BEHAVIOUR OF CONCRETE FILLED COMPOSITE

STEEL COLUMNS

Morgan Dundu, Masela S Mahlaule, Motlatsi S Mothetho

University of Johannesburg, Department of Civil Engineering,

P O Box 524, Auckland Park, 2006, South Africa

Abstract: A series of compression tests of circular concrete-filled steel tube columns are re-
ported in this paper. The columns were subjected to concentric axial loads until failure. Vari-
ables in the tests include the length, diameter, strength of the steel tubes and the strength of
the concrete. The primary mode of failure of most slender columns was overall flexural buck-
ling. Local buckling, crushing of the concrete and yielding of steel tube were experienced in
some of the short columns. Observed results from the tests shows that both the Eurocode 4
(EC4) and the South African standard (SANS 10162-1) provide reasonable prediction of the
axial capacity of the composite columns.

1. INTRODUCTION

Concrete filled steel tube (CFST) columns are valuable structural members when separate
reinforced concrete columns or steel hollow columns cannot carry loads effectively. CFST
columns combine the best characteristics of both steel and concrete materials, that is, they
have high strength, high ductility, and high stiffness. These properties and the large energy
absorption capacity have ensured that composite columns are used in structures to resist seis-
mic loads [1]. The steel tube acts as permanent formwork, and provides lateral confinement or
lateral reinforcement to the concrete whilst local buckling which is normal a problem with
thin-walled steel tube is delayed due to the presence of the concrete infill. This means that
local buckling will occur at higher loads than what will happen with steel only. What is also
important with these members is that the buckling mode of the steel tube is modified since
inward buckling, which is common with steel is prevented. Hence the wall of the steel tube
tends to buckle outwards only. Furthermore, the column sizes can be reduced if composite
columns are used resulting in increased floor spaces and lower costs.
Despite the numerous publications about CFST columns all over the world, this investiga-
tion were motivated by the need to verify the applicability of the formulas used in both SANS
10162-1 [2] and EC4 [3]design codes over a wide range of column lengths and diameters.
SANS 10162-1 [2] is based on the Canadian Code, CAN/CSAS16.1-M01 [4]. The findings of
these investigations are to be used to determine what adjustments, if any, should be made to
the formulae in the respective codes.
2 The 6th International Conference on Thin Walled Structures

2. EXPERIMENTAL PROGRAMME

2.1 Material properties

The average material properties for the steel tubes and concrete are given in Table 1. Since
these investigations were undertaken at two different times and of different material proper-
ties, the first group of tests is referred as Series 1 and the second one as Series 2.

Table 1: Average material properties

Series Diameter Length (m) Steel tube Concrete
(mm) fy (MPa) fu (MPa) E (GPa) fu (MPa) E (GPa)
Series 1 114.3 1.0-2.50 354.05 432.35 206.50 40.3 31.1
127.0 1.0-2.50 345.20 430.40 209.00
139.9 1.0-2.50 361.95 457.85 208.05
Series 2 152.4 1.0-2.0 488.20 549.60 206.70 30.9 28.27
152.4 2.50 394.30 480.20 206.70
165.1 1.0-2.0 438.20 500.90 204.60
165.1 2.50 430.30 480.15 201.60
193.7 1.0-2.0 398.80 479.10 207.70
193.7 2.50 392.20 470.80 206.80

SANS10162-1 requires the concrete strength for axially loaded CFST columns to range
from 25MPa to 100MPa [2]. To promote ductility the standard also limits the specified yield
stress (fy) of the steel to 700MPa. In the case of EC4, the standard requires the characteristic
concrete cylinder strength (fcu) to be at least 20MPa and not more than 50MPa [3]. The yield
stress (fy) of the steel for this standard must be at least 235MPa and not more than 360MPa.
As shown in Table 1, all concrete and steel strength requirements satisfy both SANS10162-1
[2] and EC4 [3]. However, the application range of SANS 10162-1 [2] is wider in terms of
concrete strength and steel yield strength.

2.2 Specimen preparation and test procedure

Steel plates of 6mm thickness were welded at the bottom of the steel tubes to contain the
concrete during casting. Each circular hollow section (CHS) was filled with concrete in 4 lay-
ers. After each layer the concrete was compacted by a poker vibrator to ensure adequate com-
paction and eliminate air pockets in the concrete. An extra 15 mm layer was cast at the top of
the columns to account for possible concrete shrinkage. The finished CFST columns were
sealed with plastic sheeting at the top to retain moisture in the concrete. This was done to en-
sure that the hydration of cement can continue properly without premature hardening of the
concrete. The specimens were allowed to cure for 28 days before testing could be conducted.
After 28 days the plastic sheeting were removed from the columns and the excess concrete at
the top of the columns was ground flat to make the concrete surface level with the steel tube.
This was done to ensure the load was applied simultaneously to the concrete core and the
steel tube during testing.
The testing of the specimens took place at the Council of Scientific and Industrial Research
(CSIR) mechanical laboratory in Johannesburg, South Africa. A Morh and Federhaff com-
pression testing machine which has a load capacity of 9000kN and maximum loading rate of
30mm/min was used for testing specimens. However, the loading rates for these tests were
2mm/min for Series 1 tests, and 3mm/min for Series 2 tests. This lower loading rate assisted
in identifying critical information during testing and also in ensuring that sufficient data was
Theme (by the C.C.) 3

collected. The upper and lower plates of the Mohr & Federhaff were designed so as to pro-
duce simply supported end conditions. A thick stiff plate was placed at the top of the CFST
columns to ensure that the load is distributed simultaneously to the concrete core and the steel
tube. The specimens were instrumented with four linear variable displacement transducers
(LVDTs) attached at the mid-length of the columns at right angles to each other to measure
lateral deflections. These deflections were used to try and quantify the magnitudes of the sec-
ond order moments. A photograph of the test setup is given in Figure 1. The test was stopped
one the load started to drop at a faster rate.

Fig. 1: Typical test setup

3. FAILURE MODES

Various failure modes of composite columns were observed in the different column
lengths. In Series 1 tests, all CFST columns failed by overall flexural buckling; with minor
local buckling occurring close to the centre of the columns. The slenderness ratios were not
large enough to prevent overall instability of the columns. However, in Series 2 tests, the
1.0m and 1.5m length CFST columns failed by crushing of the concrete accompanied by
yielding of the steel tube. The presence of the concrete prevented the occurrence of the in-
ward buckling. Bulges near the top end of the columns in Figure 2(a) and (b) are clear evi-
dence of outward local buckling. This crushing failure mode was characterized by the steel
wall being pushed out by the concrete core. Where local buckling was found it was noticed
that the concrete had crushed first, followed by buckling of the steel tube. Similarly to Series
1 tests, the slender columns in Series 2 (2.0m and 2.5m) exhibited insufficient radial strain in
the steel section to confine the concrete, causing the columns to fail by overall flexural buck-
ling. Composite tubes with large d/t ratios experienced more local buckling combined with
concrete crushing compared to those with small d/t ratios. This difference accounted for the
strain hardening behaviour of the composite columns.
4 The 6th International Conference on Thin Walled Structures

(a) 1.0m (b) 1.5m (c) 2.0m (d) 2.5m

Fig. 2: Typical failure modes of Series 2 columns

4. EXPERIMENTAL AND CODE PREDICTED RESULTS

4.1 Code requirements

The slenderness ratio (L/D) of the composite column ranged from 5.16 to 22. This range is
greater than 4, making the columns intermediate to slender [5]. In SANS10162-1, the provi-
sions for estimating the compressive resistance are applicable only to columns with a slender-
ness ratio (L/D) less than 25 [2]. EC4 does not have this limitation [3]. These lengths were
specifically chosen to represent columns that exhibit behaviour that can be expected in prac-
tice.
To avoid local buckling of composite columns, the two design codes were consulted. Ac-

ratios of circular hollow structural sections in compression are d t ≤ 28000 f y and

cording to SANS 10162-1:2005 [2] and EC4 [3], the maximum outside diameter-to-thickness

d t ≤ 21150 f y , respectively. From these limits it can be seen that EC4 has more stringent
local buckling limits than SANS 10162-1 [2]. The values of the diameter-to-thickness ratios
for Series 1 and series 2 composite columns ranged from 38.38 to 46.40 and 50.80 to 55.34,
respectively. According to SANS 10162-1 [2], these values suggest that all composite col-
umns will yield. However, this disputed by the Series 2 values, calculated using EC4 [2].
Theme (by the C.C.) 5

4.2 Analysis of results

To validate the applicability the South African code [2] and the European composite code
[3], the experimental results of circular concrete-filled steel tube columns subjected to axial
loads are compared with the predicted compressive resistance calculated from the two codes
(see Table 2). In this table NCHS is compressive resistance of the circular hollow section
(CHS), calculated using SANS10162-1 [2], NTEST is the maximum load from the tests, NSANS
is the compressive resistance calculated using SANS10162-1 and NEC4 is the compressive re-
sistance calculated using EC4. As expected the compressive resistance of all columns de-
creased with increase in slenderness ratio. The effect of confinement effect is reduced with
increasing slenderness of the column, since the lateral deflection prior to failure increases the
bending moment and reduces the mean compressive strain in the concrete. The test loads for
Series 1 are on average 4.3% greater than the loads predicted by EC4. At most, EC4 over-
estimated the capacity of columns in this group by 14.6%, and this was for the 2.5m column
of 114.85mm diameter. In general, SANS10162-1 is more conservative than EC4. On average
the tests loads are 8.7% greater than the loads predicted by SANS 10162-1. The most conser-
vative value is also for the 114.3mm diameter columns with a length of 2.5m.

Table 2: Test and calculated resistance of CFST columns

Series D L NCHS NTEST NSANS NEC4 NTEST NTEST NTEST
(m) (mm) (kN) (kN) (kN) (kN) NSANS NEC4 NTUBE
114.85 1000 358.64 806.4 747.7 712.8 1.078 1.131 2.25
114.85 1500 333.52 688.2 644.9 666.1 1.067 1.033 2.06
114.85 2000 298.06 632.2 547.0 593.6 1.156 1.065 2.12
114.85 2500 258.07 566.1 459.6 494.1 1.232 1.146 2.19
127.3 1000 392.74 912.1 861.3 843.1 1.059 1.082 2.32
127.3 1500 371.97 848.5 762.8 786.3 1.112 1.079 2.28
Series 1
127.3 2000 341.10 715.8 664.9 721.8 1.077 0.992 2.10
127.3 2500 303.97 638.8 572.7 629.6 1.116 1.015 2.10
139.2 1000 453.38 1059.8 1041.6 1026.6 1.017 1.032 2.34
139.2 1500 433.05 941.9 927.4 944.0 1.016 0.998 2.18
139.2 2000 402.09 868.3 819.5 879.3 1.059 0.987 2.16
139.2 2500 363.65 750.7 715.2 787.5 1.050 0.953 2.06
152.4 1000 668.13 1463.3 1337.0 1214.9 1.095 1.204 2.19
152.4 1500 633.69 1209.1 1191.1 1103.0 1.015 1.096 1.91
152.4 2000 582.28 1167.3 1045.3 1043.8 1.117 1.118 2.00
152.4 2500 446.21 968.9 815.7 872.1 1.188 1.111 2.17
165.1 1000 656.27 1549.5 1397.9 1324.8 1.108 1.170 2.36
165.1 1500 632.04 1338.0 1268.4 1186.5 1.055 1.128 2.12
Series 2
165.1 2000 594.23 1234.5 1138.0 1138.6 1.085 1.084 2.08
165.1 2500 538.21 1232.0 999.4 1052.1 1.233 1.171 2.29
193.7 1000 708.63 1999.6 1843.2 1819.2 1.085 1.099 2.82
193.7 1500 806.50 1817.1 1708.2 1626.2 1.064 1.117 2.25
193.7 2000 777.05 1796.3 1572.6 1532.8 1.142 1.172 2.31
193.7 2500 726.35 1620.8 1423.8 1465.3 1.138 1.106 2.23
6 The 6th International Conference on Thin Walled Structures

In Series 2, the average test loads are greater than the average loads predicted by EC4, by
13%. This percentage is significantly greater than the one obtained for slender columns in Se-
ries 1 tests, implying that as the column becomes stockier, EC4 tends to be more conserva-
tive. The test-to-predicted value of 11% in this series of tests, obtained using SANS10161-1,
compares favourably with the value obtained for Series 1 tests. The difference between the
test results and the failure loads predicted by SANS 10162-1 for all columns ranged from 6 to
23%, whereas EC4 predictions ranged from 10 to 20%. Both codes consider the effect of con-
crete confinement in circular tubes but do it in different ways, which partly accounts for the
differences in the predicted compression resistances of individual columns. Due to confine-
ment on concrete, the compressive resistance of the composite column has doubled, compared
to the compressive resistance of an ordinary steel tube or circular hollow steel column (see
Table 2). The ratio of the test load to the compressive resistance of the empty tube for both
series of tests shows that the effect of confinement increases with increase in diameter of the
composite column. The readings from the linear variable displacement traducers (LVDTs)
were inconclusive and will not be discussed.

5. LOAD-AXIAL DEFLECTION GRAPHS

The load-axial deflection graphs for all diameters are shown in Figure 3-8. From these
graphs it can be seen that the shorter CFST columns has a higher stiffness and load carrying
capacity. As the length of the composite columns increases, the load carrying capacity de-
creases. All the CFST columns displayed a ductile behaviour; however, the shorter CFST col-
umns displayed more ductile behaviour. In addition, composite columns with larger diameters
exhibited far much more ductility than the composite columns with smaller diameters. The
increased load carrying capacity in the 193.70 mm diameter CFST columns is attributed to the
hoop or circumferential stress.

900 1000

800 900

700 800

700
600
Axial load (kN)
Axial load (kN)

600
500
500
400
400
300
300
200 114.3 mm-1.0 m 200 127 mm-1.0 m
114.3 mm-1.5 m 127 mm-1.5 m
100 114.3 mm-2.0 m 100 127 mm-2.0 m
114.3 mm-2.5 m 127 mm-2.5 m
0 0
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 40
Axial displacement (mm) Axial displacement (mm)

Fig. 3: 114.3 mm diameter columns Figure 4: 127 mm diameter columns

 Theme (by the C.C.) 7

1200 1600

1100
1400
1000
900 1200

800

Axial load (kN)

Axial load (kN)

1000
700
600 800

500
600
400
300 400
139.7 mm-1.0 m 152.4mm-1.0 m
200 139.7 mm-1.5 m 152.4 mm-1.5 m
200
100 139.7 mm-2.0 m 152.4 mm-2.0 m
139.7 mm-2.5 m 152.4 mm-2.5 m
0 0
0 5 10 15 20 25 30 35 40 0 10 20 30 40 50 60 70 80
Axial displacement (mm) Axial displacement(mm)

Fig. 5: 139.7mm diameter columns Fig. 6: 152.40mm diameter columns

1600 2000

1800
1400
1600
1200
1400
Axial load (kN)
Axial load (kN)

1000 1200

800 1000

800
600
600
400
165.1 mm-1.0 m 400 193.7 mm-1.0 m
165.10 mm-1.5 m 193.7 mm-1.5 m
200 200 193.7 mm-2.0 m
165.10 mm-2.0 m
165.1 mm-2.5 m 193.7 mm-2.5 m
0 0
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
Axial displacement (mm) Axial displacement (mm)

Fig. 7: 165.10 mm diameter columns Fig. 8: 193.70 mm diameter columns

6. CONCLUSION

The behaviour and the load capacity of 24 CFST columns of varying diameters and lengths
using 30 and 40MPa concrete were investigated. The failure mode of CFST columns in Series
1 tests was largely flexural buckling with no sign of local buckling. This mode of failure was
caused by the large slenderness ratio of the columns. In Series 2 tests, the 1.0m and 1.5m
specimens failed by the crushing of the concrete core and the yielding of the steel tube. Bulg-
ing of steel tube was also observed in these columns.
The 193.70 mm diameter composite columns achieved higher ultimate load capacities
when compared to the other diameters of the same lengths. This difference in ultimate load
capacities is attributed to the hoop or circumferential stress. A higher hoop or circumferential
8 The 6th International Conference on Thin Walled Structures

stress results in significant increased load carrying capacity of the CFST columns. On aver-
age, loads predicted by SANS 10162-1 are conservative by 8.7%, and loads predicted by EC4
are conservative by 4.3 % for Series 1 tests. The average predictions of test failure loads by
SANS 10162-1 and EC4 are conservative by 11.0 and 13.1% for Series 2, respectively. In
general, it can be concluded that both codes (SANS 10162-1 and EC4) were able to predict
the test results within an acceptable margin and that all columns behaved in a fairly ductile
manner. The investigation concluded that both codes provide reasonable predictions of the
axial capacity for all lengths of columns.

References

[1] Lee E-T, KH, Yun BH, Shim HJ, Chang KH, Lee GC. “Torsional Behavior of Con-
crete-Filled Circular Steel Tube Columns”, Journal of Structural Engineering,
135(10), 1250-1258, 2009.

[2] SANS 10162-1. The structural use of steel, Part 1: Limit-state design of hot-rolled
steelwork, Standards South Africa, Pretoria, 2005.

[3] EC4. Design of composite steel and concrete structures, Part 1-1: general rules and
rules for buildings, British Standards Institution, London, 2004.

[4] CAN/CSA-S16-01. Limit States Design of Steel Structures, Canadian Standards Asso-
ciation, Rexdale, Ontario, Canada, 2001.

[5] Kuranovas A, Goode D, Kvedaras AK, Zhong S. “Load-bearing capacity of concrete-

filled steel columns”, Journal of Civil Engineering and Management, 15(1), 21-33,
2009.