Journal of Constructional Steel Research 120 (2016) 3344

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Finite element analysis of circular concrete lled tube connections

Hazem M. Ramadan , Maha M. Hassan, Mohamed A. Mooty, Sherif A. Mourad
Faculty of Engineering, Structural Engineering Department, Cairo University, Gamaet El Qahera St., Giza, Egypt

a r t i c l e

i n f o

Article history:
Received 29 January 2015
Received in revised form 6 December 2015
Accepted 9 December 2015
Available online xxxx
Keywords:
Finite element
Concrete lled tube
Connections

a b s t r a c t
A parametric study based on nite element technique is performed to investigate the performance of different
connection congurations between circular concrete lled steel tube (CCFT) columns and gusset plates subjected
to axial compression loadings. The study focuses on the effect of the pipe and gusset plate dimensions on the
connection behavior. The modeling assumptions and techniques used to perform the analysis are detailed. The
models are veried using experimental test data performed earlier by the authors. A notable effect was observed
on the behavior of the connections due to its detailing changes with respect to failure mode, yield and ultimate
capacity, stress distribution and initial and nal stiffness.
2015 Elsevier Ltd. All rights reserved.

1. Introduction
Many researchers presented modeling procedures in order to investigate the behavior of different composite systems under both static and
dynamic loading cases. Several difculties arise while dealing with
modeling of composite sections that bring together two distinct materials, the ductile steel and the brittle concrete. Modeling of such elements should capture the relative stiffness of each material properly.
Extensive experimental research on moment connections to circular
concrete lled tube (CCFT) columns is available in literature. Alostaz and
Schneider [1] investigated the momentrotation behavior of different
connection details for wide ange shapes to CCFT columns. The studied
parameters included diameter-to-thickness ratios and moment-toshear values. Elremaily and Azizinamini [2] performed an experimental
study in order to comprehend the behavior of through beam connections that showed the capability of through beam connections to
develop the full plastic exural capacity of the beam. Vulcu et al. [3]
conducted an experimental program in order to characterize the behavior of moment resisting joints to CCFT columns in multi-storey frames.
The specimens were tested under cyclic and monotonic loading. Roeder
et al. [4] provided an overview of the seismic demands on the connection of the CFT braced frame systems through an experimental program.
Performed tests showed that the effect of bond stresses is less dominant
when the diameter of column increases, and that the main load transfer
mechanism in connections with gusset plates penetrating through the
column is direct bearing. MacRae et al. [5] performed an experimental
study in order to evaluate the bearing stress on concrete under the
Corresponding author.
E-mail addresses: Drhazem2003@yahoo.com (H.M. Ramadan),
mahamoddather@eng.cu.edu.eg (M.M. Hassan), mamooty@aucegypt.edu (M.A. Mooty),
Smourad2006@eng.cu.edu.eg (S.A. Mourad).

http://dx.doi.org/10.1016/j.jcsr.2015.12.023
0143-974X/ 2015 Elsevier Ltd. All rights reserved.

gusset plate at bracebeamcolumn connections. The authors found

that the force was transferred from the diagonal member to the composite column by mainly two mechanisms: the bearing under the gusset
plate and friction between steel plate and concrete or steel tube. The
percentage of the force carried by the second mechanism is 30% of the
total transferred force. However, this part is usually neglected as
quantifying the force transferred by friction in actual situations is hard
due to the shrinkage of concrete. Moreover, it was observed that the
effect of shear studs is negligible in transferring force between steel
and concrete.
Bond stresses are present in connections with gusset plate penetrating through the CFT column. Zhang et al. [6] investigated that the factors
affecting the bond stresses are found to be dependent on the shape of
the steel tube, dimensions of the tube, surface preparation of the interior
of the steel tube, shrinkage potential of the concrete, concrete and steel
material strengths, and eccentricity of loading. Such stresses are
estimated using push-out tests, push-off tests, or connection tests. The
performed tests showed that larger diameter tubes and tubes with
large D/t ratio are more likely to require mechanical shear transfer
mechanisms than smaller diameter tubes. It is also observed that
shrinkage of concrete causes severe deterioration of the natural bond
capacity. Hajjar [7] reported bond stress values varying between 0.1
1.0 MPa. Typically, the distribution of bond stress is not uniform within
the connection zone. The value is expected to reach its maximum estimate at the location where the loading is applied and decays within a
certain distance. Leon et al. [8] reported that it is not clear what percentage of the CFT perimeter is typically engaged for transferring the load.
There is a wide range of connection types that can be used with composite construction. The effect of connection exibility must be considered in the analysis since it affects the distribution of moment between
beam and column and the overall drift of the structure. Designers
usually face difculties and insufcient design data when dealing with

34

H.M. Ramadan et al. / Journal of Constructional Steel Research 120 (2016) 3344

connections of CFT columns. Meanwhile, the used design formulas may

yield connections that might have capacities dissimilar to what was
anticipated. Connections to CFT columns may be beam-to-column
moment connections or beam-brace-column bracing connections.
Deciding the appropriate detailing of the connection depends upon
the applied straining actions, the construction capabilities, and the
type of structural system.
Numerical and analytical methods provide an economical and reliable alternative in many situations. Extensive analytical research work
has been conducted in order to arrive at a proper modeling for composite steelconcrete structures that reects reality. There are many issues
concerning the modeling of CFT columns due to material nonlinearities
of composite materials since concrete and steel have different characteristics. On one hand, steel is a ductile material with the same response
in tension and compression. On the other hand, concrete is a brittle
material with different responses in tension and compression.
Hu et al. [9] conducted a numerical study in order to analyze the behavior of gusset plate CFT-to-bracing connections under an axial compressive force. Failure of the connection was generally observed under
the connection area. It was found that increasing the thickness of gusset
plate or introducing cutouts has a small effect on the ultimate strength
of CFT column; yet they cause more local bulged shapes on the steel
tube below the connection area. The behavior of composite beam connection to circular CFT column under cyclic loading was investigated
by Wang et al. [10].

Modeling of connections usually includes modeling of shear studs.

This requires knowing the capacity of shear studs and the loadslip
relationship. Several equations have been created expressing the shear
connector capacity based on concrete properties and the ultimate
tensile strength of stud shear connector [11,12]. Makino [13] found
that the cyclic strength of studs is approximately 50% of the ones
predicted by Ollgaard et al. [12]. Oehlers et al. [11] derived the stiffness
of the stud shear connector under static and dynamic loads from 116
push-out test results using linear regression analyses. Dabaon et al.
[14] conducted 22 push-out tests in order to investigate the behavior
of different types of shear connectors in normal and high strength
concrete. The authors found that the failure mode depends on the
shape of shear connector, detailing, and strength of concrete. The size
and shape of the gusset plate connection are found to greatly affect
the structural response of concentrically braced frames.
In this paper, the nite element analysis of circular concrete lled
tube connections was conducted. The analysis incorporated geometric
and material nonlinearities. The nonlinear behavior of concrete including crushing and cracking was included in the analysis. Numerical results were veried by comparison to published tests in literature. The
veried model was used to investigate the effect of the geometric
dimensions and gusset plates as well as connection conguration on
the connection behavior including loaddeection curves, yield load, ultimate load, and initial and post-yield stiffness. Numerical results were
used also to set design recommendations of such type of connections.

Fig. 1. General view of test set-up.

H.M. Ramadan et al. / Journal of Constructional Steel Research 120 (2016) 3344

35

Table 1
Summary of tested subassemblies.
Test group

Detail no.

Test unit

D (mm)

t (mm)

tg (mm)

Concrete inll

Fcu Mpa

Fc Mpa

Loading direction

1
2
1
2
3
4
4
1
2
3
4
4
1
2
3
4
4

SP-H1
SP-H2
H-1
H-2
H-3
H-4
H-5
V-1
V-2
V-3
V-4
V-5
H-6
H-7
H-8
H-9
H-10

168.3
168.3
114.3

Filled

37.8
40.6
16.7

Perpendicular [horizontal]

Filled

47.2
50.8
21.2

114.3

None
Filled

21.2

16.7

Parallel [vertical]

114.3

12
12
12
12
6
12
12
12
12
6
12
12
12
12
12
12
12

None
Filled

26.3

20.7

Perpendicular [horizontal]

None

2. Numerical modeling
ANSYS nite element program [15] was used to build a threedimensional model representing the connection of CCFT subassemblies.
The 3-D models are built to resemble the subassemblies tested in an experimental program carried out by the author [16]. The program included a series of seventeen units designed to investigate the behavior, load
capacity and failure mode of brace-to-CCFT columns connections. The
load is applied on the subassemblies in two different directions, parallel
and perpendicular to the specimen in order to simulate the vertical and
horizontal components of the brace forces as shown in Fig. 1. The tests
were divided into four groups, A, B, C, and D depending on the loading
direction, used pipe size, and compressive strength of concrete core as
shown in Table 1. The connection details are divided into four types,
nos. 1, 2, 3, and 4. Type 1 refers to connections with gusset plate welded
directly to the steel tube shell with provision of shear connectors
welded within the connection zone and embedded inside concrete.
Type 2 refers to connections with gusset plate tting within slots in
the steel tube. Type 3 refers to connections similar to type 2 with an additional gusset plate welded perpendicular to the penetrating gusset

Perpendicular [horizontal]

plate. Type 4 refers to connections with gusset plate welded directly

to the steel tube shell which is considered as pilot connections. The connections with different shapes and conguration are attached to the
specimens' ends to avoid the effect of its deformation at mid-span on
connection behavior. Fig. 2 shows the main dimensions of the tested
specimens that consist of a small part of circular pipe fabricated
representing a part of CCFT column and connected to stiffened two
angles and thick plate at the end of gusset plate welded to the pipe to
x the actuator. It is shown that all test units are provided with two
end plates welded to the pipe to allow its xation to the lab rigid
oor. The total length of specimens is xed to 720 mm and the distance
between the gusset plate edge and all the specimens' ends is chosen
90 mm which represents the minimum probable distance in order to
decrease the moment applied on the steel tube to the least value.
The main elements included in the nite element model are the steel
tube, the conned concrete, the interface between the steel and the concrete as well as the end seats. The selection of elements and meshing
size was decided based on a sensitivity analysis [16,17] in order to
obtain accurate results with reasonable computing time. The steel
seats that were used to support the specimens are modeled using

Fig. 2. Main dimensions of tested group specimens.

36

H.M. Ramadan et al. / Journal of Constructional Steel Research 120 (2016) 3344

Fig. 5. Finite element modeling of CFT assembly.

Fig. 3. Solid modeling of the specimen components.

eight-node 3D isoparametric solid elements, SOLID45. The steel tubes

were modeled using four-node 3D nite strain isoparametric shell
elements, SHELL181. The concrete core was modeled using concrete
isoparametric solid elements, SOLID65. The interface between steel
tube and concrete inll was modeled using contact pairs: a target
element, TARGE170, and a contact element, CONTA173. Such target
and contact elements will transfer compression however the two
surfaces are allowed to separate. Shear studs were modeled using
truss elements, LINK8, which is a fair assumption since the lateral deformation of the shear connectors is fully restrained by the surrounding
concrete. Nonlinear spring elements, COMBIN39, were used to model
the loadslip curve of shear connectors. Figs. 3 & 4 show the model components. Fig. 5 shows a general view of the nite element model including mesh divisions.
3. Material properties

capability of capturing cracking in tension and crushing in compression:

uni-axial tensile cracking stress (ft), uni-axial compressive stress ( fc),
biaxial compressive stress ( fcb), ambient hydrostatic stress state (h),
biaxial crushing stress under the ambient hydrostatic stress state ( f1),
and uni-axial crushing stress under the hydrostatic stress state ( f2).
Cracking of the concrete and stressstrain relation in tension is modeled
by a linear elastic tension stiffening relationship. In the present study
the ability of concrete to transfer shear forces across the crack interface
was accounted for by using two different shear retention factors (): for
cracked shear modulus, it was assumed to be 0.2 for opened crack
representing a fairly smooth crack and 0.6 for closed crack representing
a moderately rough crack. Stiffness multiplier for cracked tensile condition was taken as 0.6. The tensile stress relaxation after cracking was
considered to allow gradual reduction of stress and overcome convergence problems resulting from the drop of stresses to zero at nodes
when a crack forms. The tensile strength was taken 0.1 of the concrete
compressive strength. The study used the equivalent uniaxial stress
strain relationship for the concrete compression by using the equation
suggested by Desayi and Krishnan [18] to compute the multilinear
isotropic stressstrain curve as shown in Fig. 6 as follows:

The material model of pipe and shear studs was assumed elastic
perfectly plastic with yield stress of 240 and 900 MPa, respectively.
The modulus of elasticity, Es, and Poisson's ratio, s, were assumed
20,000 MPa and 0.3, respectively. Von-Mises yield criteria were used
to dene the yield surface and strain-hardening was neglected. Six
different stress parameters are dened to allow concrete element

Ec
 2
1 o

where: () is the stress value at any given strain , Ec is the elastic modulus of concrete calculated as per the ACI 318m-08 [19] code and it is
equal to 26,017 MPa for fc of 30 Mpa and o is the strain at the ultimate
compressive stress taken equal to (2fc/Ec) where fc is the compressive

Fig. 4. Connection conguration.

H.M. Ramadan et al. / Journal of Constructional Steel Research 120 (2016) 3344

Fig. 6. Simplied compressive uniaxial stressstrain curve for concrete [18].

37

Fig. 9. Loaddisplacement curve for specimen H-1.

Fig. 7. Shear forceslip relationship [21].

strength of concrete cylinder after 28 days. The coefcient of friction, ,

was chosen equal to 0.20 as per Hu et al. [20]. Fig. 7 shows a general
view of the used loadslip relation that was proposed by Oehlers and
Coughlan [21].
4. Loads and boundary conditions
The CCFT assemblies were connected to the end steel seats using
bolts. Bolts were modeled by coupling nodes located at the hole interface at both ends to have the same displacements and rotations. Two anchor bolts were used to x the seats to the concrete oor. They were

Fig. 10. Loaddisplacement curve for specimen H-5.

modeled by constraining the mid-point of the anchor bolt to allow for

seat rotation around this point as observed during the experimental
test. The forces are applied across the tip of the gusset plate in order
to simulate the experimental situation. Fig. 8 shows load application
on the gusset plate. The static analysis type was used to solve the different models. The load application was applied on small increments to facilitate convergence. The NewtonRaphson approach was used to solve
the nonlinear problem. At each load step, the program calculates the difference between the applied loads and loads corresponding to element

Fig. 8. Connection loading and boundary conditions.

38

H.M. Ramadan et al. / Journal of Constructional Steel Research 120 (2016) 3344

experimental and computational results of different specimens. Fig. 13

shows the deformed shape after the testing compared to the model output. Table 2 summarizes the main experimental and numerical results
for the specimens used in the verication process. The average error
between the numerical and experimental results is 2.04%, 3.38%,
3.76%, and 3.02% for the yield load, ultimate load, initial stiffness, and
post-yield stiffness, respectively. The results indicate good correlation
between the experimental and numerical results.
6. Parametric study

Fig. 11. Loaddisplacement curve for specimen V-2.

The parametric study is performed using the calibrated models to investigate the effect of different parameters which have not been covered by the experimental works [16]. The parameters included herein
are: slenderness ratio of steel tube dened as pipe diameter to thickness
ratio, D/t, length of gusset plate and connection congurations. The
parametric study was established by varying the value of each parameter separately whereas all other parameters were kept unchanged. The
parametric study focused on the component of loading acting perpendicular to the CCFT columns only [22]. The parameters were varied
within practical limits as listed in Table 3.
7. Analysis of results

Fig. 12. Loaddisplacement curve for specimen V-5.

stresses to check that the convergence criteria were satised. The

maximum numbers of iteration per step were limited to 200. In addition, the force label convergence criteria are softened (normalized
value is set from 1% to 5%). Displacement control strategy is adopted
due to the softening behavior of concrete.
5. Model verication
Hence, the numerical models are veried using the available
experimental results. Then, the models are developed in order to
expand the parametric study. In this section, the numerical and experimental results are compared to prove the model validity for the analysis.
Figs. 9 through 12 show loaddisplacement curves obtained from

The analysis of more forty concrete lled tube connections was

conducted herein to investigate the effect of geometric properties &
connection conguration on behavior. The results include the yield
and ultimate capacities of the connection as well as the initial and
post-yield axial stiffness. The ultimate connection capacity, Pu, is dened
as proposed by Lu et al. [22] as the load corresponding to a deformation
of 3% of the diameter which is acceptable for connections that do not
exhibit a peak load. The yield connection load, Py, is dened as the
load at which the tangent stiffness reduces to one third of the initial
stiffness as proposed by Kamba and Taclendo [24].
7.1. Gusset plate length, L
Figs. 14 to 17 illustrate the inuence of increasing the gusset plate
length, L, on the behavior and capacity of the connection. Assessment
of numerical results revealed that parameters chosen to assess the connection behavior are satisfactory. Fig. 14 shows that the ultimate load,
Pu, increased with L for all connection details. The increase in (Pu) was
more pronounced in detail (3) at which (Pu) increased by 83% when L
increased from 200 to 400 mm. However, the percentage increase did
not exceed 20% in other connection details. This was expected since
using two intersecting plates will create a rigid part at the connection
zone and the stiffness increases as the gusset plate length increases.
Fig. 15 shows that the yield load, Py, was proportional to L for all

Fig. 13. Deformed shape for specimen H-5 at failure.

H.M. Ramadan et al. / Journal of Constructional Steel Research 120 (2016) 3344

39

Table 2
Comparison of experimental and numerical results.
Test unit

H-1
H-2
H-3
H-4
H-5
V-1
V-2
V-3
V-4
V-5

Yield load

Ultimate load

Initial stiffness

Post-yield stiffness

Exp. (kN)

Num. (kN)

% error

Exp. (kN)

Num. (kN)

% error

Exp. (kN/mm)

Num. (kN/mm)

% error

Exp. (kN/mm)

Num. (kN/mm)

% error

74.0
78.1
72.0
76.0
42.9
102.0
94.9
97.2
99.0
32.5

73.8
82.0
74.0
75.0
41.0
101.8
95.0
100.9
100.0
32.0

0.3
4.8
2.7
1.3
4.6
0.2
0.1
3.7
1.0
1.6

105
103.7
96
115
51.5
127.7
126.0
120.8
114.3
33.6

100.0
105.0
102.5
114.8
52.2
129.6
133.1
125.9
120.5
34.5

5.0
1.2
6.3
0.2
1.3
1.5
5.3
4.1
5.1
2.6

29.1
31.4
51.0
31.0
23.0
20.1
27.6
22.2
19.8
13.4

30.3
32.5
49.8
31.9
25.0
20.9
29.2
22.8
20.5
13.4

4.0
3.4
2.4
2.8
8.0
3.8
5.5
2.6
3.4
0.0

11.9
17.8
17.5
11.1
9.3
12.8
17.2
19.1
19.1
7.8

12.3
17.5
16.8
11.5
9.1
13.5
18.0
18.7
18.5
7.8

3.3
1.7
4.2
3.5
2.2
5.2
4.4
2.1
3.2
0.0

Table 3
Values of parameters in parametric study.
Parameter

Values

Slenderness ratio, D/t

Diameter, D (mm)
Length of gusset plate, L (mm)
Connection Conguration

20, 30, 40, 50, 60

114.3, 200, 300 and 400 mm
200, 300, and 400 mm
Detail (1) through detail (4) as shown in Fig. 2

connection details. It was noticed also that Py of detail (3) was the most
sensitive to L and increased by 56% when L increased from 200 to
400 mm. On the other hand, the percentage increase in (Py) did not exceed 35% in other connection details. Figs. 15 & 16 show that the gusset
plate length did not have a signicant inuence on the initial and postyield stiffness of the connection in the studied range. This may be attributed to the fact that the initial stiffness in CFT members is affected by the
steel tube properties and existence of the concrete inll rather than the
properties of the attachment plates.

percentages decreased to 36%, 29%, 50% and 113% with comparison to

detail (2). The precedent results indicate clearly that the connection
characteristics increase considerably by the increase of CCFT pipe diameter depending on connection details which is conrmed with the results reported by Yoo [25]. The ultimate and yield capacities of CCFT
connections with 200 mm pipes diameter are 2 to 3 times the corresponding characteristics of CCFT connections using 114.3 mm pipes.
The range reaches 5 to 7 for the initial stiffness and changes to 15 to
20 for post-yield stiffness.

7.3. Effect of pipe diameter, D

Figs. 22 to 25 illustrate the effect of CCFT diameter, D, on the properties of the CCFT connections considering xed D/t ratio equal to 40 for
the four studied connection details. It is shown that all connection
details have high sensitivity to the pipe diameter since all their properties increase considerably as the pipe diameter increases by a rate

7.2. Effect of pipe slenderness, D/t

The numerical simulations were performed for the four studied
connection details using the same range of D/t values shown in
Table 2. Figs. 18 through 21 show the inuence of D/t ratio on the
CCFT connection main characteristics using two pipe diameters 114.3
and 200 mm. It is shown that all connection characteristics decrease
as the D/t ratio increases by different rates depending on D/t values. At
lower values of D/t ratios ranging from 10 to 20, the connection characteristics decrease by 25% to 35% however such decrease is ranging from
5% to 15% for ratios 50 to 60. Connection with detail (3) generally
exhibits the highest loading capacity followed by detail (2) however detail 1 and 4 main characteristics are almost similar. The characteristics of
detail (3), Pu, Py, initial stiffness and post-yield stiffness increased by
50%, 83%, 100%, 175%, respectively when compared to detail 1. Such

Fig. 14. Ultimate loadgusset plate length behavior for CCFT 114.3 3 mm.

Fig. 15. Yield loadgusset plate length behavior for CCFT 114.3 3 mm.

Fig. 16. Initial stiffnessgusset plate length behavior for CCFT 114.3 3 mm.

40

H.M. Ramadan et al. / Journal of Constructional Steel Research 120 (2016) 3344

Fig. 17. Post-yield stiffnessgusset plate length behavior for CCFT 114.3 3 mm.

Fig. 18. CCFT ultimate loadD/t ratio.

depending on the pipe size and the connection conguration. This indicates that the behavior of CCFT connection details is dominated by its
pipe diameters. All gures show that connection details (3) & (2) developed the highest properties, while the lowest properties were developed by connection details (1) & (4). It is worth mentioning that the
nonlinear behavior observed from the numerical analysis did not necessarily indicate connection ductility. Figs. 20 and 21 show that the ultimate and yield strengths of connection details (3) & (2) are almost
25% & 8%, respectively higher than those of connection detail (4). Therefore, doubling the connecting plates at the pipe resulted in about 17% increase in the connection strength. Also it is shown that the connection
strengths increased 7 times when CCFT diameter increased from 114.3
to 400 mm. Fig. 24 shows the low rate increase of initial stiffness at
lower values of pipe diameter. Fig. 25 shows that the post-yield stiffness
is linearly proportional to pipe diameter.

Fig. 19. CCFT yield loadD/t ratio.

Fig. 20. Initial stiffnessD/t ratio.

Fig. 21. Post-yield stiffnessD/t ratio.

7.4. Effect of concrete strength on the connection behavior

Figs. 26 through 29 show the concrete strength effect on the properties of the CCFT connections considering xed D/t ratio equal to 40 and
gusset plate length equal to 300 mm for the four studied connection details. The considered values for the characteristic strength of concrete
cylinders were 25, 30, and 40 MPa. It appears that within the studied
range, the compressive strength of concrete does not have a remarkable
inuence on the different characteristics of the connection.

7.5. Effect of connection conguration on steel & concrete stress distribution

The models were used to evaluate the Von-Mises stresses at the connection zone for the studied connection congurations. Fig. 30 shows
Von-Misses patterns at the steel tube at failure. It is apparent that the

Fig. 22. Ultimate loaddiameter relation for CCFT tube with D/t = 40.

H.M. Ramadan et al. / Journal of Constructional Steel Research 120 (2016) 3344

41

Fig. 26. Effect of Fcu on ultimate load.

Fig. 23. Yield loaddiameter relation for CCFT tube with D/t = 40.

Fig. 27. Effect of Fcu on yield load.

Fig. 24. Initial stiffnessdiameter relation for CCFT tube with D/t = 40.

stresses at the plate tips are more than twice its value near the middle of plate. The high stress concentration at the plate tip makes the
pipe wall susceptible to fracture which agrees with the experimental
results. Fig. 31 shows also stress concentration in the concrete core
for the four connections at the plate tips. The concrete core stresses
extend at the front and rear sides of the connection in detail (2). Similarly, for connection detail (3), stresses increase at the location of
the gusset plate. Meanwhile, in connection with detail (4), the bearing stresses are dominant at the front side of the connection. These
observations indicate that the portion of the section participating
in transferring the load at the connection zone depends on the

Fig. 25. Post-yield stiffnessdiameter relation for CCFT tube with D/t = 40.

conguration of the connection. For connections where the gusset

plate is welded directly to the steel tube, the facing half of the column is the main part inuenced by the applied load. Meanwhile, in
connections including gusset plate penetrating through the section,
the whole section of the column is inuenced by applied loads to
the connection.
8. Conclusions
Main results from the 3-D analysis constructed to analyze the
subassemblies using ANSYS analysis program [15] are illustrated in
this section. The major conclusions are summarized as follows:
1. CCFT connection with detail (3), two perpendicular gusset plates
penetrating through the pipe, generally exhibits the highest loading capacity followed by detail (2), one gusset plate is only

Fig. 28. Effect of Fcu on initial stiffness.

42

H.M. Ramadan et al. / Journal of Constructional Steel Research 120 (2016) 3344

2.

3.
Fig. 29. Effect of Fcu on post-yield stiffness.

provided. However the main characteristics of CCFT connections

with details (1) and (4) that have gusset plates welded directly
to the steel tube with or without shear connectors are almost
similar with the lowest values. The percentage of increase in the

4.
5.

ultimate strength ranges from 10% to 60% depending upon the direction of load application, dimension of column, and existence of
concrete inll. When comparing between different details, it was
found that using connections with gusset plates penetrating the
columns show higher strength ranging from 10% to 40% or 30%
to 50% compared to the other connection details when using
CCFT columns or steel columns, respectively.
The ultimate, Pu, and yield load capacity, Py, of CCFT connection increased with the gusset plate length, L. Such increase was more pronounced in connections with details (2) & (3) since the intersecting
plates will create a rigid part at the connection zone and the stiffness
increases as the gusset plate length increases.
The gusset plate length did not have a signicant inuence on the
initial stiffness of the connection in the studied range. This may be
attributed to the fact that the initial stiffness in CCFT members is
affected by the steel tube properties and existence of the concrete
inll rather than the properties of the attachment plates.
All CCFT connection characteristics decrease as the D/t ratio increases
by different rates depending on D/t values.
All CCFT connection details have high sensitivity to the pipe diameter
since all their properties increase considerably as the pipe diameter

Fig. 30. Von-Mises stresses at steel tube for different connection congurations.

H.M. Ramadan et al. / Journal of Constructional Steel Research 120 (2016) 3344

43

Fig. 31. Von Mises stresses at concrete core for different connection congurations.

increases by a rate depending on the pipe size and the connection

conguration.
6. The compressive strength of concrete does not have a remarkable
inuence on the different characteristics of the connection.
References
[1] Y.M. Alostaz, S.P. Schneider, Analytical behavior of connections to concrete-lled
steel tubes, J. Constr. Steel Res. 40 (2) (1996) 95127.
[2] A. Elremaily, A. Azizinamini, Experimental behavior of steel beam to CFT column
connections, J. Constr. Steel Res. 57 (1993) 10991119.
[3] C. Vulcu, A. Stratan, D. Dubina, Seismic resistant welded connections for MRF of CFT
columns and I beams, 7th International Workshop on Connections in Steel Structures, 30 May02 June, 2012.
[4] C.W. Roeder, G.A. MacRae, C.A. Gunderson, D.E. Lehman, Seismic design criteria for
CFT braced frame connections, National Center for Research on Earthquake Engineering, Proceedings of the International Workshop on Steel and Concrete Composite Construction (IWSCCC-2003), Taipei, Taiwan October 89, 2003, pp. 97106.
[5] G.A. MacRae, C.W. Roeder, C.A. Gunderson, Y. Kimura, Bracebeamcolumn connections for concentrically braced frames with concrete lled tube columns, J. Struct.
Eng. (ASCE) (February, 2004).
[6] J. Zhang, D.D. Mark, J.F. Hajjar, X. Lu, Bond behavior of concrete-lled steel tube
(CFT) structures, Eng. J. (2012) 169185.
[7] J.F. Hajaar, Concrete-lled steel tube columns under earthquake loads, J. Progr.
Struct. Eng. 2 (2000) 7281.

[8] R.T. Leon, T. Perea, J.F. Hajjar, M.D. Denavit, Concrete-lled tubes columns and
beam-columns: a database for the AISC 2005 and 2010 Specications, Festschrift
Gerhard Hanswille, Germany Oct., 2011, pp. 203212.
[9] H.T. Hu, C.W. Chen, M.Y. Huang, Nonlinear nite element analysis of CFT-to-bracing
connections subjected to axial compressive forces, J. Eng. Struct. (2011).
[10] Y. Wang, Y. Xuan, P. Mao, Nonlinear nite element analysis of the steelconcrete
composite beam to concrete-lled steel tubular column joints, Int. J. Nonlinear Sci.
9 (3) (2010) 341348.
[11] D.J. Oehlers, R.P. Johnson, The strength of stud shear connections in composite
beams, J. Struct. Eng. 65B (2) (1987) 4448.
[12] J.G. Ollgaard, R.G. Slutter, J.W. Fisher, Shear strength of stud connectors in lightweight and normal-weight concrete, J. Eng. (AISC) 8 (2) (1971) 5564.
[13] M. Makino, C.W. Roeder, Design of frame steel structures with inll reinforced concrete walls, Compos. Mixed Constr. (ASCE) (1984) 279287.
[14] M. Dabon, M. Fahmy, Behavior of shear connectors using normal and high strength
concrete: part (I): experimental study, 11th International Colloquium on Structural
and Geotechnical Engineering, ICSGE, Cairo, Egypt, May, 2005.
[15] Swanson Analysis Systems, ANSYS Release 11.0, Houston, USA, 1998.
[16] M.M. Hassan, Cyclic Behavior of BeamBracingCFT Column Connections in Concentrically Braced FramesPh.D. Cairo University, Faculty of Engineering, 2013.
[17] M.M. Hassan, H.M. Ramadan, M. Abdel-Mooty, S.A. Mourad, Behavior of concentrically loaded CFT braces connections, J. Adv. Res. Cairo Univ. 5 (2013)
243252.
[18] P. Desayi, S. Krishnan, Equation for the stressstrain curve of concrete, J. Am. Concr.
Inst. 61 (1964) 345350.
[19] ACI 318, Building Code Requirements for Reinforced Concrete, American Concrete
Institute, Detroit, Michigan, 2008.

44

H.M. Ramadan et al. / Journal of Constructional Steel Research 120 (2016) 3344

[20] J.W. Hu, Y.S. Kang, D.H. Chio, T. Park, Seismic design, performance, and behavior of
composite-moment frames with steel beam-to-concrete lled tube column connections, Int. J. Steel Struct. 10 (2) (May, 2010) 177191.
[21] D.J. Oehlers, C.G. Coughlan, The shear stiffness of stud shear connections in composite beams, J. Constr. Steel Res. 6 (1986) 273284.
[22] L.H. Lu, G.D. de Winkel, Y. Yu, J. Wardenier, Deformation limit for the ultimate
strength of hollow sections, Proceeding of 6th International Symposium on Tubular
Structures, Melbourne, Australia 1994, pp. 341347.

[23] Y. Kurobane, J.A. Packer, J. Wardenier, N. Yeomans, Design Guide for Structural Hollow Section Column Connections, CIDECT, TUV-Verlag GmbH, koln, Germany,
2004.
[24] T. Kamba, C. Taclendo, CHS column connections without stiffeners, Proceeding of
8th International Symposium on Tubular Structures, Singapore 1998, pp. 567576.
[25] J.H. Yoo, Analytical Investigation on the Seismic Performance of Special Concentrically Braced Framesa thesis submitted in partial fulllment of the Ph.D. University
of Washington, Seattle, 2006.