Beruflich Dokumente
Kultur Dokumente
Heriot-Watt University
School of Energy, Geoscience, Infrastructure and Society
December 2015
The copyright in this thesis is owned by the author. Any quotation from the dissertation
or use of any of the information contained in it must acknowledge this dissertation as the
source of the quotation or information.
i
ProQuest Number: 10692994
All rights reserved
INFORMATION TO ALL USERS
The quality of this reproduction is dependent upon the quality of the copy submitted.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.
ProQuest 10692994
Published by ProQuest LLC (2017 ). Copyright of the Dissertation is held by the Author.
All rights reserved.
This work is protected against unauthorized copying under Title 17, United States Code
Microform Edition © ProQuest LLC.
ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, MI 48106 - 1346
DECLARATION STATEMENT
I Nabil Shouki Attari, confirm that this work submitted for assessment is my own and is
expressed in my own words. Any uses made within it of the works of other authors in any
form (e.g. ideas, equations, figures, text, tables, programmes) are properly acknowledged
at the point of their use. A full list of the references employed has been included.
ii
TABLE OF CONTENTS
iii
LISTS OF FIGURES AND TABLES
Chapter 1 : Introduction...................................................................................................................... 1
Fig. 1.1 Different types of shear wall system ...................................................................................... 2
Fig. 1.2 Different types of frame system ............................................................................................. 2
Fig. 1.3 Dual System .......................................................................................................................... 3
Fig. 1.4 Stiffened and Un-stiffened Steel Plate Shear Wall Systems .................................................... 4
Fig. 1.5 L.A. Live Hotel and Residences,An Innovative SPSW Solution (ASIC 2008)......................... 4
Chapter 2 : Literature review .............................................................................................................. 8
Fig. 2.1 Stress-strain curves for brittle and ductile materials (en.wikipedia.org) ................................... 8
Fig. 2.2 Ductile performance and Brittle collapse (Murty n. d.) ........................................................... 9
Fig. 2.3 Experimental load-displacement curves for stiff shear wall. (Bekő and Roško 2013) .............11
Fig. 2.4 Comparison of buckling behaviour of elements under compression. (Batikha 2008) ..............12
Fig. 2.5 Plate buckling coefficient kc compression, ks shear and kb bending. (Pollock 1993)..............13
Fig. 2.6 Plate buckling coefficients for plates in pure shear. (Ziemian 2010).......................................14
Fig. 2.7 Shear and tension fields for square plate. (Ziemian 2010) ......................................................15
Fig. 2.8 Plate buckling patterns and tension fields shape. ...................................................................16
Fig. 2.9 Shear behaviour of slender plate. (Alinia et al. 2009).............................................................17
Fig. 2.10 Shear behaviour of slender plates. (Alinia et al. 2009). ........................................................18
Fig. 2.11 Shear behaviour of moderate thick plates. (Gheitasi 2009)...................................................18
Fig. 2.12 Idealized tension field action in a typical steel plate (Sabelli and Bruneau 2007).. ................19
Fig. 2.13 Two proposed strut models: (a) 45-degree angle. (b) Multi-angles. (Astaneh-Asl. 2000). .....20
Fig. 2.14 Tension strip model: (a) Single-directional (Shishkin et al. 2009), (b) Double-directional.. ..21
Fig. 2.15 Combined strip model. (Guo et al. 2012).............................................................................22
Fig. 2.16 An orthotropic membrane model of a steel plate. (Sabelli and Bruneau 2007). .....................22
Fig. 2.17 Finite-element model of test specimen: Deflected shape. (Driver et al. 1998). ......................24
Fig. 2.18 Comparison of finite-element model and test results: Panel 1. (Driver et al. 1998). ..............25
Fig. 2.19 Plane frame strip model of test specimen. (Driver et al. 1998) .............................................25
Fig. 2.20 Comparison of strip model analysis with test results. (Driver et al. 1998).............................26
Fig. 2.21 Four-story SPSW Frame: (a) Specimen by Driver et al. (1997). (b) Detailed model .............27
Fig. 2.22 Comparison of response curves for test specimen, detailed model, modified strip model, and
basic strip model(Shishkin et al. 2009)..................................................................................29
Fig. 2.23 Comparison of response curves of shell model and double-directional tension strip model ...30
Fig. 2.24 Comparison of hysteretic curves for various modelling approaches. (Guo et al. 2012)..........32
Fig. 2.25 Comparison between analytical results and experimental results. (Guo et al. 2012 ...............33
Fig. 2.26 Proposed equivalent braced model. (Chatterjee et al.2015). .................................................34
Fig. 2.27 Comparison of response curves for SPSW model and an equivalent braced model. ..............34
Chapter 3 : Finite element analysis procedure ...................................................................................36
Fig. 3.1 Steel plate models and boundary conditions. .........................................................................37
Fig. 3.2 Assumed stress-strain relationship of mild steel. (Alina and Dastfan 2007) ............................37
Fig. 3.3 Abaqus “S4” element. (ABAQUS 2014) ...............................................................................37
Fig. 3.4 Time history of drift applied to the verified model. (Alina and Dastfan 2007) ........................38
Fig. 3.5 Three different patterns for plate mesh ..................................................................................38
iv
Fig. 3.6 Mesh convergence for the verified model..............................................................................39
Fig. 3.7 Shear force-displacement curves of the plate under cyclic loads. ...........................................40
Fig. 3.8 Shear force-displacement curves for the plate. (Alina and Dastfan 2007) ...............................40
Fig. 3.9 Envelopes of hysterical curves for the verified model, and Alina and Dastfan (2007) model. .41
Fig. 3.10 Maximum in-plane principal stress on deformed shape at end of various loading cycles .......41
Fig. 3.11 Shear force-displacement curves of the verified model under cyclic and pushover loads…..42
Fig. 3.12 Steel plate models and boundary conditions. .......................................................................43
Fig. 3.13 Shear force-displacement curves of the modified model under pushover loading..................44
Fig. 3.14 Out-of-plane deformations at the plate center against the lateral displacement......................44
Fig. 3.15 The behaviour of modified model under pushover loading...................................................45
Fig. 3.16 Varying the tension fields number through the main analysis steps. .....................................47
Fig. 3.17 Abaqus “T3D2” element. (ABAQUS 2014) ........................................................................47
Fig. 3.18 Abaqus “MPC type beam” link. (ABAQUS 2014) ..............................................................48
Fig. 3.19 Truss-elements models. ......................................................................................................49
Fig. 3.20 Comparing the shear force-displacement curves of various truss-elements models under
pushover loading. .................................................................................................................50
Fig. 3.21 Relationship between the yielding force and the width-to-diagonal length “w/d” ratio for
models group. ......................................................................................................................52
Fig. 3.22 Comparing the shear force-displacement curves of 1-Truss, 3-Trusses, 5-Trusses Models and
Plate Model..........................................................................................................................53
v
ACKNOWLEDGMENTS
I am deeply grateful to Dr. Mustafa Batikha for his suggestions, encouraging help and
support in the preparation of this dissertation.
vi
DEDICATION
I would like to acknowledge my wife, for her seemingly infinite patience and support.
vii
ABSTRACT
Since 1970’s, steel plates have been used as shear walls participating the frames in
resisting lateral loads due to wind and earthquakes, referred to as Steel Plate Shear Walls
(SPSWs). Modelling of SPSW is required a high level software which engineers are not
familiar to use, as well long time for analysis is needed. Therefore, an equivalent
analytical methods had been investigated and implemented to simplify the analysis
process by means of conventional analysis software, such as: using strip modelling and
equivalent bracing (truss) elements to replace the infill plate.
It was found that the truss-elements model required an odd number of trusses (1, 3 or 5),
with one truss connecting the two opposite corners (diagonal truss), as well the required
truss width was slightly more than the truss spacing in the model. On the other hand, the
initial stiffness of the truss-elements models was generally lower than that at the plate
model.
viii
GLOSSARY OF SYMBOLS
a Length of plate
b Width of plate
t Thickness of plate
E Elastic modulus
ν Poisson’s ratio
σc Buckling stress of plates (elastic critical stress)
τc Shear buckling stress of plates
σVM “Von Mises” stress
σ1 Principal stress in 1 direction
σ2 Principal stress in 2 direction
σ3 principal stress in 3 direction
k Buckling coefficient for flat, rectangular plates
kc Compression buckling coefficient for flat, rectangular plates
ks Shear buckling coefficient for flat, rectangular plates
kb Bending buckling coefficient for flat, rectangular plates
α Aspect ratio for flat, rectangular plate
α Angle of inclination in strip model
Ic Moment of inertia of the column in strip model
Ab Cross-sectional areas of the beams in strip model
Ac Cross-sectional areas of the columns in strip model
L Length of the frame in strip model, measured from the center lines
h Height of the frame in strip model, measured from the center lines
λ Panel height-to-thickness ratio
β Panel aspect ratio (length-to-height ratio)
d diagonal length of square shape (plate)
n number of trusses in model.
ix
Chapter 1 : Introduction
1
a) Reinforced Concrete Shear Wall b) Steel Plate Shear Wall
Fig. 1.1 Different types of shear wall system
2) Moment-resisting frame system: A frame in which members and joints are capable
of resisting forces by flexure in the members and joints. Joints are designed and
constructed to be completely rigid, therefore any lateral deformation of the frame occurs
from the bending of beams and columns. There are three types of moment resisting frames
according to its ductility which depends on joint detail; the first type is the special
moment-resisting frame (SMRF), which must be specifically detailed to provide ductile
behaviour. The second type is the intermediate moment-resisting frame (IMRF), has less
restrictive detailing requirements than special moment-resisting frame. The last type is
ordinary moment-resisting frame (OMRF), which does not meet the special detailing
requirements for ductile behaviour.
3) Dual system: A system has complete space frames that resist the vertical loads and
combine moment-resisting frames acting in conjunction with shear walls to resist lateral
loads. This frames must resist at least 25% of the base shear independently (UBC 1997).
The design base shear is distributed between two systems in proportion to their relative
rigidities.
2
Figure 1.3 Dual System
Since 1970’s, steel plates have been used as shear walls participating the frames in
resisting lateral loads, referred to as Steel Plate Shear Walls (SPSWs). SPSWs system
composes of vertical steel plate (infill plate) surrounded by Vertical and Horizontal
Boundary Elements (VBE and HBE, Fig.1.1.b) over the full height of frame in one or
more bays. There are several variables for infill plate such as: thin and thick, stiffened
and unstiffened (Fig.1.4), perforated and non-perforated, pinned and fixed boundary
conditions, rigid and shear beam-to-column connections, thickness to width ratio and
several others.
3
a) Stiffened Steel Plate Shear Walls b) Un-stiffened Steel Plate Shear Walls
Figure 1.4 Stiffened and Un-stiffened Steel Plate Shear Wall Systems
SPSWs are now the primary lateral load resisting system, even for high-rise building,
used in Japan, Canada and United States. For instance, L.A. Live Hotel and Residences
(54-story) in Los Angeles, opened in 2010, has an innovative SPSW solution by replacing
heavy concrete shear walls (762 mm thick) with light SPSWs (6.35 to 9.525 mm thick)
and free valuable real estate space, reduce seismic design force and foundation size by
eliminating 35% of the weight of the structure, compress the construction schedule and
budget, and allow for simplified and more efficient construction (ASIC 2008).
Figure 1.5 L.A. Live Hotel and Residences,An Innovative SPSW Solution (ASIC 2008)
4
2) Reduce building weight: Building with SPSWs has much lighter weight comparing
to building with concrete shear walls. For instance, use SPSWs system at L.A. Live
project eliminating 35% of the weight of the structure, thus reduce seismic design
force and foundation size (ASIC 2008).
3) Fast Construction: Using of SPSW system reduces construction time,not only fast
erection, but also no need curing period.In additional, using field-bolted,shop-welded
SPSWs can speed-up the erection process and reduce the construction cost, field
inspection and quality controal making these systems even more efficeint.
4) Increase ductility: The properly designed and detailed SPSWs system is very ductile,
and has excellent post-buckling capacity and relatively large energy dissipation
capability. As a result, SPSWs system can be very economical and efficient lateral
load resisting system.
5) Limiting drift : The SPSWs system has relatively high initial stiffness,thus very
effective in limiting the drift; the system can survive up to 4% drift without
experiencing significant damage.
6) Easier retrofit: Using SPSWs system in seismic retrofitting for existing buildings can
be much easier and faster to construct than using concrete shear walls, which will
increase total building weight and overall seismic loads, and require additional mass
foundation.
7) Tested system: At least two buildings include SPSWs system have undergone
significant earthquake, both buildings survived with insignificant structural damage.
The system has been recognized in Seismic Provisions of American Institute of Steel
Construction (AISC) in 2005 and in National Building Code of Canada (NBCC) since
1994 (Seilie and Hooper 2005).
On the other hand, the disadvantages of using SPSW system to resist lateral loads are:
(Seilie and Hooper 2005)
1) Stiffness: SPSW systems are stiff, but usually more flexible than reinforced concrete
shear walls, primarily due to their flexural flexibility. The designer must provide
additional flexural stiffness, especially in tall buildings.
2) Construction sequence: Excessive initial compressive load on steel plate may delay
the development of tension-field action. Therefore, the construction sequence must
be designed properly to avoid pre-compression of SPSW due to dead load.
5
3) Unfamiliarity: Currently, relative unfamiliarity with SPSW system might result in
higher costs for fabrication and erection. Involving the contractor in the design phase
may help.
The study is based on the pushover analysis method using ABAQUS (ver.6.14) software,
to obtain a good prediction of the inelastic behaviour of unstiffened thin SPSW. The study
6
is limited to the analysis of frameless unstiffened SPSW with relatively thin infill plate
without openings.
7
Chapter 2: Literature review
2.1 Introduction
This chapter describes theoretically the structural behaviour of steel plates under
compression and shear loads. The focus is on the shear behaviour of steel plates, buckling
stress, types of steel plates in terms of thickness, developed tension fields and equivalent
strip model.
For consistency with the objectives of the current research, this chapter provides also a
selected literature survey for past analytical studies about the various modelling
approaches of unstiffened thin SPSW in analysis.
Fig. 2.1 Stress-strain curves for brittle and ductile materials (en.wikipedia.org)
8
Depending on that classification, ductile or brittle, two failure modes can be defined; the
first mode is ductile failure, referred as ductile performance, associated with extensive
plastic deformations ahead of cracking, then stable cracks resist further extension unless
applied force is increased; the second mode is brittle failure associated with relatively
little plastic deformations, then unstable crack propagates rapidly without increase in
applied force (Zhigilei 2010).
Fig. 2.2 illustrates the ductile performance of a building which is associated with large
horizontal displacements (lateral sway), as illustrated by the green curve, while non-
ductile behaviour is associated with smaller horizontal displacements, and causes a brittle
damage or collapse to the building or its components, as illustrated by the red curve.
9
have flexural behaviour similar to a cantilever beam, while short shear walls have shear
behaviour (Bekő and Roško 2013).
RCSWs have various failure modes can occur depending on different parameters such as
boundary conditions, properties of reinforcing steel, concrete compressive strength,
reinforcement detailing and quantities, and the geometry of cross-section. The main
failure modes for RCSWs are flexure, shear, crushing, sliding and rocking; only flexure
and rocking are truly ductile where the others are not. RCSWs have also complex
destructive phenomena starting with interaction between steel and concrete, then concrete
cracking, through steel yielding and concrete crushing in compression (Bekő and Roško
2013).
An experimental test on stiff shear wall under quasi-static cyclic loading, shows an
interaction between two phenomena: sliding across the shear cracks and yielding of the
reinforcement until collapse, both mechanisms contribute to the plastic behaviour of the
wall. The features of these phenomena can be observable in the load-displacement curves
of this test at Fig. 2.3. This figure shows that the resistance will increases until the first
crack occurs at load of (3000 KN) associated with negligible displacement, followed by
steel yielding at (7200 KN) load and (6 mm) displacement, and then the collapse at
(8000KN) load and (16 mm) displacement. In other word, the stiff shear wall starts
cracking when 37% of failure load is reached, then cracking will continue progressively
with more displacement until the steel starts yielding at 90% of failure load, finally the
resistance will increase slightly (10% more) followed by rapid collapse associated with
dramatically decreasing in the resistance and increasing in the displacement.
One more phenomenon can be seen also at Fig. 2.3, a pinching around the origin which
is due to sliding between the already cracked surfaces of concrete before they come in
full contact. The collapse for this type of shear wall was due to concrete crushing in the
middle of the wall (Bekő and Roško 2013).
10
Fig. 2.3 Experimental load-displacement curves for stiff shear wall. (Bekő and Roško 2013)
Fig. 2.4 illustrates the fundamental differences in buckling behaviour under compression
for three different elements: column, plate and cylindrical thin plate. This figure presents
a curve of the relationship between two ratios for each member; the first one is
dimensionless strain on horizontal axis (a ratio of member out-of-plane strain to classical
“critical” bucking strain); the other is dimensionless stress on vertical axis (a ratio of
member stress to classical “critical” buckling stress).
11
Fig. 2.4 Comparison of buckling behaviour of elements under compression. (Batikha 2008)
Noticeably, the behaviour of all members is linear until reaching the classical buckling
stress and strain (ratio of 1 in Fig. 2.4), then the behaviour of each member will be totally
different after buckling. Where the column will keep the resistance constant, but with
large deformations; while the cylinder shell will be subjected to sharp decline in the
resistance till 50% less (ratio of 0.5 in Fig. 2.4); whilst the plate resistance will increase
slightly after buckling due to redistribution of stresses at areas where buckling did not
occurred (post buckling).
(Eq. 2.1)
12
ks
kc (b) a/b
kb
Figure (2.5.a) shows that the compression buckling coefficient kc for a plate with loaded,
clamped edges is at least equal or higher than that for plate with loaded, simply supported
edges, thus higher elastic critical stress (higher strength). For instance, kc for plate has
simply supported edges (type C on Figure) and a/b ratio of 4, is equal to 4.
Figure (2.5.b) shows that the shear buckling coefficient ks for a plate with clamped edges
is higher than that for plate with hinged edges, thus higher elastic critical stress (higher
strength). For instance, k s for plate has a/b ratio of 4, is equal to 5.6 for hinged edges, and
9.6 for clamped edges. However, shear buckling coefficient ks can be calculated based on
empirical formulas (Eq. 2.2 to Eq. 2.5) according to aspect ratio (α =a/b) and conditions
of support edge.
Figure (2.5.c) shows that the bending buckling coefficient kb according to plate aspect
ratio and the amount of edge rotational constraint. Whenever the amount of edge
rotational constraint is incremented, then the kb value will increase, thus higher elastic
critical stress (higher strength). For instance, kb for plate has simply supported edges
(amount of edge rotational constraint “ε” is zero) and a/b ratio of 4, is equal to 30.
13
Thus, it can be concluded, for a plate with simply supported edges, that the compression
buckling coefficient kc has the smallest value among the buckling coefficients, while the
bending buckling coefficient k b has the highest value, and the shear buckling coefficient
ks has value in between the both.
When a rectangular plate is subjected to pure shear stresses as shown in Fig.2.6.b, the
buckling coefficient ks for this type of loading has been evaluated for four different
conditions of edge support, and plotted with respect to b/a (1/α) ratio in Fig.2.6.a. This
figure demonstrates that the buckling coefficient k s has the highest value in case of four
clamped edges, and the lowest value in case of four simply supported edges; while it has
some in-between values for the other two mixed edge support cases. For instance, ks value
for a plate has b/a (1/α) ratio of 0.8, is equal to 12.8 as maximum for four clamped edges,
and 7.9 as minimum for four simply supported edges.
Fig. 2.6 Plate buckling coefficients for plates in pure shear. (Ziemian 2010)
14
The following formulas for ks can be used according to conditions of edge support, where
a: length of plate, b: width of plate, α: aspect ratio equals to a/b (Ziemian 2010).
(Eq. 2.2.a)
(Eq. 2.2.b)
(Eq. 2.3.a)
(Eq. 2.3.b)
3) Plate clamped on long edges and simply supported on the other two edges:
(Eq. 2.4.a)
(Eq. 2.4.b)
4) Plate clamped on short edges and simply supported on the other two edges:
(Eq. 2.5.a)
(Eq. 2.5.b)
If a rectangular thin plate is subjected to lateral shear load, gradually increased from zero
to a magnitude beyond the plate's capacity, and when the applied load is relatively low,
the infill plate experiences elastic shear deformations, and both principal tension and
compression stresses will develop, equal in magnitude to the shear stress and inclined at
45°. (Fig.2.7.a)
Fig. 2.7 Shear and tension fields for square plate. (Ziemian 2010)
15
With the increase of applied load, principal compression stress exceeds the elastic critical
buckling stress, hence the infill plate will buckle (inelastic buckling stage) and tension
fields will develop (Fig.2.7.b). During this stage, the plate behaves geometrically
nonlinear and a big loss of stiffness occurs.
After development of diagonal tension field , infill plate can carry additional loading with
more deformations (post buckling stage), known as post-buckling deformations which
may be either elastic or inelastic; These post-bucking deformations continue until
localized yield points occur within the plate. The infill plate responses in both material
and geometrical nonlinear and large post-buckling deformations develop.
With the increase of loading, yield points spread out to form diagonal yield zones at the
tension field and an effective width will arise for each that tension field (Fig.2.7.c); when
the “Von Mises” stress along tension field reach the yielding stress, this tension field will
be totally yielded (plasticity state). Stiffness is almost constant until the formation of yield
zones, followed by a noticeable fall (Habashi and Alinia 2010).
The “Von Mises” stress is an equivalent stress defined on the basis of the Von Mises yield
criterion, used to convert the principal stresses to one equivalent stress in one direction,
could be adopted for comparison against material's yield stress. In case of principal plane
stress (where σ3=0), “Von Mises” stress (σVM) is expressed in the following equation;
(Case et al. 1999)
(Eq. 2.6)
Where σ1 is the principal stress in 1 direction, and σ2 is the principal stress in 2 direction.
According to the plate geometry (ratio of width to thickness b/t), one tension field could
be developed at the plate (Fig.2.8.a), or multi tension fields can be formed (Fig.2.8.b).
Whenever the number of formed tension fields is more, the plate strength will increase
accordingly.
a. b.
1) Slender plate behaviour: Fig. 2.9 demonstrates a typical shear behaviour of slender
plate, in which four obvious stages can be noticed through the typical relationship
between applied shear (on vertical axis) and out-of-plane deflection (on horizontal axis).
The curve in Fig. 2.9 shows that the stiffness will increase linearly without any out-of-
plane deformations, until reaches the elastic buckling load; then it starts increasing non-
linearly due to buckling and developing of diagonal zone “tension field”, until reaches
the ultimate load after diagonal zone yielding. Then after, a gradual decrease in stiffness
due to softening will occur with large out-of-plane deflection until failure. Accordingly,
slender plates have a low buckling capacity, followed by a large post-buckling reserve.
2) Stocky plate behaviour: Fig. 2.10 demonstrates a typical shear behaviour of stocky
plate, in which three obvious stages can be noticed through the typical load-deflection
curve. The curve in Fig. 2.10 shows that the stiffness will increase linearly without any
out-of-plane deformations, until material yielded; then it increases slightly as non-linear
17
shape due to plastic buckling, which occurs under constant shear load; after that a gradual
decrease in stiffness due to softening will occur with large out-of-plane deflection until
failure. Accordingly, stocky plate yields without elastic buckling, and it has some post-
yield capacity.
3) Moderate thick plate behaviour: Fig. 2.11 demonstrates a typical shear behaviour of
moderate thick plate, in which two obvious stages can be noticed through the typical
relationship between ratio of applied shear to material shear limit V/Vp (on vertical axis)
and maximum out-of-plane deflection (on horizontal axis). The curve in Fig. 2.11 shows
that the stiffness will increase linearly without any out-of-plane deformations, until reach
the material proportional limit, where concurrent material yielding and geometrical
buckling occur; and then a sudden loss of stiffness due to softening occurs with large out-
of-plane deflection until failure. Accordingly, moderate thick plate neither has post-yield
nor post-buckling reserves.
Fig. 2.12 Idealized tension field action in a typical steel plate (Sabelli and Bruneau 2007).
(Eq. 2.7)
19
Where t is the thickness of the steel plate, Ic moment of inertia of the column, Ab and Ac
cross -sectional areas of the beams and columns respectively, L and h are length and
height of the frame respectively, measured from the center lines as shown in Fig. 2.14.a.
(Shishkin et al. 2009).
Eq. (2.7) is determined by taking into account the axial energy of the beams and the
columns, and the energy distributed through the tension fields developed on the steel
plate. The beams are assumed to be rigid during loading, and the columns to be in their
full actual stiffness and should not form plastic hinges. Gravity load is carried out by the
columns and has no effect on the performance of the infill plate.
Fig. 2.13 Two proposed strut models: (a) 45-degree angle. (b) Multi-angles. (Astaneh-Asl. 2000)
At first configuration (Fig.2.13.a), the steel plate is modeled as parallel, uniformly spaced,
tension-only strips pinned at both ends. The modulus of elasticity of the strips (struts) is
set equal to that of steel, and the area of each strip is set equal to the steel plate thickness
multiplied by the distance between the strips; this distance is measured along a direction
perpendicular to that of the strips (Sabelli and Bruneau 2007).
The tension strip model can adequately predict the initial, pre-yielding stiffness of a steel
plate and the forces in frame members under service loads. However, this method can be
also used in nonlinear “pushover” analysis to obtain the full force-displacement
relationship for the steel plate and the ultimate forces in the frame members. In addition,
this method has successfully been used to reasonably predict monotonic force-
20
displacement relationships of full SPSWs system, as well as the individual one for one
story. (Sabelli and Bruneau 2007)
The tension strip model has also shown to be capable of successfully predicting the quasi-
static cyclic response of SPSWs. This kind of analysis requires models that have a
symmetric layout of strip elements to account tension field action in both loading
direction. (Fig. 2.14.b)
(b)
(a)
Fig. 2.14 Tension strip model: (a) Single-directional (Shishkin et al. 2009), (b) Double-directional. (Guo
et al. 2012)
Besides, this method is recognized by some international design codes such as: Canadian
design provisions for SPSW (CSA, 2001) and the Commentary to AISC 341. The CSA
provisions require that a minimum of 10 strips be used, in order to approximate the effects
of a distributed load on the surrounding boundary elements (Sabelli and Bruneau 2007).
21
Fig. 2.15 Combined strip model. (Guo et al. 2012)
Figure 2.15 illustrates such model studied by Guo et al. (2012), where the solid lines have
the same definitions as the tension-only strips; the dashed lines denote the tension-
compression stripes.
Fig. 2.16 An orthotropic membrane model of a steel plate. (Sabelli and Bruneau 2007)
22
Nonlinear finite-element analysis
Nonlinear analysis is the most realistic method to model the steel plate, but it depends on
finite element analysis for nonlinear membrane elements, which are not available in the
common used structural analysis softwares. On the other hand, this analysis is much more
complicated comparing to that at the previous methods.
Most of the past researches show that the behaviour of steel plate can be adequately
predicated by conducting an inelastic finite element analysis, but the accurate estimates
of response quantities can be obtained only when steel plate are modeled using large
number of shell elements, such model shown at figures 2.8 to 2.11, capable of realistically
accounting for geometric and material nonlinearities. These limitations lead to
sophisticated, time-consuming models that, while suitable for research purposes, are
generally not appropriate for practical applications (Sabelli and Bruneau 2007).
Several researches and studies have been carried out since the early of 1970s, to develop
and test various SPSW modeling approaches as described in section 2.6. A selective
literature survey has been conducted to present the most related and integrated studies,
which cover most of the SPSW modeling approaches with compassions;
23
The buckle configurations are consistent in number and orientation with those observed
in the physical test.
Fig. 2.17 Finite-element model of test specimen: Deflected shape. (Driver et al. 1998)
24
Fig. 2.18 Comparison of finite-element model and test results: Panel 1. (Driver et al. 1998)
2) Strip model
Driver et al. (1998) afforded the first opportunity to evaluate the strip model method by
comparing it to a large-scale test employing a multi-storey configuration. The model
consists of series of discrete, pin-ended strips (ten strips) at each panel, inclined with the
same orientation as the tension field (angle of inclination α =45°). The strips are assigned
an area equal to the plate thickness multiplied by the width of the strip. However, the
model configuration assumes that the compression in the orthogonal direction is
negligible and that angle of inclination can be predicted reasonably well. Fig. 2.19 shows
the strip model used for the analysis of test specimen.
Fig. 2.19 Plane frame strip model of test specimen. (Driver et al. 1998)
25
Results of strip model analysis
Fig. 2.20 demonstrates a comparison among various analysis results under monotonic
loading against the test results for panel 1 (lowest one) under cyclic loading; the analysis
results are for two strip models with different values (42° and 50°) for angle of inclination,
and for the frame only.
It can be noticed from Fig. 2.20 that the maximum story shear carried out by the model
of frame only is about 1,040 KN at 60 mm story deflection; whilst it goes up to 2,960 KN
at about 30 mm story deflection (i.e. with relative increment of 1.85 times as well half
value of the story deflection) since the infill panels had been added. This demonstrates
the significant contribution of the infill panels in resisting the story shears and increasing
the stiffness of the structure.
Also, the figure illustrates that the results of two strip models gave essentially the same
response for both angles of inclination (42° and 50°). Furthermore, these results were
identical with those of physical test.
Fig. 2.20 Comparison of strip model analysis with test results. (Driver et al. 1998)
Based on the previous comparison of strip model analysis with physical test results, the
strip model represents well the load versus deflection envelope behaviour (i.e. initial
stiffness) of the test specimen, and in particular its ultimate strength.
26
2.8 Analysis of SPSWs using the modified strip model studied by (Shishkin et al.
2009)
Shishkin et al. (2009) suggested refinements to the strip model proposed by Thorburn et
al. (1983) to achieve an accurate representation of yielding and eventual deterioration of
the system. A modified version of the strip model was proposed, that was shown to be
efficient to generate while maintaining a high degree of accuracy. The parameters of the
modified strip model were generic and could be implemented in any structural analysis
program with pushover analysis capabilities. However, modelling efficiency was
evaluated against the accuracy of the solution.
The study had been conducted on two stages. The first was to define the potential
modelling improvements based on a rational approach as well as phenomena observed
during laboratory testing, then provide and examine a model (called the detailed model)
that captures all of these key improvements. The subsequent stage was to simplify the
detailed model while retaining good accuracy, thus the simpler model (called the modified
strip model) would be more desirable for use as a design tool.
Detailed model
The detailed model has been developed to represent a four-story SPSW test specimen
tested by Driver et al. (1997), same as the referred specimen at previous described
investigation. The geometry arrangement of the test specimen is shown in Fig. (2.21.a).
The key features of the experimental test, which will be implemented at the detail model
as refinements to the strip model, are special frame-joint arrangement for moment-
resisting connections, flexural plastic hinges, a pin-ended compression strut, axial hinges
include the effects of deterioration, and pushover analysis includes the consideration of
P-Δ effects. (Fig. 2.21.b)
Fig. 2.21 Four-story SPSW Frame: (a) Specimen by Driver et al. (1997). (b) Detailed model. (Shishkin et
al. 2009)
27
Modified strip model
If the detailed model could be simplified while retaining good accuracy, then the simpler
model would be more desirable for use as a design tool. To do so, several parameters of
the detailed model are examined more closely and simplified. The simplified parameters
that result in a more efficient model to generate, while not adversely affecting the
accuracy of the model significantly, are retained. The model that has the best balance
between accuracy and modelling efficiency is called modified strip model. These
simplified parameters are relocating the flexural plastic hinges to the panel nodes, reduce
the number of required nodes, and modelling the plastic hinge as rigid-perfectly plastic.
It can be obviously observed from Fig. 2.22, that all response curves are identical at the
elastic portion till a base shear of 1,540 KN at 4 mm displacement (i.e. up to 50% of the
measured ultimate strength “3,080 KN”), then the basic strip model underestimates the
initial stiffness (taken up to 60% of the measured ultimate strength “3,080 KN”) by 10%,
after which the discrepancy gradually increases as well. It also underestimates the
measured ultimate strength (3,080 KN at 43mm) of the test specimen considerably by
10% (2,770 KN at 70 mm).
On the other hand, the detailed model overestimates the measured strength very slightly
at a point just beyond the occurrence of initial yielding, then predicts a peak strength of
2,990 KN at almost the same displacement as the test specimen (43mm), only 2.9% below
the measured ultimate strength (3,080 KN).
Moreover, the modified strip model provides an excellent representation of the initial
stiffness for the test specimen up to 2,700 KN at 14mm displacement (i.e. up to 87% of
the measured ultimate strength “3,080 KN”), then predicts the ascending behaviour of the
28
test specimen near the knee of the curve slightly more accurately than the detailed model,
provides a peak strength of 2,920 KN at almost the same displacement as the test specimen
(43mm), only 5.2% below the measured ultimate strength (3,080 KN).
Fig. 2.22 Comparison of response curves for test specimen, detailed model, modified strip model, and
basic strip model. (Shishkin et al. 2009)
Although the basic strip model is conservative for use in design, the detailed model is
more accurate in predicting the elastic and inelastic behaviour of the SPSW and it
characterizes the ductility of the specimen by means of the descending branch. In
addition, the comparison of the detailed and modified strip models, indicates that little
accuracy is lost when the simplified parameters are implemented, while at the same time
rendering the model significantly more efficient in terms of modelling effort. Moreover,
a somewhat more conservative evaluation of the SPSW behaviour in terms of both
capacity and rate of deterioration is provided when the simplified parameters are used.
2.9 Hysteretic analysis of SPSWs using a modified strip model studied by (Guo et
al. 2009)
Guo et al. (2009) proposed another refinements to the strip model, to provide more
accurate predictions of the hysteretic behaviour of unstiffened SPSWs without openings.
The new proposed strip model, called combined strip model, consists of two types of
strips the traditional tension-only strips and tension-compression strips. The study had
been conducted several analysis’s and comparisons to validate the combined strip model,
which are described as following stages;
29
1) Analysis of shell element model under cyclic load
A one-bay one-story steel frame with pin-ended beam-column connections was selected
to present the proposed SPSW as shell elements. The panel aspect ratio (“β”: panel length-
to-height ratio) 1.5, and the height-to-thickness ratios “λ” of 300 and 600, were selected.
The steel yield stress was 340 MPa, the elastic modulus of steel was 2.0x105 MPa, and
the shear yield stress of steel was 169 MPa (fy / √3). The height of the frame column was
1800 mm. The panel thicknesses “t” corresponding to different height-to thickness ratios
are 6mm for λ = 300 and 3 mm for λ = 600.
Fig. 2.23 Comparison of response curves of shell and double-directional tension strip model.
(Guo et al.2009)
It can be noticed from Fig. 2.23 that the response curves are neither identical at the elastic
portion, nor beyond the occurrence of initial yielding. In case of λ = 300, the double-
directional tension strip model predicts a maximum lateral shear of 2,800 KN at story
dirty of 35 mm, while the shell model predicts 3,200 KN at same story dirty, (i.e. the
relative difference was 18%). However in case of λ = 600, the double-directional tension
strip model predicts a maximum lateral shear of 1,300 KN at story dirty of 35 mm, while
30
the shell model predicts 1,600 KN at same story dirty, (i.e. the relative difference was
12.5%). This indicates that, whenever the height-to-thickness ratios “λ” was bigger, the
response predictions would be less accurate since using the double-directional tension
strip model.
Figure 2.24 presents comparisons of analysis results under cyclic load for the different
combined strip models against the double-directional tension strip model and shell
element model. In addition, the comparisons were for three different cases: (a) β = 1.5
with λ = 100, (b) β = 1.5 with λ = 300, and (c) β = 1.5 with λ = 500. Where the vertical
axis presents the lateral shear in (× 100KN) and horizontal axis presents story drift in
(mm).
It can be noticed from Fig. 2.24 for case of λ = 100, that the Combined 3-7 model predicts
a maximum lateral shear of 8,000 KN at story dirty of 35 mm, which is almost the same
of that at the double-directional tension strip model, while it is lower than that at the shell
elements model (9,000 KN), i.e. the relative difference is 11%. However, for cases of λ =
300 and λ = 500, the Combined 2-8 model predicts a maximum lateral shear of 3,100 KN
and 1,800 KN at story dirty of 35 mm respectively, which are almost the same of those at
the shell elements model.
31
Fig. 2.24 Comparison of hysteretic curves for various modelling approaches. (Guo et al. 2012)
According to that comparison, the Combined 3-7 model was more suitable for SPSW with
height-to-thickness ratios “λ” varying from 100 to 250, and the Combined 2-8 model was
more suitable with “λ” varying from 300 to 600. This is because compression effect
become smaller with increasing the height-to-thickness ratio “λ”.
32
Fig. 2.25 Comparison between analytical results and experimental results. (Guo et al. 2012)
The Figure (2.25.a) shows that, the Combined 3-7 model predicts a maximum base shear
of 2,700 KN for top displacement of 88 mm, against 3,000 KN from the experimental
result from specimen SC6T and for same top displacement (i.e. the relative difference
was 10%). Whilst, it predicts a maximum base shear of 1,850 KN for top displacement
of 150 mm, against 1,750 KN of the experimental result from specimen FSPW2 and for
top displacement of 180mm (i.e. the relative difference was 5%).
The Figure (2.25.b) shows that, the Combined 2-8 model predicts a maximum base shear
of 50 KN for diagonal displacement of 2.8 mm, against 53 KN from the experimental
result from specimen SW2 and for same diagonal displacement (i.e. the relative difference
was 6%).
Thus, it can be concluded that the results of combined strip model coincide well with
those at shell element model and experimental results, with acceptable range of relative
differences. While, the tension-only strip model results underestimate both energy
dissipation capacity and load-carrying capacity.
33
2.10 Simplified equivalent braced frame model for SPSW by (Chatterjee et al.
2015)
Chatterjee et al. (2015) propose an equivalent braced model to predict the overall global
(not detailed) performance of SPSW, through introducing factors that accounts the
bending action of the boundary frame and the pinching effects of infill plate. In this model
as shown in Fig. 2.26, beams and columns are modeled as of beam-member rigidly
connected at beam-column joints and non-concentric diagonal trusses are modeled as
truss-members.
Figure 2.27 shows a comparison of response curves under pushover loading, between for
SPSW model with square beams (80 mm × 80 mm), square columns (200 mm × 200 mm)
and (3 mm thick) infill plate, and an equivalent braced model (truss model). Where the
vertical axis presents the base shear in (KN) and horizontal axis presents top displacement
in (mm).
If can be observed from Fig. 2.27 that the equivalent bracing model underestimates the
initial stiffness of plate model up to 1,750 KN at 15mm displacement (i.e. up to 83% of
the peak strength “2,100 KN”), then predicts the ascending behaviour of the plate model
near the knee of the curve, and gives almost similar plastic deformations. Furthermore, it
provides same peak strength of the plate model (2,100KN) but at less displacement of
45mm comparing with 75mm for the plate model.
Fig. 2.27 Comparison of response curves for plate model and truss model. (Chatterjee et al.2015)
34
The proposed equivalent braced model had been verified also under cyclic loading,
against the experimental results, and for multistory samples. It provided acceptable
estimation for the initial stiffness of the SPSW, and final strength with an average relative
difference of less than 5%.
Besides, four integrated studies with different SPSW modeling approaches was presented,
with focusing on using of strip model as a simplified method to represent the global shear
behaviour of unstiffened thin SPSW under lateral load.
However, Most of the past studies, that verified the strip model approach, presented the
whole SPSW system in one model, i.e. the frame elements and the infill plate. The current
study verify the strip model (truss elements) approach for the infill plate only to predict a
better simulation.
35
Chapter 3: Finite element analysis procedure
3.1 Introduction
This chapter describes the numerical analytical study to represent the behaviour of an
unstiffened thin steel plate under lateral monotonic “pushover” load using the finite
element method by means of ABAQUS (Ver. 6.14). Geometrically and Materially Non-
Linear Analysis (GMNLA) had been followed and large-deflection effects were taken
into account.
A shell element model of the plate, mentioned as verified model, had been verified by
generating the shear force-displacement curves under cyclic loads and comparing against
that at previous numerical research for the same plate, including the effect of the mesh
convergence.
Base on the verified model, a revised shell elements model with some additional boundary
conditions, mentioned as modified model, had been generated and analysed to present the
behaviour of the plate under later monotonic “pushover” loading.
Base on the modified model, a set of truss-elements models different in number of trusses
and their effective width, had been generated and analysed. The relationship between the
yielding force and the effective width of those models had been compared to conclude
the suitable truss-elements models that predicts the same yielding force as the modified
model.
36
added in the verified model as shown in Fig. (3.1.b). The steel plate will be modelled and
analysed by ABAQUS (Ver. 6.14) depending on the finite element method (FEM).
Material properties
Fig.3.2 demonstrates the assumed stress-strain relationship for the plate material (mild
steel).The yielding stress is σy = 240 MPa, Poisson's ratio ν = 0.3, initial young’s modulus
is E1= 210 GPa and then the strength will increase slightly with E2= 2.1 GPa; therefore,
the plate material is considered to be elastoplastic.
Fig. 3.2 Assumed stress-strain relationship of mild steel. (Alina and Dastfan 2007)
Elements modelled by ABAQUS (Ver. 6.14)
“S4” element will be used in modelling the steel plate, which is a 4-node general-purpose
shell with finite membrane strains, as shown in Fig. 3.3, each node has 6 degrees of
freedom, three displacement and three rotations. (ABAQUS 2014)
37
Applied load on the verified model
The verified model of steel plate will be subjected to cyclic loading by applying a very
small lateral displacement (1.5 mm as maximum) to the top nodes. Fig. 3.4 shows the
time history of this applied displacement (drift), where the vertical axis presents the top
displacements and the horizontal one presents the time steps up to 85 second.
Fig. 3.4 Time history of drift applied to the verified model. (Alina and Dastfan 2007)
Type of analysis
Two types of analysis, cyclic and monotonic “pushover”, had been carried out by means
of geometrically and materially nonlinear analysis (GMNLA). Large-deflection effects
were taken into account.
Mesh convergence
Three different patterns for plate mesh had been proposed and analysed to study the effect
of mesh convergence. Fig. 3.5 shows the dimensions of “S4” elements for each pattern,
where the degrees of freedom (DOF) were [1536] for (12 x12 mm2) model, [2646] for
(15x15 mm2) model and [4056] for (20x20 mm2) model.
a) 12x12 mm2 (25 divisions) b) 15x15 mm2 (20 divisions) c) 20x20 mm2 (15 divisions)
Fig. 3.5 Three different patterns for plate mesh
Figure 3.6 demonstrates the effect of mesh convergence for different patterns of the plate
studied; where the horizontal axis presents the multiplicative inverse of number degrees
38
of freedom (DOF) in the model, and the vertical axis presents the magnitude variation of
yielding shear force (Vu) under cyclic load as described previously (Fig. 3.4).
20.80 (20x20 mm2)
20.70 20.75 KN
20.60
Vu (KN)
20.50
(15x15 mm2)
20.40
20.40 KN
20.30
(12x12 mm2)
20.20
20.20 KN
20.10
0.0000 0.0001 0.0002 0.0003 0.0004 0.0005 0.0006 0.0007
1/DOF
Fig. 3.6 indicates that use the shell element “S4” with dimensions of (12x12 mm2) gave
convergent results with relative difference of 1% only comparing to use same element
with dimensions of (15x15 mm2), and 3% with dimensions of (20x20 mm2). Base on that,
using of shell element “S4” with dimensions of (15x15 mm2) is sufficient to get an
accurate solution. Where using of smaller element with dimensions of (12x12 mm2) is
non-economic solution in terms of consumed time. Fig. (3.5.b) shows the final verified
model of steel plate that will be validated by ABAQUA (Ver. 6.14) software.
39
30
20 20.64 KN
10
Vu (KN)
0
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
-10
-20.67 KN -20
-30
δ (mm)
Fig. 3.7 Shear force-displacement curves of the plate under cyclic loads.
Fig. 3.8 illustrates the analytical hysteresis curves of the plate according to study of Alina
and Dastfan (2007), where the maximum base shear force was (-18.3 KN) at (-1.5mm)
displacement.
17.6 KN
-18.3 KN
Fig. 3.8 Shear force-displacement curves for the plate. (Alina and Dastfan 2007)
For the purpose of comparison the analytical results of verified model with the analytical
results of Alina and Dastfan (2007) model, Fig. 3.9 gathers the envelopes of the hysterical
curves of each one together. It can be observed an excellent representation of the initial
stiffness for the verified model up to shear force of 1,500 KN at 0.3 mm displacement,
then provides a peak strength of (+20.64 KN) and (-20.67 KN) at (+1.5 mm) and (-1.5
mm) displacement respectively, with only 15% relative difference from results of Alina
and Dastfan (2007) study.
40
30
FEA (Current study)
20.64 KN
FEA (Alina and Dastfan 2007)
20
17.60 KN
10
Vu (KN)
0
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
-10
-18.3 KN
-20
-20.67 KN
Relative difference 15%
-30
δ (mm)
Fig. 3.9 Envelopes of hysterical curves for the verified model, and Alina and Dastfan (2007) model.
Fig. 3.10 illustrates the maximum in-plane principal strain on deformed shape at end of
various loading cycles, where Fig. (3.10.a) shows the plate at end of the elastic stage up
to displacement of (δ = +0.49 mm) and base shear of (Vu= +20.64 KN); whilst Fig.
(3.10.b) shows the buckling mode and developing of tension fields which in corners had
yielded (in grey colour) at displacement of (δ = +0.98 mm) and base shear of (Vu= +20.0
KN); thus Fig. (3.10.c) shows the gradual yielding of the tension field at end of the
analysis at displacement of (δ = +1.48 mm) and base shear of (Vu= +20.58 KN). Notably,
the plate material will start yielding since the strain is more than the value of (ε =
0.001143) accounted from the assumed stress-strain relationship of mild steel at Fig. 3.2.
a) δ = +0.49 mm (Vu= +20.64 KN) b) δ = +0.98 mm (Vu= +20.0 KN) c) δ = +1.48 mm (Vu= +20.58 KN)
Fig. 3.10 Maximum in-plane principal stress on deformed shape at end of various loading cycles.
41
Under monotonic “pushover” loading
A pushover analysis was conducted on the verified model by applying very small lateral
displacement to the top nodes. The analysis was terminated when the top nodes reached
a displacement of 1.5 mm, which is the maximum displacement recorded during the cyclic
analysis. Similarly to cyclic analysis, the pushover analysis was also Geometrically and
Materially Non-Linear Analysis (GMNLA), and the large-deflection effects were taken
into account.
40
Cyclic Analysis 29.03 KN
30
Pushover Analysis 29% 20.64 KN
20
19.53 KN
Vu (KN)
10
0
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
-10
-20
-20.67 KN Relative difference 29%
-30
δ (mm)
Fig. 3.11 Shear force-displacement curves of the verified model under cyclic and pushover loads.
Results discussion
Fig. 3.11 demonstrates the shear-displacement curves of the verified model analysed
under both cyclic and pushover loads. It can be noticed that the maximum yielding force
of the pushover analysis is 29.03 KN at 0.62 mm displacement, while it was 20.64 KN at
0.5mm displacement at the cyclic analysis. Accordingly the relative difference between
both maximum yielding forces is around 29%. However, at the end of the analyses (1.5
mm displacement), the shear force was 19.53 KN, slightly less than that at cyclic analysis
(20.64 KN).
Based on the previous analysis and comparisons, the verified model is vailed to carry out
the later finite-element analysis in this study.
Accordingly, it can be concluded from the modified model response that, the plate
material had yielded prior to the elastic critical buckling occurred (because that the plate
imperfections were not considered in the analysis), yielding continues until reaching a
maximum force (ultimate strength) after which the out-of-plane deformations started; that
43
indicates that the elastic critical bucking stress was equal to the maximum force. Beyond
this stress, a post-buckling stage had started.
Fig. 3.14 shows the relationship between the out-of-plane deformations at the center point
of the plate (Uz on the vertical axis), and the lateral displacements (δ on the horizontal
axis).It is clearly observed that, at lateral displacement of (0.8mm) the out-of-plane
deformation had been started, and gradually increased up to a maximum value of
(1.06mm) for lateral displacement of (0.945 mm). Beyond this point, the out-of-plane
deformations changed in the direction, and gradually reduced due to development of the
two main tension fields.
1.2
0.8
0.6
Uz (mm)
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
-0.2
δ (mm)
Fig. 3.14 Out-of-plane deformations at the plate center against the lateral displacement.
44
Fig. 3.15 shows the maximum in-plane principal strain on deformed shape for the featured
points of the curve, where the grey colour indicates the yielded portions of plate (strain ε
≥ 0.001143).
Start yielding
45
3.5 Finite-element analysis for the equivalent truss-elements model
The idea of modelling the plate as equivalent truss-elements as aim of this study, is to
replace the shell elements by set of the truss members, that matches the number, angel of
inclination and area of those tension fields developed. As the set of tension fields will be
changed during the analysis and accordingly their area (width) as well, it must account
their possible numbers at the first development till end of the analysis. Later, a group of
truss-elements models will be evaluated through changing two parameters, the number of
the trusses and the effective width, to verify their behaviour against that at the shell
elements model (modified model); then to conclude the suitable number of truss-elements
and the effective width required to capture most of the plate behaviour under later shear
loads. However, the material properties, monotonic “pushover” loading and type of the
analysis are same as that at the modified model of the steel plate.
Accordingly, the models group was included 1-truss model, 2-trusses model, 3-trusses
model, 4-trusses model, and 5-trusses model. The trusses in each model were associated
with an effective width depending on the trusses number. However, changing the
effective width had been investigated to predict the yielding force as that in the modified
model.
46
a) At Vu=32.30 KN, δ= 0.4 mm b) At Vu=40.37 KN, δ= 0.5 mm
Fig. 3.16 Varying the tension fields number through the main analysis steps.
Initially, the area of each truss is equal to the plate thickness (1 mm) multiple by the truss
spacing, which is equal to the diagonal length of the plate “d” over “n+1”, where n is the
number of trusses in that model; the diagonal length “d” of the 300mm × 300mm square
shape is equal to 424 mm. Subsequently, various values for the truss area has been
assumed by changing the truss width to verify its effects on the analysis.
47
Boundary conditions
Since the modelling of boundary elements (i.e. the frame) is out of scope of this study,
the major issue in modelling truss-elements is that the whole trusses must act together as
one unit. To achieve that, several ways of linking by frame element, truss element, link
and spring, had been assumed and verified. The most accurate way found is to use “Multi-
point constraints (MPCs) type beam” link of ABAQUS (Ver. 6.14).
As shown in Fig. 3.18, “MPC type beam” link provides a proposed rigid beam between
two nodes to constrain the displacement and rotation at the first node to the displacement
and rotation at the second node, corresponding to the presence of a rigid beam between
the two nodes. (ABAQUS 2014)
However, the boundary conditions of the trusses nodes were all three displacement
constraints (Ux ,Uy ,Uz ) at the bottom edge, and two displacement constraints (Uy ,Uz )
at the other edges.
48
1-Truss Model 2-Trusses Model
5-Trusses Model
49
40 36.46 KN
35
30 31.72 KN
25
Vu (KN)
20 24.13 KN
15
10
5
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
δ (mm)
Fig. 3.20 Comparing the shear force-displacement curves of various truss-elements models under
pushover loading.
It can be noticed from the curves at Fig. 3.20 that the responses of the 1-Truss Model, 3-
Truss Model and 5-Truss Models are almost the same, and show a higher initial stiffness
than that for 2-Trusses and 4-Truss Models. However, the yielding force for these three
models are 36.46 KN at displacement of 0.68 mm, while it is 31.72 KN at displacement
of 0.78 mm for the 4-Trusses Model and 24.13 KN at displacement of 0.88 mm.
Comparing to the yielding force of the plate (Modified model), which was (41.58 KN) at
displacement of (0.54 mm), none of the truss-element models has reached this value. But,
the yielding forces and their related displacements for 1-Truss Model, 3- Trusses Model
and 5-Trusses Model, are closer to that at the plate than the other two models. Therefore,
an odd number for trusses, with one truss connects two opposite corner nodes (diagonal
truss), is required to predict a response closed to that in the plate (Modified
model).However, the initial area of the truss (truss width) is not sufficient to reach the
yielding force of the plate, so it need to be increased.
Subsequently, various values for truss area has been assumed by changing the truss width
at each model. Table 3.1 collects all analysis results for all models group.
50
Table 3.1 Analysis results for all models group.
Models Number of Truss Spacing Assumed truss Truss area Yielding force Displacement
Trusses “n” (mm) d/(n+1) width (mm) (mm2) Vu (NK) at Vu, (mm)
1-Truss 1 d/2 = 212 0.60 d = 254.4 254.4 43.18 0.70
Models 0.55 d = 233.2 233.2 39.58 0.70
0.50 d = 212 212 35.98 0.70
0.33 d = 141 141 23.93 0.70
0.25 d = 106 106 17.99 0.70
0.20 d = 84.8 84.8 14.39 0.70
0.15 d = 63.6 63.6 10.18 0.70
0.10 d = 42.4 42.6 7.2 0.70
2-Trusses 2 d/3 = 141 0.33 d = 141 141 23.96 0.90
Models 0.25 d = 106 106 18.01 0.88
0.25 d = 84.8 84.8 14.41 0.90
0.15 d = 63.6 63.6 10.81 0.90
0.10 d = 42.4 42.4 7.2 0.88
3-Trusses 3 d/4 = 106 0.33 d = 141 141 48.01 0.70
Models 0.30 d = 127.2 127.5 43.22 0.70
0.25 d = 106 106 36.01 0.70
0.20 d = 84.8 84.8 28.81 0.70
0.15 d = 63.6 63.6 21.61 0.70
0.10 d = 42.4 42.6 14.41 0.7
4-Trusses 4 d/5 = 84.4 0.20 d = 84.8 84.8 31.35 0.78
Models 0.15 d = 63.6 63.6 23.52 0.78
0.10 d =42.4 42.6 15.68 0.78
0.05 d = 21.2 21.2 7.84 0.78
5-Trusses 5 d/6 = 70.67 0.200 d = 84.8 84.8 43.15 0.70
Models 0.167 d = 70.67 70.67 35.96 0.70
0.143 d = 60.57 60.57 30.82 0.70
0.125 d = 53 53 26.97 0.70
0.111 d = 47.11 47.11 23.97 0.70
Fig. 3.21 demonstrates relationship between the yielding force and the width-to-diagonal
length “w/d” ratio for each models group, 1-Truss Models, 2-Trusses Models, 3-Trusses
Models, 4-Trusses Models, and 5-Trusses Models; in addition, the steel plate yielding
force of the steel plate (modified model) is shown for comparison. It can be noticed that
the 2-Trusses Models shows same relationship as the 1-Truss Models; and the 4-Trusses
Models shows a relationship closed to that at the 5-Trusses Models. However, the 1-
Truss, 3-Trusses and 5-Trusses Models can reach the yielding force of the plate model
51
(modified Model) at reasonable ratios for width-to-diagonal length “w/d”, and at same
displacement of 0.7 mm (as shown at Table 3.1).
60
50
48.01
43.15 43.18
41.58 43.22
41.58
40
35.96 39.58
36.01 35.98
Vu (KN)
30.82 31.35
30
26.97 28.81
23.52 23.96
23.97 23.93
21.61 18.01
20
15.68 14.41 17.99
14.41
14.39
10.81
10 7.84 10.18
7.20 7.20
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
"w/d" Ratio
Fig. 3.21 Relationship between the yielding force and the width-to-diagonal length “w/d” ratio for models
group.
Table 3.2 provides the required “w/d” ratios to reach the yielding force of the plate model
(modified Model). It can be concluded that the required truss width is slightly more than
the truss spacing in the model.
Table 3.2 Required “w/d” ratios to reach the yielding force of the plate model.
Model Truss spacing Width-to-diagonal Yield force Displacement at
ratio length “w/d” ratio Vu, (KN) Vu, (mm)
1-Truss Model 0.5 0.578 41.58 0.7
3-Trusses Model 0.25 0.289 41.58 0.7
5-Trusses Model 0.167 0.192 41.58 0.7
Fig. 3.22 demonstrates the shear-displacement curves of the modified model of the steel
plate and these three models (i.e. 1-Truss Model, 3- Trusses Model and 5-Trusses Model),
where the required “w/d” ratio to reach the yielding force of the plate model are used. It
can be noticed that the initial stiffness of the plate model is higher than that for the truss-
elements models, where the displacements in which the yielding starts are convergent
with relative difference of 23%.
52
45 23%
41.58, 0.54 41.58, 0.70
40
35
30
Vu (KN)
25
20
15
10
Relative difference 23%
5
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
δ (mm) Vu , δ
Fig. 3.22 Comparing the shear force-displacement curves of 1-Truss, 3-Trusses, 5-Trusses Models and
Plate Model.
53
Chapter 4: Conclusions
At the end of this study, the following results has been concluded:
2) To predict a response closed to that in the plate model, the truss-elements model
required an odd number of trusses (1, 3 or 5), with one truss connecting the two opposite
corners (diagonal truss).
3) To reach the maximum yielding force of the plate model, the required truss width was
slightly more than the truss spacing in the model. For instance, the 3-Truss Model has
required a truss width of 0.289 of the diagonal length while the truss spacing was 0.25.
4) The initial stiffness of the truss-elements models was lower than that at the plate model,
where the relative difference was 23% for displacements.
54
2) Change the geometry of the steel plate.
3) Appling different loading scenario (cyclic and dynamic).
4) Take the imperfection of the plate into account, so the elastic buckling load can be
shown in shear force-displacement curve.
55
References and Bibliography
- ABAQUS (2014) Abaqus ver. 6.14 Online Documentation, Build ID: 2014-04-18-
11.27.5544621, Dassault Systèmes.
- ASIC (2008) L. A. Live Hotel & Residences: An Innovative Steel-Plate Shear Wall
Solution, American Institution of Steel Construction [online] available from
<http://www.aisc.org/content.aspx?id=16012> [21 October 2015].
- Astaneh-Asl, A (2000) 'Steel Plate Shear Walls.' U.S. - Japan Workshop on Seismic
Fracture issues in Steel Structures, San Francisco, (12).
- Astaneh-Asl, A (2001) Seismic Behaviour and Design of Steel Shear Walls. USA:
Steel TIPS Report, Structural Steel Education Council.
- Astaneh-Asl, A (2002) Seismic Behaviour and Design of Composite Steel Plate Shear
Walls. USA: Steel TIPS Report, Structural Steel Education Council.
- Bekő, A and Roško, P (2013) Chapter 10 Testing of Stiff Shear Walls. In: Krims-
Steiner, N, (ed.) IRIS Industrial Safety and Life Cycle Engineering. Vienna:
VCE Vienna Consulting Engineers ZT GmbH.
- Case, J, Chilver, L and Ross, C T F (1999) 4th edn. Strength of Materials and Structures.
London: Arnold.
56
- Driver, R G, Kulak, G L, Elwi A E and Kennedy, D J L (1998) 'FE and Simplified
Models of Steel Plates Shear Walls.' Journal of Structural Engineering ASCE
124, (2) 121-130.
- Guo, L, Li, R, Zhang, S and Yan, G (2012) 'Hysteretic Analysis of Steel Plate Shear
Walls (SPSWs) and A Modified Strip Model for SPSWs.' Advances in
Structural Engineering 15, 1751- 1764.
- Murty, C V R (n. d.) Earthquake Tip 8: What is the Seismic Design Philosophy for
Buildings? [Online] available from <http://www.iitk.ac.in/nicee/EQTips/EQTip
08.pdf> [30 October 2015].
- Sabelli, R and Bruneau, M (2007) Steel Design Guide 20: Steel Plate Shear Walls,
American Institute of Steel Construction, Inc.
- Seilie, I F and Hooper, J D (2005) Steel Plate Shear Walls: Practical Design and
Construction. North American Steel Construction Conference, Modern Steel
Construction.
57
- Shishkin, J J, Driver, R G and Grondin, G Y (2009) 'Analysis of Steel Plate Shear
Walls Using the Modified Strip Model.' Journal of Structural Engineering ASCE
135, (11) 1357-1366.
- Ziemian, R D, ed. (2010) 6th edn. Guide to Stability Design Criteria for Metal
Structures, Structural Stability Research Council (SSRC), New Jersey: John
Wiley and Sons.
Bibliography
- Eddin, R J (2014) ‘the behaviour of RC frame infilled with stiffened steel shear wall
under seismic loads’, MSc dissertation Damascus University, unpublished.
- Student Guide (2012), ‘Harvard Citing and Referencing Guide Version 1.1’, Heriot-
Watt University Information Services.
58