Structures 8 (2016) 6374

Contents lists available at ScienceDirect

Structures
journal homepage: www.elsevier.com/locate/structures

Study on the Bending Capacity of Cold-formed Stainless Steel

Hollow Sections
Baofeng Zheng a,, Ganping Shu a, Lianchun Xin b, Ran Yang c, Qinglin Jiang d
a

School of Civil Engineering, Key Lab of Concrete and Prestressed Concrete Structures of Ministry of Education, Southeast University, No. 2 Sipailou Nanjing, China
Architects & Engineers Co., Ltd. of Southeast University, Nanjing, China
Anhui Huadian Engineering Consultation & Design Co. Ltd., Hefei, China
d
Jiangsu Dongge Stainless Steel Ware Co., Ltd., China
b
c

a r t i c l e

i n f o

Article history:
Received 8 May 2016
Received in revised form 12 July 2016
Accepted 19 August 2016
Available online 24 August 2016
Keywords:
Cold-formed
Stainless steel
Hollow sections
Bending capacity
Tests
Direct strength method

a b s t r a c t
Experimental and numerical studies on the bending capacity of cold-formed stainless steel rectangular and circular hollow sections are reported. Eight four-point bending tests were carried out. Material properties, geometric imperfections, bending capacities, moment-curvature curves were obtained in the tests. Finite element
models were developed, and were veried using the test data. Material nonlinearity and geometric imperfections
were considered in the modeling. Parameter studies were conducted to expand test data with different material
properties and section slenderness. Based on the test results and the nite element analysis results, direct
strength formulas were proposed for rectangular and circular hollow sections, respectively. Test data were collected from literatures as references, and compared with predictions of the Eurocode, the American code, the
Australian/New Zealand code, the Chinese code, and the continuous strength method (CSM). Comparisons
show that the Eurocode, the American code, the Australian/New Zealand code, and the Chinese code are generally conservative. The Chinese code provides most conservative predictions, due to the neglect of strain hardening effect and section plasticity, especially for stocky cross sections. The performance of CSM method is better
than the design codes. The proposed formulas show better accuracy in predicting bending capacities of coldformed stainless steel hollow sections.
2016 Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.

1. Introduction
Stainless steel has become an attractive material in building constructions due to its durability, favorable mechanical properties, and esthetic appearance. Researches have been carried out on the behaviors of
stainless steel members, and design codes have been issued in some
countries. Nevertheless, stainless steel contains chromium and nickel,
which are more expensive than ferrum in the carbon steel. The high initial cost of the material is the most important issue that restraints its
wide application. Using the cross-section with excellent structural behaviors, and improving the accuracy of predictions of design codes are
critical points to be considered. In this paper, study was focused on
the bending behaviors of cold-formed stainless steel hollow sections.
Johnson and Winter [1] carried out tests on cold-formed stainless
steel hat sections and box sections to study the effectiveness of compression anges and webs. Rasmussen and Hancock [2], Talja and
Salmi [3], Real and Mirambell [4], Gardner et al. [5,6], Zhou and Young
[7], Theodanous and Gardner [8], Zheng et al. [9], Huang and Young
Corresponding author.
E-mail addresses: [email protected] (B. Zheng), [email protected]
(G. Shu).

[10], Afshan and Gardner [11], Bock et al. [12], Theofanous et al. [13],
Zhao et al. [1416] conducted tests on cold-formed stainless steel rectangular hollow sections in bending. Compared to the number of tests
on the rectangular hollow sections, few tests were founded in literature
on the circular hollow sections. Rasmussen and Hancock [2], Kiymaz
[17], Way [18] performed tests on cold-formed stainless steel circular
hollow sections in bending. In these experiments, several grades of
stainless steel were considered, including austenitic stainless steel 03
methods were proposed for the predictions of bending capacity of stainless steel hollow sections. The Eurocode 3 [19] employs the concept of
cross-section classication in the treatment of local buckling, and allows
to use section plasticity for stocky cross sections. In the American code
[20], section plasticity can be used in the design of rectangular hollow
sections, but it is not applicable to the design of circular hollow sections.
The provisions in the Australian/New Zealand code [21] are very similar
to those in American code, except that partial plasticity design is permitted for the circular hollow section bending members. In the Chinese
code [22], plastic design is not allowed for cold formed stainless steel
members, and the effective width method is adopted to account for
the effectiveness of slender sections. Experimental studies show that
these design code are conservative in predicting the bending capacity.
To improve the performance of the design codes, Gardner's group

http://dx.doi.org/10.1016/j.istruc.2016.08.007
2352-0124/ 2016 Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.

64

B. Zheng et al. / Structures 8 (2016) 6374

[2326] developed a new method to utilize the strain hardening property of stainless steel, named the continuous strength method (CSM).
Huang and Young [10] modied the effective width method in the
Eurocode 3 [19], and the direct strength method (DSM) formula in the
North American Cold-formed Steel Specication [27].
The objective of this paper is to investigate the prediction performance of the Chinese code [22] and to develop the direct strength method (DSM) for stainless steel hollow sections to utilize the strain hardening
properties of stainless steel. A series of tests on cold-formed stainless steel
hollow sections are rstly presented. Finite element models were developed and validated, and then used to perform parameter analysis. DSM
formulas were proposed based on the nite element results. Test data
in literatures were collected and compared with the predictions of design
codes, the continuous strength method, and the direst strength method.
2. Experimental study
Experimental investigations consist of material tests on cold-formed
tubes and the virgin plate, four four-point bending tests on rectangular
hollow sections, and four four-point bending tests on circular hollow sections. All the test specimens were cold-rolled from stainless steel plate of
austenitic steel S30408 (1.4301) [28]. The steel plate was initially rolled
into a circular hollow section and then crushed into a rectangular hollow
section. The nominal cross-sections of the rectangular hollow sections are
70 mm 3 mm and 100 mm 3 mm. The nominal cross-sections of the
circular hollow sections are 89 mm 3 mm and 127 mm 3 mm.
2.1. Material tests
A series of material coupon tensile tests were carried out. To obtain
the full information of the material properties around the cross section,
the cross sections were cut into several strips. The positions of the strips
are shown in Fig. 1. The width and the length of the strips are 10 mm
and 150 mm, respectively.
Sixteen coupons from the at part and two coupons from the corner
part were tested for the SHS 70 mm 3 mm, respectively. For SHS
100 mm 3 mm, twenty coupons were tested for the at part, and
two coupons were tested for the corner part. Six coupons were tested
for each circular hollow section. Besides the coupons from the nished
tubes, three coupons were cut from the virgin plate.
Coupon tensile tests were performed in a 100 kN SANS universal
testing machine. The tests were conducted in accordance with
the Chinese standard GB/T 228.12010 [29]. The gauge length of the

Fig. 1. Denitions of symbols and locations of coupons in cross-sections.

coupons was 50 mm. Load rate was about 0.5 mm/min before the strain
reached 0.01, and 5 mm/min after that. Prior to testing, a pair of strain
gauges were pasted on each face of the coupon to obtain the strainstress curve. To get rid of introducing moment to the corner coupons,
a pair of corner coupons were gripped symmetrically around the steel
bar which was located at the end with the same radius as the internal
radius of corner in the test specimens. Typical corner coupon is shown
in Fig. 2.
Material test results were processed according to the strain-stress
model proposed by Gardner [5]. The results of the material test are
shown in Tables 1 and 2 for virgin plates and the tubes, respectively.
In these two tables, E0 is the material Young's modulus, 0.2 is the material 0.2% proof stress, n is a strain hardening exponent, 1.0 is the material 1.0% proof stress, n0.2,1.0 is a strain hardening coefcient
representing a curve that passes through 0.2 and 1.0, u is the ultimate
tensile stress, is the percentage elongation after fracture.
From Tables 1 and 2, it is concluded that: (1) the nominal yield
strength 0.2 in the hollow sections are much higher than 0.2 of the virgin plates due to the cold working in the production process. The 0.2
(489.03 MPa) in the SHS70 3 is over two times of that in the virgin
material (243.26 MPa); (2) the ultimate strength is also increased due
to cold working, but the degree of increase is not as higher as that of
the nominal yield strength; (3) compared to the percentage elongation
in the virgin plates, the percentage elongation in the hollow sections
was lower. The percentage elongation in the corner part is still higher
than 20%, which shows very good ductility. (4) Because the coupons
of virgin plates were curved when taken off from the coil, the elastic
Young's modulus is a little bit lower than that of coupons of the hollow
sections.
2.2. Beam tests
A total of eight specimens, (two specimens for each section), were
tested under four-point bending condition. The dimensions of the specimens were measured, shown in Table 3, using the nomenclatures dened in Fig. 1. Both the local and global geometric imperfections were
measured. The global imperfections were measured using a Vernier caliper and one string. The tightened string was used to form a perfect
straight line between the two ends, along the longitudinal direction of
the specimen. The gap between the string and the edge of the specimen
was measured using the Vernier caliper. The maximum value of the gap
is regarded as the global imperfection. For the rectangular hollow sections, the local imperfections were examined using an arrangement
comprising a dial gauge and a milling machine [5]. For the circular hollow sections, the local imperfections were measured using Vernier caliper. Local imperfections for the circular hollow sections are the
average difference between the diameter at the weld and the diameter
at the direction 90 to the weld.
All specimens with a total length of 1900 mm were simply supported between two steel rollers which allowed axial displacement of the
beam's ends. The rollers were located 150 mm inward from the ends
of the specimens and, resulting in a span of 1600 mm. The specimens
were loaded symmetrically at two points by using a spreader beam.
The distance between the two loading point was 600 mm. Two knifehinge supports were settled between the spreader beam and the loading points to allow the rotation of the specimens during tests. Loads
were applied by using a 300 kN hand hydraulic bottle jack. To avoid
the premature failure of the specimens at the loading points due to
the concentrated loads, steel plates and steel rings were used to
strengthen the cross-section at the loading points for rectangular and
circular hollow specimens, respectively. The test conguration is
shown in Fig. 3. The steel plate and the steel ring for strengthening are
shown in Fig. 4.
A load cell was arranged at the top of the jack, below the bottom of
the loading frame to record the total force applied on specimens. At
each end of specimens, two displacement transducers (LVDT) were

B. Zheng et al. / Structures 8 (2016) 6374

65

Fig. 2. Corner coupons.

set on the top surface of specimens to measure the rotation of the ends.
Another three displacement transducers were placed at the two loading
points and the central point of the specimens to measure the vertical
displacements. A data acquisition system was used to record the applied
loads and the readings of displacement transducers.
Test results are listed in Table 4. In this table, Ft is the maximum load
applied using the hydraulic jack, Mt. is the ultimate bending moment.
For the slender sections, SHS100 3 and CHS127 3, the major failure
is the local buckling of the ange in compression. The compression
ange buckles inwards with consistent deformation at the web. This
type of failure mode is shown in Fig. 5(a) and (b) for rectangular and circular hollow sections, respectively. After reaching ultimate load, the
stiffness of the specimen became negative, while the vertical displacement increased very fast. For the stocky sections, SHS70 3 and
CHS89 3, no cross-section buckling failure occurred during the tests.
For SHS70 3 specimens, at the last stage of the loading, the load
reached a plateau, while the vertical displacement increased rapidly.
For CHS 89 3 specimens, the circular cross-section between loading
points became elliptical at the ultimate state, while the height of the
cross section considerably decreased. Fig. 5(c) shows the failure mode
of CHS89 3-1900a.
Fig. 6 shows the relationship between the moment at the central section and the curvature at the pure bending part. In this gure, the vertical axis is the moment at the center of the specimen, while the
horizontal axis is the curvature of the specimen at the pure bending
part. The curvatures were calculated using Eq. (1) [11]



8 Dmidspan Daverage
k

2
4 Dmidspan Daverage L2midspan

where Dmidspan is the vertical deection at mid span, Daverage is the average vertical displacement at the loading points, and Lmidspan is the length
between the loading points.
3. Finite element modeling and validation
The nite element models of test specimens were developed in
ABAQUS [30] using S9R5 element. Two different models (Model 1 and
Model 2) for every specimen were built. In Model 1, the full length of
test specimens were built with pinned-roller constraint at the bearings
shown in Fig. 7. To simulate the enhancement at the loading points and
the actual constraint at the bearings, nodes of the cross-sections at the
loading points and the supports were coupled with a set of reference
points with rigid body constraint. Enhanced corner material properties
were considered, and the enhanced region is assumed to extend as far

as two times the thickness into the adjacent at parts [5]. In Model 2,
only the pure bending part of the test specimen was built. This model
was simply-supported, and moments were applied at the two ends.
Rigid body coupling was applied at the loading points.
In these two models, the geometric imperfections were introduced
into model by using the rst eigen buckling mode scaled by the measured magnitude of local imperfection to the perfect structural model.
For the cold formed sections, the major type of residual stresses is bending residual stress [31]. During tensile coupon tests, the coupons are
straightened, which effectively re-introduces the bending residual
stresses into the coupons. Therefore, provided the material properties
are established using coupons cut from the cross-section, the effects of
bending residual stresses are inherently present, and do not need to
be dened explicitly in the numerical models [2,5,32].
Analysis results of these two models were compared with test results, shown in Table 5. In this table, Mt. is the test bending strength,
M1u and M2u are the predicted bending strength of Model 1 and
Model 2, respectively. It indicates that bending strength from the test
is slightly (about 5%) higher than that predicted by Model 1 on average.
That is due to the friction between the knife-hinge and the steel plates
(the steel rings for CHS) at the loading points. The test specimen was
strengthened by the spreader beam due to the friction force. Although
ne sand was used to reduce the friction, it was observed that the friction force pushed the knife-hinge together with the steel plates (the
steel rings for CHS) moving towards the supports. Compared the last
two columns in Table 5, it indicates that Model 1 and Model 2 provide
the same predictions for the test specimens. These two models also provide similar eigen buckling modes, and failure modes. Considering the
computation efforts, Model 2 is used in the following analyses.
Fig. 8 shows the failure modes predicted by the nite element results
(Model 1). Compared with Fig. 5, it can be concluded that nite element
model can predict failure modes correctly. Fig. 9 shows the momentcurvature curves obtained from tests and nite element models. Over
all the moment-curvature curves predicted by nite element model
agree well with the test curves. However, at the ultimate strength
stage, the test curves are a little bit higher than the curves predicted
by nite element models. That is mainly because of the friction force between the knife-hinge and the steel plates.
4. Proposed method
4.1. Parameter analysis
The parameter analysis was based on the numerical simulation
using Model 2. To expand the data on the bending capacity of coldTable 2
Weighted average tensile material properties of nished tubes.

Table 1
Weighted average tensile material properties of virgin plates.

Cross section

Position

E0

0.2

1.0

n0.2,1.0

Flat
Corner
Flat
Corner

196,838
195,303
193,096
194,591
201,807
203,186

489.03
620.24
408.27
560.63
385.35
329.02

586.71
781.52
487.31
679.45
440.13
375.84

771.86
895.98
712.46
822.84
678.77
648.44

3.90
4.30
4.17
4.49
5.44
6.22

3.94
4.09
3.67
3.73
3.13
2.61

0.42
0.25
0.50
0.31
0.59
0.60

Cross section

E0

0.2

1.0

n0.2,1.0

SHS70 3

SHS70 3
CHS89 3
SHS100 3
CHS127 3

176,472

243.26

297.85

636.57

3.77

2.25

0.60

SHS100 3

187,566

231.78

279.35

627.61

4.97

2.11

0.63

CHS89 3
CHS127 3

66

B. Zheng et al. / Structures 8 (2016) 6374

Table 3
Measured dimensions of test specimens.
Specimens

SHS70 3-1900a
SHS70 3-1900b
SHS100 3-1900a
SHS100 3-1900b
CHS89 3-1900a
CHS89 3-1900b
CHS127 3-1900a
CHS127 3-1900b

Dimensions

Imperfections/mm

H/mm

B/mm

t/mm

ro/mm

L/mm

Local

Overall

71.4
71.3
102.2
102.0
89.3
89.15
127.8
127.8

71.8
71.8
102.5
102.8

3.1
3.1
3.09
3.09
3.11
3.11
3.04
3.04

6.25
6.25
5.88
5.88

1900
1900
1900
1901
1900
1900
1900
1900

0.086
0.096
0.11
0.11
0.12
0.12
0.30
0.30

1.3
1.6
2.7
1.0
0
0.5
0
1.3

formed stainless steel hollow sections, bending members over a wider

range of material properties and section slenderness were calculated.
Thirty types of material parameters were used. The nominal yield
strength considered includes 200 MPa, 300 MPa, 400 MPa, 500 MPa,
and 600 MPa. The strain hardening exponent was set as 3, 4, 5, 6, 7,
and 8. Two-stage material model proposed by Gardner [5] was used
and material parameters for the second stage of the material model
were predicted using approximate formulas [33]. These material properties could represent the commonly used austenitic and duplex stainless steel. Ferritic stainless steel, which has higher strain hardening
exponent and lower u/y ratio, was not covered in this paper. The consideration of enhanced corner material properties was simplied as
specifying the average value of material properties in the at and corner
regions among the full cross section.
For the rectangular hollow sections, eighteen types of section slenderness were considered, ranging from 0.25 to 2.0. It should be noted
that the section slenderness is calculated using the local buckling stress
of the whole cross section and the nominal yield stress. The denition of
section slenderness is given in Eq. (2).

s
My
M crl

where My is the nominal yield moment and, Mcrl is the eigen buckling
moment of the whole cross section in bending. The overall prole of
the cross-section is 100 mm 100 mm and the thickness varied according to the section slenderness. The outer corner radius is two times of
the thickness.
For the circular hollow sections, twenty types of D/t ratio were considered ranging from 10 to 300, corresponding to the section slenderness from 0.1 to 1.5. The outer diameter of the cross-section for all
CHS sections is 200 mm, and the thickness varied according to the section slenderness. The length of the members in the calculations was
1000 mm and 2000 mm for rectangular and circular hollow sections, respectively, which is long enough to minimize the inuence of the rigid
body coupling at the ends to the bending capacity.

(a) Steel plate

Fig. 4. Steel plate and steel ring at the loading point.

Local imperfections were introduced into nite element models

using the rst eigen buckling mode scaled by the experienced imperfection magnitudes shown in Eq. (3) [5].

w=t


0:023 y = cr
0:2

for RHS
for CHS

where w is the imperfection magnitude, t is the thickness of cross section, y is the nominal yield stress, and cr is the elastic buckling stress.
The residual stresses were not considered in the nite element model.
Totally 540 models were analyzed for rectangular hollow sections,
and 600 models were analyzed for circular hollow sections. Results of
parameter analysis are shown in Figs. 10 and 11 (the black dots in
these gures). It indicates that the bending capacity ratio (Mu/My) is related to the section slenderness as well as material property. For the
members with the same section slenderness, the maximum variation
on the bending capacity ratio reaches up to 15% due to the difference
of material properties. In order to propose simple and practical design
formulas, the inuence of material properties on the bearing capacity
is neglected and the proposed formulas will be aimed to be applicable
to the commonly used stainless steel grades.
4.2. DSM formulas
Direct strength method (DSM) is one of the most popular design
methods in predicting capacities of thin-walled members. DSM is initially developed to replace the effective width method, which is too
complex in predicting the distortional buckling capacity. Recently,
DSM has been utilized in the design of stainless steel structural members. In this paper, the DSM method was employed to develop a prediction formula for the local buckling capacity of stainless steel hollow
sections.
DSM formulas were obtained by tting the nite element results
shown in Figs.10 and 11. Design formula for rectangular and circular
hollow sections are shown in Eqs. (4) and (5), respectively.
Mu

Mu

Fig. 3. Layout of the four-point bending test.

(b) Steel ring

1:10:171:82
0:32
10:62
0:33

M y u W p

My u W p

where Mu is the predicted bending capacity, is the section slenderness,

My is the yield moment, u is the ultimate tensile strength, and Wp is the
plastic section modulus. It should be noted that uWp was set as the
maximum of the designed bending capacity to avoid mathematic
error for compact cross sections.

B. Zheng et al. / Structures 8 (2016) 6374

67

Table 4
Results of four-point bending tests.
Specimens

Ft/kN

Mt/kNm

Failure mode

SHS70 3-1900a

53.84

13.46

SHS70 3-1900b
SHS100 3-1900a
SHS100 3-1900b

54.82
85.53
85.83

13.71
21.38
21.46

CHS89 3-1900a
CHS89 3-1900b
CHS127 3-1900a
CHS127 3-1900b

42.96
40.73
66.86
66.56

10.74
10.18
16.72
16.64

Outwards bulge in compression ange

adjacent to the steel plate
Extensive plastic deformation
Central inwards bulge in compression ange
Outwards bulge in compression ange
adjacent to the steel plate
Extensive plastic deformation
Extensive plastic deformation
Central inwards bulge in compression portion
Outwards bulge in compression portion
adjacent to the steel plate

When using the DSM formulas, section slenderness is an important

factor. According to the denition of section slenderness (Eq. (2)), critical local buckling moment Mcrl should be determined in advance. For
rectangular hollow sections, critical local buckling moment Mcrl can be
expressed by Eq. (6).
Mcrl k f

 2
42 E
t
W
121v2 b

where, E is the elastic Young's modulus, v is the Poisson ratio; t is the

thickness, W is the elastic section modulus, b is the width of the ange
measured from the mid-thickness (B t), and kf is the interactive factor, which indicates the contribution of the web stiffness to the local
buckling of the ange.

In order to develop an expression for the interactive factor kf, sixteen

rectangular hollow sections with different aspect ratio B/H were analyzed using the nite strip software CUFSM [34,35]. In the CUFSM, a
mode identication technique is utilized to obtain the pure local buckling mode. Typical calculation result is shown in Fig. 12.
Fig. 13 shows the analysis results along with the aspect ratio B/H. It
indicates that as the ratio of B/H increases, the interactive factor kf increases as well. The interactive factor is more sensitive to the ratio of
B/H as the ratio of B/H is less than 0.5. Based on the calculation results,
the interactive factor kf is expressed in a two-stage expression, i.e.
Eq. (7).
8
B
B
>
< 3:29 0:37 0:1b b0:5
H
H
kf
>
: 0:27 B 1:14 1:0N B 0:5
H
H

where B is the overall width of the ange, and H is the overall height of
the cross section.
Comparisons among the analysis results by CUFSM, the predictions
of the Chinese code [22], and Eq. (7) are shown in Fig. 13. The proposed
formula provides more accurate predictions than the Chinese design
code in calculating the interactive factor. The Chinese code [22] gives
unsafe predictions as the ratio of B/H is higher than 0.8.
For circular hollow sections, the critical local buckling moment Mcrl
can be approximately calculated using Eq. (8) [36].

(a) Rectangular hollow section (SHS1003-1900a)

E
t
Mcrl p W
31v2 R

where Mcrl, E, v, t, W are consistent with those in Eq.(6), and R is the radius of the cross section.
5. Comparisons of the test strengths with the design strengths
5.1. Current design codes

(b) Circular hollow section (CHS1273-1900a)

(c) Circular hollow section (CHS893-1900b)

Fig. 5. Failure modes of test specimens.

5.1.1. Eurocode (EN 1993-1-4:2006/A1:2015)

In the Eurocode [19], cross sections are classied into four classes according to their plastic behaviors. For Class 1 and Class 2, bending capacity was calculated using the plastic section modulus. For Class 3, the
elastic section modulus is used. For Class 4 whose failure is dominated
by local buckling, the effective cross section modulus is used. Eq. (9)
shows the design formula.

Mc;Rd

8
W pl f y
>
>
>
for Class 1 and Class 2
>
>

>
>
< W M0 f
el; min y

for Class 3
>
M0
>
>
>
>
W eff ; min f y
>
>
for Class 4
:
M0

B. Zheng et al. / Structures 8 (2016) 6374

22
20
18
16
14
12
10
8
6
4
2
0
0.0000

18

SHS1003-1900b

CHS1273-1900a

16

SHS1003-1900a

14

SHS703-1900a

CHS1273-1900b

12
M (kN.m)

M (kN.m)

68

SHS703-1900b

CHS893-1900a

10
CHS893-1900b

8
6
4
2

0.0002

0.0004

0.0006

0.0008

0
0.00000 0.00015 0.00030 0.00045 0.00060

k (mm-1)
(b) Circular hollow section

k (mm-1)

(a) Rectangular hollow section

Fig. 6. Moment-Curvature curves of test specimens.

where Mc,Rd. is the design bending capacity, Wpl is the plastic section
modulus, Wel,min is the minimum elastic section modulus, Weff,min is
the minimum effective section modulus, fy is the nominal yield stress,
and M0 is the partial factor. Table 6 shows the classication limits. In
this table, c is the width of the web or the ange, t is the thickness; d
is the diameter of the circular cross section; is the material factor, calculated using(235 E/fy/210,000)0.5.

is beyond a certain limit (see Eq. (11)).

5.1.2. American code (SEI/ASCE 8-02)

In the American code [20], two design methods are specied for
rectangular hollow section. One is based on the initial yielding assumption, while the other one utilizes the inelastic reserve capacity brought
about by the spread of plasticity through the section. The second method was employed in this paper. In the second method, the maximum
compression strain is dened by Cy fy/E0. Cy is the compression strain
factor, calculated by using Eq. (10).

where D is the diameter of the circular hollow section, Sf is the elastic

section modulus of full, unreduced cross sections, and Kc is the reduction factor calculated by Eq. (12).

8
>
>
<

w=t 1
 3

w=t1
1 bw=tb2
C y 32
>
2 1
>
:
1
w=t 2

Mn

Kc

8
>
>
< f yS f
>
>
: Kc f yS f

D 0:112E

t
fy
0:112E D 0:881E

fy
t
fy



1C E=f y
8:93c D=t

11

5:882C
8:93c

12

where C is the ratio of the proportionality stress-to-yield stress, and c is

equal to 3.048C.
10

where w is the width of the compression ange, t is the thickness of the

ange, 1 is calculated using 1.11/(fy/E)0.5, and 2 is calculated using
1.28/(fy/E)0.5. The bending capacity is then calculated by integrating
the stress along the cross-section based on the elastic-perfect-plastic
material assumption and plane section assumption. The maximum
bending capacity should not exceed 1.25Me. For circular hollow sections, plastic design is not allowed, and a linear interpolation is used
to account for the effectiveness of the cross-section when the D/t ratio

5.1.3. Australian/New Zealand code(AS/NZS 4673-2001)

In Australian/New Zealand code [21], the design process of the bending capacity is similar to that in the American code. For rectangular hollow sections, the upper limits of bending capacity is the plastic moment
(fySp) instead of 1.25Me for the stocky cross sections with w/t less than
1. For circular hollow sections, the upper limits of bending capacity is
the plastic moment (fySp). An interpolation formula is used to consider
the effective cross-section properties (see Eq. (14)). Eqs. (13) and (14)
show formulas used to calculate bending capacity of circular hollow sections in the Australian/New Zealand codes. The symbols in these

Table 5
Comparisons of test results and nite element analysis results.

Fig. 7. Boundary conditions in the nite element models.

Specimens

Mt
kNm

M1u
kNm

M2u
kNm

Mt/M1u

Mt/M2u

SHS70 3-1900a
SHS70 3-1900b
SHS100 3-1900a
SHS100 3-1900b
CHS89 3-1900a
CHS89 3-1900b
CHS127 3-1900a
CHS127 3-1900b

13.46
13.71
21.38
21.46
10.74
10.18
16.72
16.64

13.00
13.00
20.42
20.42
9.84
9.84
16.17
16.17

13.01
13.01
20.42
20.42
9.85
9.85
16.25
16.25
AVG.
COV.

1.04
1.05
1.05
1.05
1.09
1.03
1.03
1.03
1.05
0.02

1.03
1.05
1.05
1.05
1.09
1.03
1.03
1.02
1.05
0.02

B. Zheng et al. / Structures 8 (2016) 6374

69

2.0

MuuWp

1.8

Fem
Austenitic
Duplex
Ferritic
Proposed

1.6
M u/M yc

1.4

(a) Rectangular hollow section (SHS1003-1900a)

1.2
1.0
0.8
0.6
0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Fig. 10. Comparisons of the test data and the nite element results with the predictions of
the proposed DSM formula for rectangular hollow sections.

(b) Circular hollow section (CHS1273-1900a)

listed in Table 7, the reduction factor should be calculated according to

Eq. (15).
8
q
bc bc
>
>
2:411:63

p
>
>
t
>
>
!t
>
>
<

be

>
t
>
>
>
>
>
>
>
:

(c) Circular hollow section (CHS893-1900b)

Fig. 8. Failure modes predicted using FEM.

equations have the same meanings as those in Eqs. (11) and (12).
8
>
>
f y Sp
>
>
>
>
>
<
Mn f y S f
>
>
>
>
>
>
>
: Kc f yS f

D 0:078E

t
fy
0:078E D 0:31E

fy
t
fy
0:31E D 0:881E

fy
t
fy

13

0:178C
Kc

3:226c D=t 3:226c

14

5.1.4. Chinese code (CECS 410:2015)

In the Chinese code [22], the plastic design for RHS is not allowed
currently. Effective width method is adopted to account for the effectiveness of the slender section. When the b/t ratio is beyond the limits

!
0:092p 0:65 bc
t
p

15

p N1:5

16

where b is the total width of the plate, k is the buckling factor (4 for the
ange, and 23.87 for the web), and k1 is the interactive factor between
the web and the ange calculated using Eqs. (17) and (18).

k1

8
>
>
>
<
>
>
>
: 0:11

SHS1003-1900a

1
p

0:93

1:1
2

0:05

17

N1:1

MuuWp

2.0
Fem
Test-Austenitic
Test-Duplex
Proposed

1.8

FEM

1.6

CHS1273-1900a
SHS703-1900a
CHS893-1900a

Mu / My

M (kN.m)

p 1:0

 
kk1 2 E0 t 2
121v2 b

2.2
22
20
18
16
14
12
10
8
6
4
2
0
0.0000

where bc is the width of the plate in compression, be is the effective

width of the plate, t is the thickness of the plate, p is the slenderness
of the plate, equal to (/cr)0.5, is the maximum stress in the plate,
and cr is the critical buckling stress of the plate, calculated according
to Eq. (16).
cr



1C E=f y

3:15p 1:86

p 1:0

1.4
1.2
1.0
0.8
0.6
0.4

0.0002

0.0004

0.0006

0.0008

k (mm-1)
Fig. 9. Comparison of experimental and simulation (Model 1) moment-curvature curves.

0.0

0.2

0.4

0.6

0.8

1.0

1.2

Fig. 11. Comparisons of the test data and the nite element results with the predictions of
the proposed DSM formula for circular hollow sections.

70

B. Zheng et al. / Structures 8 (2016) 6374

Table 6
The classication limits for cold-formed stainless steel hollow sections in the Eurocode.
Limits

RHS

Class 1
Class 2
Class 3

where c, tc, and kc are the width, the thickness, and the buckling factor of
the plate next to the plate being calculated. For example, when calculating the interactive factor of the web to the ange, c, tc, and kc are the
width, the thickness, and the buckling factor of the web, respectively.
For the CHS, yield moment is used for sections within the limits given
in Table 7.
5.1.5. DSM method
Huang and Young [10] did tests and numerical simulations on the
lean duplex stainless steel rectangular cross section and improved the
DSM formula in North American Specication for the Design of Coldformed Steel Structural Members [27], as shown in Eq. (19).

MDSM

1:10:776
1 1M y

 !

Mcrl 0:4
Mcrl 0:4

10:15
1:1M y
>
:
1:1M y
1:1M y

l 0:776
l N0:776

19

where My is the yield moment of the cross-section, Mcrl is the critical

elastic local buckling moment, and l is the section slenderness.
5.1.6. CSM method
Gardner's group [2326] developed the continuous strength method
(CSM) to utilize the strain hardening property of stainless steel. The
CSM method is a strain based design approach featuring two key components: (1) a base curve that denes the level of strain that a cross-

1.8
1.6
1.4
1.2

kf

1.0
CUFSM
CECS410:2015
Proposed

0.8
0.6
0.4

c/t 33
c/t 35
c/t 37

d/t 502
d/t 702
d/t 2802

csm
y

20

8

3 
>
>
> 4:44  10 15; C 1 u
0:3
>
<
y
4:5
c
!

>
0:224
1
>
>
0:3 b0:6
1 0:342
>
:
c
0:342
c


E
fy



E Esh y
Nfy

21

22

f u f y
C 2 u y

23



fy
C4
u C 3 1
fu

24

Esh

where, csm is the maximum compression strain that the cross section
can carry, y is the yield strain (fy/E), u is the ultimate tensile strain,
p is the slenderness of rectangular hollow sections, c is the slenderness of circular hollow sections, Esh is the tangent elastic modulus of
the material as the stress is beyond the nominal yield stress, and C1,
C2, C3, and C4 are parameters, related to the types of material. For austenitic and duplex stainless steel, C1, C2, C3, and C4 are taken as 0.1,
0.16, 1.0, and 0, respectively. For ferritic stainless steel, C1, C2, C3, and
C4 are 0.4, 0.45, 0.6, and 0, respectively.
5.2. Comparisons
Test data on bending capacities of rectangular and circular hollow
sections were collected from literatures and compared with current design codes including, the Eurocode, the American code, the Australian/
New Zealand code, and the Chinese code, as well as the CSM method,
and the proposed method. Several notes should be made before the
comparisons: (1) The partial factors (M0 and ) were set to 1.0. (2) Tensile material properties from corresponding tests were used in calculations. (3) When using the Eurocode, the corner enhancement of
rectangular hollow section was considered in the capacity prediction
for Class 1, Class 2, and Class 3 cross-sections. (4) When using the
American code, the increase of yield stress in the corner region was
Table 7
The classication limits for cold-formed stainless steel hollow sections in the Chinese code.

0.2
0.0
0.0

c/t 72
c/t 76
c/t 90



csm 0:25
C 1 u
3:6 min 15;
y
y
p
18

8
>
<

Flange

section can carry in a normalized form, and (2) a material model,

which allows for strain hardening and, in conjunction with the strain
measurement, can be used to determine the cross-section bending resistance. The base curve for rectangular and circular hollow sections is
given by Eq. (20) [25] and Eq. (21) [26], while the material model is
shown in Eqs. (22), (23) and (24).

Fig. 12. Typical output of the CUFSM for RHS subjected pure bending.

s
c tb k

t c b kc

CHS

Web

Grades

0.2

0.4

0.6

0.8

1.0

B/H
Fig. 13. Comparisons of the analysis results and the predictions of the design code and the
proposed formula for the interactive factor.

S30403, S31603
S30408, S31608
S22053, S22253

RHS

CHS

Web

Flange

121
110
74

49
44
30

100
90
50

B. Zheng et al. / Structures 8 (2016) 6374

71

Table 8
Comparisons of test results and predictions of the current design methods and the proposed design method for rectangular hollow sections.
Ref.

Specimens

Grade

Mt
Men

Mt
Masce

Mt
Mau

Mt
Mcn

Mt
Mcsm

Mt
Mh

Mt
Mp

2
3

S1B1
RHS-1 B1
RHS-1 B2
RHS-1 B3
RHS-2 B1
RHS-2 B2
RHS-2 B3
RHS-3 B1
RHS-3 B2
RHS-3 B3
SHS80 80
SHS80 120
SHS80 80 4B1
SHS100 100 2-B1
SHS100 100 3-B1
SHS100 100 4-B1
SHS100 100 8-B1
RHS60 40 40-B1
RHS100 50 2-B1
RHS100 50 3-B1
RHS100 50 4-B1
RHS100 100 3-A
RHS120 80 3-A
RHS140 60 3-A
RHS100 100 3-C850
RHS120 80 3-C850
RHS140 60 3-C850
N40 40 2
N40 40 4
N80 80 2
N80 80 5
N100 50 2
N100 50 4
N120 60 2
N120 60 4
H40 40 2
H50 50 1.5
H150 150 3
H150 150 6
H140 80 3
H160 80 3
H200 110 4
100 100 4-B1
100 100 4-B2
80 80 4-B1
80 80 4-B2
60 60 3-B1
60 60 3-B2
80 40 4-B1
80 40 4-B2
SHS-50 2-B1
SHS-80 3-B3
SHS-80 3-B4
SHS-80 4-B3
SHS-80 4-B4
SHS-100 3-B3
SHS-100 4-B2
SHS-100 4-B3
RHS-75 45 2-B3
RHS-75 45 2-B4
RHS-95 45 2-B3
RHS-95 45 2-B4
RHS-100 50 2-B3
RHS-100 50 2-B4
RHS-120 60 3-B3
RHS-120 60 3-B4
50 30 2.5L900
30 50 2.5L900
50 50 1.5L900
50 50 2.5L900
70 50 2.5L1100
50 70 2.5L1100
100 50 2.5L1500

304L
1.4301

0.56
0.29
0.27
0.29
0.64
0.65
0.64
0.34
0.34
0.34
0.64
0.52
0.46
1.07
0.71
0.57
0.25
0.24
0.59
0.39
0.30
0.73
0.61
0.51
0.83
0.69
0.58
0.43
0.22
0.82
0.33
0.50
0.27
0.67
0.34
0.48
0.77
1.17
0.57
0.56
0.65
0.66
0.63
0.65
0.52
0.52
0.49
0.50
0.29
0.28
0.71
0.63
0.63
0.48
0.49
0.79
0.60
0.60
0.64
0.63
0.65
0.63
0.75
0.74
0.53
0.53
0.30
0.45
0.81
0.50
0.51
0.68
0.54

1.28
1.23
1.20
1.24
1.26
1.26
1.26
1.24
1.23
1.23
1.43
1.00
1.12
1.20
1.39
0.96
1.01
1.49
0.96
1.15
1.30
1.42
1.11
1.16
1.42
1.37
1.12
1.14
1.25
1.28
1.37
1.30
1.47
1.25
1.47
1.22
1.37
1.22
1.13
1.09
1.01
0.90
1.25
1.19
1.14
1.12
1.08
1.11
1.17
1.28
1.74
1.50
1.46
1.29
1.18
1.57
1.27
1.24
1.39
1.15
1.06
1.02
1.36
1.45
1.22
1.26
1.24
1.11
1.35
1.21
1.13
1.42
1.10

1.45
1.33
1.26
1.34
1.29
1.29
1.29
1.61
1.61
1.60
1.39
1.18
1.27
1.12
1.34
1.06
1.46
1.72
1.09
1.28
1.48
1.37
1.24
1.32
1.35
1.35
1.26
1.31
1.34
1.22
1.51
1.55
1.75
1.19
1.83
1.23
1.31
1.14
1.25
1.21
1.09
1.09
1.24
1.21
1.19
1.17
1.17
1.20
1.30
1.43
1.69
1.35
1.32
1.42
1.30
1.51
1.55
1.52
1.46
1.39
1.31
1.26
1.34
1.42
1.32
1.35
1.38
1.20
1.29
1.35
1.29
1.38
1.25

1.43
1.31
1.24
1.32
1.29
1.29
1.29
1.58
1.57
1.57
1.39
1.14
1.26
1.12
1.34
1.05
1.44
1.68
1.07
1.25
1.45
1.37
1.22
1.29
1.35
1.35
1.23
1.29
1.32
1.22
1.49
1.51
1.71
1.19
1.80
1.21
1.31
1.14
1.23
1.18
1.07
1.07
1.24
1.21
1.17
1.16
1.16
1.19
1.26
1.39
1.69
1.35
1.32
1.40
1.28
1.51
1.55
1.52
1.46
1.39
1.28
1.23
1.34
1.42
1.29
1.33
1.35
1.19
1.29
1.33
1.27
1.38
1.23

1.68
1.62
1.52
1.62
1.48
1.49
1.48
1.92
1.92
1.92
1.37
1.38
1.49
1.02
1.30
1.23
1.77
2.13
1.31
1.57
1.83
1.32
1.47
1.63
1.25
1.35
1.55
1.53
1.66
1.14
1.80
1.87
2.17
1.40
2.26
1.44
1.22
1.05
1.45
1.45
1.32
1.31
1.41
1.35
1.39
1.37
1.38
1.42
1.63
1.79
1.64
1.50
1.46
1.66
1.52
1.43
1.78
1.74
1.71
1.64
1.57
1.51
1.33
1.41
1.60
1.64
1.71
1.41
1.19
1.58
1.54
1.34
1.52

1.30
1.07
1.07
1.08
1.18
1.19
1.18
1.04
1.03
1.03
1.07
0.96
1.06

0.98
0.82
1.23
1.01
1.03
1.05

1.18
1.14

1.16
1.06
1.11

1.16
1.28
1.20
1.22
1.21
1.17

1.15
1.12
1.17
1.03
1.17
1.14
1.12
1.10
1.05
1.08
1.13
1.24

1.26
1.23
1.24
1.14

1.33
1.30
1.32
1.25
1.19
1.12

1.23
1.26
1.20
1.08

1.17
1.11

1.11

1.22
0.98
0.92
0.99
1.14
1.14
1.14
1.19
1.19
1.19
1.04
0.95
1.01
1.00
1.07
0.91
1.02
1.24
0.98
1.02
1.12
1.11
1.12
1.14
1.15
1.10
1.15
1.01
0.97
1.05
1.12
1.29
1.29
1.11
1.39
1.01
1.10
1.07
1.07
1.07
1.05
1.01
1.09
1.05
1.00
0.99
0.96
0.99
0.98
1.08
1.33
1.16
1.13
1.15
1.06
1.22
1.32
1.29
1.30
1.23
1.19
1.14
1.12
1.19
1.15
1.18
1.04
0.96
1.09
1.11
1.08
1.08
1.09

1.19
0.94
0.95
0.95
1.07
1.07
1.07
0.98
0.98
0.97
0.94
0.85
0.98
0.95
1.04
0.90
0.73
1.08
0.95
0.98
1.02
1.08
1.09
1.11
1.10
1.09
1.12
0.97
0.97
1.01
1.08
1.20
1.11
1.08
1.18
1.09
1.12
1.04
1.06
1.06
1.06
0.93
1.05
1.02
1.05
1.03
0.97
1.00
1.03
1.13
1.24
1.14
1.11
1.15
1.05
1.09
1.21
1.18
1.20
1.14
1.07
1.02
1.05
1.12
1.18
1.21
1.11
1.00
1.04
1.09
1.04
1.05
1.06

4
5

10

304
1.4301

1.4318

1.4318-C850

304

Duplex
HSA
Duplex
HSA
1.4162

304

1.4162

(continued on next page)

72

B. Zheng et al. / Structures 8 (2016) 6374

Table 8 (continued)
Ref.

11

12

13

14

15
16
This paper

Specimens

50 100 2.5L1500
150 50 2.5L1500
50 150 2.5L1500
RHS120 80 3-4PB
RHS60 40 3-4PB
SHS80 80 3-4PB
SHS60 60 3-4PB
RHS120 80 3-3PB
RHS60 40 3-3PB
SHS80 80 3-3PB
SHS60 60 3-3PB
60 60 2-3P
80 40 2-3P
60 60 2-4P
70 50 2-4P
70 50 2-4P
80 40 2-4P
80 40 2-4P
100 40 2-4P
100 40 2-4P
SHS50 50 3
SHS60 60 3
SHS100 100 3
RHS60 40 3-MA
RHS60 40 3-MI
SHS100 100 5
SHS120 120 5
RHS150 100 6
RHS150 100 8
SHS150 150 8
SHS40 40 2
SHS50 50 2
SHS60 60 3
SHS100 40 2
SHS70 3-1900a
SHS70 3-1900b
SHS100 3-1900a
SHS100 3-1900b

Grade

1.4003

1.4509
1.4003

1.4509
1.4003

1.4301
1.4307

1.4301
1.4571
1.4307
1.4404
1.4162
1.4509
1.4003
304

Mt
Men

Mt
Masce

Mt
Mau

Mt
Mcn

Mt
Mcsm

Mt
Mh

Mt
Mp

0.94
0.70
1.30
0.56
0.28
0.56
0.43
0.57
0.29
0.56
0.44
0.64
0.77
0.65
0.57
0.72
0.47
0.77
0.50
0.94
0.42
0.47
0.75
0.36
0.49
0.46
0.50
0.36
0.36
0.45
0.44
0.57
0.44
1.00
0.50
0.50
0.68
0.68

1.42
1.33
1.42
1.01
1.04
1.02
1.05
1.07
1.15
1.02
1.13
1.04
1.23
1.16
0.98
1.21
0.97
1.19
0.78
1.31
1.25
1.22
1.47
1.27
1.24
1.30
1.19
1.16
1.07
1.09
1.08
0.98
1.06
1.47
1.19
1.22
1.42
1.42
1.22
0.16

1.30
1.06
1.26
1.10
1.18
1.10
1.11
1.17
1.30
1.10
1.20
0.94
1.16
1.07
1.05
1.16
1.06
1.11
0.84
1.20
1.43
1.38
1.42
1.52
1.43
1.49
1.42
1.60
1.44
1.43
1.23
1.11
1.17
1.33
1.33
1.36
1.39
1.39
1.31
0.17

1.30
1.04
1.26
1.08
1.15
1.08
1.10
1.15
1.27
1.08
1.18
0.94
1.16
1.07
1.05
1.16
1.06
1.11
0.84
1.20
1.41
1.36
1.42
1.49
1.42
1.46
1.40
1.57
1.42
1.41
1.21
1.10
1.16
1.33
1.31
1.34
1.39
1.39
1.30
0.17

1.04
1.35
0.95
1.30
1.43
1.27
1.30
1.38
1.58
1.27
1.40
1.03
1.04
1.15
1.35
1.10
1.42
1.00
1.13
0.93
1.68
1.61
1.35
1.84
1.66
1.73
1.65
1.93
1.68
1.69
1.44
1.29
1.38
1.02
1.55
1.58
1.37
1.37
1.48
0.26

1.04
1.01
1.05
1.05
1.11
1.12
1.05
1.13
0.88

1.00
1.03

0.96

0.79

1.14
1.16

1.06
1.20
1.23
1.17
1.01
0.94
1.02
1.05
1.02
1.04

1.16
1.19
1.24

1.12
0.11

1.04
1.09
0.97
0.96
0.86
0.94
0.88
1.02
0.96
0.94
0.94
0.81
0.91
0.91
1.00
0.93
0.96
0.87
079
0.94
1.11
1.10
1.14
1.16
1.15
1.17
1.14
1.19
1.05
1.11
0.96
0.95
0.92
1.07
1.09
1.11
1.10
1.10
1.07
0.11

0.99
1.09
0.97
0.96
0.99
0.99
1.02
1.02
1.10
0.99
1.10
0.80
0.88
0.90
0.98
0.91
0.92
0.85
0.77
0.90
1.06
1.07
1.07
1.04
1.10
1.12
1.06
0.94
0.87
0.94
1.03
0.93
1.02
1.03
1.07
1.09
1.08
1.08
1.04
0.10

AVG.
COV.

Note: Other two continuous beam tests are not shown in this table, as the bending capacities of the specimens cannot be determined directly by the load force Section slenderness is
higher than 0.68, which is beyond the applicable limit of CSM method.

ignored for all the cross sections according to the related provisions in
the code. (5) When using the Chinese code, corner enhancement was
neglected for all the cross sections. (6) When calculating the yield moment My for the DSM method proposed by Huang, cold working effect
in the corner region was neglected, since the factor 1.1 in Eq.(19) has
been used to include the corner enhancement. (7) When calculating
the section slenderness for the CSM and DSM methods proposed by
this paper, the cold working effect in the corner region was included
in the yield moment Myc. (Myc = f Wf + c Wc, in which, f and
c are the nominal yield stress of the at and corner region, respectively,
and Wf and Wc are the elastic section modulus of the at and corner region, respectively.).
Calculation results are shown in Tables 8 and 9 for rectangular and
circular hollow sections, respectively. In these two tables, Men, Masce,
Mau, Mcn, Mcsm, Mh, and Mp are the predicted bending capacity by
using the Eurocode, the American code, the Australian/New Zealand
code, the Chinese code, the CSM method, the DSM formula proposed
by Huang, and DSM formula proposed by this paper, respectively.
From Table 8, it can be concluded that: (1) All the seven design
methods are conservative in predicting the bending capacity of rectangular hollow sections. (2)The Chinese code gives the highest ratio of the
test results over the predicted values, with the mean value of 1.48 and
large scatter of 0.26. The American code, the Australian/New Zealand
code, the Eurocode, and the CSM method show good performance in
the ascending order. (3) The DSM formula proposed by Huang provides
good predictions with the mean value 1.07. The proposed DSM method

gives the most accurate predictions. The major difference between

these two DSM formulas is the upper limit of the bending capacity.
The upper limit of Eq. (19) is 1.85My, while Eq. (3) adopts uWu as the
maximum value of bending capacity.
Fig.14 shows the ratios of the test results over the predictions by
Eurocode and Chinese code. It indicates that: (1) Chinese code shows
good predictions for specimens with slenderness higher than 0.75,
which is the limit beyond which the effectiveness of the section should
be considered. (2) Chinese code gives more conservative predictions for
Class 1 and Class 2 sections. That is because Chinese code uses the yield
moment, while the Eurocode adopts the plastic moment for Class 1 and
Class 2 sections. (3) For Class 3 and Class 4 sections with slenderness
ranging from 0.62 to 0.75, the two codes show similar performance.
Test data collected were also plotted in Fig. 10 together with the nite element results and the proposed strength curve. It can be concluded that: (1) Most of the test data points lie above of the line Mu/Myc =
1.0 which demonstrates that the strain hardening considerably contributes to the bending capacity of the stainless steel members. (2) Test data
points of members of austenitic and duplex stainless steel show the
same trend, a little bit above the proposed strength curve, while test
data points of ferritic stainless steel members lie below the proposed
strength curve. So, the proposed curve gives unsafe predictions for ferritic stainless steel members. Compared with austenitic stainless steel
and duplex stainless steel, ferritic stainless steel has low ratio of u/y,
which is the main reason for the different performance of austenitic
and duplex stainless steel members to the ferritic stainless steel

B. Zheng et al. / Structures 8 (2016) 6374

73

Table 9
Comparisons of test results and predictions of the current design methods and the proposed design method for circular hollow sections.
Ref

Specimens

Grade

Mt
Men

Mt
Masce

Mt
Mau

Mt
Mcn

Mt
Mcsm

Mt
Mp

2
17

C1B1
TB01
TB02
TB03
TB04
TB05
TB06
TB07
TB08
TB09
TB10
TB11
CHS89 3-1900a
CHS89 3-1900b
CHS127 3-1900a
CHS127 3-1900b

304L
304

0.24
0.37
0.47
0.47
0.41
0.33
0.33
0.42
0.36
0.20
0.27
0.30
0.21
0.21
0.23
0.23

1.15
1.24
1.17
1.06
1.25
1.32
1.30
0.96
1.33
1.21
1.37
1.25
1.21
1.15
1.07
1.07
1.19
0.11

1.50
1.56
1.74
1.48
1.66
1.44
1.44
1.23
1.59
1.59
1.37
1.25
1.59
1.51
1.40
1.39
1.48
0.14

1.15
1.24
1.17
1.06
1.25
1.32
1.30
0.96
1.33
1.21
1.37
1.25
1.21
1.15
1.07
1.07
1.19
0.11

1.50
1.24

1.32
1.30

1.59
1.37
1.25
1.59
1.51
1.40
1.39
1.41
0.13

1.15
1.31
1.29
1.17
1.34
1.37
1.35
1.04
1.40
1.14
1.14
1.28
1.14
1.08
1.07
1.06
1.21
0.12

1.10
1.16
1.28
1.17
1.24
1.15
1.14
0.98
1.23
1.07
1.06
1.04
1.09
1.04
1.02
1.01
1.11
0.09

18

This
paper

2205

1.4541
1.4435
304

AVG.
COV.
Note: D/t ratios of these test specimens are beyond the limit in Chinese code.

members. (3) Due to different strengthening methods used in tests to

prevent premature local buckling at the loading points, test data points
with similar section slenderness scatter in a considerable range.
From Table 9, we can conclude that: (1) the American code provides
the most conservative predictions followed by Chinese code, with the
ratio of test versus predictions as 1.48 and 1.41, respectively. These
two codes do not allow plastic design for circular hollow sections. And
the effectiveness of the cross-section is considered when the D/t ratio
is beyond certain limits according to the American code. (2) The
Eurocode and the Australian/New Zealand code provide the same predictions for all the test specimens. That is because the section slenderness limit for Class 2 and Class 3 in the Eurocode is almost the same
with the limit for the plastic/elastic design in the Australian/New
Zealand code. (3) CSM method gives good predictions for specimens
with section slenderness higher than 0.3, while for the case when the
slenderness is less than 0.3, this method shows very conservative predictions. (4) The proposed method provides good predictions on the
safe side, with the mean ratio (test over predictions) of 1.11 and the
scatter of 0.09.
Test data of circular hollow members are also plotted in Fig. 11 together with nite element analysis results and the proposed strength
curve. Due to the strain hardening of stainless steel, the ratios of the
test results over yield moments are higher than 1.0 for most of the
test specimens. The proposed curve lies on the lower boundary of the

2.5
Internal element
EN
CN

M t/M pred

2.0
1.5
1.0
Class 1,2

A series of tests on the bending capacity of stainless steel rectangular

and circular hollow sections were reported. Material properties, geometric imperfections, bending capacities, and moment-curvature
curves were obtained. Finite element models were built and validated
for the test specimens. Two kinds of nite element models based on
full-length members and pure bending segment of members provide almost the same bending capacity, which is about 5% lower than the test
results due to the friction between the knife-hinge and the strengthen
rig in the test. Parameter analysis were conducted to consider the inuence of section slenderness and material properties on bending capacity. DSM formulas were proposed by tting the nite element results for
rectangular and circular hollow sections, respectively. Test data collected from literatures were compared with the predictions of current design codes and the proposed method. Comparisons show that the
Eurocode, the American code, the Australian/New Zealand code, and
the Chinese code provide conservative predictions for both rectangular
and circular hollow sections. CSM method shows better predictions
than the design codes. The Chinese code provides very good predictions
for slender rectangular sections, but generates the most conservative
predictions for stocky cross sections due to the neglect of strain hardening effect and section plasticity. The proposed method provides safe and
accurate predictions for the bending capacity of austenitic and duplex
stainless steel hollow members. The average ratio and the standard deviation of the test results over the predicted strengths are 1.04 and 0.10
for rectangular hollow members, and 1.11and 0.09 for circular hollow
members. The behaviors of ferritic stainless steel members are different
from that of austenitic and duplex stainless steel members and the proposed formulas were not applicable. More work need to be done on the
bending capacity of ferritic stainless steel members.
Acknowledgements

0.62
0.65

0.0
0.0

6. Conclusions

Class 3
Class 4

0.5

test data. It provides better predictions for the test specimens with section slenderness less than 0.35.

0.2

0.4

0.6

0.75

0.8

1.0

Slenderness p

1.2

1.4

Fig. 14. Comparisons of the Eurocode and the Chinese code with the test data.

The research work described in this paper is supported by the Priority Academic Program Development of Jiangsu Higher Education Institutions and by the National Science Foundation of China through the
projects No. 51178098, No. 51378105 and 51578134. The nancial support is highly appreciated. Special thanks to the Jiangsu Dongge Stainless steel Ware Co., Ltd. for providing the test specimens.

74

B. Zheng et al. / Structures 8 (2016) 6374

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.istruc.2016.08.007.
References
[1] Johnson AL, Winter G. Behavior of stainless steel columns and beams. J Struct Div
ASCE 1966;92(5):97118.
[2] Rasmussen KJR, Hancock GJ. Design of cold-formed stainless steel tubular members.
II: beams. J Struct Div ASCE 1993;119(8):236886.
[3] Talja A, Salmi P. Design of stainless steel RHS beams, columns and beam-columns research note 1619. Finland: VTT Building Technology; 1995.
[4] Real E, Mirambell E. Flexural behavior of stainless steel beams. Eng Struct 2005;27:
146575.
[5] Gardner L. A new approach to stainless steel structural design. [PhD Thesis] London:
Structures Section. Department of Civil and Environmental Engineering, Imperial
College; 2002.
[6] Gardner L, Talja A, Baddoo NR. Structural design of high-strength austenitic stainless
steel. Thin-Walled Struct 2006;44:51728.
[7] Zhou F, Young B. Tests of cold-formed stainless steel tubular exural members. ThinWalled Struct 2005;43:132537.
[8] Theofanous M, Gardner L. Experimental and numerical studies of lean duplex stainless steel beams. J Constr Steel Res 2010;66:81625.
[9] Zheng BF, Shen XM, Xin LC, Shu GP. Test on the bending capacity of cold-formed
stainless steel rectangular hollow sections. Eighth international conference on
steel and aluminum structures, Hong Kong, China; 2016 [in press].
[10] Huang Y, Young B. Experimental and numerical investigation of cold-formed lean
duplex stainless steel exural members. Thin-Walled Struct 2013;73:21628.
[11] Afshan S, Gardner L. Experimental study of cold-formed ferritic stainless steel hollow sections. J Struct Eng 2013;139(5):71728.
[12] Bock M, Arrayago I, Real E. Experiments on cold-formed ferritic stainless steel slender sections. J Constr Steel Res 2015;109:1323.
[13] Theofanous M, Saliba N, Zhao O, Gardner L. Ultimate response of stainless steel continuous beams. Thin-Walled Struct 2014;83:11527.
[14] Zhao O, Rossi B, Gardner L, Young B. Behavior of structural stainless steel crosssections under combined loading-part I: experimental study. Eng Struct 2015;89:
23646.
[15] Zhao O, Rossi B, Gardner L, Young B. Experimental and numerical studies of ferritic
stainless steel tubular cross sections under combined compression and bending. J
Struct Eng 2016;142(2):04015110.
[16] Zhao O, Gardner L, Young B. Buckling of ferritic stainless steel members under combined axial compression and bending. J Constr Steel Res 2016;117:3548.

[17] Kiymaz G. Strength and stability criteria for thin-walled stainless steel circular hollow section members under bending. Thin-Walled Struct 2005;43:153449.
[18] Way J. Structural design of stainless steel circular hollow sections. Report to ECSC.
The Steel Construction Institute; 2000.
[19] EC3. Eurocode 3: Design of steel structures-part 14: General rules-supplementary
rules for stainless steels. European Committee for Standardization; 2015[EN 19931-4:2006/A1:2015, CEN, Brussels].
[20] ASCE. Specication for the design of cold-formed stainless steel structural members.
Am Soc Civil Eng 2002 [SEI/ASCE-8-02, Reston, Virginia].
[21] AS/NZS. Cold-formed stainless steel structures. Australian/New Zealand-standard,
AS/NZS 4673:2001, standards Australia, Sydney, Australia; 2001.
[22] CECS. Technical specication for stainless steel structures. China Association for Engineering Construction Standardization, CECS 410:2015. Bejing China: China Planning Press; 2015.
[23] Ashraf M, Gardner L, Nethercot DA. Structural stainless steel design: resistance based
on deformation capacity. J Struct Eng 2008;134(3):40211.
[24] Gardner L, Theofanous M. Discrete and continuous treatment of local buckling in
stainless steel elements. J Constr Steel Res 2008;64:120716.
[25] Afshan S, Gardner L. The continuous strength method for structural stainless steel
design. Thin-Walled Struct 2013;68:429.
[26] Buchanan C, Gardner L, Liew A. The continuous strength method for the design of
circular hollow sections. J Constr Steel Res 2016;118:20716.
[27] AISI. North American specication for the design of cold-formed steel structural
members. AISI S100-2007, North American cold-formed steel specication.
Washington, DC: American Iron and Steel Institute; 2007.
[28] GB/T 20878-2007. Stainless and heat-resisting steels-designation and chemical compositions. Beijing: China Standard Press; 2007[in Chinese].
[29] GB/T 2281-2010. Metallic materials-tensile testing-part 1: method of test at room
temperature. Beijing: China Standard Press; 2010[in Chinese].
[30] ABAQUS. ABAQUS/standard user's manual, version 67 Pawtucket, RI; 2007.
[31] Gardner L, Cruise RB. Modeling of residual stresses in structural stainless steel sections. J Struct Eng 2009;135(1):4253.
[32] Ashraf M, Gardner L, Nethercot DA. Finite element modelling of structural stainless
steel cross-sections. Thin-Walled Struct 2006;44:104862.
[33] Quach WM, Teng JG, Chung KF. Three-stage full-range stress-strain model for stainless steels. J Struct Eng 2008;134(No 9):151827.
[34] Schafer BW, Adany S. Buckling analysis of cold-formed steel members using CUFSM:
conventional and constrained finite strip methods. Proceedings of the eighteenth international specialty conference on cold-formed steel structures. FL, USA: Orlando;
2006. p. 3954.
[35] Li Z, Schafer BW. Application of the nite strip method in cold-formed steel member
design. J Constr Steel Res 2010;66(89):97180.
[36] Timoshenko S. Theory of elastic stability. 2nd edition. McGraw-Hill; 1961.