Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Engineering Structures 301 (2024) 117263

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Predicting the shear capacity of composite steel plate shear wall with the
application of RSM
Maryam Bypour a, Mohammad Yekrangnia b, Mahdi Kioumarsi c, d, *
a
Structural Engineering, Department of Civil Engineering, Semnan University, Semnan, Iran
b
Department of Civil Engineering, Shahid Rajaee Teacher Training University, Tehran, Iran
c
Department of Engineering, Østfold University College, Fredrikstad, Norway
d
Department of Built Environment, OsloMet – Oslo Metropolitan University, Oslo, Norway

A R T I C L E I N F O A B S T R A C T

Keywords: This research aims to predict the maximum shear capacity of Composite Steel Plate Shear Walls (CSPSWs)
Finite Element Analysis through nonlinear Finite Element (NLFE) analyses and Response Surface Method (RSM). The variables are
Composite Steel Plate Shear Wall concrete thickness and compressive strength, steel infill plate thickness, and yield stress. A parametric study was
Response Surface Method
performed on CSPSW comprising a concrete layer on one side of the infill plate and the corresponding models
Linear Regression
Response characteristics
having the concrete layer between the two plates. The results show that RSM is a reliable practice for precisely
predicting the maximum shear capacity of the studied models. Furthermore, comparing the RSM results with
those acquired by Linear Regression (LR) indicated that RSM is more accurate than LR. Moreover, in the models
with the internal concrete layer, the concrete contributed more to the shear capacity of the specimens, and this
difference expands with an increase in the thickness of the concrete layer.

1. Introduction infill plates [8]. The latter type has advantages over the former ones in
the case of concrete casting and providing additional confinement for
Using steel plate shear walls (SPSWs) as a dependable lateral force- the concrete, leading to superior service performance and construction
resisting system has demonstrated commendable efficacy in construct­ convenience [9].
ing and rehabilitating existing structures. This is mainly due to their Though requirements for designing CSPSW are presented in various
significant lateral stiffness and ductility, enabling high-rise buildings to design codes such as AISC-16 [14], there is no unique formula applicable
withstand severe seismic forces and winds optimally. Interestingly, even for acquiring the capacity of CSPSW. The code refers to different re­
in concrete structures, applying SPSW has shown a superior load- quirements to design the components of CSPSW, which consist of a steel
carrying capacity and inelastic deformation capacity than concrete infill plate, boundary frame, and concrete panel separately. Further­
shear walls [1]. That is why, in structures that require retrofitting more, previous studies show that the considerable differences between
practices, the use of SPSW can be prioritized [2,3]. Due to its desirable lateral-load capacities predicted numerically and the one calculated
performance, SPSW has been appealing to numerous researchers to using AISC-16 are a result of ignoring the impact of applying axial load,
enhance the structural response to lateral loads by employing different the wall aspect ratio, and the ratio of reinforcement used in the concrete
methods such as using stiffeners [4,5] to increase shear capacity, panel with the AISC-16 relations [15]. Hence, it is essential to propose
opening for controlling the steel plate capacity [6,7], and using rein­ an accurate and viable relation to estimate the capacity of this system.
forced concrete (RC) panel to enhance strength and ductility. Composite The major purpose of this numerical study is to propose reliable
Steel Plate Shear Wall (CSPSW) is a structural system including a steel equations to predict the maximum shear capacity (Vmax ) of the CSPSW
plate and RC either on one side, as shown in Fig. 1(a), or two sides, as system accurately. In this regard, the data obtained from a wide range of
depicted in Fig. 1(b)-(d), of the plate. Studs or bolts are mechanical shear Nonlinear Finite Element (NLFE) analyses and Response Surface Method
connectors that connect the RC to the plate. In another variant of (RSM) were used. This numerical study can pave the way for experi­
CSPSW, concrete is positioned between two steel plates, as illustrated in mental research for selecting the characteristics of experimental speci­
Fig. 1(e)-(f). The major role of the RC is to prevent early buckling of steel mens based on the required Vmax . The variable parameters in this study

* Corresponding author at: Department of Built Environment, OsloMet – Oslo Metropolitan University, Oslo, Norway.
E-mail address: mahdi.kioumarsi@oslomet.no (M. Kioumarsi).

https://doi.org/10.1016/j.engstruct.2023.117263
Received 25 July 2023; Received in revised form 31 October 2023; Accepted 26 November 2023
Available online 12 December 2023
0141-0296/© 2023 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
M. Bypour et al. Engineering Structures 301 (2024) 117263

Fig. 1. Different types of CSPSW: (a) Concrete located on one side of the infill plate, (b)-(d) Steel infill plate is embedded entirely into concrete, (e)-(f) Concrete is
placed between two steel plates.

are the thickness (tp ) and yield stress (fy ) of steel infill plate as well as reduced thickness in the CSPSW system with opening rather than
concrete thickness (tc ) and concrete compressive strength (f ′c ). Each considering the openings [17]. Experimental and FE analyses were
parameter is examined by five values. The research comprises two conducted into CSPSW by considering various thicknesses for concrete
phases: models in phase one (I) include CSPSW with concrete covering under lateral loading. As a result of parametric study data, formulas
one side of the steel infill plate, and in phase two (II) models, concrete is were proposed for the yield shear force as well as the stiffness of such a
placed between two steel plates. Since all the parameters remained system [10].
constant for both stages and as in phase II models, two plates are used to A new model for assessing the in-plane shear capacity of CSPSW was
confine the concrete, the thickness of each plate in these models is half proposed using the lower-bound plastic limit theory. The validity of the
that of phase I models. Based on NLFE results, equations have been formulas was checked by comparing the results obtained by the equa­
derived for calculating the Vmax of both types of investigated CSPSWs tions with those of collected experimental data [18]. Based on
using RSM. In addition, using the same FE results, other equations were comprehensive numerical research, a reliable Macroscopic Fiber-based
obtained by employing Linear Regression (LR). Then, the accuracy of the Model was introduced to predict the nonlinear cyclic response of
equations acquired through the two methods has been checked and CSPSW subjected to seismic loading. The derived equations and
compared. Finally, the sensitivity of the Vmax to different variables was modeling guidelines effectively simulate the seismic performance of
evaluated. CSPSW structures with aspect ratios of 1 or greater [19]. A study
investigated the experimental examination of the influence of different
2. Background levels of corrosion on the axial compressive behavior of a
double-steel-plate composite shear wall (DSCW) with binding bars. The
In recent decades, numerous studies have been carried out on the results demonstrated that the increase in corrosion rate has an adverse
analysis, design, and assessment of the performance of CSPSWs, exper­ effect on the capacity of DSCW to withstand axial compressive forces,
imentally and numerically. Several experimental and numerical studies ductility, and stiffness. Consequently, a formula was introduced to es­
resulted in proposing equations [16]. The results of a study on CSPSW timate the axial compressive load-bearing capacity of corroded DSCW
containing steel plates filled with plain concrete indicated that the finite with binding bars [20].
element (FE) method, as well as mechanics-based equations, can be The in-plane performance of the CSPSW system was investigated
employed to calculate the effective stiffness [13]. Analyzing a wide experimentally by employing the mechanics-based behavior model,
range of concrete-filled steel plates (CFSP) containing concrete columns which is based on composite plate theory. In this regard, a range of
with steel tubes resulted in proposing reasonably accurate formulas for parameters was evaluated, showing a linear relationship between the
predicting the drift capacity and ductility of CFSP [9]. Furthermore, an reinforcement ratio and the in-plane shear capacity of this system [21].
innovative Multi-Expression Programming (MEP) algorithm was used to Furthermore, evaluating the out-of-plane seismic behavior of CSPSW
propose an experiment-based relation to estimate the equivalent through experimental research indicated the significant impact of the

2
M. Bypour et al. Engineering Structures 301 (2024) 117263

Fig. 2. Schematic view of the experimental specimen (CS) for verification, after [8].

applied vertical load and the plate thickness on the maximum capacity dissipate energy [27].
of this system. In contrast, the concrete strength showed a negligible Numerical and experimental investigations on CSPSW with double
effect [22]. steel plate were conducted by considering variables such as material
Some research has been conducted to assess the seismic performance (steel and concrete) strength and other parameters relating to the
of CSPSW [23,24]. A novel composite wall containing steel plate and RC specimens’ geometry to evaluate failure mode, hysteresis curve, and
(SPRC) was tested under simulated earthquake conditions, incorpo­ skeleton curve. Based on the results, this type of composite wall has
rating steel coupling beams and columns. The columns, subjected to indicated satisfactory seismic performance with high bearing capacity
shear forces transferred from the coupling beams, demonstrated elastic and ductility [28]. An innovative type of CSPSW containing welded
behavior and protected the SPRC wall by providing a stable overturning reinforcement grids and partially encased composite columns was
moment resistance [25]. experimented with to investigate a range of parameters associated with
An experimental study assessed the seismic behavior of a novel steel plate and boundary frame. The test results showed excellent
double steel plate composite shear wall filled with a unique combination bearing and deformation capacity, acceptable energy dissipation ca­
of iron tailings and recycled aggregate concrete, referred to as ITRAC- pacity, and stiffness [11]. The CSPSW system was numerically investi­
CSW. Experimental testing involved subjecting it to cyclic lateral and gated by NLFE analysis under gravity and fire load. The parameters
constant axial loads. The findings from the analysis demonstrated that evaluated were the strength and geometric details of the steel plate,
the ITRAC-CSW displayed favorable load-bearing capacity, energy reinforcement, axial load ratio, and boundary conditions. The results
dissipation capacity, and stiffness, making it a promising choice for real- indicate that reducing the thickness of the infill plate and decreasing the
world construction projects [26]. The seismic performance of a new rotational fixity of the boundary conditions significantly shortened the
composite steel plate shear wall, which incorporates Engineered time to failure for these composite walls [29].
Cementitious Composite (ECC) panels for restraint, was investigated Many studies were conducted to assess the difference between the
experimentally. The test findings demonstrated that, in contrast to a behavior of CSPSW and corresponding conventional SPSW. For
conventional buckling-restrained SPSW, the innovative one exhibited example, testing was conducted on an innovative CSPSW incorporating
adequate ductility, high initial stiffness, and a commendable ability to rubber-coated uplift restrained (RCUR) studs to compare the behavior

3
M. Bypour et al. Engineering Structures 301 (2024) 117263

Fig. 4. Stress-strain relation for steel in FE simulation.

Table 2
Utilized material parameters for the CDP model of concrete in FE simulation.
Dilation angle Eccentricity fb0/fc0 Kc Viscosity Parameter

36.5 0.1 1.16 0.67 0.001

beam-column connections. The specimen contains a steel frame, a steel


infill plate with a thickness of 2 mm, a concrete layer with a thickness of
30 mm, four bolts, and a steel mesh as reinforcement placed on concrete
(see Fig. 2). It should be noted that in the CS specimen, the RC panel was
Fig. 3. Modeling of the benchmark specimen (the meshing size).
placed with 11.25 mm gap from the boundary frame.

Table 1 3.2. Numerical FE simulation


Employed characteristics for the steel material of the CS in numerical simulation
[8]. 3.2.1. Element and connection type
Section type Yield stress (Fy ) Ultimate strength (Fu ) In this study nonlinear finite element (FE) modeling with nonlinear
(MPa) (MPa) static analysis under cyclic loading was performed. Fig. 3 shows a 3D FE
Steel infill plate (thickness 268 415 model of the benchmark specimen (CS). The element type for the steel
=2 mm) infill plate is a 4-node shell element with reduced integration (S4R),
IPE 100 used in beam flange 308 479 suitable for both thick and thin shells. An 8-node solid element with
IPE 100 used in beam web 285 446 reduced integration (C3D8R) is utilized for the concrete layer, steel
Fish plates 297 406
Reinforcement 336 492
frame, and stiffeners, as this selection provides the necessary precision
for modeling these components. Reinforcements and bolts are modeled
with Truss elements (T3D2).
with the conventional unstiffened SPSW. The novel CSPSW has indi­ Surface-to-surface contact interaction between the steel infill plate
cated superior seismic behavior compared to the conventional SPSW and the concrete layer is used. In this regard, hard contact in the normal
[30]. Furthermore, the results of FE analyses by ABAQUS to compare direction and penalty friction in the tangential direction are specified. As
one-story CSPSW with the corresponding SPSW proved the superiority reinforcement mesh and bolts are embedded in the concrete layer,
of CSPSW over SPSW by developing tension fields, which leads to a embedded region constraint is employed as the interaction between
considerably higher shear capacity [31]. As steel–concrete composite these members and the concrete. A tie constraint is applied to establish
shear walls typically damage at the corner zones due to seismic loads, an the connection between the bolts and the steel frame with the steel infill
innovative corner design for this structural system was suggested ac­ plate.
cording to the design criteria of plastic damage relocation. The study
experimentally evaluated energy dissipation, crack development, hys­ 3.2.2. Loading
teretic curve, failure mode, and stiffness decay. Regarding the experi­ According to ATC24 [30], the lateral displacement was applied to the
mental database, the novel CSPSW showed more desirable seismic top beam using a cyclic quasi-static loading protocol.
performance than conventional SPSW [32]. Furthermore, half-scale
vertical and horizontal corrugated and flat SPSWs and the correspond­ 3.2.3. Material properties
ing CSPSW were experimented to compare the failure modes and Table 1 reports the characteristics of steel members of the CS. The
deformation capacity of these two specimens under cyclic loading. The modulus of elasticity of steel members is considered 203 GPa in nu­
findings have shown that the lateral stiffness, energy dissipation ca­ merical simulation. Kinematic hardening was applied for the steel ma­
pacity, and ductility of the CSPSW system were superior to that of the terials in both cyclic and pushover analyses. In this regard, elastoplastic
SPSWs [12]. material with linear strain hardening, in which the plastic zone slope (E′)
was considered equivalent to one percent of the elastic modulus of
3. Verification and modeling procedure elasticity (E), was used for the stress-strain behavior of steel (see Fig. 4).
The compressive strength and modulus of elasticity of concrete uti­
3.1. Benchmark specimen lized for the CS are 43 and 35,000 MPa, respectively. The Concrete
Damaged Plasticity (CDP) model was employed to simulate the behavior
In this study, to verify the assumptions and method used for the of concrete material in ABAQUS. Table 2 shows the considered param­
simulation of CSPSW using ABAQUS software [33], the CS experimental eters for the CDP model.
specimen tested by Arabzadeh et al. [8] was used. The selected experi­ Fig. 5 shows the concrete behavior in uniaxial load in tension and
mental specimen, CS, is a 1:4 scale, one-story CSPSW with rigid compression according to ABAQUS/Standard [29]. Eqs. (1)–(2)

4
M. Bypour et al. Engineering Structures 301 (2024) 117263

Fig. 5. Concrete behavior in uniaxial loading: (a) tension and (b) compression according to ABAQUS/Standard [33].

50
5
45
Compressive stress (MPa)

4.5
40 4
Tensile stress (MPa)

35 3.5
30 3
25 2.5
20 2
15 1.5
10 1
5 0.5
0 0
0 0.001 0.002 0.003 0.004 0 0.0005 0.001 0.0015
Strain Strain
(a) (b)
Fig. 6. Stress–strain curves of concrete: (a) compression and (b) tension in the CDP model.

represent the stress-strain relationships of concrete subjected to uniaxial numerical model at ultimate displacement, respectively. The figures
compression and tension, respectively. reveal that yielding has occurred in all parts of the steel infill plate.
( pl )
σ c = E0 εc − ̃εc (1 − dc ) (1) 4. Application of response surface method (RSM)
( pl )
σ t = E0 εt − ̃εt (1 − dt ) (2) The preceding part has shown the NLFE approach to be a precise
methodology for evaluating the capacity of CSPSWs. However, the
in which dc and dt are the uniaxial damage parameters for compression analysis using this approach can be time-consuming due to the need to
and tension, respectively, E0 is the initial (undamaged) elastic stiffness assess the interactions between fundamental essential parameters that
pl pl influence CSPSW capacity. The Response Surface Method (RSM), a sta­
of the concrete, ̃ εc and ̃εt are the compressive and tensile equivalent
tistical method, can be employed to overcome this limitation. RSM fa­
plastic strains, respectively [33].
cilitates the development and optimization of processes [34] and has
Based on the CDP model, stress-strain curves utilized for simulating
gained significant recognition in various fields.
concrete in tension and compression are illustrated in Fig. 6(a)-(b).
Compressive and tensile damages were also considered for concrete.
4.1. RSM background
3.2.4. Comparison of the experimental data with FE modeling results
Although RSM has been widely used in Environmental Engineering
Fig. 7 compares the force-displacement curve acquired from the FE
to optimize the effect of process variables such as emissions character­
analysis with the corresponding experimental specimen (CS) under cy­
istics, there have been limited attempts to utilize this methodology to
clic and pushover loads. The two curves, experimental and numerical,
foresee the structural response and capacity of various structural sys­
have exhibited an excellent coincidence, which means the FE analysis
tems. In order to examine the cyclic shear behavior of buckling-
has an acceptable accuracy in predicting the response characteristics of
restrained SPSWs (BRSPSW) with beam-connected only, RSM was
the studied wall. Fig. 8(a)-(c) show the deformation of the experimental
applied to forecast the shear capacity of this system. The outcomes
specimen, the maximum plastic strains, and von Mises stress of the
demonstrated that the RSM prediction model effectively and accurately

5
M. Bypour et al. Engineering Structures 301 (2024) 117263

Fig. 7. Experimental and numerical force-displacement curves of the CS.

Fig. 8. CS in ultimate limit states: (a) experimental, (b) maximum plastic strains, and (c) von Mises stress in FE simulation.

Table 3
Variables along with their respective levels.
No. Variables Abbreviation Unit Levels

Axial Factorial Axial

( − 2) Low (-1) Center (0) High (1) ( + 2)

1 Concrete compressive strength f′c MPa 20 25 30 35 40


2 Yield strength of steel infill plate fy MPa 100 150 200 250 300
3 Concrete thickness tc mm 20 25 30 35 40
4 Thickness of steel infill plate tp mm 1 1.5 2 2.5 3

estimated the shear capacity of the system [35]. In an FE-based nu­ piers without boundary elements, employing a three-level fractional
merical study, RSM was employed to propose equations to predict the factorial design technique. The findings revealed that the aspect ratio
shear capacity of stiffened steel plate shear wall with rectangular notably influences the lateral load-carrying capacity and stiffness. In
openings. The findings indicated that RSM is such a reliable method to contrast, the slenderness ratio of the faceplate had a relatively minor
predict the capacity of the studied models, and formulas were derived to impact on strength and initial stiffness [15].
calculate the shear capacity of this system [7]. Statistical analyses were To achieve multi-objective optimization for a composite orthotropic
conducted to explore the primary design parameters’ impact on the bridge, RSM was utilized. In the research, regression models were
stiffness and lateral strength of steel-plate concrete composite shear wall developed using RSM to predict stress ranges for fatigue-sensitive

6
M. Bypour et al. Engineering Structures 301 (2024) 117263

Table 4 CFRP-strengthened ECC. The experimental data was then analyzed


Characteristics of models in numerical analyses. using RSM to confirm the alignment of the proposed model with the
Run Order f′c (MPa) fy (MPa) tp (mm) tc (mm) experimental results. This indicated a strong correspondence between
the model and experimental outcomes [37]. An investigation was un­
1 35 150 2.5 35
dertaken to assess the accuracy of predicting the residual compressive
2 30 200 2 40
3 25 250 2.5 35 strength of concrete after being subjected to high temperatures. This
4 25 150 1.5 25 study utilized data from previous experiments and employed RSM. The
5 40 200 2 30 RSM analysis highlighted that heating temperatures had the most pro­
6 30 200 3 30 nounced impact on the residual compressive strength of concrete, and
7 35 150 2.5 25
the RSM model exhibited a robust connection with the validation
8 35 250 1.5 35
9 35 250 2.5 35 datasets [38].
10 30 300 2 30 RSM has been increasingly employed for parametric design and
11 30 200 2 20 analysis, structural design, and reliability performance [39–41]. By
12 20 200 2 30
employing RSM, the detection of parameter interactions becomes more
13 35 250 1.5 25
14 25 250 1.5 25 efficient, enabling a more streamlined assessment of CSPSW capacity.
15 35 250 2.5 25 This integration of RSM into the analysis process can enhance the pre­
16 25 150 1.5 35 cision and effectiveness of evaluating and optimizing CSPSW structures.
17 30 100 2 30 This study employs the RSM to estimate the association between the
18 25 250 2.5 25
maximum capacity of the CSPSW (Vmax ), referred to as the "response"
19 25 150 2.5 25
20 25 150 2.5 35 and the "variables," including thickness (tp ) and yield strength (fy ) of
21 25 250 1.5 35 steel infill plate as well as concrete compressive strength (f ′c ) and
22 35 150 1.5 25
23 30 200 2 30
thickness (tc ). The function that approximates this correlation is called
24 30 200 1 30 the "response surface".
25 35 150 1.5 35

4.2. Second-order polynomial function


components and the overall structural weight. An analysis of correlation
and variance indicated that RSM is both statistically significant and The second-order polynomial function is commonly employed to
capable of accurately forecasting the desired response variables [36]. An predict the response, taking into account the interactions between two
experimental investigation was conducted to study the bond behavior of factors or parameters. This function, denoted as Eq. 3, applies to k
variables. By considering the second-order terms, the model accounts for

Fig. 9. CS-CB model: (a) configuration of CS-CB models (dimension scale: 2:1), (b) von Mises stress, and (c) maximum plastic strains of model number 10,
see Table 4.

7
M. Bypour et al. Engineering Structures 301 (2024) 117263

Table 5
Comparison of the Vmax taken from FE analyses with RSM and LR for both models.
Run Order Vmax (kN)

CS-CO CS-CB

FEM RSM LR Ratio* Ratio** FEM RSM LR Ratio* Ratio**

1 569.019 569.155 572.77 1.000 0.993 722.681 712.650 711.403 1.014 1.016
2 545.401 544.360 543.93 1.002 1.003 710.586 742.300 721.311 0.957 0.985
3 616.736 617.765 613.95 0.998 1.005 772.979 761.900 751.983 1.015 1.028
4 473.497 472.785 467.96 1.002 1.012 530.270 532.100 518.872 0.997 1.022
5 544.130 543.940 543.42 1.000 1.001 629.798 643.700 645.438 0.978 0.976
6 641.558 640.150 642.79 1.002 0.998 717.552 731.200 742.074 0.981 0.967
7 566.511 566.310 570.61 1.000 0.993 621.575 620.150 628.856 1.002 0.988
8 509.309 509.550 514.59 1.000 0.990 660.395 651.750 655.346 1.013 1.008
9 619.497 620.395 615.60 0.999 1.006 783.439 772.700 758.656 1.014 1.033
10 587.743 585.910 584.61 1.003 1.005 670.256 683.800 686.018 0.980 0.977
11 539.703 540.100 539.62 0.999 1.000 585.410 577.300 556.217 1.014 1.052
12 541.116 540.660 540.13 1.001 1.002 619.704 629.500 632.090 0.984 0.980
13 507.876 507.995 512.43 1.000 0.991 570.624 571.150 572.799 0.999 0.996
14 507.125 507.205 510.79 1.000 0.993 567.592 568.050 566.126 0.999 1.003
15 617.383 617.890 613.44 0.999 1.006 674.621 674.700 676.110 1.000 0.998
16 474.821 474.540 470.11 1.001 1.010 608.795 599.100 601.419 1.016 1.012
17 499.060 500.250 498.95 0.998 1.000 577.671 587.800 591.511 0.983 0.977
18 614.831 615.400 611.80 0.999 1.005 671.070 672.000 669.436 0.999 1.002
19 563.985 563.960 568.97 1.000 0.991 618.045 617.150 622.182 1.001 0.993
20 566.597 566.665 571.12 1.000 0.992 711.085 701.550 704.729 1.014 1.009
21 508.232 508.620 512.94 0.999 0.991 648.110 640.550 648.672 1.012 0.999
22 474.251 473.435 469.60 1.002 1.010 533.993 535.500 525.546 0.997 1.016
23 542.979 542.940 541.78 1.000 1.002 627.051 627.900 638.764 0.999 0.982
24 437.367 438.130 440.77 0.998 0.992 515.291 525.200 535.454 0.981 0.962
25 475.714 475.330 471.76 1.001 1.008 620.517 610.600 608.093 1.016 1.020
*
Ratio = FEM/RSM.
**
Ratio = FEM/LR.

applied to establish Eq. 3 using the collected response data, i.e., Vmax . In
Table 6 the case of k variables, the overall number of design points in the central
The properties of the chosen specimens to validate the results obtained from composite design comprised of 2k factorial points along with 2k axial
RSM and LR. points and an additional center point. By employing half fractional
Run Order f′c (MPa) fy (MPa) tp (mm) tc (mm) factorial points, the quantity of fractional points is decreased to 2k− 1 . To
26 20 300 1 30 determine the coded distance between the axial points and the center
27 25 250 2 25 point, Eq. 4 is employed.
28 30 200 1.5 20 √̅̅̅̅̅
29 35 100 2.5 35 α = 4 2k (4)
30 40 150 3 40
31 20 200 1 30
32 25 300 2 30 4.3. Selected parameters in RSM
33 30 100 1.5 20
34 35 300 2.5 25 The objective of employing RSM is to estimate Vmax of CSPSW by
35 40 250 3 35
taking into account the impact of multiple parameters and their inter­
36 20 100 1 30
37 20 150 1.5 20 action effects on Vmax . In this study, RSM and NLFE analyses were uti­
38 25 200 2 25 lized to study the influence of four parameters, namely, the f ′c , fy , tc , and
39 25 250 2.5 20
tp . To assess the combined influence of these parameters, FE analyses
40 30 300 3 30
41 30 100 1 30 were conducted by varying the combinations of parameters. In the CCD
42 35 150 1.5 20 method, the studied variables were assigned five levels each, including
43 35 200 2 35 the zero level (Xi = 0), the one level (Xi = ± 1), and α level (Xi = ± α).
44 40 250 2.5 40 For all studied parameters, α was set to 2. The levels considered for each
45 40 300 3 40
variable are presented in Table 3.

nonlinear relationships that may exist among the variables, providing a 4.4. Overview of the central composite design
more realistic approximation of the response.
∑k ∑k ∑∑k Based on the CCD and RSM, conducting 25 numerical experiments is
Y = β0 + β Xi +
i=1 i
β X2 +
i=1 ii i
β Xi Xj
i<j ij
(3) necessary when dealing with four variables. Table 4 displays the specific
numerical analyses that need to be carried out based on the variables
In the Eq.3, Y is the RSM response (Vmax ), k represents the variable
numbers, and Xi denotes the coded level of the ith design variable. The β0 mentioned above, i.e., f ′c , fy , tp , and tc .
is a constant coefficient in the equation, while βi , βii and βij are the This study compares the maximum shear capacity of specimens with
regression coefficients for the linear, quadratic, and interaction effects, concrete placed on one side of the CSPSW system and specimens with
respectively. concrete positioned between two steel plates in the CSPSW system. The
The central composite design (CCD) is a widely employed approach latter structural configuration is modeled and analyzed with equivalent
for constructing a second-order model. In this research, the CCD is characteristics to the former one. As the experimental specimen used for
verification was CS, the terms ‘CS-CO’ and ‘CS-CB’ were utilized to refer

8
M. Bypour et al. Engineering Structures 301 (2024) 117263

Table 7
Validation of the equations derived from RSM and LR.
Run order Vmax (kN)

CS-CO CS-CB

FEM RSM LR Ratio1 Ratio2 FEM RSM LR Ratio* Ratio**

26 461.150 463.360 481.951 0.995 0.957 543.155 563.700 576.034 0.964 0.943
27 563.054 562.253 561.292 1.001 1.003 624.100 619.950 617.781 1.007 1.010
28 490.426 489.120 489.118 1.003 1.003 533.752 534.575 504.562 0.998 1.058
29 542.496 543.640 551.354 0.998 0.984 692.410 688.550 687.777 1.006 1.007
30 618.175 615.570 625.174 1.004 0.989 837.238 852.425 807.669 0.982 1.037
31 437.256 437.550 439.121 0.999 0.996 511.21 526.4 528.780 0.971 0.967
32 585.739 584.860 583.784 1.002 1.003 667.085 682.575 682.681 0.977 0.977
33 456.187 454.600 446.288 1.003 1.022 498.602 509.425 457.309 0.979 1.090
34 638.672 643.785 634.858 0.992 1.006 701.125 707.9 699.736 0.990 1.002
35 672.924 674.228 666.927 0.998 1.009 846.662 845.225 813.648 1.002 1.041
36 412.932 412.020 396.291 1.002 1.042 494.157 504.900 481.527 0.979 1.026
37 472.786 470.605 466.056 1.005 1.014 513.088 529.375 474.262 0.969 1.082
38 542.679 540.753 539.877 1.004 1.005 598.687 595.275 594.154 1.006 1.008
39 611.161 613.685 610.719 0.996 1.001 652.350 650.975 628.162 1.002 1.039
40 695.855 700.140 685.617 0.994 1.015 777.632 806.000 789.328 0.965 0.985
41 412.789 412.460 397.938 1.001 1.037 489.142 504.000 488.201 0.971 1.002
42 473.626 471.955 468.526 1.004 1.011 517.680 521.875 484.273 0.992 1.069
43 544.244 544.523 543.679 0.999 1.001 698.619 684.875 683.375 1.020 1.022
44 620.764 621.985 617.500 0.998 1.005 899.320 859.575 803.267 1.046 1.120
45 701.226 705.150 689.419 0.994 1.017 904.210 966.500 878.549 0.936 1.029
1
Ratio = FEM/RSM.
2
Ratio = FMR/LR.

1000 1000
Predicdted Vmax (RSM) (MPa)

Predicted Vmax (RSM) (MPa)

y=x y=x
900 Training set 900 Training set
Testing set Testing set
800 800

700 700

600 600

500 500

400 400
400 500 600 700 800 900 1000 400 500 600 700 800 900 1000
FE Vmax (MPa) FEM Vmax (MPa)
(a) CS-CO (b) CS-CB

1000 1000
y=x y=x
Predicted Vmax (LR) (MPa)
Predicted Vmax (LR) (MPa)

900 Training set 900 Training set


Testing set Testing set
800 800

700 700

600 600

500 500

400 400
400 500 600 700 800 900 1000 400 500 600 700 800 900 1000
FEM Vmax (MPa) FEM Vmax (MPa)
(c) CS-CO (d) CS-CB

Fig. 10. Comparison of the FE analyses results (Vmax ) with predicted responses acquired by RSM and LR.

to the analyzed models. CO implies concrete on one side, and CB in­ 4.5. Results and discussions
dicates placing concrete between two plates. Fig. 9 shows the scheme of
CS-CB models. To obtain comparable results, the element types, in­ 4.5.1. RSM results
teractions, boundary conditions, and mesh sizes for CS-CB models were This part provides the second-order response equations (Vmax ) pre­
selected the same as the CS-CO ones (see Fig. 9(b)-(c)). Considering both dicted by the RSM using coded variables, as illustrated in Eqs. (5)–(6).
CS-CO and CS-CB, 50 specimens have been modeled and analyzed. The factors are f ′c , fy , tp , and tc with the output of Vmax . One notable
advantage of employing these regression models is their simplicity,

9
M. Bypour et al. Engineering Structures 301 (2024) 117263

Fig. 11. Vmax versus tc in CS-CO models considering fy = 100, 150, 200, 250 and 300 MPa for: (a)-(e) tp = 2 mm and (f)-(j) 3 mm.

10
M. Bypour et al. Engineering Structures 301 (2024) 117263

Fig. 12. Vmax versus tc in CS-CB models: tp = 2 mm for (a)-(e) and 3 mm for (f)-(j) and fy = 100, 150, 200, 250 and 300 MPa.

avoiding unnecessary complexity. 2


As a result of performing RSM analysis, the regression model utiliz­ Vmax = 737 − 6.8f ′c − 0.37fy + 13tp − 17.9tc + 0.087f ′c + 0.00079fy 2 + 0.3tp 2
ing coded variables for CS-CO models is derived, offering a simplified + 0.319tc 2 − 0.0003f ′c fy − 0.04f ′c tp + 0.081f ′c tc + 0.189fy tp
and practical approach, see Eq. (5). Furthermore, the regression model + 0.0055fy tc + 1.74tp tc
using coded variables for CS-CB models is presented in Eq. (6).
(6)
2
Vmax = 301.2 + 0.138f ′c + 0.0883fy + 74.22tp + 0.475tc − 0.0064f ′c
4.5.2. Linear regression
+ 0.000014fy 2 − 3.8tp 2 − 0.0071tc 2 + 0.00014f ′c fy + 0.170f ′c tp (5)
In order to simplify the relations for calculating maximum shear
+ 0.0014f ′c tc + 0.1702fy tp − 0.00034fy tc + 0.095tp tc capacity (Vmax ), Linear Regression (LR) is used to estimate the rela­
tionship between a dependent response and independent variables.
Equations for both groups of models are proposed using Eq. 7.

11
M. Bypour et al. Engineering Structures 301 (2024) 117263

Fig. 13. Variation of Vmax versus variation of tc with constant f ′c = 35 MPa: (a)-(b) CS-CO and (c)-(d) CS-CB models for fy = 100, 200 and 300 MPa.


hθ (x) = θ0 + θi Xi i = 1, 2, 3, 4 Vmax = 69.97523062 + 0.6673917f ′c + 0.4725325fy + 103.31008339tp
(9)
+ 8.2546917tc
hθ (x) = θ0 + θ1 X1 + θ2 X2 + θ3 X3 + θ4 X4
To compare the responses acquired by RSM and LR, Table 5 provides
θ = (X T X)− 1 X T y (7) the data obtained from the FE results along with RSM and LR responses
using Eqs. (5–6) and Eqs. (8–9). The results of both methods strongly
where hθ (x) is the predicted response, which is Vmax , θ0 is the intercept agree with FE results.
coefficient; θ1 , θ2 , θ3 , and θ4 are the slope coefficients relating to inde­
pendent variables X1 , X2 , X3 , and X4 , respectively. These independent 4.5.3. Validity check of RSM results
variables are representative of f ′c , fy , tp , and tc , respectively. The In order to validate the equations derived from the RSM and LR, a
parameter y is the Vmax obtained by FE modeling. To compare the results series of FE analyses were conducted. These analyses involved utilizing
of the two methods (RSM and LR), FE results of the models selected by random values within the specified ranges for the parameters of interest.
RSM (see Table 4) are utilized for LR practice. By solving the matrix in Subsequently, the results attained from the FE analyses were compared
Eq. 7, the values obtained for θ and linear equations for both categories with the results predicted by the RSM and LR equations (Eqs. (5)–(6) and
of models are as follows: Eqs. (8–9)). This comparative analysis aimed to assess the accuracy and
For CS-CO models: reliability of employed prediction methodologies. Twenty sets of
random parameters were selected to ensure a comprehensive validation
θ0 242.691796 process. The details of these parameter sets can be found in Table 6. Each
⎡ ⎤ ⎡ ⎤
θ1 0.164700121 set represents a unique combination of parameter values within the
⎢ ⎥ ⎢ ⎥
⎢ ⎥ ⎢ ⎥
θ = ⎢ θ2 ⎥ = ⎢ 0.428300008 ⎥ established ranges. By utilizing different sets of parameters, a diverse
⎣ ⎦ ⎣ ⎦ range of scenarios and conditions were considered, enhancing the
θ3 101.009667
robustness of the validation process. A comparison between the Vmax
θ4 0.215516788 values obtained from FE analyses and those predicted by the RSM and
LR equations are presented in Table 7. The results demonstrate the ac­
Vmax = 242.691796 + 0.164700121f ′c + 0.428300008fy + 101.009667tp curate prediction of both methods in estimating Vmax of the specimens.
(8)
+ 0.215516788tc To evaluate and compare the performance of the two methods (RSM
and LR) employed for predicting the Vmax of the models, more precisely,
For CS-CB models: the comparison of the predicted responses with FE results for both CS-
θ0 69.97523062 CO and CS-CB models are presented in Fig. 10. In this regard, the data
⎡ ⎤ ⎡ ⎤ employed for proposing equations (training set, Table 5) and the ones
θ1
⎢ ⎥ ⎢
0.6673917
⎥ utilized for checking the results (testing set, Table 7) were used. Looking
⎢ ⎥ ⎢ ⎥
θ = ⎢ θ2 ⎥ = ⎢ 0.4725325 ⎥ at Fig. 10 (a)-(d), the excellent coincidence of the FE results with
⎣ ⎦ ⎣ ⎦
θ3 103.31008339 calculated ones can be perceived, which indicates the acceptable accu­
racy of both methods. However, the predicted responses with RSM are
θ4 8.2546917
more accurate in both groups of models as the data is distributed in an

12
M. Bypour et al. Engineering Structures 301 (2024) 117263

Fig. 14. Vmax versus tp for CS-CO models with fy = 100, 150, 200, 250 and 300 MPa: tc = 30 mm for (a)-(e) and tc = 20 mm for (f)-(j).

excellent coincidence with line y = x, while the data distribution in LR To be more precise about the effect of f ′c on the models with different
illustrates higher dispersion. characteristics, the maximum and the minimum values obtained for
each group of models are presented on graphs (see Fig. 11). Fig. 11(a)
5. Effect of variables on V max
shows, in models with fy = 100 MPa and tp = 2 mm, by increasing f ′c
from 20 MPa to 40 MPa, Vmax marginally varies between 2.72 and
Figs. 11 and 12 show the correlation between the maximum shear
3.28 MPa, which is a negligible value compared to the Vmax of approx­
capacity (Vmax ) and concrete thickness (tc ) in the analyzed models of
imately 500 MPa for all CS-CO models. The same is true for the other
phase I (CS-CO models) and phase II (CS-CB models), respectively. To
models. This variation in models with tp = 3 mm is almost double that of
avoid redundancy, only the results of the models with the infill plate
tp = 2 mm. Similarly, variations in Vmax in all the models with tc between
thicknesses of 2 and 3 mm are presented in this section. It is evident that
20 and 40 mm indicates an insignificant value. Fig. 13(a)-(b) illustrate
there is an increase in Vmax by increasing tc from 20 to 40 mm. In
this by a constant value of f ′c = 35 MPa and tp = 2 and 3 mm for fy =
addition, models with higher f ′c have shown greater Vmax .
100, 200, and 300 MPa. As shown in these figures, in all models by

13
M. Bypour et al. Engineering Structures 301 (2024) 117263

Fig. 15. Vmax versus tp for CS-CB models with fy = 100, 150, 200, 250 and 300 MPa: tc = 30 mm for (a)-(e) and tc = 20 mm for (f)-(j).

increasing the concrete thickness by 5 mm (e.g., tc from 20 to 25) Vmax in Fig. 13 (c)-(d), yields consistent results from CS-CO models. It is
increases by a minimum of 0.1% and a maximum of 0.4%, which are evident that, despite varying fy from 100 to 300 MPa and tc from 20 to
relatively small values. These results are indicative of the trivial impact 40 mm, the resulting ΔVmax remains relatively unchanged in the models.
of tc on Vmax in CS-CO models. Upon examining Fig. 12, a comparable For instance, in CS-CO models, when tc varies from 20 to 25 mm, the
trend to the CS-CO models reveals the minimal influence of f ′c on Vmax in corresponding ΔVmax changes by only 0.4% for fy = 100 MPa and 0.3%
CS-CB models. for fy = 200 and 300 MPa. These findings strongly indicate the insigni­
To evaluate the effect of tc on Vmax , the same methodology as the CS- ficance of fy on influencing the increase of Vmax .
CO models is employed, which is shown in Fig. 13 (c) and (d). These To evaluate the effect of tp on Vmax , the results of CS-CO and CS-CB
figures clearly illustrate that ΔVmax experiences a moderate increase, models with the concrete layer thickness (tc ) of 30 and 20 mm are
ranging between 3% and 10%, in CS-CB models. These values are sig­ presented in Figs. 14 and 15, respectively. It is evident from these figures
nificant when compared to models possessing similar characteristics to that, there is a linear relationship between Vmax and tp in both groups of
CS-CO. These outcomes imply that by placing concrete between two models. Additionally, while in CS-CO models, Vmax varies between 400
steel plates, the concrete contributes to the shear capacity of the spec­ and 750 MPa, the figure for the corresponding CS-CB models shows
imen significantly more than the corresponding specimens with concrete values from 500 to 900 MPa. This indicates that in CS-CB models,
layer on one side. In other words, CSPSWs with a concrete layer between confining concrete by steel infill plate enhances the performance of the
steel plates (CS-CB models) can lead to a higher Vmax . specimens and leads to a higher shear capacity. That is to say, in CS-CB
Moreover, the sensitivity analysis of CS-CB models to fy , as depicted models, the impact of tc on Vmax is more than CS-CO models with the

14
M. Bypour et al. Engineering Structures 301 (2024) 117263

Fig. 16. Comparison of the Vmax of CS-CO and CS-CB models by the variation of tc with constant fc′ = 35 MPa: (a)-(b): tp = 2 mm and, (c)-(d): tp = 3 mm.

same properties. These figures further confirm the marginal influence of procedure of both CS-CO and CS-CB models are represented in a flow­
f ′c on Vmax , as consistent results are obtained from five different values of chart, see Fig. 17. The flowchart can be a useful tool for estimating the
shear capacity of both types of the investigated CSPSWs, offering a
f ′c . This finding is in accordance with previous experimental studies [8,
straightforward method.
16,24], which have reported similar observations.
By considering the insignificant effect of tc and f ′c on CS-CO models, 7. Summary and conclusions
Fig. 14 can be used as a guideline for estimating Vmax of CSPSW speci­
mens with concrete on one side of the steel infill plate, with fy ranging This study aimed to predict the maximum shear capacity (Vmax ) of
from 100 to 300 MPa and tp between 1 and 3 mm. For instance, for Composite Steel Plate Shear Walls (CSPSWs). Nonlinear finite element
acquiring Vmax > 600 MPa, steel infill plate with fy > 150 MPa is (NLFE) analyses were conducted, which consisted of Phase I, models
required. with a concrete layer on one side of a steel infill plate, and Phase II,
Fig. 15 can be a useful criterion for selecting the characteristics of models with a concrete layer between two plates. The investigated pa­
CSPSW specimens with concrete between two steel plates. As these rameters included concrete thickness (tc ) and compressive strength (f ′c )
specimens are more sensitive to tc than the corresponding CS-CO
as well as steel infill plate thickness (tp ) and yield stress (fy ), with five
models; in addition to fy and tp , the parameter tc is of importance. For
different values for each parameter. The NLFE data was used to develop
example, for fy = 100 MPa and tp = 3 mm, Vmax is slightly below
regression equations and graphs by applying Response Surface Meth­
700 MPa for models with tc = 30 mm (see Fig. 15(a)), whereas the figure odology (RSM) and Linear Regression (LR). These equations and graphs
for tc = 20 mm is just over 600 MPa, (see Fig. 15(f)). allow for the prediction of Vmax based on the characteristics of the
specimens, including tp and fy for Phase I specimens and additionally tc
6. Comparison of the results of CS-CB and CS-CO models
for Phase II specimens. Consequently, these tools enable the selection of
optimal dimensions for the infill plate and concrete thicknesses, as well
Fig. 16 provides a comparison between CS-CO and CS-CB models in
as the yield strength of the infill plate, based on the desired Vmax .
terms of the relationship between Vmax and tc . In this regard, the results
Similarly, if the required Vmax is known, the graphs can be utilized to
of models with tp = 2 and 3 mm, as well as fy = 200 and 300 MPa,
determine the corresponding specifications of the components. The key
are presented by considering the constant value of f ′c = 35 MPa for all findings of this research can be concluded as follows:
the models. Comparing Vmax of the models in this figure indicates that in
similar characteristics, Vmax of CS-CB models are greater than that of CS- - Comparison of the Vmax values of FEM and RSM in both CS-CO and
CO in all thicknesses. Interestingly, by increasing the thickness of con­ CS-CB models indicated negligible differences between values ac­
crete, the difference between Vmax of two categories of models relatively quired by these two methods. This means that RSM can be used as a
increases. This outcome reaffirms the positive impact of placing con­ reliable method for predicting Vmax of the specimens accurately.
crete between the two steel plates as CS-CB models with the concrete - Employing LR method resulted in acquiring simpler equations with
layer thickness (tc ) of 40 mm, leads to an increase of Vmax by over 35% acceptable accuracy. However, comparing the results of the two
compared to the corresponding CS-CO. methodologies, RSM and LR, indicated that RSM is more precise than
In order to use the results obtained from RSM and LR, the design LR.

15
M. Bypour et al. Engineering Structures 301 (2024) 117263

Fig. 17. Design procedure using the drived equations from RSM and LR.

- The effect of f ′c on Vmax is negligible either in CS-CO or CS-CB models CRediT authorship contribution statement
as roughly equal results are obtained for Vmax by five different values
Kioumarsi Mahdi: Conceptualization, Data curation, Funding
of f ′c .
acquisition, Investigation, Methodology, Project administration, Soft­
- The results indicated that placing a concrete layer between the two
ware, Writing – original draft, Writing – review & editing. Yekrangnia
steel plates (CS-CB models) contributes considerably more to the
Mohammad: Data curation, Investigation, Software, Writing – review &
specimens’ shear capacity than the corresponding models with a
editing. Bypour Maryam: Data curation, Investigation, Methodology,
concrete layer on one side (CS-CO models). This means that CSPSWs
Software, Visualization, Writing – original draft, Writing – review &
with concrete between two steel plates can result in a greater Vmax .
editing, Conceptualization, Validation.
- The models of both phases show the same sensitivity to fy which is
indicative of the trivial effect of this parameter on Vmax .
Declaration of Competing Interest
- There is a linear correlation between Vmax and tp in both categories of
models.
The authors declare that they have no known competing financial
- Comparison of Vmax of CS-CB models with corresponding CS-CO ones
interests or personal relationships that could have appeared to influence
indicates that by increasing the thickness of the concrete layer, the
the work reported in this paper.
difference between Vmax of the models of two phases relatively in­
creases. Furthermore, this difference is over 35% for models with
Data availability
40 mm concrete thickness.
Data will be made available on request.

16
M. Bypour et al. Engineering Structures 301 (2024) 117263

References [22] Li X, Li X. Steel plates and concrete filled composite shear walls related nuclear
structural engineering: experimental study for out-of-plane cyclic loading. Nucl
Eng Des 2017;315:144–54.
[1] Choi I-R, Park H-G. Cyclic loading test for reinforced concrete frame with thin steel
[23] Wang Y, et al. Axial and hysteretic behavior of T-shaped steel–concrete composite
infill plate. J Struct Eng 2011;137(6):654–64.
shear walls. Structures. Elsevier,; 2022.
[2] Bypour M, et al. Nonlinear analysis to investigate effect of connection type on
[24] Todea V, et al. Experimental investigations on the seismic behavior of composite
behavior of steel plate shear wall in RC frame. Eng Struct 2019;179:611–24.
steel concrete coupled shear walls with central openings. Structures. Elsevier,;
[3] Bypour M, Kioumarsi B, Kioumarsi M. Investigation of failure mechanism of thin
2021.
steel plate shear wall in RC frame. Key Eng Mater 2019;803:314–21.
[25] Ma Z-b, et al. Seismic behavior of steel plate and reinforced concrete composite
[4] Tan J-K, et al. Interaction behaviour of buckling-restrained steel plate shear wall
wall coupled to steel side columns. J Build Eng 2023;65:105820.
and boundary composite frame. J Constr Steel Res 2022;191:107189.
[26] Zhang Y, et al. Material properties of ITRAC and cyclic behavior of double steel
[5] Bypour M, Kioumarsi M, Zucconi M. Effect of stiffeners on behavior of steel plate
plate composite shear wall filled with ITRAC. Constr Build Mater 2023;394:
shear wall with rectangular openings. AIP Conference Proceedings. AIP
131635.
Publishing,; 2020.
[27] Yang X, Xu L, Pan J. Experimental investigation on the seismic behavior of
[6] Sabouri-Ghomi S, Mamazizi S. Experimental investigation on stiffened steel plate
composite steel plate shear wall restrained by ECC panels. Eng Struct 2023;297:
shear walls with two rectangular openings. Thin-Walled Struct 2015;86:56–66.
116946.
[7] Bypour M, Kioumarsi M, Yekrangnia M. Shear capacity prediction of stiffened steel
[28] Zhang Z, et al. Seismic performance of grid tubular-double steel plate concrete
plate shear walls (SSPSW) with openings using response surface method. Eng Struct
composite shear wall. J Constr Steel Res 2022;189:107077.
2021;226:111340.
[29] Anvari AT, et al. Performance of composite plate shear walls/concrete filled (C-
[8] Arabzadeh A, Soltani M, Ayazi A. Experimental investigation of composite shear
PSW/CF) under fire loading: a numerical investigation. Eng Struct 2022;271:
walls under shear loadings. Thin-Walled Struct 2011;49(7):842–54.
114883.
[9] Hu H-S, Nie J-G, Eatherton MR. Deformation capacity of concrete-filled steel plate
[30] Feng X, Yu J, Shen J. Seismic behavior of composite steel plate shear walls with
composite shear walls. J Constr Steel Res 2014;103:148–58.
rubber-coated uplift-restrained studs. J Constr Steel Res 2021;182:106683.
[10] Qi Y, Gu Q, Wang H. Study on the shear mechanism and resistance of single-side
[31] Shafaei S, Ayazi A, Farahbod F. The effect of concrete panel thickness upon
concrete-encased composite plate shear wall. Thin-Walled Struct 2022;180:
composite steel plate shear walls. J Constr Steel Res 2016;117:81–90.
109876.
[32] Wang D, et al. Study on seismic behaviors of steel–concrete composite shear walls
[11] Ke X-j, Li N, Tang Z-k. Seismic behavior of steel plate concrete composite shear
with novel corner designs. J Build Eng 2023;70:106339.
walls with PEC columns and WRGs. Eng Struct 2023;277:115412.
[33] Abaqus -6.14, Analysis users’s guide. 2013.
[12] Wang W, et al. Experimental study of the hysteretic behaviour of corrugated steel
[34] Kioumarsi MM, Hendriks M, Geiker MR. Quantification of the interference of
plate shear walls and steel plate reinforced concrete composite shear walls.
localised corrosion on adjacent reinforcement bars in a concrete beam in bending.
J Constr Steel Res 2019;160:136–52.
Nord Concr Res (NCR) 2014;49:39–57.
[13] Varma AH, Shafaei S, Klemencic R. Steel modules of composite plate shear walls:
[35] Tan J-K, et al. Shear capacity prediction of buckling-restrained SPSWs with beam-
behavior, stability, and design. Thin-Walled Struct 2019;145:106384.
connected only using response surface method. Structures. Elsevier,; 2023.
[14] AISC341 -16, seismic provisions for structural steel buildings. 2016.
[36] Xiang Z, Zhu Z. Multi-objective optimization of a composite orthotropic bridge
[15] Epackachi S, Whittaker AS, Aref A. Seismic analysis and design of steel-plate
with RSM and NSGA-II algorithm. J Constr Steel Res 2022;188:106938.
concrete composite shear wall piers. Eng Struct 2017;133:105–23.
[37] Lye HL, et al. Bond behaviour of CFRP-strengthened ECC using response surface
[16] He W, et al. Experimental study on seismic behaviors of the welded L-shaped
methodology (RSM). Case Stud Constr Mater 2020;12:e00327.
double steel plate-concrete composite shear wall. J Constr Steel Res 2021;187:
[38] Ho, C. et al., Prediction of concrete residual compressive strength under elevated
106944.
temperatures: response surface methodology (RSM) approach. Materials Today:
[17] Meghdadian M, et al. Proposition of an equivalent reduced thickness for composite
Proceedings, 2023.
steel plate shear walls containing an opening. J Constr Steel Res 2020;168:105985.
[39] Myers RH, Montgomery DC, Anderson-Cook CM. Response Surface Methodology:
[18] Wang X, Gong J, Sun Y. A plastic-based model for in-plane shear strength of steel
Process and Product Optimization Using Designed Experiments. John Wiley &
plate-concrete shear walls. J Constr Steel Res 2023;200:107676.
Sons,; 2016.
[19] Asgarpoor M, et al. Nonlinear modeling for composite plate shear walls-concrete
[40] Hooshmandi S, et al. Application of response surface method (RSM) on sensitivity
filled structures. J Build Eng 2023;63:105383.
analysis of reinforced concrete bridge pier wall. in Nordic concrete research.
[20] Li J, et al. Axial compressive behavior of corroded double-steel-plate composite
Proceedings of the XXIII Nordic Concrete Research Symposium. Norsk
shear wall with binding bars. J Build Eng 2023;79:107767.
Betongforening,; 2017.
[21] Seo J, et al. Steel-plate composite (SC) walls: in-plane shear behavior, database,
[41] Kallias A, Rafiq MI. Performance assessment of corroding RC beams using response
and design. J Constr Steel Res 2016;119:202–15.
surface methodology. Eng Struct 2013;49:671–85.

17

You might also like