EUROSTEEL 2014, September 10-12, 2014, Naples, Italy

STRENGTH CURVES FOR WEB CRIPPLING DESIGN

First approach for stainless steel hat sections based on numerical analyses
Marina Bock, Esther Real, Enrique Mirambell
Universitat Politcnica de Catalunya (UPC), Department of Construction Engineering, Spain
[email protected], [email protected], [email protected]

INTRODUCTION
Hat sections belong to the family of cold-formed steel structural members and are often used as
secondary structural members under local transverse forces while supporting roof or wall cladding.
The webs of such sections have high height-to-thickness ratio; hence web crippling usually controls
their design. During last years, stainless steel structural applications have been growing due to their
combination of favourable material properties, durability and aesthetic appeal. Significant progress
has also been made in the development of design rules for this material, which is covered by the
European stainless steel design code EN1993-1-4 [1]. However, this code provides no formulation
for web crippling design and refers to the carbon steel cold-formed members provisions EN1993-13 [2]. Web crippling is a rather complex local instability governed by many geometric and material
parameters as well as type of loading: interior one flange (IOF), exterior one flange (EOF), interior
two flange (ITF) and exterior two flange (ETF). As a result, the design provisions to predict web
crippling strength provide empirical equations based on test data adjustment that lack theoretical
background. Although some theoretical models are available to predict the web crippling strength
on carbon steel hat sections [3, 4], they are too cumbersome for hand calculation purposes. Some
background related to the web crippling of stainless steel cross-sections can be found in [5] and [6].
On the other hand, the treatment of most of the failure modes within the European design provisions
are based on the so called strength curves, providing different slenderness-based functions ( ) for
a given instability. A recent investigation [7] has proven that such a slenderness based approach is
adequate for web crippling design of plate channel steel beams. The present study intends to extend
the method by proposing ( ) strength curves to predict web crippling strength of ferritic stainless
steel hat sections under IOF and EOF loading. The investigation has been based on numerical and
experimental data tested in other institutions [8]. A comparison between this proposal and predicted
resistances by EN1993-1-3 [2] and American standards [9] is also presented.
1

FINITE ELEMENT MODEL

1.1 Modeled tests and FE description

The finite element software Abaqus was employed to model 8 tests on ferritic stainless steel hat
sections subjected to web crippling (4 tests under IOF loading and 4 tests under EOF loading)
carried out at VTT Technical Research Centre of Finland [8]. It is important to mention that the
configuration of these tests was intended to reproduce the web crippling response in the vicinity of
the intermediate support (IOF) and the end support (EOF) of a continuous beam in the most general
case where the lips of the hat section are oriented upwards. Fig. 1 presents the dimensions of the hat
sections where h is the web height, b is the flange width, t is the thickness, c is the lip length and ri,
rm and R are the internal, centreline and external bending radius respectively. Table 1 presents the
reported material properties in [8] where E is the material Young modulus, 0.2 is the 0.2% proof
stress, u and u are the ultimate stress and its corresponding ultimate strain and finally, n and m are
the first and second strain-hardening parameters respectively. The centreline dimensions, the
specimens length, L, the support dimensions, S, the bearing plate length, ss, as well as the clear
distance between the steel plate under the force and the end support, e, (see Fig. 2) of these crosssections [8] are presented in Table 2 where important information is provided by the beam labeling.
Considering ITH_10 as an example, I is the load configuration, TH stands for Top Hat and 10 is ten
times the nominal thickness of the cross-section in mm. The geometry of these ferritic stainless

steel hat sections was discretized by using the four-node doubly curved shell element with reduced
integration S4R. The employed mesh size used in the model was optimized to achieve accurate
results whilst minimizing computational time. The whole stress-strain (engineering) material data
curve was obtained employing the compound two-stage Mirambell and Real model [10]. The true
stress and plastic logarithmic strain were specified in the numerical model. The model was intended
to simulate the experimental test configurations [8] as close as possible and therefore, different
boundary conditions were applied to both loading conditions (IOF and EOF). More details are
provided in the following sections.
c
Table 1.

Material properties of the modeled specimens [8]

Nominal
E
0.2
u
n
m
thickness (mm) (GPa) (MPa)
(MPa)
1
200
359 23.1 479 1.46
1.5
191
322 26.1 475 1.21
2
193
372 23.0 489 1.30
3
180
297 23.5 445 1.22

rm

u
0.0170
0.0160
0.0164
0.0160

h
ri

R
b
Fig. 1. Definition of symbols in the cross-section

Table 2.
Beam
ITH_10
ITH_15
ITH_20
ITH_30
ETH_10
ETH_15
ETH_20
ETH_30

Centreline dimensions and length of the modeled specimens. Experimental [8] and numerical results

h
(mm)
71.09
70.73
70.08
69.95
71.05
70.84
70.52
69.39

b
(mm)
72.89
70.56
69.72
68.86
72.85
70.47
69.65
68.86

c
t
rm
L
S
ss
e
Fu,test Ru,test
(mm) (mm) (mm) (mm) (mm) (mm) (mm) (kN) (kN)
24.17 0.99 1.65 399
50
25
10.01 5.00
24.11 1.53
1.9
399
50
25
20.73 10.37
24.02 1.99
2.4
399
50
25
34.84 17.42
23.82 2.95 4.25 399
50
25
55.01 27.51
24.15 0.99 1.65 399
50
25
75 10.05 3.59
24.03 1.53
1.9
399
50
25
75 21.06 7.52
23.98 1.99
2.4
399
50
25
75 36.29 12.96
23.74 2.94 4.25 399
50
25
75 58.90 21.04

Fu,num
(kN)
10.19
21.04
34.99
57.89
9.96
20.36
33.91
53.72
Mean
COV

Ru,num
(kN)
5.09
10.70
17.50
28.95
3.56
7.27
12.11
19.18

Fu,num/
Fu,test
1.018
1.032
1.004
1.052
0.991
0.967
0.934
0.912
0.989
0.046

1.2 Boundary conditions in the Internal One Flange (IOF) loading

The steel plate (ss) that applies the transverse force was assumed to be a rigid plate controlled by a
reference point (RP) in which all degrees of freedom were restrained except the vertical
displacement. The interface between the flange (slave surface, extended up to the corners) of the
stainless steel section and the bearing plate (master surface) was modeled as a contact pair (surfaceto-surface) assuming frictionless response in the tangential direction and hard response in the
normal one. The transverse compressive load was applied by means of an imposed vertical
displacement of the reference point (RP) of the bearing plate. Additional rigid plates (S) were
placed on both edges in contact with the lips of the cross section to model support conditions. Those
plates were also controlled by a reference point provided with appropriate boundary conditions to
allow rotation due to bending. To avoid possible local instabilities and slide in those support
regions, the ends of the flat parts of both the web and flange were restrained against vertical and
horizontal displacement as well as rotation in the x-axis along the same support length to model the
wooden blocks placed in such parts during testing (see Fig. 2a).
1.3 Boundary conditions in the External One Flange (EOF) loading
The steel plate (S) that applies the transverse force was modeled similarly to the IOF loading. This
plate was fixed to the lips of the hat section by screw clamps in the experimental test [8] which
were modeled by tying the surfaces in contact. The end bearing support (ss) was modeled as a rigid

surface and a contact pair was used to model the interface with the specimen. In the experimental
test [8], the distortional deformation of the further end support was restrained by placing a wooden
block between the webs of the cross-section which was accounted for in the FE model by defining
the hat section as a rigid body with its corresponding reference point in its center of mass.
Appropriate boundary conditions were applied to the corresponding controlling reference points of
the rigid surface and the rigid body which were free to rotate in the y-axis (see Fig. 2b).
F

ss

Wooden
block in test

F S

ss

Additional BC
Master surface

Rigid body (wooden

block in test)

RP

Tied surfaces
RP

Slave surface
Master surface
RP
Support region
Loading region

Slave surface
RP
Master surface
End support region
Loading region

a)

b)

Fig. 2. FE model details for the a) IOF loading and b) EOF loading
60

60
50

ITH_30

40

Load (kN)

Load (kN)

50

ITH_20

30
20

ITH_15

10

ITH_10

0
0

Displacement (mm)

Test
FE
Fu,test
Fu,num
10

ETH_30

40
ETH_20

30
20

ETH_15

Test
FE
Fu,test
Fu,num

10
ETH_10
0

12

a)

Displacement (mm)

b)

Fig. 3. Load-displacement response of the beams subjected to a) IOF and b) EOF loading

a)

b)

Fig. 4. Typical web crippling failure modes in a) IOF loading and b) EOF loading

1.4 FE model validation

Fig. 3 presents a comparison between the experimental and the numerical load-displacement
response in which the corresponding ultimate loads have also been highlighted whereas Table 2
presents the applied ultimate experimental load, Fu,test, the web crippling resistance per web, Ru,test,
the ultimate numerical load, Fu,num, the ultimate numerical resistance per web, Ru,num, as well as the
numerical to experimental ratio. In line with other studies [11, 12], a strong discrepancy between
the experimental and numerical stiffness is observed in Fig. 3, due to the FE model sensitiveness to

the boundary conditions. On the other hand, overall good agreement between test and numerical
ultimate loads was achieved with an average value of the FE to experimental ultimate load ratio of
0.989 and a coefficient of variation (COV) of 0.046. Hence this FE model is deemed reliable and
suitable to conduct further parametric studies. Some experimental and numerical failure modes are
presented in Fig. 4 for both IOF and EOF loading.
2

PARAMETRIC STUDIES

2.1 Purpose and numerical models

Having validated the numerical model, parametric studies were conducted with the aim to study the
influence of the different parameters affecting web crippling capacity and determine the adequacy
of a slenderness-based approach formulation for hat sections subjected to web crippling under both
IOF and EOF loading conditions. The adopted material in this parametric study was ferritic
stainless steel with the following material properties: E=200 GPa, 0.2=350 MPa, n=15, u=450
MPa, m=3 and u=0.15. A total of 7 different sections subjected to both loading conditions (IOF and
EOF) were considered. The modeled centreline dimensions in mm were (hxbxcxtxrm):
30x30x17x1x1.5, 50x50x20x1.5x2, 70x70x25x1.5x2, 80x50x20x1.5x2, 100x50x20x1.5x2,
50x80x20x1.5x2 and 100x100x25x1.5x2. The specimens length (L), the bearing plate (ss) and the
support plates (S) were 400, 25 and 50 mm respectively. Additional beams were modeled to study
the influence of: the bending radius (rm= 1.5, 2.5 and 3 mm); the length (L=600 and 800 mm); the
bearing length (ss=12.5 and 50 mm for IOF whereas 10 and 35 mm for EOF); the thickness (t=0.5,
0.75, 1 and 2 mm); and the clear distance (e) for the EOF loading condition, which originally was
25 mm, was increased up to 75 mm.
2.2 Types of analyses
The essence of the slenderness-based approaches provided in current design rules is that a member
fails in a way involving buckling and yielding. On the basis of this principle, different strength
curves are provided. These curves depend on the relative slenderness and the so called reduction
factor given in Eq.(1). Consequently, 3 types of analyses were performed in this study to
determine the corresponding resistances per web: an elastic buckling analysis to obtain the elastic
critical resistance Rcr; a first order plastic analysis (non-linear material) to determine the first order
plastic resistance Rpl; and a fully material and geometrical non-linear analysis to determine the web
crippling strength Ru. In total, about 190 numerical analyses were conducted.

Ru
;=
R pl

R pl
Rcr

(1)

2.3 Results
Fig. 5 presents the variation of the reduction factor with the relative slenderness based on
numerical results for both modeled and tested [8] beams. For both loading configurations, the
results exhibit values of the reduction factor less than 1.0 which indicates that the first order plastic
load, Rpl, is suitable to define this variable for web crippling design. Moreover, the reduction factor
decreases for higher relative slenderness which evidences the contribution of elastic critical
buckling resistance, Rcr, to the web crippling strength and confirms that the given expression in Eq.
(1) for the relative slenderness is appropriate, in line with [7]. Hence these observations lead to
conclude that such a slenderness based approach based on ( ) strength curves is also adequate for
web crippling design of ferritic stainless steel hat sections subjected to IOF and EOF loading.
3

PROPOSED STRENGTH CURVES AND COMPARISON WITH OTHER

FORMULATIONS

A first approach is presented in this section by providing two strength curves given in Eq. (2) and
Eq. (3) for IOF and EOF loading respectively where is determined according to Eq. (1). These

curves are plotted in Fig. 5 in which strength curves for plain steel channels [7] were also depicted
for comparison purposes. Although this first proposal requires numerical analyses to determine both
first order plastic resistance Rpl and critical buckling resistance Rcr, it is strongly believed that
further theoretical predictive models could achieve good agreement with the numerical data.
1.2
1.0

Plain steel channels [7]

Proposed design curve - Eq.(3)
Tests [8]
Parametric

1.0

0.8

0.8

0.6

0.6

1.2

Plain steel channels [7]

Proposed design curve - Eq.(2)
Tests [8]
Parametric

0.4

0.4

0.2

0.2

0.0

0.0

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

a)

b)

Fig. 5. Numerical results and proposed strength curves for a) IOF and b) EOF loading

0.270
Ru , pro = R pl 0.78 for IOF loading

0.355
Ru , pro = R pl 0.90 for EOF loading

Table 3.
Beam
ITH_10
ITH_15
ITH_20
ITH_30

Rcr
(kN)
8.62
30.94
67.70
195.97

(2)
(3)

Experimental to predicted resistance ratios by different methods for the tested specimens [8]
Rpl
(kN)
21.73
38.12
53.41
67.20

Ru,EC Ru,ASCE Ru,pro Ru,test/ Ru,test/ Ru,test/

Rcr
Beam
(kN)
(kN) (kN) (kN) Ru,EC Ru,ASCE Ru,pro
3.20 3.32 4.09 1.56 1.50 1.22 ETH_10 4.17
6.19 6.80 9.49 1.67 1.53 1.09 ETH_15 15.21
10.29 11.65 15.82 1.69 1.50 1.10 ETH_20 33.05
15.27 17.65 27.54 1.80 1.56 0.99 ETH_30 95.76
Mean 1.68 1.52 1.10
COV 0.05 0.02 0.07
30

25

25

20

20

Ru,test (kN)

Ru,test (kN)

30

Rpl Ru,EC Ru,ASCE Ru,pro Ru,test/ Ru,test/ Ru,test/

(kN) (kN) (kN) (kN) Ru,EC Ru,ASCE Ru,pro
17.91 1.39 1.53 3.29 2.58 2.35 1.09
25.71 2.85 3.49 7.21 2.64 2.151 1.04
34.29 4.97 6.23 11.97 2.61 2.079 1.08
40.48 8.56 12.26 21.72 2.46 1.716 0.99
Mean 2.57 2.07 1.05
COV 0.03 0.11 0.03

15

15
10

10

EN1993-1-4
ASCE
Proposal

5
0
0

10

15

Ru,pred (kN)

20

25

EN1993-1-4
ASCE
Proposal

5
0

30

a)

10

15

Ru,pred (kN)

20

25

30

b)

Fig. 6. Predicted resistances by different formulations for a) IOF and b) EOF loading

Table 3 presents the obtained numerical values of Rpl and Rcr for the tested beams [8] as well as the
experimental resistance Ru,test to predicted strength by EN1993-1-3 [2] Ru,EC, ASCE [7] Ru,ASCE and
proposal by Eqs. (2-3) Ru,pro ratio. The results show overly conservative strength predictions
provided by design provisions, whereas more accurate results and smaller scatter are achieved by

this proposal for both loading conditions. A graphical comparison of this improvement is given in
Fig. 6 for the experimental data considered [8]. The mean values and COV for the numerical,
Ru,num, to predictive formulations ratio are presented in Table 4.
Table 4.

Numerical to predicted resistance ratio by different methods for the parametric study
IOF
Ru,num/ Ru,num/
Ru,EC Ru,ASCE
Mean 1.83
1.81
COV 0.22
0.26

Ru,num/
Ru,prop
1.15
0.11

EOF
Ru,num/ Ru,num/
Ru,EC Ru,ASCE
2.23
2.18
0.18
0.20

Ru,num/
Ru,prop
1.06
0.11

ACKNOWLEDGMENTS

The research leading to these results has received funding from the European Communitys
Research Fund for Coal and Steel (RFCS) under Grant Agreement No. RFSR-CT-2010-00026,
Structural Applications for Ferritic Stainless Steels and from Ministerio de Ciencia e Innovacin to
the Project BIA 2012-36373. The first author is grateful to the Secretaria dUniversitats i de
Recerca del Departament dEconomia i Coneixement de la Generalitat de Catalunya i del Fons
Social Europeu for their financial contribution. The authors gratefully acknowledge all the
experimental information provided by Asko Talja and Petr Hradil from VTT. They would also like
to thank Professor Feng Zhou from Tongji University for the provided support in the FE model.

REFERENCES
[1] EN1993-1-4, 2006. Eurocode 3: Design of steel structures - Part 1.4: General rules - Supplementary
rules for stainless steel, CEN.
[2] EN1993-1-3, 2006. Eurocode 3: Design of steel structures - Part 1.3: General rules - Supplementary
rules for cold-formed members and sheeting, CEN.
[3] Bakker MCM., Stark JWB., 1994. Theoretical and Experimental Research on Web Crippling of ColdFormed Flexural Steel Members, Thin Walled Structures, Vol. 18, No. 4, pp. 261-290.
[4] Hofmeyer H., Kerstens JGM., Snijder HH., Bakker MCM., 2001. New prediction model for failure of
steel sheeting subject to concentrated load (web crippling) and bending, Thin Walled Structures, Vol.
39, No. 9, pp. 773796.
[5] Zhou F., Young B., 2008. Web Crippling of Cold-Formed Stainless Steel Tubular Sections, Advances
in Structural Engineering, Vol. 11, No. 6, pp 679-691.
[6] Bock M., Arrayago I., Real E., Mirambell E., 2013. Study of web crippling in ferritic stainless steel
cold formed sections, Thin Walled Structures, Vol. 69, No. 4, pp 29-44.
[7] Duarte APC., Silvestre N., 2013. A New Slenderness-based Approach for the Web Crippling Design
of Plain Channel Steel Beams, International Journal of Steel Structures, Vol. 3, No. 3, pp. 421-434.
[8] Talja A., Hradil P., 2011. SAFSS Work package 2: Model calibration tests - Test Report, VTT
Technical Research Centre of Finland.
[9] American Society of Civil Engineers (ASCE), 2002. Specification for the design of cold-formed
stainless steel structural members (SEI/ASCE 8-02).
[10] Mirambell E., Real E., 2000. On the calculation of deflections in structural stainless steel beams: an
experimental and numerical investigation, Journal of Construction Steel Research, Vol. 54, No. 1, pp.
10933.
[11] Hofmeyer H., 2000. Combined web crippling and bending moment failure of first-generation
trapezoidal steel sheeting, PhD-thesis, Eindhoven University of Technology, Faculty of Architecture,
Department of Structural Design, The Netherlands, ISBN 90-6814-114-7.
[12] Kaitila, O, 2004. Web crippling of cold-formed thin-walled steel cassettes, Doctoral dissertation,
Helsinki University of Technology Laboratory of Steel Structures publications TKK-TER-30, Finland.