Ultimate Anchorage Capacity of Concrete Filled Steel Box
Connection as Footing
Muhammad Aun Bashir
Candidate for the Degree of Doctor of Philosophy
Supervisor: Prof. Tamon UEDA
Division of Built Environment
Introduction
Damage due to recent earthquakes has resulted in an
increased conservatism in the design of piles and pile-
to-pile cap connections. In contrast, current
recommendations produce connection detailing which
result in high levels of congestion of steel reinforcement
in pile cap, which make extremely difficult to construct
according to designer’s recommendation.
By considering this point, a new type of footing,
concrete filled steel box (CSFB) connection as footing,
is presented. Steel concrete sandwich structure is a
relatively new form of structural system. This form of
structure has the potential to fully utilize strength of
both steel and concrete with help of composite action
such as confinement. It allows the prefabrication of
large section in a factory, and enables rapid installation
into main structure, dramatically reducing the
fabrication cost and construction time. Steel faces act as
permanent formwork during the construction and
provide impermeable skins for the structure upon
completion.
Our purpose in this study is to develop design
method for the CSFB footing by using a 3-D nonlinear
finite element program CAMUI, which the authors’
group has developed. Therefore we can take part in the
development of simple and economical hybrid structure
with excellent mechanical properties.
First of all, numerical simulation, with the help of
3D non linear FEM program, of experimental specimens
was carried out. The experimental program included
five specimens were conducted in two phases. First
phase included two specimens subjected to axial loading
on column. The variable parameter for the first phase is
concrete depth under the column. Second phase
included three specimens subjected to monotonic
loading perpendicular to longitudinal axis of column.
The variable parameter for the second phase is column
insertion length. After confirming the reliability of 3D
non linear FEM program, parametric analysis was
carried out and these parameters are insertion length,
concrete strength, box size, column size, column length
and box tube thickness. In this paper the effect of these
important parameters are closely observed that will be
useful for the development of design method of this
type of footing.
FEM ANALYSIS
Constitutive Model of Materials
For un-cracked concrete and steel, elasto-plastic fracture
model is used. The elasto-plastic fracture model divides
concrete nonlinearity into continuum damage and
plasticity, and this model is also very suitable for steel.
The adopted failure criteria that acted in agreement with
Niwa’s model in tension-compression zone and Aoyagi
and Yamada’s model in tension-tension region were
extended to three-dimensional criteria by satisfying
boundary conditions. For cracked concrete until peak,
Vecchio and Collins model is used but for post peak,
fracture energy concept was used. Fracture energy
concept will be modified to predict the post peak
behavior of the confined concrete with steel.
When crack occurs, a local coordinate system based
on each crack plane is defined. In the case of 2 cracks
occurring, two local coordinate systems arranged to
share a parallel axis at the intersection line between the
two crack planes. Constitutive models are applied in the
direction parallel as well as normal to the crack and to
shear slip along the crack planes. Global stresses are
calculated by superposing the stress calculated in each
local coordinate.
Modification of compression softening
model for cracked concrete
The numerical solution of the finite element method
using the smeared crack model is generally affected by
the element size when the strain softening material
model is adopted. This poses a severe problem to solve
the post peak behavior of concrete structures. One
solution of the problem was suggested by Nakamura is
to adjust the stress strain relation according to the
fracture energy balance in terms of element size. This is
why; the fracture energy is treated as a material property
and is kept constant in localized element regardless of
element size. As we also know that post peak behavior
of confined concrete varies with level of confinement.
So by keeping these points in mind the fracture energy
equation is modified in such a way that it should be a
function of lateral confining stress. After peak stress, the
effect of crack on compression softening is considered
by the linear descending line. Reduced compressive
stress has a limit that is 10% of the confined
compressive strength. The gradient of strain softening is
defined by compressive fracture energy.
( a + bf l / f c ' )
G fc = 8 . 8 × ( f ' cc ) (1)
Where,
f’cc : Confined concrete stress = fc’ + 4.1* fl
fl : Lateral confining stress (average of two lateral
stresses from 3D analysis)
Eq. (1) is modified form of Nakamura’s equation.
The modification is change of fc’ to f’cc and “0.5” is
changed to variable form i.e. (a +b fl/ fc’). These two
terms are modified because of following reasons.
Fracture energy increases with the increase of
confinement. The equation should be valid for both
confined and unconfined concrete e.g. for confined
concrete, the second term (b fl/ fc’) will be added to “a”
(because “fl” will be compressive and sign convention
is positive for compression). But on the other hand, for
unconfined concrete, the second term (b fl / fc’) will be
subtracted from “a” (because “fl” will be tensile and
sign convention is negative for tension). “a” and “b”
variable of the test specimens was concrete depth under
the column. The steel box and column dimension are the
same for all the specimens. Fig. 4 presents the
dimensions of the tested specimens of Phase 1. Axial
loads were applied through inserted square mortar filled
steel tube as the column. Phase 2 consists of three
specimens. The notations A1, A2 and A3 will be used
1400
1200

Axial Force (KN)

are the parameters which are determined by trial and 1000
error technique. In order to find “a” and “b” rather 800
Experimental
simple cases of confined concrete cases are chosen. 600 a = 0.5
400 a = 0.67
Concrete-filled steel tube (CFT) cases subjected to axial a = 0.86
200 a=1
compression are an ideal case to see the applicability of
0
the post peak behavior with confinement. The limit 0 0.005 0.01 0.015 0.02
strain for compression is calculated from Nakamura et
Strain (No unit)
al.
2G ε Fig.1 Axial force versus axial strain for Su-29
εu = fc
+
p
(2) 4000
σ peak × l eq 2

Axial Force (KN)

3000
 ε − ε u 
σ = σ   (3) 2000
 ε u − ε d 
peak
Experiement
1000 a = 0.5
a = 0.86
Due to symmetry, only one eighth of the CFT column is 0
analyzed. Symmetric boundary conditions enforced on 0 0.005 0.01
the symmetric planes. The corners of CFT with square Strain (No unit )
sections are assumed to be exact 90° and corner radii are
not considered. The uniform compressive loading is Fig.2 Axial force versus axial strain for Su-70
applied to the top surface of the column directly. 2500
Specimens from previous studies, that show concrete
Axial Force (KN)

2000
crushing dominant failure, are analyzed by varying the 1500
parameters “a” and “b” in the proposed fracture energy Experimental
equation. In order to find the value of parameters “a” 1000 a = 0.5
a = 0.86
and “b”, first of all value of “b” is assumed and “a” is 500
varied. So in order to make our equation suitable for 0
both confined and unconfined concrete, the value of 0 0.005 0.01 0.015
“b” was assumed in such a way that (a+ b fl / fc’) Strain (No unit)
should become close to “0.5” value for unconfined
concrete. Parameters that precisely simulate the post Fig.3 Axial force versus axial strain for Su-17
peak behavior of confined concrete are selected and
then verified for other specimens. Furthermore, it can be
30 mm
observed from Fig. 1, Fig. 2 and Fig. 3 that with the ts = 4.5 mm
increase of confinement, post peak part of the load
200 mm
strain curve becomes more ductile. So it means fracture
energy increases with increase of confinement level. d
tb = 25 mm
This was the basic reason of the modification of the
300 mm
fracture energy equation. After analysis it can be 100 mm 100 mm

observed from Fig.1, Fig. 2 and Fig. 3 that for a = 0.86, 500 mm
best simulation for post peak behavior of CFT column is 320 mm
For S_37 d = 37 mm
achieved. So after trial and error method value of “a” For S_106 d = 106 mm
and “b” were decided that is 0.86 and 7 respectively. 300 mm 300 mm

Experimental program Fig. 4 Test set up and specimen details for Phase 1

Experimental program consists of two phases. First Table 1. Specimen details for Phase 2
phase was carried out to conduct the reliability of non Specimens
Insertion length Box width (mm)
linear finite element program (CAMUI) for axial (mm)
A1 63.8 342.8
capacity and second phase was carried out to conduct
A2 97.8 342.8
the reliability of the program for moment capacity. A3 142.8 342.8
Phase 1 consists of two specimens. The notations S_37
to S_106 will be used for the specimens. The principal
 Load
Inner pile 500 load capacity. The higher stiffness of the analytical
250 250
Outer pile specimens A2 and A3 in comparison with the tested
171.4
stiffness is probably because of the unexpected rotation

Box size
342.8
of the outer pile due to the incomplete filling between
171.4
228.6
the inner and outer pile. This difference of stiffness
200 715 L becomes more prominent with the increase of peak load.
Insertion length 547.2 It was also observed that with the increase of column
718.6 insertion length, the ultimate capacity of the connection
226.8
increased. This increase of peak load is because of
increase in resistive moment.
30

50 50 Strain (no unit)

Fixed
-0.03 -0.025 -0.02 -0.015 -0.01 -0.005 0
600
Units: mm 0

Fig. 5 Test set up and specimen details for Phase 2

Compressive stress (N/mm )

2
S_37 S_106
for the specimens. The principal variable of the test -10
specimens was insertion length of the pile. The details
of experimental specimens for Phase 2 are given in
Table 1. The column size and material strength were the -20
same for all the specimens. Fig. 5 presents the
dimension of the tested specimens of Phase 2. Bending
moment was applied through the inserted pile. The inner -30
pile was mortar filled circular steel tube, while the outer
pile was circular steel tube. The concrete filled steel box
Fig. 6 Compression softening of concrete under
together with the inserted inner pile was prepared first
column
and the outer pile was set after wards. The gap between 1000 Experimental (S_37)
the inner and outer pile was carefully filled by mortar. 900 Experimental (S_106)
In order to assure easy handling when the pile was fixed 800
Analytical (S_37)
Analytical (S_106)
at the target location in the steel box we introduced the 700
outer pile. The steel box was fixed by the connected 600
Load (kN)

steel beam which was fixed at its end. 500

400
300
Numerical simulation of experimental 200
specimen 100
0
In order to show the reliability of the FEM simulation 0 10 20 30

by CAMUI, the experimental results of Phase 1 and Displacement (mm)

Phase 2 were compared with the simulation results. Fig. 7 Comparison of analytical and experimental
As the first step failure mechanism of the footings was results of Phase 1
observed. Crushing of concrete under the column is the 60
A1(Exp) A1(Analysis) A2(Exp)
dominant factor for the failure of the both specimens
A2(Analysis) A3(Exp) A3(Analysis)
S_37 and S_106 as shown in Fig. 6 in which the stress-
strain relationship at the gauss point 7 mm away from 40
Load (kN)

the column bottom show compression softening. From

Fig. 7, it can be observed that experimental and
analytical load displacement relationship shows good 20
agreement in terms of ultimate load capacity but post
peak part of the experimental specimen is very ductile.
This is probably because of the extensive local bending 0
deformation of bottom steel plate took place underneath 0 10 20 30 40 50
the column, which cannot be simulated by the FEM Displacement (mm)

ignoring geometrical nonlinearity. It was also observed Fig. 8 Comparison of analytical and experimental
that peak load of specimen with the greater concrete results of Phase 2
layer depth under the column was higher than that of
specimen with the smaller concrete depth under the Parametric study on axial capacity of
column. This increase of peak load is due to higher peak concrete filled steel box footing
stress on concrete under the column.
Crushing of concrete surrounding the column is the
The analytical specimen is pile footings which
dominant factor for the failure of the all specimens of
comprises of concrete filled steel tube (CFT) column
Phase 2. From Fig. 8, it can be observed that
inserted in concrete filled steel box and are supported by
experimental and analytical load displacement
four circular piles. The analytical models, as shown in
relationship shows good agreement in terms of ultimate
Fig. 9, are prepared by using commercially available
software (3D Sigma). The specimen is symmetric, so in
order to reduce the mesh number, one quarter of the
specimen is modeled. In one analysis one parameter is
variable and the rest of the parameters are kept constant.
200 mm
Diameter = 200 mm
Effect of steel box size
In order to see the effect of box size on the strength of
footing, four cases with different box size were
analyzed. The box sizes for these cases were (800 x
800) mm, (1100 x 1100) mm, (1250 x 1250) mm and
(1400 x 1400) mm respectively.

100 mm 1400
200mm
1200
320 mm

1000

Load (kN)
800
300 mm 150 mm
Fixed_Condition
600
200 mm 200 mm Roller_with bond element
400

Pile thickness = 20 mm 200

150 mm
Column thickness = 10 mm
400 mm 0
Box thickness = 10 mm
800 mm 0 10 20 30 40
Displacement (mm)
Fig. 9 Specimen details for concrete filled steel box
footing under axial load Fig. 10 Comparison of load displacement relationship
for different boundary conditions
Effect of boundary conditions 1400
1200
Peak load (kN) 1000
First of all, the pile footing was analyzed for two 800
extreme boundary conditions. For one case piles were 600
fixed and for 2nd case bond link elements are inserted 400
200
between the pile and roller support to allow horizontal 0
movement and rotation of the pile. After analysis, no 0 50 100 150 200 250
difference was observed in terms of load displacement Concrete depth (mm)
characteristics as shown in Fig. 10. So after this
observation, the pile fixed at bottom was adopted for all Fig. 11 Relationship between axial capacity and
analytical work of the pile footing under axial load. concrete depth under the column
1800
Effect of concrete depth under the column 1600
1400
Peak load(kN)

1200
In order to see the effect of concrete depth under the 1000
column on the strength of pile footing, three cases with 800
100mm, 150 mm and 200 mm concrete depth were 600
400
analyzed. After analyzing the specimens, it was
200
observed that strength of the connection decreases with 0
the decrease of the concrete depth under the column as 0 10 20 30 40 50 60
shown in Fig. 11. This reduction of axial capacity is Concrete strength (N/mm )
2

because of the reduction in peak stress of concrete under

the column. The peak stress decreases with the decrease
in the concrete depth is due to the increase in tensile
strain in the direction normal to the peak compressive
stress. It is believed that the tensile strain is caused as
bending action in the steel box.
Effect of concrete strength
In order to see the effect of concrete strength on the
strength of footing, four cases with different concrete
strength were analyzed. The compressive strength of
concrete for these cases was 24 MPa, 30 MPa, 40 MPa
and 50 MPa respectively. After the analysis, it was
observed that axial capacity of the footing increases
with the increase of concrete strength as seen in Fig. 12.
This increase of peak load is because of the increase in
peak stress. Peak stress of concrete increased due to
increase in concrete strength.
Fig. 12 Relationship between axial capacity and
concrete strength
After the analysis, it was observed that axial
capacity of the footing increases with the decrease of
box size as seen in Fig. 13. This increase of capacity is
because of the decrease in tensile strains due to bending
of the beam action with span between the supports in
the box. The tensile strain due to the bending in the box
decreases with decrease of span to depth ratio. Due to
the decrease of tensile strain as confining stress, the
peak stress of concrete increases. This reduction of peak
stress of concrete under the column is the cause of the
capacity reduction of larger box size.
Effect of column size
In order to see the effect of column size on the strength
of footing, three cases with different column sizes were
analyzed. The column sizes for these cases were (150 x
150) mm, (200 x 200) mm and (250 x 250) mm
respectively. After the analysis, it was observed that
axial capacity of the footing increases with the increase
of column size as seen in Fig. 14. This increase of peak
load is because of the increase in bearing area of
concrete in which the compression crushing occurs.
bearing area of concrete in which compression crushing
takes place and bearing area is increased due to increase
in length. Moment arm is also increased due to increase
of insertion length. Increase of resistive force and
moment arm can be seen in Fig. 17.

1400
1200
Axial load (kN)

1000
800
600
400
200
Fixed at bottom
0
0 500 1000 1500
Box size (mm)
Fig. 15 Specimen details for concrete filled steel box
Fig. 13 Relationship between axial capacity and box footing connection of column
450
size
400
1600 350
1400 Peak load (kN) 300
1200
250
Axial load (kN)

200
1000
150
800
100
600 50
400 0
200 0 50 100 150 200 250 300
0
Insertion length (mm)
0 50 100 150 200 250 300
Column size (mm) Fig. 16 Relationship between peak load and insertion
length for moment anchorage
Fig. 14 Relationship between axial capacity and
30
column size Insertion_260 Insertion_210 Insertion_160
Moment arm is increasing with the
20 increase of insertion length
Parametric study on moment anchorage
Compressive stress (N/mm )
2

capacity of concrete filled steel box footing 10

R2
F
F = R1 -R2
The analytical specimens were footings which 0
comprises of concrete filled steel tube (CFT) column
Column
inserted in concrete filled steel box and was fixed at 0
bottom. Analytical models, as shown in Fig. 15, were
prepared by using commercially available software (3D -10
R1
Sigma). The specimen is symmetric, so in order to
reduce the mesh number, one half of the specimen was -20

modeled. In one analysis one parameter is variable and

the rest of the parameters are kept constant.
Effect of insertion length
In order to see the effect of insertion length on the
strength of connection, four cases with different
insertion length were analyzed. The insertion length for
these cases was 160 mm, 200 mm, 210 mm and 260 mm
respectively. After the analysis, it was observed that
anchorage capacity of the footing increases with the
increase of insertion length as seen in Fig. 16. This
increase of peak load is because of the increase in
resistive moment. Resistive moment increases with the
increase of resistive force and moment arm. The
increase in resistive force is because of the increase in
-30 R1 is increasing with increase of insertion length
Fig. 17 Stress distribution around pile
Effect of concrete strength
In order to see the effect of concrete strength on the
strength of connection, four cases with different
concrete strength were analyzed. The compressive
strength of concrete for these cases was 18 MPa, 24
MPa, 36 MPa and 48 MPa respectively. After the
analysis, it was observed that anchorage capacity of the
footing increases with the increase of concrete strength
as seen in Fig. 18. This increase of peak load is because
of the increase in peak stress with the increase in
concrete strength.
Effect of column size
In order to see the effect of column size on the strength
of connection, three cases with different column sizes
were analyzed. The column sizes for these cases were
(150 x 150) mm, (200 x 200) mm and (250 x 250) mm
respectively. After the analysis, it was observed that
anchorage capacity of the footing increases with the
increase of column size as seen in Fig. 19. This increase
of peak load is because of the increase in bearing area of
concrete where crushing in compression takes place.
700
600
Peak load (kN )
stress is increased and as a result ultimate capacity
increased.
Effect of column length
Since the loading point of the CFT column is at its end,
a longer column gives a greater ratio of bending
moment to shear force in the column at the location
where the column insertion starts. In order to see the
effect of column length on the strength of connection,
four cases with different column length were analyzed.
The column length, outside the footing, for these cases
was 400 mm, 800 mm, 1200 mm and 1600 mm
respectively. After the analysis, it was observed that for
longer pile length, failure (concrete crushing around the

500
column) takes place due to moment and for shorter pile
400
length shear force is the dominant cause of failure. This
300 can be observed in the Fig. 22 in which with the
200 increase of column length, ultimate moment capacity of
100 the connection becomes constant.
0
0 10 20 30 40 50 60 350
2 300
Concrete strength (N/mm ) Peak load (kN)
250
Fig. 18 Relationship between peak load and concrete 200
strength for moment anchorage 150
100
450
400 50
350 0
Peak load (kN)

300 0 200 400 600 800 1000 1200 1400

250 Box width (mm)
200
150 Fig. 20 Relationship between peak load and box width
100
for moment anchorage
50
0
0 50 100 150 200 250 300 400
350
Column size (mm)
300
Peak load (kN)

Fig. 19 Relationship between peak load and column 250

size for moment anchorage 200

150
100
Effect of box size 50
0
In order to see the effect of box size on the strength of 0 5 10 15 20
connection, four cases with different box width were Box thickness (mm)
analyzed. The ratio of width to column section size for
these cases was 2, 3, 4 and 6 respectively. After the Fig. 21 Relationship between peak load and box
analysis, it was observed that anchorage capacity of the thickness
footing remains almost same with the increase of box 300
width as seen in Fig. 20. 250
Moment (kN-m)

200
Effect of box thickness
150
100
In order to see the effect of box tube thickness on the
strength of connection, three cases with different box 50

tube thickness were analyzed. The thickness for these 0

cases was 5 mm, 10 mm and 15 mm respectively. After 0 500 1000 1500 2000
the analysis, it was observed that anchorage capacity of Column length (mm)
the footing increases with the increase of box tube
thickness as seen in Fig. 21. This increase of capacity Fig. 22 Relationship between moment anchorage and
is because of the better confinement of concrete column length
surrounding the pile. Due to better confinement, peak