Professional Documents
Culture Documents
Khanouki (2016)
Khanouki (2016)
a r t i c l e i n f o a b s t r a c t
Article history: Nonlinear finite element models of through beam connection to concrete filled circular steel tube (CFCST) col-
Received 13 May 2015 umn with three types of connection failures (column, beam and joint shear failures) under monotonic loading
Received in revised form 7 December 2015 were proposed by ABAQUS programme. The connection detail composed of a steel I-beam which is completely
Accepted 1 January 2016
passed through the circular steel tube column and welded to it by fillet or full penetration weld and the tube is
Available online 29 January 2016
then filled with concrete. The connection parameters investigated included different ratios of column-to-beam
Keywords:
flexural strength, fillet or full penetration weld to connect the beam to the tube, using rebar inside of the column
Concrete filled tube columns and effect of beam web and concrete core inside the joint. Buckling analysis, concrete damage, weld fractured and
Through beam connection tube tearing were defined in the model. Good agreement was achieved between the model and existing test re-
Finite element analysis sults in terms of the beam tip force–storey drift relationship, joint distortion and joint shear strain. As a result, the
Steel and concrete damage models could correctly predict the linear, nonlinear and post-failure behaviours of the connection. In addition,
Weld fractured other parameters of the connection were investigated for the specimens. Finally, the effect of column axial
Buckling analysis load level for the connection behaviour in the three types of failures was investigated as a parametric study.
© 2016 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jcsr.2016.01.002
0143-974X/© 2016 Elsevier Ltd. All rights reserved.
M.M.A. Khanouki et al. / Journal of Constructional Steel Research 121 (2016) 144–162 145
penetration weld and after that the tube is filled by concrete. The spec-
Nomenclature
imens were divided in three groups based on the type of failures as
following:
bf beam flange width
Group one (Column failure) included specimens NSF1 and NSF2. In
d total beam depth
these specimens, fillet weld was used to attach the beam to the tube
dp penetration measure
wall and the column was weaker than the beam such that the failure
Ec modulus of elasticity of the concrete core
took place in the column. In addition, specimen NSF2 had no concrete
Ecm modulus of elasticity of the concrete
inside the joint in order to investigate the effect of the concrete core in
Es modulus of elasticity of the steel
the joint. The results were the weld fractured and tearing of the tube
fbo biaxial concrete compressive strength
aligned the weld.
fco uniaxial concrete compressive strength
Group two (Beam failure) included specimens NSF4, NSF5 and
f'c cylinder concrete compressive strength
NSF8 in which the column was stronger than the beam such that
ft concrete tensile strength
the column to beam flexural strength ratio (r) were 1.7, 1.65 and
Fyb minimum specified yield stress of the steel beam
2.17, respectively. These ratios were bigger than 1.2, which is recom-
H distance between two reaction points (mid height) of
mended in the design of beam to column connections [17]. The weld
the column
between beam and the tube was full penetration for specimens NSF4
Kdflt default contact stiffness
and NSF5 while fillet weld was used for specimen NSF8. The only dif-
Ki stiffness for each segment
ference between specimens NSF4 and NSF5 was the use of four re-
i segment number
bars (grade 60) inside of the column for specimen NSF4. For these
L distance between two loading points of the beam
specimens, the results were buckling of the beam flange and web
Mp beam plastic moment
outside the tube without any local buckling or signs of yielding in
Paxial axial load applied on the column during the test
the tube. This is the same behaviour that is expected from through
Pno nominal pure axial compressive strength of the column
beam connections in case of beam failure.
r column to beam flexural strength ratio
Group three (Joint failure) included specimens NSF6 and NSF7. The
S geometric scale factor
only difference between the specimens was existence of a thin web in-
S0 initial scale factor
side the joint in specimen NSF7 to investigate the effect of the web in
tf beam flange thickness
joint. In these specimens both column and beam were much stronger
tw beam web thickness
than the joint such that the failure was governed to be occurred in the
Vcsn nominal shear strength of the concrete core
joint in order to understand the distribution of the joint shear force be-
Vf nominal shear strength of the flange
tween the joint component i.e. the tube, as well as the beam web and
Vj horizontal shear force of the joint
the concrete core inside the joint.
Vn nominal joint shear strength
Vtn nominal shear strength of the steel tube
Vwn nominal shear strength of the beam web
3. Finite element models
Wc density of concrete
ϕ reduction factor of nominal shear strength
To study the behaviour of through beam connection to CFCST col-
εc elastic compressive strain corresponding to 0.4fc′
umn, three finite element models were proposed using the ABAQUS Ex-
εc1 compressive strain corresponding to fc′
plicit programme [18]. The models exhibit the linear, nonlinear and post
εt elastic tensile strain corresponding to ft
failure behaviours of the connection in three cases of connection failure;
γ material density
column, beam and joint shear failure. Specimens NSF1 and NSF2 as well
υ Poisson ratio
as NSF4 to NSF7 were selected for this study. As previously described in
σc concrete compressive stress
Section 2, the specimens were divided in three groups based on the
∑M p the sum of plastic moment of the beam
types of failure.
∑M c the sum of the nominal flexural strength of the columns
The cruciform–shape (as in the experimental programme) was se-
lected to represent the interior joint in moment resisting frames sub-
jected to lateral loads which the moment distribution causes inflection
points to be formed at approximately the mid-span of the beams and
and effect of beam web or concrete core inside the joint. The objectives the mid-height of the columns. These points were subjected to shear
of the study are to investigate the behaviour of the connection under force. Fig. 2 shows the deformed joint with vertical shear forces applied
three different failure modes: column, beam and joint shear, in terms at the inflection points of the beam in opposite directions while the in-
of connection capacity to develop the full flexural capacity of the flection points of the column were fixed for translational movement.
beams, contribution of the joint components in resisting the joint This system simulates the joint deformation in the laterally loaded
shear force in the panel zone as well as load transferring from beam frame.
flange to the joint through finite element analysis and verification The typical components of the connection for specimen NSF7 are
with existing experimental results. Finally, the connection behaviour shown in Fig. 3. The components included the steel tube and jacket,
under different axial load level on the column is investigated through beam with attached threaded rods, rigid caps for column top and
parametric studies. bottom, concrete core and an interface. The interface was a compo-
nent without thickness which was attached inside the tube as well
2. Summary of experimental test programme as around the beam flange and web inside the tube for defining inter-
action between steel and concrete. In addition, fillet weld between
Experimental investigation of through beam connection to CFCST tube and beam was modelled for specimens NSF1, 2, 8 as shown in
column under monotonic loading was conducted by Elremaily and Fig. 4. No full penetration weld between the tube and the beam
Azizinamini [8].The test programme included seven specimens which was modelled but they were joined together by common nodes at
were symbolized as NSF1 and NSF2 as well as NSF4 to NSF8 as shown the appropriate places. No adhesion was also considered between
in Fig. 1 and Table 1. In all specimens, the steel beam was completely the steel and the concrete core as it generates a small force with no
passed through the steel tube and welded to it by fillet or full significant effect. The following sections present the development
146 M.M.A. Khanouki et al. / Journal of Constructional Steel Research 121 (2016) 144–162
Fig. 1. Specimen details: (a) NSF1; (b) NSF2; (c) NSF4 & 5; (d) NSF8; (e) NSF6 &7 [8].
of nonlinear three-dimensional finite element models of the Concrete core and fillet weld: Eight-node solid element (C3D8R)
specimens. with three translational degrees of freedom at each node; this element
is able to consider concrete cracking, crushing and weld fractured in
three orthogonal directions at each integration point.
3.1. Element types Rebar or threaded rod: Two nodes linear beam element (B31) with
three translational and three rotational degrees of freedom at each
Fig. 3 also shows the typical component meshes for specimen NSF7. node;
The element size was considered such that it could satisfy a good accu- Interface: Surface element (SFM3D4R). This element had three
racy and a reasonable computational time. In addition, reduced integra- translational degrees of freedom at each node with no stiffness or
tion and appropriate hourglass control were considered for the strength while appropriate mass density should be assigned to it.
elements. The components were meshed by appropriate type of ele-
ments as follows: 3.2. Boundary condition and loading
Steel tube, jacket and beam: Four-node shell element (S4R) with
three translational and three rotational degrees of freedom at each The boundary condition and loading on the model are shown in
node; Fig. 3. The references points at column caps were constrained for
M.M.A. Khanouki et al. / Journal of Constructional Steel Research 121 (2016) 144–162 147
Table 1
Summary of the test specimens.
NSF1 Fillet welds W18*50 500.4 ϕ305 ∗ 6.4 890 0.17 301.4
NSF2 Fillet welds W18*50 500.4 ϕ305 ∗ 9.5 979 0.16 400.9
NSF4 Full pen. weld + rebar W18*35 374.2 ϕ406 ∗ 6.4 1423 0.18 637.6
NSF5 Full pen. weld W18*35 390 ϕ406 ∗ 6.4 1512 0.19 642.1
NSF8 Fillet welds W16*31 295.9 ϕ406 ∗ 6.4 1690 0.21 641.8
NSF6 No beam web inside the pipe Built-up 1211.4 ϕ406 ∗ 6.4* 1690 0.22 1285.3
NSF7 Thin beam web inside the pipe Built-up 1211.4 ϕ406 ∗ 6.4* 1468 0.19 1282.0
*: the tube outside the joint was strengthened by a steel jacket of 6.4 mm thickness.
Step 3. Ratio of column to beam flexural strength and ratio of joint shear force to joint component shear strength.
ϕVn (kN)
r ¼ ∑M c
Group number Specimen VJ (kN) Vf Vn (kN) VJ
ϕV n
∑M p
Vwn Vtn Vcsn
translational and rotational movements except rotational movement Similar experimental loading conditions were applied on the
around Y axis to allow full rotation at column ends in plane of load- model. In step one; the axial load was applied at references point at
ings. In addition, the movement in the vertical direction (Z) for the the top cap and it was constant during the step two. In step two,
top cap was considered to be free due for application of the axial two equal and vertical velocity loads were applied at the beam
load on the column. The nodes on the beam at the lateral bracing end's reference nodes at two opposite directions. A special
places were also restrained against movement in the lateral direction 300 mm/s rate of velocity was monotonically used for the loading
(Y). rate. Velocity controlled loading provided a more stable system in
the nonlinear stage rather than force controlled loading. This rate
was achieved by comparing the kinematic energy with the internal
energy when the effect of dynamic analysis could be neglected. The
real rate, which was much smaller than this rate, resulted in longer
analysis with little effect on the accuracy of the results. In order to
define the beam failure, linear buckling analysis was employed
using the original perfect geometry of the specimen in accordance
with the buckling shapes resulted from the eigen buckling analysis.
Fig. 3. Typical meshing, boundary condition and loading on the model of specimen NSF7.
value was considered based on property of electrode E7018 which the EC2 [19] as shown in Fig. 6a. The modulus of elasticity of the con-
was used in designing of the specimen NSF8 [16]. crete (Ecm) is given in Eq. (1).
To model behaviour of concrete in compression and tension, an
equivalent uniaxial stress–strain relationship was used according to
0 1 0
Ecm ¼ 9:5 f c þ 8 3 Ecm in GPa and f c in MPa ð1Þ
The compression curve was divided into three parts i.e. the elastic
range, the nonlinear parabolic portion and the descending slope. The
value of the first part was the proportional limit up to stress of 0.4fc′
and strain of εc, where, fc′ is the cylinder concrete compressive strength
Fig. 4. Typical meshing of fillet weld for specimens NSF1, 2 & 8. Fig. 5. Stress–strain relationship for steel materials.
M.M.A. Khanouki et al. / Journal of Constructional Steel Research 121 (2016) 144–162 149
Table 2
Mechanical properties of tested steel coupons [8].
of the concrete. For the nonlinear parabolic part, the relation between
concrete stress (σc) and strain (εc) was obtained as follows:
assumed to be 2350 kg/m3 and 0.2, respectively. Compressive strength
of concrete cylinder samples (fc′) as a representative of the concrete
kn−n 2
0 εc εc1
σc ¼ f ; n¼ ; k ¼ 1:1 Ecm 0 ð2Þ core of the column is presented in Table 3.
1 þ ðk−2Þn c εc1 fc
Parameter and tensile behaviour of concrete.
Table 3 column, beam and joint shear failure. In group column failure, be-
Concrete strength of tested samples [8]. cause the column is weaker than the beams, the failure was occurred
Specimen fc′(MPa) in the column. Defining concrete cracking and fracturing of fillet
NSF1 41.6
weld between beam flange and tube wall as well as tube wall tearing
NSF2 38.9 along the fillet weld are the most important failure criteria in this
NSF4 33.9 group. In group beam failure, buckling of the beam flange and web
NSF5 34.3 outside the tube should be defined to study post failure of the con-
NSF6 & 7 32.4
nection. As the column remains in elastic range, defining concrete
NSF8 33.1
cracking and crushing have no significant effect on the behaviour of
the connection. For joint shear failure, concrete crushing and tube
fracture in the joint are the most important failure criteria in the
connection.
define general contact, some text comment, such as *CONTACT 3.6. Analysis solution
CLEARANCE ASSIGNMENT and *CONTACT CONTROL ASSIGNMENT
were defined in the programme. Abaqus Explicit was employed for analysis of the specimens,
Both tangential and normal contacts were employed to define which is suitable for nonlinear materials and geometry, large defor-
contact properties between the interface and the concrete core. Dif- mation, concrete damage, contact, and discontinuous parts [18].
ferent friction coefficient value of 0.2–0.6 between steel and con- Available analysis, such as general static in the Abaqus Standard
crete was used by other researchers [27–32]. As the composite was used for initial analysis. However, it caused a convergence prob-
column is not sensitive for the value of friction coefficient in steel lem and stopped the analysis at the beginning of the second step. The
and concrete [32], the friction contact between the parts was defined results of the RIKS method also resulted in a convergence problem
using coulomb friction with friction coefficient of 0.3 [32] which [33].
agrees with test results [8] and a finite element model of through
beam connection [16]. Geometry scaling, which is a form of softened 4. Numerical results
pressure-over closure relationship was used to model the normal
contact between the parts as shown in Fig. 8. This model provides The behaviour of the specimens was presented in terms of beam tip
a simple interface to increase the default contact stiffness (k dflt ) force–storey drift, beam tip force–joint distortion and shear strain of the
when a critical penetration is exceeded. The penetration measure panel zone relationships, connection rigidity and distribution of axial
(dp) is defined either directly or as a fraction of the minimum ele- stress at beam flange of the specimens. The results were verified against
ment length in the contact region. Each time the current penetration the experimental test results where it is possible. Discussion on the re-
exceeds a multiple of this penetration measure, the contact stiffness sults was carried out by comparison of the finite element results be-
is scaled by a geometric scale factor (S), Ki = S0 ⋅ kdflt ⋅ Si − 1 is the stiff- tween the specimens.
ness for each segment where, i is segment member and S0 is initial
scale factor. 4.1. Verification of the beam tip force versus storey drift
The beam tip force–drift curves from the finite element analysis
3.5. Types of failure
were compared to the experimental test to ensure the accuracy of the
model as shown in Fig. 9. In all specimens, good agreement was
In order to investigate the post failure behaviour of through beam
achieved between the model and experimental results. The model pre-
connection to CFCST column, types of failure based on the connection
dicted the maximum capacity of the specimens accurately. In the anal-
groups and experimental test results [8] should be correctly defined
ysis, the beam loading was continued to identify the post failure
in the model. As previously described in Section 2, this paper is going
behaviour of the specimens. As shown in Fig. 9, the post failure behav-
to study the connection behaviour under three types of failure:
iour in the analytical model was similar to that of the experimental
test for all specimens.
In addition, the failure mode of specimens NSF1, NSF5 and NSF7
(as a representative of each group) are compared between the FEA
and experimental test results. The results showed that the models
could predict the failure mode of the specimens. As shown in
Fig. 10, the fillet weld between beam flange and tube wall was frac-
tured in specimen NSF1, so the failure took place in the column.
Due to yielding and buckling of the beam flange, most of the defor-
mation in specimen NSF5 was related to hinge formation in the
beam outside the tube. This indicated that the failure of the speci-
mens took place in the beam. Significant panel deformation, tube
tearing and concrete crushing in the joint of specimens NSF6 and 7
showed the joint failure.
The beam tip force versus joint distortion from the analytical
model was compared for the specimens NSF1 and NSF5 as shown
in Fig. 11 as well as for specimens NSF6 and NAS7 in Fig. 12. The
force was obtained as the average force at the loading points of the
beam. Joint distortion was defined as the deviation of the corner an-
Fig. 8. Softened pressure–overclosure relationship — geometry scaling. gles of the panel zone from the right angle and calculated using the
M.M.A. Khanouki et al. / Journal of Constructional Steel Research 121 (2016) 144–162 151
Fig. 9. Comparison of FEA and experimental beam tip force versus storey drift for the specimens.
displacements of the panel corner nodes in the analytical model. As specimen NSF5. For more validation of the model, compressive diag-
shown in Fig. 12 and comparing results between FEA and experi- onal strain of the concrete in the joint was compared between FEA
mental test, it can be seen the model is able to exhibit correctly and experimental test for specimen NSF7 as shown in Fig. 13. The
the joint distortion and shear strain at the joint in specimen NSF6 value of the compressive diagonal strain of the concrete in the
and 7. Also the results show that the value of the joint distortion joint shows the significance of compression strut of the concrete in
and shear strain of the joint is much smaller than allowed value in the joint.
152 M.M.A. Khanouki et al. / Journal of Constructional Steel Research 121 (2016) 144–162
Fig. 10. Comparison of failure mode between FEA and experimental for the specimens.
4.3. Distribution of the joint shear force to the joint components were comprised of the tube, beam web inside the tube, the beam
flange and the concrete core in the panel zone. Fig. 14 shows the dis-
The horizontal shear force was distributed between the joint tribution of joint shear force components versus storey drift. The
components at the mid height of the joint. The joint components joint shear force component was normalized against the joint shear
M.M.A. Khanouki et al. / Journal of Constructional Steel Research 121 (2016) 144–162 153
Fig. 11. Comparison of joint distortion and shear strain of the joint.
Fig. 12. Comparison of joint distortion and shear strain of the joint between FEA and
force of each component which is obtained from available equation experimental test.
[11] as shown in Table 1. Based on the results, the beam web was
the first component to yield. The flat portion at the end of the curve
related to the web indicated the beam web yielding for all speci-
mens. After web yielding, the additional shear force was resisted by
the tube and concrete core. The tube and concrete core reached to
maximum value only in specimens NSF6 and NSF7 which indicate
the joint failure was occurred. In these specimens, parts of the web
outside the joint also participated in resisting the joint shear force
as shown in Fig. 14-c and d. Fig. 15 shows the distribution of the
shear stress in the tube wall and web outside the joint for specimen
NSF7. Based on the results, when the tube shear strength is de-
creased, the shear stress of the web outside the joint is increased.
Mechanism of load transfer from the beam flange to the joint was
also evaluated through finite element analysis. The distribution of the
longitudinal stress in the top beam flange for specimen NSF1, NSF5
and NSF7 is shown in Fig. 16a, b and c. The comparison between the
value of the longitudinal stress at beam flange inside and outside of
the tube confirmed that tensile beam flange forces were transferred sig- Fig. 13. Comparison of compressive diagonal strain of the concrete in the joint between
nificantly to the beam flange inside the tube especially in specimens FEA and experimental test for specimen NSF7.
154 M.M.A. Khanouki et al. / Journal of Constructional Steel Research 121 (2016) 144–162
Fig. 14. Distribution of joint shear force versus storey drift for the specimens.
NSF5 and NSF7. Also Fig. 17 shows the distribution of axial stress at top 4.5. Connection rigidity
and bottom level of the top beam flange of specimen NSF7. Based on the
results, the axial stress values adjacent to the tube wall indicate the for- The rigidity of the connection for the specimen NSF5 was investigat-
mation of plastic hinge at the beam flange. ed through finite element analysis. The results were determined
M.M.A. Khanouki et al. / Journal of Constructional Steel Research 121 (2016) 144–162 155
Fig. 15. Distribution of shear stress in tube wall and web outside the joint for specimen
NSF7.
5. Parametric studies
Fig. 17. Distribution of longitudinal stress in the top beam flange at positive and negative
surfaces for specimen NSF7.
Fig. 22a, b and c show column tip force versus storey drift under dif-
ferent axial load for specimen group in column failure, beam failure and
joint shear failure respectively. As shown in the figures for specimens in
column failure group, ductility of the specimens are increased with the
increasing of the column axial load level. As previously mentioned, the
connection failure in these specimens is fillet weld fractured between
the tube wall and beam flanges. Higher axial load on the column caused
lower tension stress at the fillet weld, therefore the weld fractured is de-
layed and ductility of the specimens is increased. In specimens of beam
failure group, both ductility and column tip force are significantly de-
creased with increasing axial load level on the column. However as
Fig. 18. Connection rigidity for specimen NSF5. Fig. 19. Distribution of connection rotation between the column, joint and beam.
M.M.A. Khanouki et al. / Journal of Constructional Steel Research 121 (2016) 144–162 157
Table 4
Specimen detail with different column axial load levels.
1 NSF1-0 0 0
NSF1-0.2 0.2 1047
NSF1-0.4 0.4 2094
NSF1-0.6 0.6 3141
2 NSF5-0 0 0
NSF5-0.2 0.2 1592
NSF5-0.4 0.4 3183
NSF5-0.6 0.6 4776
3 NSF7-0 0 0
NSF7-0.2 0.2 1545
NSF7-0.4 0.4 3090
NSF7-0.6 0.6 4636
indicates that the specimens in beam failure group with different axial
load level on the column are rigid. For specimens in joint shear failure
group, the column tip force value as well as the ductility is approximate-
Fig. 20. Deflected interior joint subjected to lateral load at column top. ly the same for the specimens. The results show that the axial load level
has no significant effect for the specimens.
shown in Fig. 23, the normalized moment at column face versus storey 5.2. Panel distortion
drift has same behaviour for the specimens. The moment at column face
were normalized against plastic moment of the beam based on Table 1. Fig. 25 shows comparison of column tip force versus panel distor-
Fig. 24 shows evaluation of connection rigidity of the specimens. It tion for the specimens in the three groups. The allowed value of
Fig. 21. Typical meshing, boundary condition and loading on the model of specimen NSF7-0 to 0.6.
158 M.M.A. Khanouki et al. / Journal of Constructional Steel Research 121 (2016) 144–162
Fig. 23. Comparison of normalized moment at column face versus storey drift for
specimens NSF5-0.0 to NSF5-0.6.
Three main joint components to resist the joint shear force of the
specimens are tube, web and concrete core inside the joint. However
the beam flange and parts of beam web outside the tube are also
contribute to resist the joint shear force of the specimens. Due to
its significant value, these two components are only investigated
for specimens in joint shear failure group. It is noted that the value
of flange shear force depends on the beam flange thickness which
is very thick only for specimens in joint shear failure group.
Figs. 26–29, show the comparison of normalized tube, web, core
and flange shear force for the specimens. The shear force of the
joint components is normalized against their related values based
on Table 1. Based on the results, all three main components contrib-
ute to resist joint shear force of the specimens in the three groups
under different axial load level on the column. The tube shear force
reached to maximum capacity only for specimens in joint shear fail-
ure group. The value of the tube shear capacity is 15% more than the
Fig. 22. Comparison of column tip force versus storey drift for the specimens.
shear strain for the tube and web is also shown in the figure. It is
noted that the maximum value of panel distortion should be limited
to four times of shear strain of the tube and web [34]. Based on the Fig. 24. Comparison of connection rigidity for specimens NSF5-0.0 to NSF5-0.6.
M.M.A. Khanouki et al. / Journal of Constructional Steel Research 121 (2016) 144–162 159
Fig. 26. Comparison of normalized tube shear force versus storey drift for specimens.
Fig. 25. Comparison of column tip force versus panel distortion for the specimens.
value obtained from the formula based on Table 1, while the web is variable from 95% to 125% of the maximum capacity based on
reached up to 95% of the maximum capacity in both column and Table 1 for specimen of NSF7-0.0 to 0.06 respectively. Fig. 30 shows
joint shear failure groups. The core shear capacity of the joint is in- the distribution of the axial force on the column between tube web
creased with increasing axial load level on the column such that it and core at the centre of the joint for the specimens. Based on the
160 M.M.A. Khanouki et al. / Journal of Constructional Steel Research 121 (2016) 144–162
Fig. 27. Comparison of normalized web shear force versus storey drift for specimens.
Fig. 28. Comparison of normalized core shear force versus storey drift for specimens.
results, the force is gradually decreased in the tube and the web 5.4. Distribution of flange axial stress to the joint
while is increased for core concrete in the joint. Higher axial load of
the core has increased the core shear capacity. As shown in Fig. 31, Fig. 32 shows comparison of distribution of flange axial stress at
the shear force of the beam web outside the joint is increased with drift 2% for specimens NSF5-0.0 to NSF5-0.6. Based on the results,
increasing the axial load level for specimens NSF7-0.0 to NSF7-0.6. the axial load level had no significant effect on the distribution of
M.M.A. Khanouki et al. / Journal of Constructional Steel Research 121 (2016) 144–162 161
Fig. 29. Comparison of normalized flange shear force versus storey drift for specimens
NSF7-0.0 to NSF7-0.6.
tension axial stress of the beam flange to the joint, while the com-
pression axial beam flange force which is directly transferred to the
beam flange inside the joint is increased with increasing of the
axial load level.
6. Conclusion