Thin-Walled Structures 90 (2015) 4960

Contents lists available at ScienceDirect

Thin-Walled Structures
journal homepage: www.elsevier.com/locate/tws

Aluminum tubular sections subjected to web crippling

Yu Chen a,b, Xixiang Chen c,n, Chaoyang Wang b
a
b
c

School of Urban Construction, Yangtze University, Jingzhou 434023, China

College of Civil Engineering, Huaqiao University, Xiamen 361021, China
College of Technology & Engineering, Yangtze University, Jingzhou 434023, China

art ic l e i nf o

a b s t r a c t

Article history:
Received 6 November 2014
Received in revised form
7 January 2015
Accepted 7 January 2015
Available online 30 January 2015

This paper presents the details of experimental and numerical research study on web crippling property
of aluminum tubular under concentrated web crippling loadings. A total of 48 aluminum square hollow
sections with different boundary conditions, loading conditions, bearing lengths and section heights
were tested. The experimental scheme, failure modes, loaddisplacement curves and strain intensity
distribution curves were also presented. The investigation was focused on the effects of different
boundary conditions, loading conditions, bearing lengths and web slenderness on web crippling
ultimate capacity and ductility of aluminum square hollow sections. The results obtained from the
experiments are shown that the effect of bearing length on the web crippling ultimate capacity under
End-One-Flange (EOF) and End-Two-Flange (ETF) loading and boundary conditions is more obvious than
those under Interior-One-Flange (IOF) and Interior-Two-Flange (ITF) boundary and loading conditions.
The web crippling ultimate capacities under EOF and ETF loading conditions decreased as the
slenderness ratio increased. As the bearing length was 150, the web crippling ultimate capacity under
IOF and ITF loading conditions reached its peak when the value of the web slenderness was minimum.
The web crippling ultimate capacities of aluminum tubular with bearing length 50 mm and 100 mm
under IOF, ITF, EOF and ETF boundary and loading conditions decreased progressively. The web crippling
ultimate capacity of aluminum tubular with bearing length 150 mm was approximately equal. Finite
element models were developed to numerically simulate the tests performed in the experimental
investigations. Based on the results of the parametric study, a number of design formulas proposed in
this paper can be successfully employed as a design rule for predicting web crippling ultimate capacity of
aluminum tubular sections under four loading and boundary conditions.
& 2015 Published by Elsevier Ltd.

Keywords:
Aluminum tubular sections
Web crippling
Ultimate capacity
Finite element analysis
Design formulas

1. Introduction
The use of aluminum alloys in construction has been permanently increasing during the last years due to its characteristics in
terms of light weight, corrosion resistance, high strength to weight
ratio and ease of production. However, the modulus of elasticity of
aluminum is approximately 1/3 of steel. Therefore, the web of
aluminum beam is susceptible to buckling phenomena due to its
lower elastic modulus. Therefore, web crippling needs to be
considered in designing aluminum beams.
A considerable amount of experimental investigations has been
carried out on thin walled hollow sections subjected to web crippling
over many years by numerous researchers. An experimental study was
conducted by Stephens and Laboube [1] to establish the web crippling
strength of both box and I-beam headers for an interior-one-ange

Corresponding author. Tel.: 86 13872386804.

E-mail address: chenyuelite@sina.com (X. Chen).

http://dx.doi.org/10.1016/j.tws.2015.01.009
0263-8231/& 2015 Published by Elsevier Ltd.

(IOF) loading condition. Based on the results of this study, design

recommendations were proposed. Finite element models of the hatsection were made by Hofmeyer [2]. The quality of the nite element
models for describing moving yield lines was veried. A major
improvement of the ultimate failure model of web crippling was
presented by Hofmeyer [3]. The ultimate failure model using the
ctitious strain method performed equally well as Eurocode3, but
provided more insight in the failure behavior. Experimental and
numerical investigations of aluminum alloy square hollow sections
with a circular hole in the webs subjected to web crippling were
presented by Zhou and Young [4]. Web crippling strength reduction
factor equations were proposed for the ETF and ITF loading conditions.
Macdonald et al. [5] presented the results of an investigation into web
crippling behavior-conducted on cold-formed thin-walled steel lipped
channel beams subjected to Interior-One-Flange (IOF), Interior-TwoFlange (ITF), End-One-Flange (EOF) and End-Two-Flange (ETF) loading
conditions as dened by the American Iron and Steel Institute. Web
crippling strength predicted from the Eurocode3 were also compared
with the experimental results and the comparisons indicated

50

Y. Chen et al. / Thin-Walled Structures 90 (2015) 4960

Nomenclature
EOF
IOF
ETF
ITF
Pcr
PcrC
PcrE
PFEA
PcrRE
PEOF
PIOF
PETF

End-one-ange boundary and loading condition

Interior-one-ange boundary and loading condition
End-two-ange boundary and loading condition
Interior-two-ange boundary and loading condition
experimental web crippling ultimate capacity
web crippling ultimate capacity obtained by using
Chinese steel structures design code (GB50013-2003)
web crippling ultimate capacity obtained by using
European design of steel structures (Eurocode 3)
web crippling ultimate capacity obtained by using
nite element analysis
web crippling ultimate capacity obtained by using
formulas the paper put forward
web crippling ultimate capacity under EOF condition
web crippling ultimate capacity under IOF condition
web crippling ultimate capacity under ETF condition

considerable underestimations for the range of specimens under EOF

and ETF loading conditions. A combination of experimental tests and
non-linear elasticplastic nite element analyses were used to investigate the effect of holes on web crippling under interior-two-ange
(ITF) loading conditions by Uzzaman et al. [6]. Design recommendations in the form of web crippling strength reduction factors were
proposed. A design rule which was based on a theoretical and
numerical model of strength of cold-formed steel lipped channel
beams was put forward by Macdonald and Heiyantuduwa [7]. Based
on the results of the parametric study, a design rule was developed
which was much more exible to adapt for new types of sections and
ranges of dimensions. Zhou and Young [8] numerically investigated
cold-formed high strength stainless steel square and rectangular
hollow sections subjected to web crippling at elevated temperatures.
A unied web crippling equation for cold-formed high strength
stainless steel square and rectangular hollow sections at elevated
temperatures was proposed.
There is little experimental and numerical research being
carried out on the behavior of aluminum beams under four
loading and boundary conditions subjected to web crippling.
Therefore, the ultimate capacity, failure modes and ductility of
aluminum beams subjected to web crippling need further investigation. In this study, a series of web crippling tests of the
aluminum square hollow sections was conducted. The test specimens were tested under EOF, IOF, ETF and ITF boundary and
loading conditions. The effects of bearing lengths, web slenderness
and boundary and loading conditions on the ultimate capacity and
initial stiffness of aluminum beams subjected to web crippling
were investigated. Furthermore, using the calibrated nite element model, a parametric study was conducted to comprehensively investigate the effects of some important geometric
parameters on the ultimate capacity of aluminum beams subjected
to web crippling. The design formulas of ultimate capacity were
also proposed for aluminum beams subjected to web crippling at
the end of the paper.

PITF
fy
fu

web crippling ultimate capacity under ITF condition

tensile yield stress (0.2% tensile proof stress)
ultimate tensile stress

Poisson's ratio

elongation after fracture

E
elastic modulus
H
overall height of aluminum tubular section
B
overall width of aluminum tubular section
R
Internal arc radius
T
wall thickness of aluminum tubular section
L
overall length of aluminum tubular section
A
bearing length
(ht  2t)/t web slenderness
i
strain intensity
1
rst principal strain
2
second principal strain
3
third principal strain

with different boundary conditions, loading conditions, section

heights, and bearing lengths were tested in Table 1. The specimen
had the nominal thickness ranging from 1.1 to 2.6 mm, the
nominal heights of webs ranging from 65 to 110 mm, and the
ange widths ranging from 25 to 100 mm. The measured ratio of
the height to the thickness (web slenderness) of the webs ranged
from 30 to 88. The symbols of aluminum tubular sections are
dened in Fig. 1.
The bearing plates were fabricated with Chinese Standard Q345
steel having the nominal thickness of 30 mm. All the bearing
plates were machined to specied dimensions, and the length was
300 mm. The bearing plates were designed to act across the full
ange widths of the specimen sections, so as to ensure the overall
displacement loading.
In the paper, the specimens were tested in four loading conditions, namely, End-One-Flange (EOF), Interior-One-Flange (IOF),
End-Two-Flange (ETF) and Interior-Two-Flange (ITF). In order to
eliminate the boundary effect, the distance from the edge of the
bearing plate to the end of the member was set to be at least
1.5 times the overall depth of the web. Schematic sketch of web
crippling tests in four boundary and loading conditions are considered in Fig. 2. Fig. 3 shows photographs of web crippling tests in
four boundary and loading conditions.

2.2. Specimen labeling

In Table 1, all specimens were labeled to easily identify material
type of the specimens, section geometry, and the loading condition, as well as length of the bearing could be identied from the
label. For example, the labels AL60  65-EOF-N100 is dened as
the following specimens:

 The rst letters indicates the material type of the specimens,

where AL refers to aluminum.

 The following number symbols are the nominal dimension of

2. Experimental investigation
2.1. Test specimens
Experimental investigations were designed to examine the
inuence of various boundary and loading on web crippling
ultimate capacity. A total of 48 aluminum square hollow sections




the specimens in mm, where 60  65 (b  h) means the

nominal ange width (b) of 60 mm, and the nominal height
(h) of 65 mm.
The following three letters indicate the loading and boundary
condition, where EOF refers to exterior-one-ange test.
The last part of the label N100 indicates the bearing plate
width in mm, where 100 means the nominal bearing length
(a) of 100 mm.

Y. Chen et al. / Thin-Walled Structures 90 (2015) 4960

51

Table 1
Parameters and ultimate capacity of aluminum tubular sections subjected to web crippling.
Boundary and loading conditions

Specimen number

a (mm)

b (mm)

h (mm)

L (mm)

t (mm)

(h  2t)/t

r (mm)

Pcr (kN)

EOF

AL60  65-EOF-N50
AL60  65-EOF-N100
AL60  65-EOF-N150
AL65  110-EOF-N50
AL65  110-EOF-N100
AL65  110-EOF-N150
AL26  76-EOF-N50
AL26  76-EOF-N100
AL26  76-EOF-N150
AL100  100-EOF-N50
AL100  100-EOF-N100
AL100  100-EOF-N150

50
100
150
50
100
150
50
100
150
50
100
150

60.00
60.02
60.02
65.00
64.88
64.92
25.34
25.40
25.40
100.02
100.04
100.00

65.02
65.00
65.02
110.90
110.40
110.80
76.10
76.24
76.20
100.00
100.06
100.00

400
400
400
400
400
400
400
400
400
400
400
400

2.00
2.00
2.00
2.46
2.60
2.60
1.40
1.30
1.36
1.10
1.20
1.20

30.51
30.50
30.51
43.08
40.46
40.62
52.36
56.65
54.03
88.91
81.38
81.33

4.00
4.00
4.00
4.92
5.20
5.20
2.80
2.60
2.72
2.20
2.40
2.40

15.05
27.05
37.20
11.50
18.40
30.40
2.65
4.65
6.30
2.10
3.10
3.65

IOF

AL60  65-IOF-N50
AL60  65-IOF-N100
AL60  65-IOF-N150
AL65  110-IOF-N50
AL65  110-IOF-N100
AL65  110-IOF-N150
AL26  76-IOF-N50
AL26  76-IOF-N100
AL26  76-IOF-N150
AL100  100-IOF-N50
AL100  100-IOF-N100
AL100  100-IOF-N150

50
100
150
50
100
150
50
100
150
50
100
150

60.02
60.00
60.00
64.96
64.90
64.92
25.36
25.40
25.36
100.02
100.02
100.00

65.00
65.00
65.00
110.62
110.60
110.72
76.10
76.22
76.20
100.00
100.04
100.04

600
600
600
600
600
600
400
400
400
400
400
400

2.02
2.00
2.00
2.46
2.50
2.52
1.40
1.40
1.36
1.18
1.20
1.20

30.18
30.50
30.50
42.97
42.24
41.94
52.36
52.44
54.03
82.75
81.37
81.37

4.04
4.00
4.00
4.92
5.00
5.04
2.80
2.80
2.72
2.36
2.40
2.40

20.60
29.70
42.90
22.00
31.80
33.80
5.65
6.90
8.25
5.20
5.95
5.80

ETF

AL60  65-ETF-N50
AL60  65-ETF-N100
AL60  65-ETF-N150
AL65  110-ETF-N50
AL65  110-ETF-N100
AL65  110-ETF-N150
AL26  76-ETF-N50
AL26  76-ETF-N100
AL26  76-ETF-N150
AL100  100-ETF-N50
AL100  100-ETF-N100
AL100  100-ETF-N150

50
100
150
50
100
150
50
100
150
50
100
150

60.04
60.00
60.02
64.92
64.90
64.94
25.38
25.38
25.40
100.02
100.04
100.02

65.00
65.02
65.00
110.62
110.68
110.76
76.16
76.24
76.20
100.02
100.04
100.00

400
400
400
600
600
600
400
400
400
400
400
400

2.00
2.00
2.00
2.50
2.52
2.54
1.40
1.40
1.36
1.10
1.18
1.16

30.50
30.51
30.50
42.25
41.92
41.61
52.40
52.46
54.03
88.93
82.78
84.21

4.00
4.00
4.00
5.00
5.04
5.08
2.80
2.80
2.72
2.20
2.36
2.32

12.90
26.00
35.95
10.95
19.80
31.20
2.45
4.65
8.30
1.70
2.65
3.50

ITF

AL60  65-ITF-N50
AL60  65-ITF-N100
AL60  65-ITF-N150
AL65  110-ITF-N50
AL65  110-ITF-N100
AL65  110-ITF-N150
AL26  76-ITF-N50
AL26  76-ITF-N100
AL26  76-ITF-N150
AL100  100-ITF-N50
AL100  100-ITF-N100
AL100  100-ITF-N150

50
100
150
50
100
150
50
100
150
50
100
150

60.00
60.02
60.00
64.96
64.98
64.94
25.34
25.38
25.38
100.02
100.00
100.02

65.00
65.02
65.02
110.62
110.68
110.80
76.10
76.24
76.24
100.02
100.04
100.00

400
400
400
600
600
600
400
400
400
400
400
400

2.00
1.98
2.00
2.50
2.50
2.54
1.30
1.30
1.36
1.10
1.18
1.16

30.50
30.84
30.51
42.25
42.27
41.62
56.54
56.65
54.06
88.93
82.78
84.21

4.00
3.96
4.00
5.00
5.00
5.08
2.60
2.60
2.72
2.20
2.36
2.32

18.70
31.20
37.20
19.80
26.70
34.70
4.65
5.65
7.80
3.55
4.15
4.65

Fig. 1. Denition of symbols of aluminum tubular sections.

were taken from the center of the web plate in the longitudinal
direction of the untested specimens. The nominal coupons were
prepared and tested according to Chinese Metallic MaterialsTensile testing at ambient temperature (GB/T228-2002) [9], The
coupons were tested in an MTS displacement controlled the
testing machine using friction grips. The strain gauges and a
calibrated extensometer were used to measure the longitudinal
strain. A data acquisition system was used to record the load and
strain at regular intervals during the tests. The material properties
obtained from the tensile coupon tests are summarized in Table 2,
including the tensile yield stress (fy), the ultimate tensile stress
(fu), Poisson's ratio (), the elongation after fracture () and the
elastic modulus (E).

2.3. Material properties

2.4. Test program

The material properties of the aluminum specimens were

determined from a series of tensile tests. The tensile coupons

Two displacement gauges D1 and D2 were located at the

surface of the bearing plates on the top ange of aluminum square

52

Y. Chen et al. / Thin-Walled Structures 90 (2015) 4960

Fig. 2. Schematic sketch of web crippling tests under four boundary and loading conditions. (a) EOF, (b) IOF, (c) ETF and (d) ITF.

Fig. 3. Photos of web crippling tests under four boundary and loading conditions. (a) EOF. (b) IOF, (c) ETF and (d) ITF.

hollow sections in order to record the vertical displacement during

the test, as shown in Fig. 4a. Five strain gauge rosettes (T1T5),
which enabled strain values to be measured simultaneously, were
distributed at the same interval on the web of aluminum square
hollow sections, as shown in Fig. 4b.

3. Test results

Table 2
Result of aluminum tube material characteristic coupe test.
Member

fy (Mpa)

fu (Mpa)

fy/fu

(%)

E (GPa)

60  65  2.0
65  110  2.6
26  76  1.4
100  100  1.2

190
185
165
160

216
201
185
184

0.88
0.92
0.89
0.87

0.32
0.31
0.33
0.34

8.8
10.2
9.6
8.4

66
75
70
69

3.1. Failure modes

3.2. Effect of bearing lengths
The failure modes of the aluminum tubular section specimens
subjected to web crippling were webs out-of-plane buckling,
obvious plastic hinge zone formed in the middle parts of the webs
and concave deformation in one loading ange under IOF boundary and loading condition, concave deformation in two anges
under ITF, EOF and ETF boundary and loading condition, as shown
in Fig. 5.

Table 3 shows the ultimate capacity of aluminum tubular sections

subjected to web crippling with different bearing lengths. It is shown
that as the bearing length increased, the web crippling ultimate
capacity also increased under the four loading conditions. In general,
the web crippling ultimate capacity under EOF and ETF boundary and
loading conditions increased faster than those under IOF and ITF

Y. Chen et al. / Thin-Walled Structures 90 (2015) 4960

boundary and loading conditions as the bearing length increased. For

example, increasing the bearing length from 50 to 100 and 150 mm,
the mean values of web crippling ultimate capacity increased by 66%,
131% under EOF boundary and loading condition, and 82%, 177% under
ETF boundary and loading condition, compared with 31%, 55% under
IOF boundary and loading condition, and 35%, 68% under ITF boundary
and loading condition, as shown in Table 3, respectively. It is shown
that the effect of bearing length on web crippling ultimate capacity for

Displacement transducers

D1

D2
D1

Hydraulic jack

D2
Hydraulic jack

ETF/ITF

EOF/IOF

Hydraulic jack
Hydraulic jack

Strain gauge

ETF/ITF

EOF/IOF

Fig. 4. Arrangement of displacement and strain gauges. (a) Displacement transducers and (b) strain gauges.

53

EOF and ETF boundary and loading conditions was more obvious than
those under IOF and ITF boundary and loading conditions.
In Fig. 6, it is observed that the web crippling ultimate capacity
under EOF and ETF boundary and loading conditions followed a
bearing length linear function, and the web crippling ultimate
capacity under IOF and ITF boundary and loading conditions
followed a bearing length nonlinear function. It is shown that as
the bearing length increased, the web crippling ultimate capacity
enhancement under ETF boundary and loading condition was
higher than those under EOF boundary and loading condition.
Similarly, the web crippling ultimate capacity enhancement under
ITF loading condition was higher than those under IOF boundary
and loading condition as the bearing length increased. The web
crippling ultimate capacity enhancement under ETF boundary and
loading condition was the maximum, the web crippling ultimate
capacity enhancement under IOF boundary and loading condition
was the minimum. The effect of bearing length on the web
crippling ultimate capacity decreased as the slenderness ratio
increased in most cases, as shown in Table 3.
3.3. Effect of web slenderness
Fig. 7 plotted the web crippling ultimate capacity versus the web
slenderness curves under four boundary and loading conditions. The
web slenderness ratio of the aluminum square hollow sections ranged
from 30 to 88. It is shown that as the slenderness ratio increased, the
web crippling ultimate capacity decreased under EOF and ETF
boundary and loading conditions. Under the IOF and ITF boundary
and loading conditions, the values of web crippling ultimate capacity
of the aluminum square hollow sections was minimum as the web
slenderness ratio reached its peak. It is shown that as the bearing
length was 150 mm, the values of web crippling ultimate capacity of
the aluminum square hollow sections with the web slenderness
ratio30 reached its peak under IOF and ITF boundary and loading
conditions. However, as the bearing length was 50 mm, the values of
web crippling ultimate capacity of the aluminum square hollow

Fig. 5. Photos of failure modes under four boundary and loading conditions. (a) EOF, (b) IOF, (c) ETF and (d) ITF.

54

Y. Chen et al. / Thin-Walled Structures 90 (2015) 4960

Table 3
Comparison of ultimate capacity of aluminum tubular sections subjected to web crippling with different bearing lengths.
Boundary and loading condition

Member (slenderness)

Pcr (N 50) (kN)

Pcr (N 100) (kN)

Pcr (N 150) (kN)

Pcr (N 100)/Pcr (N 50)

Pcr (N 150)/Pcr (N 50)

EOF

AL60  65(30)
AL65  110(43)
AL26  76(52)
AL100  100(88)
Mean
COV

15.05
11.50
2.65
2.10

27.05
18.40
4.65
3.10

37.20
30.40
6.30
3.65

1.80
1.60
1.75
1.48
1.66
0.089

2.47
2.64
2.38
1.74
2.31
0.171

IOF

AL60  65(30)
AL65  110(43)
AL26  76(52)
AL100  100(88)
Mean
COV

20.60
22.00
5.65
5.20

29.70
31.80
6.90
5.95

42.90
33.80
8.25
5.80

1.44
1.45
1.22
1.14
1.31
0.117

2.08
1.54
1.46
1.12
1.55
0.258

ETF

AL60  65(30)
AL65  110(43)
AL26  76(52)
AL100  100(88)
Mean
COV

12.90
10.95
2.45
1.70

26.00
19.80
4.65
2.65

35.95
31.20
8.30
3.50

2.02
1.81
1.90
1.56
1.82
0.106

2.79
2.85
3.39
2.06
2.77
0.197

ITF

AL60  65(30)
AL65  110(43)
AL26  76(52)
AL100  100(88)
Mean
COV

18.70
19.80
4.65
3.55

31.20
26.70
5.65
4.15

37.20
34.70
7.80
4.65

1.67
1.35
1.22
1.17
1.35
0.167

1.99
1.75
1.68
1.31
1.68
0.167

3.00

2.50

AL60h65

AL60h65
AL26h76

2.00

AL100h100

1.50
1.00

AL26h76
1.50

AL100h100

1.00
0.50

0.50
0.00

AL60h110

2.00

AL65h110
Pcr /Pcr (N=50)

Pcr/Pcr (N=50)

2.50

50

100

150

0.00

200

Bearing length/mm

100

150

200

Bearing length/mm

4.00

2.50

AL60h65

AL60h65

3.50

AL65h110

3.00

AL26h76

2.50

AL100h100

2.00

AL65h110

2.00
Pcr /Pcr (N=50)

Pcr /Pcr (N=50)

50

1.50
1.00

AL26h76
1.50

AL100h100

1.00
0.50

0.50
0.00

50

100

150

200

Bearing length/mm

0.00

50

100

150

200

Bearing length/mm

Fig. 6. Comparison of capacity of aluminum tubular sections subjected to web crippling with different bearing lengths. (a) EOF, (b) IOF, (c) ETF and (d) ITF.

sections with the web slenderness ratio43 reached its peak under
IOF and ITF boundary and loading conditions.
Fig. 8 shows that typical loaddisplacement curves of aluminum tubular sections subjected to web crippling with different
web slenderness under four boundary and loading conditions. It
can be observed from Fig. 8 that as the web slenderness ratio was

small (30 and 43), the ultimate capacity was high, but after
reaching peak, the loaddisplacement curves fell fast and the
ductility was poor. Similarly, it can be observed from Fig. 8 that as
the web slenderness ratio is large, the specimens had low ultimate
capacity and good ductility. It is shown that as the web slenderness was largest ratio (88), the specimens had lowest ultimate

Y. Chen et al. / Thin-Walled Structures 90 (2015) 4960

100
90
80
70
60
50
40
30
20
10
0
30

80
70

N50

N100

N150

50

N50

40
30
20
10
0
30

43

52

88

43

N150

52

88

(h-2t)/t

(h-2t)/t

80

80

N50

70

N100

N150

60

60

50

50

40

30

20

20

10

10

43

52

0
30

88

N100

N150

40

30

0
30

N50

70

N/kN

N/kN

N100

N/kN

N/kN

60

55

43

52

88

(h-2t)/t

(h-2t)/t

Fig. 7. Comparison of capacity of aluminum tubular sections subjected to web crippling with different web slenderness. (a) EOF, (b) IOF, (c) ETF and (d) ITF.

90

80

AL60h65-EOF-N150(30)
AL65h110-EOF-N150(43)
AL26h76-EOF-N150(52)
AL100h100-EOF-N150(88)

70
60

70
60
N/kN

N/kN

50
40

40
30

20

20

10

10
02

10 12
/mm

14

16

18

20

10

15

20

25

80

AL60h65-ETF-N150(30)
AL65h110-ETF-N150(43)
AL26h76-ETF-N150(52)
AL100h100-ETF-N150(88)

70
60

60
50

40

40

30

30

20

20

10

10

8
/mm

10

12

AL60h65-ITF-N150(30)
AL65h110-ITF-N150(43)
AL26h76-ITF-N150(52)
AL100h100-ITF-N150(88)

70

N/kN

50

/mm

80

N/kN

50

30

AL60h65-IOF-N150(30)
AL65h110-IOF-N150(43)
AL26h76-IOF-N150(52)
AL100h100-IOF-N150(88)

80

14

10

12

14

16

/mm

Fig. 8. Loaddisplacement curves of aluminum tubular sections subjected to web crippling with different web slenderness. (a) EOF, (b) IOF, (c) ETF and (d) ITF.

Y. Chen et al. / Thin-Walled Structures 90 (2015) 4960

70

EOF
ETF

IOF
ITF

EOF
ETF

60

IOF
ITF

50
40

N/kN

50
45
40
35
30
25
20
15
10
5
0
30

N/kN

N/kN

56

30
20
10

43

52

0
30

88

43

52

(h-2t)/t

88

100
90
80
70
60
50
40
30
20
10
0
30

EOF
ETF

43

52

IOF
ITF

88

(h-2t)/t

(h-2t)/t

Fig. 9. Comparison of capacity of aluminum tubular sections subjected to web crippling under different boundary and loading conditions. (a) N50, (b) N100 and (c) N150.

90

80

AL60h65-EOF-N150
AL60h65-ETF-N150
AL60h65-IOF-N150
AL60h65-ITF-N150

80
70

60
50

50

N/kN

N/kN

60
40
20

20

10

10
0

8
/mm

10

12

14

16

14
12

10

12

14

AL100100-EOF-N150
AL100100-ETF-N150
AL100100-IOF-N150
AL100100-ITF-N150

12
10
N/kN

10
8
6

8
6
4

2
0

14

AL26h76-EOF-N150
AL26h76-ETF-N150
AL26h76-IOF-N150
AL26h76-ITF-N150

16

/mm

18

N/kN

40
30

30

AL65h110-EOF-N150
AL65h110-ETF-N150
AL65h110-IOF-N150
AL65h110-ITF-N150

70

/mm

10

12

14

16

18

20

22

/mm

Fig. 10. Loaddisplacement curves of aluminum tubular sections subjected to web crippling under different boundary and loading conditions. (a) AL60  65-N150,
(b) AL65  110-N150, (c) AL26  76-N150 and (d) AL100  100-N150.

capacity, but after reaching peak, the loaddisplacement curve fell

slowly and the ductility was good.
3.4. Effect of boundary and loading conditions
The comparison of the ultimate capacity of aluminum tubular
sections subjected to web crippling under different boundary and
loading conditions is shown in Fig. 9. It is shown that as the
bearing length was 50 and 100 mm, the web crippling ultimate
capacity decreased progressively under Interior-One-Flange (IOF),
Interior-Two-Flange (ITF), End-One-Flange (EOF) and End-TwoFlange (ETF) boundary and loading conditions. However, as the
bearing length was 150 mm, the web crippling ultimate capacity
under IOF, ITF, EOF, ETF boundary and loading conditions was
almost equal.
The loaddisplacement curves of aluminum tubular sections
subjected to web crippling under different loading conditions are

plotted in Fig. 10. It is observed that the initial axial compressive

stiffness of the specimens under four different boundary and
loading conditions was relatively same, and the specimens under
IOF and ITF boundary and loading conditions had high ultimate
capacity and good ductility.
3.5. Loadstrain intensity on web face curves
Strain distributions at the face of web region were derived from
the readings of three-element rosettes strain gauges and the
failure mechanism of aluminum tubular sections subjected to
web crippling was studied. The strains at the measuring points
of strain gauges under different load levels of typical specimens
are plotted in Fig. 11, in which the horizontal axis represents the
measuring point of strain gauges (as shown in Fig. 4b), the vertical
axis represents the strain (i), and the dash line represents the
strain corresponding to the yield load. The strain (i) could be

Y. Chen et al. / Thin-Walled Structures 90 (2015) 4960

3500

3000
30kN
50kN

3000
y=2466

40kN
55kN

2000

2000
1500

T2

T3
Ti

T4

1500

0
T1

T5

3000

T2

T3
Ti

T4

T5

4500
y=2466

2500

40kN
60kN

4000

30kN
50kN

3500

40kN
56kN

3000
i /

2000

i /

50kN
65kN

500

500

1500
1000

50kN
65kN

y=2466

2500
2000
1500
1000

500
0
T1

40kN
60kN

1000

1000

0
T1

y=2466

2500

i /

i /

2500

57

500
T2

T3

T4

T5

0
T1

Ti

T2

T3

T4

T5

Ti

Fig. 11. Strain intensity distribution of web. (a) AL65  110-EOF-N150, (b) AL65  110-IOF-N150, (c) AL65  110-ETF-N150 and (d) AL65  110-ITF-N150.

calculated as follows [10,11]:

pq
2
1  2 2 2  3 2 3  1 2
i
3

where 1, 2 and 3 are the rst, second and third principal strains,
respectively, which were obtained from three-element rosettes
strain gauges along the joint intersection region.
Under the EOF boundary and loading condition, the upper part
of web had high strain intensity, and all measuring points of T1
and T3 rstly entered plasticity. Under the ETF boundary and
loading condition, all measuring points have almost the same
strain intensity values, and the measuring point of T1 rstly
entered plasticity. Under the IOF boundary and loading condition,
all measuring points are in elastic phase in the ultimate limit state,
and the measuring point of T4 had the maximum strain intensity.
Under the ITF boundary and loading condition, strain intensity of
T3 located at the centerline of web kept the maximum value.

code (PcrC) is far larger than the experimental web crippling ultimate
capacity (Pcr), because the ultimate capacity reduction caused by outof-plane buckling of the thin web and effects of loading and boundary
conditions on web crippling ultimate capacity are not considered in
Chinese steel structures design code. The mean values of PcrE/Pcr ratio
are 0.51, 0.74, 0.55 and 0.82 with the corresponding COV of 0.410,
0.267, 0.442 and 0.260 under EOF, IOF, ETF and ITF boundary and
loading conditions, respectively. It is shown that the calculated result
obtained by using European steel structures design code is very
conservative. The experimental web crippling ultimate capacity is
relatively close to calculated result obtained by using European steel
structures design code under IOF and ITF boundary and loading
condition.

5. Finite element analysis

5.1. General

4. Comparison of web crippling ultimate capacity

The experimental web crippling ultimate capacity (Pcr) and the
calculated value using specication under EOF, IOF, ETF and ITF
boundary and loading condition are given in Table 4, respectively.
The specication values are calculated using geometry size of
aluminum tubular sections and the measured yield strength. PcrC
and PcrE are the web crippling ultimate capacity of aluminum
tubular sections beams obtained by using Chinese steel structures
design code (GB50017-2003) [12] and European design of steel
structures (Eurocode 3) [13], respectively.
The mean values of PcrC/Pcr ratio are 4.60, 2.88, 4.78 and 3.33 with
the corresponding COV of 0.501, 0.465, 0.524 and 0.501 under EOF, IOF,
ETF and ITF boundary and loading conditions, respectively. The
calculated result obtained by using Chinese steel structures design

The nite element software ABAQUS version 6.11 [14] was used to
develop nite element models for aluminum tubular sections subjected to web crippling loading. The model was based on centerline
dimensions (see Fig. 1) determined from measured geometry reported
in Table 1.The bearing plates, the aluminum square hollow section and
the interfaces between the bearing plates and the aluminum square
hollow section have been modeled. In the nite element model, the
measured cross-section dimensions and material properties obtained
from the tests were used. Specic modeling issues are described in the
following section.
5.2. Element mesh and material
The effect of element sizes in the cross-section of the aluminum
square hollow section was investigated to provide both accurate

58

Y. Chen et al. / Thin-Walled Structures 90 (2015) 4960

Table 4
Comparison of capacity of aluminum tubular sections subjected to web crippling between test and code.
Boundary and loading condition

Specimen

Pcr (kN)

PcrC (kN)

PcrC/Pcr

EOF

AL60  65-EOF-N50
AL60  65-EOF-N100
AL60  65-EOF-N150
AL65  110-EOF-N50
AL65  110-EOF-N100
AL65  110-EOF-N150
AL26  76-EOF-N50
AL26  76-EOF-N100
AL26  76-EOF-N150
AL100  100-EOF-N50
AL100  100-EOF-N100
AL100  100-EOF-N150
Mean
COV

15.05
27.05
37.20
11.50
18.40
30.40
2.65
4.65
6.30
2.10
3.10
3.65

30.40
49.40
68.40
39.55
66.86
90.91
16.40
25.63
38.24
11.70
22.66
32.26

2.02
1.83min
1.84
3.44
3.63
2.99
6.19
5.51
6.07
5.57
7.31
8.84max
4.60
0.501

IOF

AL60  65-IOF-N50
AL60  65-IOF-N100
AL60  65-IOF-N150
AL65  110-IOF-N50
AL65  110-IOF-N100
AL65  110-IOF-N150
AL26  76-IOF-N50
AL26  76-IOF-N100
AL26  76-IOF-N150
AL100  100-IOF-N50
AL100  100-IOF-N100
AL100  100-IOF-N150
Mean
COV

20.60
29.70
42.90
22.00
31.80
33.80
5.65
6.90
8.25
5.20
5.95
5.80

30.82
49.40
68.40
39.55
63.59
87.55
16.40
27.95
38.24
12.78
22.66
32.26

ETF

AL60  65-ETF-N50
AL60  65-ETF-N100
AL60  65-ETF-N150
AL65  110-ETF-N50
AL65  110-ETF-N100
AL65  110-ETF-N150
AL26  76-ETF-N50
AL26  76-ETF-N100
AL26  76-ETF-N150
AL100  100-ETF-N50
AL100  100-ETF-N100
AL100  100-ETF-N150
Mean
COV

12.90
26.00
35.95
10.95
19.80
31.20
2.45
4.65
8.30
1.70
2.65
3.50

ITF

AL60  65-ITF-N50
AL60  65-ITF-N100
AL60  65-ITF-N150
AL65  110-ITF-N50
AL65  110-ITF-N100
AL65  110-ITF-N150
AL26  76-ITF-N50
AL26  76-ITF-N100
AL26  76-ITF-N150
AL100  100-ITF-N50
AL100  100-ITF-N100
AL100  100-ITF-N150
Mean
COV

18.70
31.20
37.20
19.80
26.70
34.70
4.65
5.65
7.80
3.55
4.15
4.65

results and reduced computation time. Depending on the size of

the section, the nite element mesh sizes ranged from 2  2 mm
(length by width) to 6  6 mm were used in the anges and webs.
Dense meshing was used at the intersection region of bearing
plates and aluminum square hollow section. The bearing plates
were modeled using analytical rigid plates and the I-beams were
modeled using the C3D8I solid elements.
The material properties of the tested specimens are given in
Table 2. The measured stressstrain curve of the specimens was
used in the FEA. The material behavior provided by ABAQUS
allows the multi-linear stressstrain curve to be used. Since the
analysis of post-buckling involves large in-elastic strains, material

PcrE (kN)

PcrE/Pcr

PcrRE (kN)

PcrRE/Pcr

4.85
5.82
6.78
6.80
8.78
9.99
2.25
2.51
3.28
1.38
2.09
2.57

0.32
0.22
0.18min
0.59
0.48
0.33
0.85max
0.54
0.52
0.66
0.67
0.70
0.51
0.410

5.92
7.89
9.86
8.03
11.33
13.79
2.79
3.54
4.92
1.77
2.97
3.90

0.39
0.29
0.27min
0.70
0.62
0.45
1.05
0.76
0.78
0.84
0.96
1.07max
0.68
0.413

1.50min
1.66
1.59
1.80
2.00
2.59
2.90
4.05
4.63
2.46
3.81
5.56max
2.88
0.465

12.49
14.08
15.91
17.32
20.06
22.57
5.50
6.60
7.34
3.77
4.77
5.65

0.61
0.47
0.37min
0.79
0.63
0.67
0.97max
0.96
0.89
0.72
0.80
0.97max
0.74
0.267

12.53
14.76
17.22
17.14
20.62
23.89
5.51
6.96
8.04
3.81
5.08
6.24

0.61
0.50
0.40max
0.78
0.65
0.71
0.98
1.01
0.97
0.73
0.85
1.08max
0.77
0.276

30.40
49.40
68.40
40.47
64.24
88.39
16.40
27.95
38.24
11.70
22.22
31.07

2.36
1.90min
1.90min
3.70
3.24
2.83
6.69
6.01
4.61
6.88
8.39
8.88max
4.78
0.524

5.27
6.32
7.37
7.52
8.90
10.30
2.38
3.00
3.46
1.39
2.06
2.46

0.41
0.24
0.21min
0.69
0.45
0.33
0.97max
0.65
0.42
0.82
0.78
0.70
0.55
0.442

5.76
7.98
10.20
7.81
10.55
13.33
2.67
3.93
4.97
1.59
2.73
3.56

0.45
0.31
0.28min
0.71
0.53
0.43
1.09max
0.84
0.60
0.93
1.03
1.02
0.69
0.426

30.40
48.79
68.40
40.47
63.59
88.39
14.91
25.63
38.24
11.70
22.22
31.07

1.63
1.56min
1.84
2.04
2.38
2.55
3.21
4.54
4.90
3.30
5.35
6.68max
3.33
0.501

14.63
14.78
15.55
21.38
21.92
23.21
5.12
5.36
6.15
3.19
3.94
3.99

0.78
0.47
0.42min
1.08
0.82
0.67
1.10max
0.95
0.79
0.90
0.95
0.86
0.82
0.260

13.92
15.18
17.02
19.87
21.68
24.21
4.93
5.72
7.08
3.09
4.23
4.70

0.74
0.49
0.46min
1.00
0.81
0.70
1.06max
1.01
0.91
0.87
1.02
1.01
0.84
0.247

properties were incorporated into the FE model converting the

stressstrain (engineering) curve into true stress and logarithmic
plastic strain. The equations for true stress and plastic true strain
were specied in ABAQUS.
5.3. Loading and boundary conditions
The nodes of the aluminum square hollow section and the
bearing plate were restrained to represent the horizontal symmetry condition. The interface between the bearing plate and the
aluminum square hollow section were modeled using the surfaceto-surface contact option. The bearing plate was the target surface,

Y. Chen et al. / Thin-Walled Structures 90 (2015) 4960

59

Fig. 12. Failure mode attained from nite element analysis. (a) EOF, (b) IOF, (c) ETF and (d) ITF.

while the aluminum square hollow section was the contact

surface. The two contact surfaces were not allowed to penetrate
each other subjected to compression.
The vertical load applied to the aluminum square hollow
sections in the laboratory tests was modeled using the displacement control method. An imposed displacement was applied to
the nodes of the top bearing plate where the vertical load was
applied. The top bearing plate was restrained against all degrees of
freedom except for the translational degree of freedom in the
loading direction. In the anges fastened condition, the node
coupling method was used in the region where the anges were
connected to the bearing plates. The nodes were coupled together
in all degrees of freedom.

5.4. Verication of FEM

Finite element models were developed under EOF, IOF, ETF and
ITF boundary and loading conditions. It is necessary to verify the
FEM. In the verication of the FEM, a total of 48 aluminum square
hollow sections subjected to web crippling was analyzed. A
comparison between the experimental results and the nite
element results was carried out. The main objective of this
comparison is to verify and check the accuracy of the FEM. The
nite element models were veried against the experimental
results in terms of failure modes and ultimate capacity. The failure
modes obtained from the tests of aluminum square hollow
sections in Fig. 5 were compared with the FE predictions in
Fig. 12. It can be seen that the failure mode predicted by the FEA
was in good agreement with the laboratory test results. A
comparison of the test results (PTest) with nite element analysis
results (PFEA) of the web crippling ultimate capacities is shown in
Table 5. It can be seen that good agreement has been achieved
between both results for all specimens. The mean values of the
PTest/PFEA ratio are 1.08 with the corresponding COV of 0.084. The
minimum error was 4% and the maximum error is 12%. It is shown
that good agreement was achieved between the experimental and
nite element results for both the web crippling strength and
failure mode.

Table 5
Statistical result of PTest/PFEA.
Specimens number

Statistical item

Statistical value

48

Max
Min
Average
COV

1.12
1.04
1.08
0.084

6. Design formulas
Based on material strength failure of aluminum tubular sections under local compression, the calculated results obtained by
using Chinese steel structures design code are larger than the
experimental value. Because the small effect of bearing length on
the web crippling strength is considered in European steel
structures design code, the calculation obtained by using European
steel structures design code are generally quite conservative.
The calculation equations of web crippling ultimate capacity
under four boundary and loading condition according to European
steel structures design code are very complicated and more
conservative, so the effect of the bearing length was considered
to be improved. Based on the results of the parametric study, this
paper put forward more accurate calculation formulas (25) of
aluminum beam web crippling ultimate capacity. The design
values can be reduced appropriately according to importance of
structure. The calculation formulas can accurately predict experimental value. The mean values of ratio of the calculation values
obtained by using calculation formulas (25) (PcrRE) and experimental values (Pcr) are 0.68, 0.77, 0.69 and 0.84 with the corresponding COV of 0.413, 0.276, 0.426 and 0.247 under EOF, IOF, ETF
and ITF boundary and loading condition in Table 4.
The web crippling ultimate capacity of aluminum square
hollow sections under four boundary and loading conditions are
calculated using calculation formulas (2)(5) as follows:


h  2t =t h
ai
P EOF 0:90 6 
1 0:02  t 2 f y
2
130
t


h 2t =t h
ai
1 0:01  t 2 f y
P IOF 0:90 15 
50
t

60



h 2t =t h
ai
1 0:025  t 2 f y
P ETF 0:85 6 
60
t


h  2t =t h
ai
1 0:005  t 2 f y
P ITF 0:9 20 
16
t

Y. Chen et al. / Thin-Walled Structures 90 (2015) 4960

where PEOF web crippling ultimate capacity under EOF condition;

PIOF web crippling ultimate capacity under IOF condition; PETF
web crippling ultimate capacity under ETF condition; PITF web
crippling ultimate capacity under ITF condition; fy yield stress of
aluminum alloy tube; ht web height of aluminum alloy tube; t
web thickness of aluminum alloy tube and a bearing length.

7. Conclusions
An experimental investigation of aluminum square hollow
sections under four loading and boundary conditions subjected
to web crippling had been presented. The ultimate capacity, failure
modes, local deformations and strain intensity distributions of
typical specimens have been reported. In addition, the corresponding nite element analysis was also performed and the
validated FE model was used for the parametric study to evaluate
the effects of main geometric parameters on the behavior of
aluminum beams subjected to web crippling. Based on the
experimental and numerical investigations, the following conclusions can be drawn:
(1) The failure modes of the aluminum tubular section specimens
were webs out-of-plane buckling and concave deformation in
ange.
(2) Web crippling ultimate capacity under EOF and ETF boundary
and loading conditions followed a bearing length linear
function.
(3) The web crippling ultimate capacity decreases under EOF and
ETF loading and boundary conditions as the slenderness ratio
increases.
(4) The initial axial compressive stiffness of the specimens was relatively
close under four different boundary and loading conditions.
(5) The proposed design formulas of web crippling ultimate
capacity can be successfully employed as a design rule for
predicting web crippling ultimate capacity.

Acknowledgments
This research work was supported by the National Natural
Science Foundation of China (Nos. 51278209 and 51478047)
Program for New Century Excellent Talents in Fujian Province
University (2014FJ-NCET-ZR03), Incubation Programme for Excellent Young Science and Technology Talents in Fujian Province
Universities (JA13005) and the Research Grant for Young and
Middle-aged Academic Staff of Huaqiao University (No. ZQNPY110). The authors are also thankful to Fuan Steel Structure
Engineering Co., Ltd. for the fabrication of test specimens. The
tests were conducted in Fujian Key Laboratory on Structural
Engineering and Disaster Reduction at Huaqiao University. The
support provided by the laboratory staff is gratefully
acknowledged.
References
[1] Stephens SF, LaBoube RA. Web crippling and combined bending and web crippling
of cold-formed steel beam headers. Thin-Walled Struct 2003;41:107387.
[2] Hofmeyer H. Cross-section crushing behaviour of hat-sections (Part I: numerical modelling). Thin-Walled Struct 2005;43:114354.
[3] Hofmeyer H. Prediction of sheeting failure by an ultimate failure model using
the ctitious strain method. Thin-Walled Struct 2009;47:15162.
[4] Zhou F, Young B. Web crippling of aluminium tubes with perforated webs. Eng
Struct 2010;32:1397410.
[5] Macdonald M, Heiyantuduwa MA, KoteLko M, Rhode J. Web crippling
behaviour of thin-walled lipped channel beams. Thin-Walled Struct
2011;49:68290.
[6] Uzzaman A, Lim JB, Nash D, Rhodes J, Young B. Web crippling behaviour of
cold-formed steel channel sections with offset web holes subjected to interiortwo-ange loading. Thin-Walled Struct 2012;50:7686.
[7] Macdonald M, Heiyantuduwa MA. A design rule for web crippling of coldformed steel lipped channel beams based on nonlinear FEA. Thin-Walled
Struct 2012;53:12330.
[8] Zhou F, Young B. Web crippling behaviour of cold-formed duplex stainless
steel tubular sections at elevated temperatures. Eng Struct 2013;57:5162.
[9] Chinese Code. Metallic materials-tensile testing at ambient temperature. GB/T
228-2002. Beijing, China; 2002 [in Chinese].
[10] Wang W, Chen YY, Meng XD, Leon RT. Behavior of thick-walled CHS X-joints
under cyclic out-of-plane bending. J Constr Steel Res 2010;66(6):82634.
[11] Chen Y, Feng R, Wang J. Behaviour of bird-beak square hollow section X-joints
under in-plane bending. Thin-Walled Struct 2015;86:94107.
[12] GB50017-2003 Code for Design of Steel Structure. Beijing, China, 2003 [in
Chinese].
[13] Eurocode 3 (EC3). Design of steel structures-Part 13: General rulesSupplementary Rules for Cold-formed Members and Sheeting, EN 1993-1-8, CEN.
Brussels, Belgium; 2005.
[14] ABAQUS. Standard user's manual. Version 6.11. USA: Hibbitt, Karlsson and
Sorensen, Inc., vols. 13; 2006.