Pipeline burst–related
combined effects of anisotropy and non–homogeneity. Ships Offshore Struct 2021. ground stability in blowout condition. Transp Geotech 2021;29:100587.
[34] Keawsawasvong S, Shiau J, Ngamkhanong C, Lai VQ, Thongchom C. Undrained [52] Yodsomjai W, Keawsawasvong S, Likitlersuang S. Stability of unsupported conical
stability of ring foundations: axisymmetry, anisotropy, and non–homogeneity. Int J slopes in Hoek-Brown rock masses. Transport Infrastruct Geotechnol 2021;8:
Geomech 2022. 278–95.
[35] Nguyen DK, Nguyen TP, Keawsawasvong S, Lai VQ. Vertical uplift capacity of [53] Keawsawasvong S, Thongchom C, Likitlersuang S. Bearing capacity of strip footing
circular anchors in clay by considering anisotropy and non–homogeneity. on Hoek-Brown rock mass subjected to eccentric and inclined loading. Transport
Transport Infrastruct Geotechnol 2022;9:653–72. Infrastruct Geotechnol 2021;8:189–200.
[36] Van CN, Keawsawasvong S, Nguyen DK, Lai VQ. Machine learning regression [54] Butterfield R. Dimensional analysis for geotechnical engineering. Géotechnique
approach for analysis of bearing capacity of conical footings in heterogenous and 1999;49(2):357–66.
anisotropic clay. Neural Comput Applic 2020. [55] Lai VQ, Shiau J, Keawsawasvong S, Tran DT. Bearing Capacity of Ring Foundations
[37] Jearsiripongkul T, Lai VQ, Keawsawasvong S, Nguyen TS, Van CN, Thongchom C, on Anisotropic and Heterogenous Clays: FEA, NGI-ADP, and MARS. Geotech Geol
et al. Prediction of uplift capacity of cylindrical caissons in anisotropic and Eng 2022;40:3929–41.
inhomogeneous clays using multivariate adaptive regression splines. Sustainability [56] Pramanik R, Mukherjee S, Babu GS. Deterministic and probabilistic prediction of
2022;14(8):4456. maximum wall facing displacement of geosynthetic-reinforced soil segmental walls
[38] Lai VQ, Banyong B, Keawsawasvong S. Stability of limiting pressure behind soil using multivariate adaptive regression splines. Transp Geotech 2022;36:100816.
gaps in contiguous pile walls in anisotropic clays. Eng Fail Anal 2022;134:106049. [57] Alzabeebee S, Chapman DN. Evolutionary computing to determine the skin friction
[39] Keawsawasvong S, Ukritchon B. Design equation for stability of a circular tunnel in capacity of piles embedded in clay and evaluation of the available analytical
an anisotropic and heterogeneous clay. Underground Space 2022;7(1):76–93. methods. Transp Geotech 2020;24:100372.
[40] Lai VQ, Jamshidi CR, Banyong R, Keawsawasvong S. Undrained stability of [58] Hasthi V, Raja MNA, Hegde A, Shukla SK. Experimental and Intelligent Modelling
opening in underground walls in anisotropic clays. Int J Geomech, ASCE 2022. for Predicting the Amplitude of Footing Resting on Geocell-Reinforced Soil Bed
[41] Shiau J, Lai VQ, Keawsawasvong S. Multivariate adaptive regression splines under Vibratory Load. Transp Geotech 2022:100783.
analysis for three–dimensional slope stability in anisotropic and heterogenous clay. [59] Nguyen VQ, Tran VL, Nguyen DD, Sadiq S, Park D. Novel hybrid MFO-XGBoost
J Rock Mech Geotech Eng 2022. model for predicting the racking ratio of the rectangular tunnels subjected to
[42] Lai VQ, Shiau J, Keawsawasvong S, Seehavong S, Cabangon LT. Undrained stability seismic loading. Transp Geotech 2022;37:100878.
of unsupported rectangular excavations: anisotropy and non–homogeneity in 3D. [60] Zhang R, Wu C, Goh ATC, Böhlke T, Zhang W. Estimation of diaphragm wall
Buildings 2022;12(9):142. deflections for deep braced excavation in anisotropic clays using ensemble
[43] Lai VQ, Nguyen DK, Banyong B, Keawsawasvong S. Limit analysis solutions for learning. Geosci Front 2021;12(1):365–73.
stability factor of unsupported conical slopes in clays with heterogeneity and [61] Wu C, Hong L, Wang L, Zhang R, Pijush S, Zhang W. Prediction of wall deflection
anisotropy. Int J Comput Mater Sci Eng 2022;11(1):2150030. induced by braced excavation in spatially variable soils via convolutional neural
[44] Yodsomjai W, Keawsawasvong S, Senjuntichai T. Undrained stability of network. Gondw Res 2022.
unsupported conical slopes in anisotropic clays based on Anisotropic Undrained [62] Zhang W, Zhang R, Wu C, Goh ATC, Wang L. Assessment of basal heave stability for
Shear failure criterion. Transport Infrastruct Geotechnol 2021;8(4):557–68. braced excavations in anisotropic clay using extreme gradient boosting and
[45] OptumCE OptumG2. Copenhagen, Denmark: Optum Computational Engineering. random forest regression. Underground Space 2022;7(2):233–41.
See https://optumce.com/. Accessed 1 Dec 2020. [63] Zhang W, Zhang R, Wang W, Zhang F, Goh ATC. A Multivariate Adaptive
[46] Sloan SW. Geotechnical stability analysis. Géotechnique 2013;63(7):531–72. Regression Splines model for determining horizontal wall deflection envelope for
[47] Ciria H, Peraire J, Bonet J. Mesh adaptive computation of upper and lower bounds braced excavations in clays. Tunn Undergr Space Technol 2019;84:461–71.
in limit analysis. Int J Numer Meth Eng 2008;75(8):899–944. [64] Zhang W, Zhang R, Wu C, Goh ATC, Lacasse S, Liu Z, et al. State-of-the-art review
[48] Yodsomjai W, Keawsawasvong S, Lai VQ. Limit analysis solutions for bearing of soft computing applications in underground excavations. Geosci Front 2020;11
capacity of ring foundations on rocks using Hoek-Brown failure criterion. Int J (4):1095–106.
Geosynthetics Ground Eng 2021;7:29. [65] Zhang R, Goh ATC, Zhang W. MARS inverse analysis of soil and wall properties for
[49] Keawsawasvong S, Lai VQ. End bearing capacity factor for annular foundations braced excavations in clays. Geomech Eng 2018;16(6):577–88.
embedded in clay considering the effect of the adhesion factor. Int J Geosynthetics [66] Lai VQ, Sangjinda K, Keawsawasvong S, Eskandarinejad A, Chauhan VB, Sae-
Ground Eng 2021;7:15. Long W, et al. A machine learning regression approach for predicting the bearing
[50] Keawsawasvong S, Ukritchon B. Undrained stability of a spherical cavity in capacity of a strip footing on rock mass under inclined and eccentric load. Front
cohesive soils using finite element limit analysis. J Rock Mech Geotech Eng 2019; Built Environ 2022;8:962331.
11(6):1274–85.
16