Moratuwa Engineering Research Conference (MERCon) 2020

Study on Concrete Filled Steel Circular and Square

Tubes
P.B.M.R. Bogahawaththa K. P. Madhuranga K. Baskaran
Department of Civil Engineering Department of Civil Engineering Department of Civil Engineering
University of Moratuwa University of Moratuwa University of Moratuwa
Moratuwa, Sri Lanka Moratuwa, Sri Lanka Moratuwa, Sri Lanka
198023u@uom.lk 150380j@uom.lk baskaran@civil.mrt.ac.lk

Hidallana-Gamage H. D.
Department of Civil Engineering
University of Moratuwa
Moratuwa, Sri Lanka
hasithad@uom.lk

Abstract— Concrete filled steel tube (CFST) is a composite Several types of research have been conducted to examine
material which is increasingly used in the civil engineering the effects of various parameters on the compressive
industry. So, it is required to identify the properties of concrete- behaviour and failure modes of CFSTs under experimental
filled steel tubes as a composite material. This paper focuses on programmes. Gupta studied the effects the grade of concrete
a study conducted on concrete-filled circular steel tubes and
infill, D/t ratio (diameter/ steel tube thickness ratio), L/D ratio
galvanised square sections. Concrete filled circular steel tubes
were tested under axial compression with three different and volume of fly ash in the concrete mix [1]. The number of
diameter-to-thickness (D/t) ratios and two different concrete 81 specimens including 9 specimens of a hollow tube of the
compressive strengths since CFSTs are most effective as circular cross-section with infill concrete strength 30 MPa
compression members. Galvanised steel square sections were and 40 MPa was used. It was noted that the effect of the
also tested for compression capacity and failure mode. Finite confinement increases with the increase of infill concrete
element model was developed using the ABAQUS software and strength and smaller of D/t ratio. FE analysis has been done
two experimental programmes were carried out to find material using FE software like ANSYS and ABAQUS by previous
properties of concrete and steel for finite element modelling and research to compare and verify those with the experimental
two experimental programmes were carried out to verify the
results. Ellobody examined the behaviour and design of
finite element modelling. Galvanised iron square sections were
modelled using MIDAS gen software and verified with axially loaded circular CFST columns numerically using an
experimental results for the failure load and failure mode also. ABAQUS based FE model [2]. FE analysis results were
The axial capacity of CFST was found theoretically using two compared and verified with experimental results of tests
different codes and one equation found in the literature. conducted by Giakoumelis and Lam and Sakino. It was
Experimental as well as numerical results showed that the axial concluded that there was a good agreement with column
capacity of CFST increases as the concrete infill increases and strength, load–axial shortening curves and deformed shapes
decreases with D/t ratios. of the columns. Some of the research has compared the
compressive axial capacities of CFST obtained from
Keywords— concrete-filled steel tube; composite material;
structural and economical effectiveness; axial compression; finite
experimental study and FEM analysis with theoretical values
element modelling predicted by different international codes, standards and
suggested equations. The American Concrete Institute (ACI
I. INTRODUCTION 318) [3], American Institute of Steel Construction (AISC),
CFST is a type of composite material which is being used the Australian Standard (AS) and Eurocode 4 are the
in many countries for a wide range of construction purposes. standards and codes [4]. The equation developed by Mander
CFST consists of two main construction materials concrete can be applied to CFST columns utilizing the defined
and steel therefore properties of CFST are better than those effective lateral confining stress that depends on the structural
individual materials and has more advantages than individual steel dimensions [5].
materials. The tendency to buckle locally of steel tubes is
higher in compression. In addition to that, concrete along does II. STUDY ON CIRCULAR TUBES
not have high ductility when considering separately and steel Concrete compression test and steel tensile tests were
does not have good fire resistance also. So, considering both conducted to determine the data which is required to develop
materials as composite material weak side of both materials FE model and other main two tests which are Circular hollow
can be minimized. CFST has higher compressive strength steel (CHS)tube compression test and compression test on
because of the confining effect provided by steel. CFST has CFST were conducted to verify the FE model and the
more resistance to local buckling due to its higher stiffness. theoretical calculation to identify the behaviour of CFST
Fire resistance is also higher due to the concrete core. Since under axial compression.
steel tube behaves as the formwork during construction it is
easy to do the rapid construction and reduce the cost. A. Experimental Details
Therefore, CFST is the solution to enhance the good quality Specimens were prepared depending on the D/t ratio,
material for the industry. availability of market at that time when those were purchased.
To determine the behaviour of the composite CFST column
and CHS, respectively 6 specimens with two different

978-1-7281-9975-7/20/$31.00 ©2020 IEEE

42
Authorized licensed use limited to: Auckland University of Technology. Downloaded on October 03,2020 at 23:55:28 UTC from IEEE Xplore. Restrictions apply.
 Moratuwa Engineering Research Conference (MERCon) 2020

concrete strength, three different wall thickness for CFST and III. THEORETICAL CALCULATION
three different wall thickness for CHS were tested. Concrete Two different codes and one equation were used to
mix proportions for compressive strength 75 / , 81.5 / determine the compression capacity of CFST.
and quantity of 0.027 3 which were used are mentioned
in Table I. First steel tubes were cut by lathe machine to get A. AISC Code
both surfaces to be levelled and inner and outer surfaces of the For the axially loaded CFST columns, the design
steel tubes were cleaned, and rust was removed. The inner compressive capacity NAISC can be calculated for the
surface was moistured before filling the concrete to reduce flexural buckling limit state based on member slenderness as
water absorption from the concrete mix. Steel tubes were follows:
placed on a horizontal surface and horizontality was ensured,
using a level. Then concrete was poured in the tubes by four
layers. Each layer was compacted using a tamping rod, by NAISC = 0.685 ≤ 2. (1)
giving 25 strokes. After casting samples top surface was Pe
covered by polythene sheet for two days for curing. After 55
days CFST were prepared to test. All the details of the selected NAISC = 0.877Pe > 2.25 (2)
specimen are mentioned in Table II. Pe

TABLE I. CONCRETE MIX PROPORTIONS The nominal axial strength ( ) of compact CFST specimen,
where =0.95 for circular CFST sections.
Concrete Cemen Wate Fine Coarse Admixt Final
Com. t r Aggr. aggr. ure
Strength (kg) (kg) (kg) (ml) Slump = + (3)
( / (kg)
(mm)
) : concrete area
75 10.125 4.92 19.7 26.83 81 155 : steel area

81.5 11.205 4.68 18.28 26.28 89 160

The elastic critical buckling load ( ) can be calculated using
Euler's formula
TABLE II. DETAILS OF SELECTED SPECIMENS ( )
= (4)
No. t D L D/t fc ( )
(mm) (mm) (mm) ( / )
1 4.19 113.95 325 27.19 ---
For the tested CFST ( ) can be calculated using
2 4.19 113.95 325 27.19 75
3 4.19 113.95 325 27.19 81.5 ( ) = + (5)
4 4.76 107.76 325 22.63 ---
where accounts for the effective rigidity of filled
5 4.76 107.76 325 22.63 75
composite compression member and is calculated as follows.
6 4.76 107.76 325 22.63 81.5
7 6.5 114.65 325 17.63 ---
= 0.6 + 2 ≤ 0.9 (6)
8 6.5 114.65 325 17.63 75 +
9 6.5 114.65 325 17.63 81.5
B. Mander’s Equation
B. Experimental Results Axial compression capacity (NMan) of CFST can be found
using the following formulae which consider the confinement
The failure load of specimens both CHS and CFST that effect of the concrete. So, this formula is more accurate.
were measured during the test are shown in Table III. Failure
shape of the specimens is shown in Fig. 1.
= + (7)

Where is the compressive strength of confined concrete,

is the area of steel section, Ac is the area of the concrete
section, is the yield strength of steel and can be
determined by the following equation.

7.94
= −1.254 + 2.254 1 + −2 (8)

Where is the compressive strength of concrete, is

effective lateral confining stress on concrete which can be
determined by,
Fig. 1. Failure shape of CHS (test no.4) & CFST (test no.5)

43
Authorized licensed use limited to: Auckland University of Technology. Downloaded on October 03,2020 at 23:55:28 UTC from IEEE Xplore. Restrictions apply.
 Moratuwa Engineering Research Conference (MERCon) 2020

2
= Where = 0.1 (9) IV. FINITE ELEMENT MODEL
Two finite element models were developed using
Calculated axial compression capacities of tested ABAQUS software to predict the behaviour of hollow steel
specimens according to Mander’s formulae are mentioned in (CHS) tube and CFST under axial compression.
Table III. A. Material Properties
C. Eurocode 4 1) Concrete
The axial capacity of the CFST column can be predicted Strength and ductility of concrete in CFST are increased
based on the Eurocode 4 equation as follows: under axial compression because of confinement given by the
steel tube. So accurate concrete model is needed to describe
′ the behaviour of concrete. The method proposed by Mander
= + 1+ ′ (10) was used to find the confined compressive strength of
Υ Υ
concrete. Stress-strain curve for confined concrete compared
with unconfined concrete is shown in Fig. 2. Here
Where Υ and Υ are respectively 1.0 and 1.3 which are ′
, ′ , ′ , ′ are respectively confined concrete compressive
structural steel and concrete material partial safety factors.
Because of the pure axial loading of the CFST members strength, unconfined concrete cylinder compressive strength
which is equal to 0.8( fcu) and fcu is the unconfined concrete
tested in this study, the coefficients and are replaced
cube compressive strength, corresponding confined stain and
with and as given by the following equations
unconfined strain. Drucker-Prager criteria were used to define
respectively. the plastic properties of concrete. Key material properties such
as the angle of friction, flow stress ratio and dilation angle (ψ)
0
= 0.25(3.2 ) ≤ 1 (11) are mentioned in Table IV.
2
0
= 4.9 − 18.5 + 17 0 (12) TABLE IV. PROPERTIES OF DRUCKER PRAGER CRITERIA
Angle of friction (deg) Flow-stress Dilation angle
Where λ is the relative slenderness ratio of the CFST column ratio (ψ) (deg)
in the plane of bending and is the elastic critical load. 38 1 17

= (13)

D. Comparison of Theoretical and Experimental Results

The calculated capacities of the CFST specimens using the
above mentioned theoretical approach are compared with
experimental results and it is mentioned in Table III. It can be
identified that underestimation and overestimation of the
compression capacity of CFST columns calculated by AISC
and Eurocode 4 codes. But prediction ranges are 6% to13%
and 1% to 13% respectively for both approaches. Mander’s
have overestimated the compression capacity of CFST in all
cases ranges from 3% to 25%. The reason for underestimation Fig. 2. Stress-strain curve of unconfined and confined concrete
is less account for the concrete confinement to axial capacities
and reason for overestimation is more account for the concrete 2) Steel
confinement to axial capacities. Required material properties like yield point, Young’s
modulus is determined from the stress-strain graph obtained
TABLE III. COMPARISON BETWEEN EXPERIMENTAL RESULT AND from the steel coupon test. Young’s modulus of the steel was
THEORETICAL RESULTS determined as 210 GPa and Poisson's ratio of the steel was
Test Diff- Diff- Diff- used as 0.3. The idealized stress-strain curve shown in Fig.3
No. (kN) (kN) (kN) (kN) (2-1) (3-1) (4-1) was used to find the σ–ε model and the plastic properties for
(1) (2) (3) (4) % % % steel.
2 1205. 1279.6 1508 1361 6.1 25 12.8
6
3 1251. 1332 1586 1406 6.4 26 12.3
5
5 1519. 1377.2 1584 1501 -9 4.2 -1.2
4
6 1640. 1423 1704 1540 -13 3.8 -6.12
5
8 1591. 1484.8 1755 1636 -6.7 10.3 2.8
4
9 1650. 1534.3 1835 1678 -7.02 11 1.6
2

Fig. 3. Idealized stress-strain curve for steel

44
Authorized licensed use limited to: Auckland University of Technology. Downloaded on October 03,2020 at 23:55:28 UTC from IEEE Xplore. Restrictions apply.
 Moratuwa Engineering Research Conference (MERCon) 2020

B. Parts TABLE V. COMPARISON BETWEEN EXPERIMENTAL RESULTS AND

FEM RESULTS
8-node brick elements with three translation degrees of
Test Concrete D/t Difference
freedom at each node (C3D8R) and four-node shell elements No. Com.strength (kN) (kN) %
with reduced integration (S4R) were used to model the (MPa)
concrete core and steel tube, respectively. End plates were not 1 No fill 27.19 395.5 661 67.1
used. 2 75 27.19 1205.6 1188 -1.45
C. Interaction 3 81.5 27.19 1251.5 1222 -2.35
4 No fill 22.63 865.5 902 4.21
Surface-to-surface contact was used for the interaction 5 75 22.63 1519.4 1505 -0.94
simulation between the steel tube and concrete. The inner 6 81.5 22.63 1640.5 1533 -6.55
surface of the steel tube was selected as the master surface 7 No fill 17.63 980.3 959 -2.17
since the stiffness of steel is higher than concrete and the 8 75 17.63 1591.4 1569 -1.40
outer surface of the concrete core was selected as slave 9 81.5 17.63 1650.4 1618 -1.96
surface. For CFST columns, there is no slip between the steel
tube and concrete since they are loaded simultaneously. H. Comparison of FEM and Experimental Results
D. Contact Interaction Property The output of the FE model was compared with
experimental results under the criteria of failure load and
“Hard contact” in the normal direction which allows the failure shape. Figures 5 and 6 show a comparison of failure
separation of the interface in tension and no penetration of shapes of specimen no.4 and no.5. Table V shows a
that in compression was assigned for the interface. The comparison of failure loads from two approaches. Both CFST
tangent behaviour was used selecting the ‘penalty’ method as and CHS steel tubes have shown an outward buckling near
friction formulation. The friction coefficient was defined as the top and bottom ends in tested specimen and FEM. Failure
0.6 initially and it was calibrated depending on the load for CHS determined from both approaches do not vary
experimental result. too much except for test no.1. Results have less than 5%
E. Loading and Boundary Conditions variation for test no.4 and no.7 while having 67% variation
for test no.1. Specimen no.1 had a fracture in middle and
During the experiment specimen was placed between the
failure has occurred there. Failure loads for CFST obtained
stiff plate of the testing machine and load was applied to the
from FE analysis comply with the loads measured during the
top plate. Bottom of the specimen was restrained in all degree
of freedom defining fixed support. Top of the specimen was experiment having less than 7% variations.
restrained in all rotations and displacements except
displacement along the longitudinal direction.
F. Meshing
The size of the mesh was kept 5 mm for the CHS and 20
mm for CFST. As mentioned in the literature, even though
finer mesh provides very accurate result but coarse mesh also
provides an acceptable result for CFST.
G. Results
The failure load and shapes of the specimens are shown
in Table V and Figures 5 and 6 respectively. Load vs.
displacement graph obtained as history output for specimen
no. 8 on ABAQUS is shown in Fig. 4.

Fig. 5. Comparison of failure shape of CHS (test no.4)

1800
1600
1400
1200
Load (kN)

1000
800
600
400
200
0
0 5 10 15 20 25 30 35
Displacement (mm)

Fig. 4. Load vs. displacement graph obtained as history output for specimen Fig. 6. Comparison of failure shape of CFST (test no.5)
no. 8 on ABAQUS

45
Authorized licensed use limited to: Auckland University of Technology. Downloaded on October 03,2020 at 23:55:28 UTC from IEEE Xplore. Restrictions apply.
 Moratuwa Engineering Research Conference (MERCon) 2020

V. STUDY ON SQUARE TUBES B. Experimental Results

Commercially available galvanized square steel hollow Experimental results show that CFST sections have
sections were prepared for the experiment which showed 270 yielded a more than 100% capacity improvement in GI
N/mm2 tensile strength. Samples were taken for varying sections according to table VI. It shows about 2.5 times
length as 8 samples in total and 3 from 1.5m length and 1m strength improvement in CFST compared to hollow GI
for both hollow and CFST. sections.
A. Experimental Details TABLE VII. EXPERIMENTAL FAILURE LOADS OF GI SECTIONS
For the infilling of square tubes, a self-compacting Specimen Height (m) Hollow CFST
concrete was developed using commercially available No. Section (kN) Section (kN)
admixture (Glenium) yielding strength of 45MPa as follows.
1 1.5 60 140
TABLE VI. CONCRETE MIX PROPORTIONS 2 1.5 56 131
3 1.5 62 128
Concrete Cemen Wate Fine Coarse Admix Silica
Com. t r Aggr. aggr. ture (l) Fume 4 1 42 140
Strength (kg) (kg) (kg)
( / (kg) (kg)
C. Finite Element Modelling
)
Finite element modelling was carried out using MIDAS
45 425 233.3 965 700 2 20 GEN 2019 software for CFST elements and Abaqus software
(version 6.14) was used to simulate Local buckling failure in
Both of the hollow and CFST sections with Galvanised hollow GI sections.
iron were tested using the Amsler testing machine in the Local buckling failure was simulated by non-linear
structural testing laboratory of the University of Moratuwa analysis in ABAQUS software for hollow GI sections.
civil engineering department. Hollow sections were failed in Propagation of failure pattern is shown in Fig. 9. and the
local buckling and CFST sections were failed in global hollow section gets shorten with the increase of load due to
buckling mode as in the following figures 7 and 8. local buckling. Damage initializing location has slightly
moved upward when compared with the damage location of
the experimental test as in experimental test the damage was
initiated at the bottom-most point of the hollow GI section.
The slight dimension changes and uneven surface at the
bottom surface GI sections have caused this deviation of the
damage location.

Local
Buckling
failure

Fig. 7. Local Buckling failure of hollow GI sections

Fig. 9. Local buckle failure of a hollow GI section simulated by ABAQUS

CFST sections were modelled with the MIDAS Gen 2019

software. It includes a particular section property to model
Fig. 8. Global buckling failure of CFST GI sections CFST as SRC section. Grade 45 concrete properties were
used with the GI material properties to define the SRC

46
Authorized licensed use limited to: Auckland University of Technology. Downloaded on October 03,2020 at 23:55:28 UTC from IEEE Xplore. Restrictions apply.
 Moratuwa Engineering Research Conference (MERCon) 2020

section. Fig. 10 illustrates the first eigenmode of the buckling VI. CONCLUSION
analysis of CFST column. In this research, the effects of two main parameters that
are infill concrete compressive strength and diameter-to-
thickness (D/t) ratio were considered to investigate the axial
compressive capacity of CFST columns. It can be concluded
that the axial capacity of CFST increases as the concrete infill
increases and decreases with D/t ratios. It was also evident
that when the D/t ratio increases confinement provided by
steel tube decreases. Member stiffness and ductility also
decrease as the increase of D/t ratios. Also, in GI sections
failure mode was changed from local buckling to global
buckling when it is infilled. Capacity wise 2.5 times
increment was recorded both experimentally and
numerically.
The Comparison of theoretical and experimental results
conclude that AISC and Eurocode 4 codes can be safely used
Fig. 10. The shape of SRC section analysed by Midas Gen 2019 in the design except for Mander’s formulae. In the nonlinear
FE model analysis, the FE model was compared and verified
Both numerical simulations showed axial compression with the tested specimens. Good agreement between model's
capacity results almost closed to experimental values. The output which is tested data in terms of failure mode and the
buckling analysis was performed and the first mode buckling axial capacity is observed for the CFST. It can be concluded
failure load was determined using the software. Table VII that FE modelling can be used for further studies since it has
summarises the FEM output for both hollow and CFST GI an acceptable degree of accuracy. According to the results
sections. obtained from three approaches experimental, theoretical and
FEM analysis of those studies, it was finally concluded that
TABLE VIII. NUMERICAL ANALYSIS RESULTS FOR GI SECTIONS
CFST is most effective under axial compression.
Member section Maximum Failure mode
applicable force ACKNOWLEDGMENT
(kN)
The authors would like to thank the Senate Research
Hollow Section (kN) 62 Local buckling
Committee of the University of Moratuwa
CFST Section (kN) 140 Global buckling (SRC/LT/2019/32) for the financial assistance throughout the
research period. The authors also would like to acknowledge,
the support given by all academic and non-academic staff in
CFST sections were inspected for honeycombs by cutting the Department of Civil Engineering, University of
them along the cross-sections and fig. 11 shows the Moratuwa, Sri Lanka.
confinement inside the section.
REFERENCES

[1] P. K. Gupta, S. M. Sarda, and M. S. Kumar, "Experimental and

computational study of concrete filled steel tubular columns under
axial loads," J. Constr. Steel Res., vol. 63, no. 2, pp. 182–193, Feb.
2007.
[2] E. Ellobody, B. Young, and D. Lam, "Behaviour of normal and high
strength concrete-filled compact steel tube circular stub columns," J.
Constr. Steel Res., vol. 62, no. 7, pp. 706–715, July 2006.
[3] F. Abed, M. AlHamaydeh, and S. Abdalla, "Experimental and
numerical investigations of the compressive behavior of concrete filled
steel tubes (CFSTs)," J. Constr. Steel Res., vol. 80, p. 429–439, 2013.
[4] "AISC 360–10. Specification for structural steel buildings, Chicago.
Chicago, Illinois," American Institute of Steel Construction, 2010.
[5] "Design of composite steel and concrete structures, part 1.1—General
rules and rules for buildings. Euro code 4," European Committee for
Standardization (CEN), 1992.
Fig. 11. The final quality of infilled concrete [6] J. B. Mander, M. J. N. Priestley and R. Park, "Theoretical Stress
Strain Model for Confined Concrete," J. Structural Eng., vol. 114, no.
Both GI hollow sections and also the CFST sections show 8, p. 1804–1826, 1988.
almost the same compressive strength as well as the mode of [7] G. Tusshar and K. T. Aditya, "Finite Element Modeling of Circular
failure. According to euro code, used GI section categorises Concrete Filled Steel Tube (CFST)," Indian J. Sci. Technol., vol. 11,
no. 34, pp. 1-9, 2018.
into class 4 section and it is prone to buckle locally.
Confirming the above criteria both experimental and
numerical models also showed the same behaviour as well as
the failure load.

47
Authorized licensed use limited to: Auckland University of Technology. Downloaded on October 03,2020 at 23:55:28 UTC from IEEE Xplore. Restrictions apply.