You are on page 1of 14

Articles

Joachim Lindner DOI: 10.1002/stco.201600004


Ulrike Kuhlmann
Adrian Just*

Verification of flexural buckling according to Eurocode


3 part 1-1 using bow imperfections
The resistance of compression members may be calculated by the
design buckling resistance based on the reduction factor χ, which is made for various cross-sections, types of interaction, steel
depends on the slenderness, or by the cross-section resistance grades and buckling directions.
based on internal forces according to a second-order analysis tak- Furthermore, FEM ultimate load calculations are car-
ing into account equivalent initial bow imperfections as well. The ried out taking into account plasticizing within the
second way is especially advantageous in the case of axial forces cross-section and along the longitudinal axis, residual
and bending. Values for equivalent initial bow imperfections are stresses, the linear elastic–ideal plastic constitutive law and
given in codes such as Eurocode 3 [5] or DIN 18800 [4]. For compar- L/1000 as the bow imperfection. These “real” load-carry-
ison, initial bow imperfections are derived here from the buckling ing capacities are also compared with the different ap-
curves for compression members with different cross-sections, proaches for bow imperfections. Parts of these investiga-
buckling directions and steel grades based on different types of tions have already been published [8].
cross-sectional interaction. The results are discussed and com- In the past, initial bow imperfections were derived for
mented with regard to the design of compression members. Fur- members loaded by axial forces only. This was also the
thermore, investigations are carried out for cases of axial compres- basis for Table 5.1 in [5]. Here, extensive investigations
sion and bending moments My or Mz. Data available from ultimate were carried out where the effects of bending moments My
load calculations are used for comparison. The consequences for or Mz acting in combination with the axial compression
the assumed design verification according to second-order theory force N are taken into account.
are discussed and a proposal is presented.
2 Derivation of the bow imperfection
1 Introduction
The accurate determination of the equivalent imperfection
According to Eurocode 3, Part 1-1, a member subjected to is carried out on the basis of centrally loaded struts, see
an axial compression force may be verified by two different Fig. 1, where a sine curve for the initial bow with the max-
approaches: an analysis based on the European buckling imum value e0 is assumed.
curves given in EN 1993-1-1, section 6.3.1 (also called The equivalent bow imperfection is determined by in-
“equivalent member method”), or a cross-sectional analysis troducing the moment M, calculated by second-order the-
according to EN 1993-1-1, section 6.2, with bending mo- ory, in an appropriate interaction formula. This procedure
ments according to second-order theory assuming initial is used as an example of the linear imperfection of Eq. (1)
bow imperfections e0 according to EN 1993-1-1, section [6], [10], [12], [19].
5.3. The consistency of these two approaches is investi-
gated in the following. N M
+
(1) ≤1
In this paper it is presupposed that the buckling curves Npl Mpl
used for the equivalent member method, which depend on
a statistical evaluation of many tests and a mechanical 1 1
M = N·e0 ·
(2) = N·e0 ·
model, are “correct”. Assuming that the cross-sectional anal- 1 N
1− 1−
ysis according to second-order theory (and including an in- α cr Ncr
itial bow imperfection) also results in the same load-carrying
capacity Nu, an imperfection e0 can be derived depending using

on the relative slenderness “l”. These imperfections are com-
pared with the values given in EN 1993-1-1, Table 5.1, which N = χ ·Npl
(3)
do not depend on the relative slenderness. This comparison

Shortened version of this paper selected and reviewed by


the Scientific Committee of SDDS 2016, the International
Colloquium on Stability and Ductility of Steel Structures,
Timişoara, Romania
* Corresponding author: adrian.just@ke.uni-stuttgart.de Fig. 1. Centrally loaded member

© Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin · Steel Construction 9 (2016), No. 4 349
J. Lindner/U. Kuhlmann/A. Just · Verification of flexural buckling according to Eurocode 3 part 1-1 using bow imperfections

χ = f (λ) reduction factor for critical buckling direction (4) (Eq. (6.39) in [5]), and for bending of rectangular hollow
sections about the weak axis z-z, Eq. (10b) may be used
Npl with h instead of b and t instead of tf.
λ=
(5) normalized relative slenderness For circular hollow sections (CHS):
Ncr
ψ = 1 − χ1,7
(10d)
leads to
Non-linear plastic interaction for rectangular solid sec-
e =
(6)
(1 − χ) · (1 − χ · λ 2) · Mpl tions:
0
χ Npl
ψ = 1 − χ2
(10e)
Generally, this equation may be written as
The German code DIN 18800-1 [4] contains the following
e0 =
(7)
(
ψ · 1 − χ · λ 2 Mpl
·
) solutions:
χ Npl For bending of I- or H-sections about the strong axis
y-y:
where y is the interaction factor for the interaction for-
mula used: 1− χ
ψ=
(10f) but ψ ≤ 1.0
0.9
Mpl,N
ψ=
(8)
Mpl For bending of I- or H-sections about the weak axis z-z:

To compare the different influences, it is more appropriate 1 − χ2


ψ=
(10g) but ψ ≤ 1.0
to use the non-dimensional value 0.91

L The following proposals for improvements have been sub-


j=
(9)
e0 mitted for the new Eurocode generation, which could be
released around 2021.
instead of the absolute value of e0. In contrast to [19], no Kindmann/Stroetmann [18] made proposals to mod-
partial safety factor was taken into account in the deriva- ify the rules in EN 1993-1-1, section 6.2.9.1, as follows:
tion used here. This means that the amplification factor acr For bending of I- or H-sections about the strong axis
was chosen as the system factor, independent of γM0 or y-y:
γM1, similarly to the other derivations [1], [6], [17].
h N · 1− χ 
ψ= −
(10h)
pl ( ) (
 · 1− χ ·
Npl
)
3 Interaction factors for different interaction formulae  2 4 · b· fy  Mpl,y

For the linear plastic interaction as shown above, y has to


be taken from Eq. (10a): for χ >
(
t w · h − 2· t f )
A
ψ =1− χ
(10a)
 Mpl,y χ ·Npl  Npl
EN 1993-1-1 [5], section 6.2.9.1, contains various interac- ψ= −  ·χ·
(10i)
 χ ·Npl 4 · t w · fy  Mpl,y
tion formulae for class 1 and 2 sections:
For bending of I- or H-sections about the strong axis
y-y (Eq. (6.36) in [5]): for χ ≤
(
t w · h − 2· t f )
A
1− χ
ψ=
(10b) but ψ ≤ 1.0
1 − 0.5·a For bending of I- or H-sections about the weak axis z-z:

where:  b N · 1− χ 
ψ= −
(10j)
pl ( ) (
 · 1− χ ·
Npl
)
 2 8· t f · fy  Mpl,z
A − 2· b· t f
a =
A
t w · h + 0.7· r 2
for χ >
For bending of I- or H-sections about the weak axis z-z A
(Eqs. (6.37) and (6.38) in [5]):
 Mpl,z χ ·Npl  Npl
2
 χ − a ψ=
(10k) −  ·χ·
ψ =1− 
(10c) but ψ ≤ 1.0  χ ·Npl 4 · h · fy  Mpl,z
 1 − a 

For bending of rectangular hollow sections about the t w · h + 0.7· r 2


for χ ≤
strong axis y-y (Eq. (10b)), t may be chosen instead of tf A

350 Steel Construction 9 (2016), No. 4


J. Lindner/U. Kuhlmann/A. Just · Verification of flexural buckling according to Eurocode 3 part 1-1 using bow imperfections

Bureau/CTICM proposed a set of formulae for strong axis


buckling which are relatively close to those of the existing
code [2]:

ψ = 1.0 if ≤ min 0.1; 0.35·a′


(10l) ( )
 χ2 
ψ = 1 −
(10m)
 (
a′ · 2 − a′ 
( )
if min 0.1; 0.35·a < χ ≤ a′
)
1− χ
ψ=
(10n) if χ > a′
1 − 0.5·a′
Fig. 2. Non-dimensional representative value j = L/e0 of ini-
where: tial bow imperfections for sections associated with buckling

curve “a”, S235, depending on the relative slenderness l
a′ =
(h − t f ) · t w
A

Lindner also proposed the following for strong axis buck-


ling y-y [2]:

ψ = 1.0 if χ ≤ 0.1
(10o)

ψ = 1.087 − 0.87· χ if 0.1 < χ ≤ 0.3


(10p)

(10q) (
ψ = 1.18· 1 − χ if χ > 0.3)
For circular hollow sections (CHS), the exact solution, ir-
respective of the code, is given by Eq. (10r): Fig. 3. Non-dimensional representative value j = L/e0 of ini-
tial bow imperfections for sections associated with buckling

(
ψ = cos 0.5· χ · π
(10r) ) curve “b”, S235, depending on the relative slenderness l

4 Influence of the cross-section


The results for buckling curve “b” are given in Fig. 3.
As shown above for flexural buckling, the interaction fac- Two groups of results can be recognized: One group of
tor y depends on the reduction factor c, based on the sections associated with buckling direction z-z, e.g. HEB
relevant buckling curve. For rolled and welded I-sections, 1000 with a minimum value j ≈ 196, and another group of
the buckling curve has to be chosen from Table 6.2 [5] sections associated with buckling direction y-y, e.g. HEA
depending on their dimensions, buckling direction and 160 with a minimum value j ≈ 270. The differences be-
steel grade. tween different sections in each group are rather small.
For the normal steel grade S235, buckling curves “a”, The value j = 200 proposed in Table 5.1 of EC3 lies in the
“b” and “c” are relevant for I-sections up to a flange thick- range of the results given here.
ness tf = 100 mm. Therefore, all “normal” rolled I- and The results for the non-dimensional value j = L/e0 (Eq.
H-sections were investigated regarding these three buck- (10)) for typical rolled I-sections associated with curve “c”
ling curves for the following sections: IPE 80 to IPE 600, for buckling direction z-z lead to a minimum value j ≈ 160,
IPEo 180 to IPEo 600, IPEv 400 to IPEv 600, HEA 100 to as shown in Fig. 4. Circular hollow sections (CHS) behave
HEA 1000, HEB 100 to HEB 1000, HEM 100 to HEM similarly to I-sections with a slightly higher value j ≈ 185.
1000 and HEAA 100 to HEAA 1000. Rectangular hollow sections (RHS) behave similarly, too,
For simplicity, the results contained in this section with a value j ≈ 180. The value j = 150 proposed in Table
only refer to the interaction formulae given in EN 1993-1- 5.1 of EC3 for curve “c” lies in the range of the results
1, section 6.2.9.1 [5], i.e. Eqs. (10b) and (10c) were used. given here.
For all three buckling curves “a”, “b” and “c”, the re- The results calculated and compared support to a
sults of the non-dimensional value j = L/e0 (Eq. (10)) are great extent the values given in Table 5.1 of EC3.

given depending on the relative slenderness l for a selec-
tion of sections. 5 Influence of different interaction equations
The results for buckling curve “a” are shown in Fig. 2.
Different rolled I-sections and CHS and RHS sections be- The type of interaction equation assumed for the cross-sec-
have very similarly, with a minimum value j = 417. Addi- tion verification based on second-order analysis has a great
tional investigations involving CHS sections [12] show that influence on the calculated value of the initial bow imper-
different diameters D and ratios D/t have very little influ- fection e0 or on the relative value j = L/e0. In addition to
ence on the value of j. The value j = 250 proposed in Table the interactions mentioned in section 3, the mechanically
5.1 of EC3 is much smaller than the values derived from exact interaction is taken into account when calculating
the buckling curves depending on the slenderness. the imperfection derived from reduction factor cu based on

Steel Construction 9 (2016), No. 4 351


J. Lindner/U. Kuhlmann/A. Just · Verification of flexural buckling according to Eurocode 3 part 1-1 using bow imperfections

Fig. 4. Non-dimensional representative value j = L/e0 of ini- Fig. 7. Non-dimensional values j = L/e0 of initial bow im-
tial bow imperfections for sections associated with buckling perfections for HEB 400, buckling direction z-z, buckling

curve “c”, S235, depending on the relative slenderness l curve “b”, steel grade S235, for different types of interaction

FEM calculations (as described shortly in the introduc-


tion). In all cases, a yield strength fy = 235 N/mm2 has been
assumed. Other yield strengths will be dealt with later.
The results for all interaction formulae mentioned in
section 3 for I- and H-sections were determined and com-
pared with the values in Table 5.1 of EC3 for plastic and
elastic verification.
As examples of the results of many calculations car-
ried out, results are given for…
–– HEA 400, buckling direction y-y, buckling curve “a” in
Fig. 5
–– HEA 120 , buckling direction y-y, buckling curve “b” in Fig. 8. Non-dimensional values j = L/e0 of initial bow im-
Fig. 6 perfections for HEM 120, buckling direction z-z, buckling
–– HEB 400, buckling direction z-z, buckling curve “b” in curve “c”, steel grade S235, for different types of interaction
Fig. 7

–– HEM 120, buckling direction z‑z, buckling curve “c” in


Fig. 8

Example 1
A detailed numerical example for HEA 400 (buckling
­direction y-y, S235, buckling curve “a”) shows how the
­values can be calculated. Results may be compared with
Fig. 5.
i) Basic data:
Fig. 5. Non-dimensional values j = L/e0 of initial bow im-
perfections for HEA 400, buckling direction y-y, buckling A = 159 cm2
curve “a”, steel grade S235, for different types of interaction
Wel,y = 2310 cm3

Wpl,y = 2562 cm3

iy = 16.836 cm

Npl,k = 159 · 23.5 = 3736.5 kN

Mpl,y = 2562 · 23.5 = 60207 kNcm



l = 1.0

Fig. 6. Non-dimensional values j = L/e0 of initial bow im- Buckling curve “a”
perfections for HEA 120, buckling direction y-y, buckling
curve “b”, steel grade S235, for different types of interaction χ = 0.665603

352 Steel Construction 9 (2016), No. 4


J. Lindner/U. Kuhlmann/A. Just · Verification of flexural buckling according to Eurocode 3 part 1-1 using bow imperfections

v) Linear elastic interaction:


E 21000
λa = π · = π· = 93.913
fy,k 23.5 Wpl,y 2562
j = jplastic interaction · = 584.08· = 647.5
Wel,y 2310
Assuming
Comparing the various results in Fig. 5 to Fig. 8, it can
1 1 be stated that in nearly all cases the interaction of
Mpl,N = N·e0 · = N·e0 ·
1 N EN 1993-1-1, section 6.2.9.1 [5], leads to the minimum
1− 1−
α cr Ncr values of j. Comparable values are derived from the inter-
actions of Kindmann/Stroetmann [18], Bureau [2] and
with Lindner [2].
In general, the following conclusions may be drawn:
N = χ · Npl –– The results would remain valid even if the interaction
equation of EN 1993-1-1 [5] would in future be replaced
Here, j may be determined depending on the inter­ by one of the equations discussed so far.
action factor ψ for different interaction equations: –– The values derived from FEM calculations are more fa-
vourable.
L j –– Linear plastic interaction gives higher values of j, lead-
j= = 1
e0 ψ ing to smaller values of initial bow imperfection e0.
–– If the load-carrying capacity is limited to the yield
Npl strength, a linear elastic interaction governs. In this case
j1 = χ · λ · λ a · i y ·
((1 − χ · λ ) ·M )
2
pl
the j value for plastic linear interaction may be multi-
plied by a factor Mpl/Mel.

L = l · la · iy = 1.0 · 93.913 · 16.836 = 1581.119 cm
6 Influence of steel grade
j1 = 0.665603·1.0· 93.913·16.836
All previous results depend on the non-dimensional slen-
3736.5 –
= 195.315 derness l. In Eq. (9) the relative bow imperfection j de-
((1 − 0.665603·1 ) ·60207)
·
2
pends on L in the form j = L/e0, where L can be written as
given in Eq. (11):
ii) Linear plastic interaction:
235
L = λ · λ1 · i = λ · i· 93.91·
(11)
ψ = 1 – χ = 1 – 0.665603 = 0.334397 fy,act

j1 195.315 where:
j= = = 584.1 –
ψ 0.334397 l non-dimensional slenderness

iii) Interaction according to EN 1993-1-1, section 6.2.9.1, E 235


λ1 = π · = 93.91· ε and ε =
Eq. (6.36): fy fy,act

2· b· t 2· 30·1.9 i radius of gyration


a =1− =1− = 0.283 < 0.5
A 159
Consequently, the values of j for steel grades higher than
n = 0.6656 > a 0 0.283 S235 may be calculated based on the values of j derived
from the various interaction formulae of section 3 for S235
1− χ 1 − 0.665603 by multiplying those by a factor e, where e = 0.924 for
ψ= = = 0.3895
λ − 0.5·a 1.0 − 0.5·0.283 S275, 0.814 for S355 and 0.748 for S420. Special buckling
curves for S460 are given in EN 1993-1-1 [5]. Therefore, for
j1 195.315 these buckling curves “a0”, “a” and “b”, all rolled I- and
j= = = 501.5
ψ 0.3895 H-sections mentioned in section 4 are also investigated
here. Furthermore, cold-formed CHS and RHS sections
iv) Interaction according to DIN 18800-1: associated with buckling curve “c” have been taken into
account.
1 − χ 1 − 0.665603
ψ= = = 0.37155
0.9 0.9 Steel grades S275, S355 and S420
Fig. 9 and Fig. 10 show examples of comparisons for se-
j1 195.315 lected sections in steel grade S355.
j= = = 525.7
ψ 0.37155 Fig. 9 again shows that the results for all four interac-
tions under discussion in EN 1993-1-1 are close together
and still well above the reference value of EN 1993-1-1,
Table 5.1. Fig. 10 shows that the results from EN 1993-1-1

Steel Construction 9 (2016), No. 4 353


J. Lindner/U. Kuhlmann/A. Just · Verification of flexural buckling according to Eurocode 3 part 1-1 using bow imperfections

Fig. 9. Non-dimensional values j = L/e0 of initial bow im- Fig. 12. Non-dimensional values j = L/e0 of initial bow im-
perfections for HEA 400, buckling direction y-y, buckling perfections for sections associated with buckling curve “a”,
curve “a”, steel grade S355, for different interaction formulae steel grade S460, depending on the non-dimensional slen-

derness l

Fig. 10. Non-dimensional values j = L/e0 of initial bow im-


perfections for HEB 400, buckling direction z-z, buckling Fig. 13. Non-dimensional values j = L/e0 of initial bow im-
curve “b”, steel grade S355, for different interaction formulae perfections for sections associated with buckling curve “b”,
steel grade S460, depending on the non-dimensional slen-

derness l

are not confirmed (178 < 200), but the FEM loadbearing
capacity achieves more favourable results. can be achieved by looking at different interaction equa-
tions, and the results are given in Figs. 14 to 16.
Steel grade S460, I- and H-sections Fig. 14 shows that the results for an IPE 140 section,
Fig. 11 to Fig. 16 show results for steel grade S460 in a buckling direction y-y, associated with buckling curve “a0”
similar way to those of Fig. 2 to Fig. 8 for grade S235. As for the interactions according to EN 1993-1-1, Kindmann/
before, the comparison refers only to the interaction for- Stroetmann [18], Bureau and Lindner [2] are relatively
mulae given in EN 1993-1-1, section 6.2.9.1 [5], which close together, although not as close as for steel grade S235
means Eqs. (10b) and (10c) were used. from Fig. 6.
The influence of the type of cross-section can be seen Fig. 15 shows the results for an IPE 500 section, buck-
in Fig. 11 to Fig. 13. ling direction z-z, in steel grade S460, where this section is
The min j values in Fig. 12 (217 < 250) and Fig. 13 classified as buckling curve “a”. This considers that the
(157 < 200) are smaller than those in Table 5.1 in [5]. classification in EN 1993-1-1, Table 6.2, is incorrect and
Therefore, a correction should be discussed. More clarity must be adjusted [6], [9]; it will be changed according to a
recent amendment already decided in TC 250/SC 3 [3].
Fig. 16 shows the results for an HEA 360 section, buckling
direction z-z, where this section is classified as buckling
curve “b”.
In both cases the results according to EN 1993-1-1,
section 6.9.2.1 [5], and Kindmann/Stroetmann [18] are
close together, whereas the plastic linear interaction leads
to much higher j values. Obviously, the constant values of
EC3, Table 5.1, for the plastic verification (j = 200 for curve
“b” and j = 250 for curve “a”) do not cover the results of
the non-linear plastic interaction sufficiently.
Therefore, the comparison with the “real” values de-
Fig. 11. Non-dimensional values j = L/e0 of initial bow im- rived by FEM calculations is of interest. Regarding the
perfections for sections associated with buckling curve a0”, FEM calculations, special conditions exist for rolled I- and
steel grade S460, depending on the non-dimensional slen- H-sections. As shown earlier by ECCS, explained in [10],

derness l the residual stresses do not change their absolute values if

354 Steel Construction 9 (2016), No. 4


J. Lindner/U. Kuhlmann/A. Just · Verification of flexural buckling according to Eurocode 3 part 1-1 using bow imperfections

ble 5.2: c/t = 426/10.2 = 41.8 > 27.2 (= 38 e). Additionally,


FEM calculations were carried out taking into account
that the web is categorized as a class 4 web. The results are
also given in Fig. 15. In fact, the differences between both
FEM calculations are small and may be neglected. How-
ever, regarding Fig. 15, the j value in EC3, Table 5.1, seems
to be too optimistic and should be corrected, although the
result for the class 4 web in this special case confirms the
EC3 value (min j = 257 > 250).

Steel grade S460, CHS and RHS sections


Fig. 14. Non-dimensional values j = L/e0 of initial bow im- It was shown earlier [12] that different sections only have
perfections for IPE 140, buckling direction y-y, buckling a small influence on j values. Therefore, only the following
curve “a0”, steel grade S460 and different interaction formu- sections were investigated: CHS 711.1x20, CHS 355.6x25.0,
lae RHS 300x10 and RHS 200x6. For cold-formed sections,
they are associated with buckling curve “c”. Taking into
account the interaction of EN 1993-1-1, section 6.2.9.1,
and the range of non-dimensional slenderness between 0.5
and 3.0, the j values vary between j = 190 and min j = 129.
This min j value is smaller than that in Table 5.1 with a
value j = 150.

Load-carrying aspects
The smaller j values compared with those of EN 1993-1-1,
Table 5.1, only show a tendency for the load-carrying ca-
pacity to be smaller than it should be. Therefore, ultimate
load calculations were also carried out, where the correct
j values were used instead of those in Table 5.1 in EC3-1-1.
Fig. 15. Non-dimensional values j = L/e0 of initial bow im-
This leads to the factor hul, where values > 1.0 indicate that
perfections for IPE 500, buckling direction z-z, buckling
curve “a”, steel grade S460 and different interaction formu-
results are on the unsafe side. Results for one unfavourable
lae section associated with each buckling curve are given in
Fig. 17.
The utilization rate hul is defined by Eq. (12):

n u,j [EC3,Tab.5.1]
ηul =
(12)
n u,j [FEM]

Eq. (12) shows that values hul > 1.0 represent results on the
unsafe side; the most unfavourable result is hul = 1.056.
Regarding the influence of steel grade, the following
conclusions may be drawn:
–– For S235 the constant values of initial bow imperfec-
tions given in Table 5.1 of EN 1993-1-1 lie mainly in the
range of values calculated here, irrespective of the as-
Fig. 16. Non-dimensional values j = L/e0 of initial bow im- sumed interaction formulae. Slight deviations – see, for
perfections for HEA 360, buckling direction z-z, buckling example, buckling curve “b”, buckling direction z-z in
curve “b”, steel grade S460 and different interaction formu- Fig. 3 – may be acceptable.
lae

higher steel grades are used, but they are approximately


the same as for S235. Therefore, several FEM ultimate
load calculations for higher steel grades were carried out
again, but this time using the residual stresses of S235.
Some of the results are shown in Fig. 14 and Fig. 15.
Fig. 15 shows that the j value in EC3 for plastic capac-
ity is not confirmed. The FEM results are more favourable
and therefore no reduction in the EC3 value seems neces-
sary.
The results for IPE 500 for steel grade S460 are calcu-
lated for the full cross-section, neglecting the fact that the Fig. 17. Utilization rate hul according to Eq. (12) for differ-
c/t ratio of the web exceeds the value of EN 1993-1-1, Ta- ent sections, buckling curves and yield stresses

Steel Construction 9 (2016), No. 4 355


J. Lindner/U. Kuhlmann/A. Just · Verification of flexural buckling according to Eurocode 3 part 1-1 using bow imperfections

Table 1. Correction factor δ

Load case

d 0.273 0.032 –0.189 –0.090 –0.420 –0.650 –0.210

–– For steel grades higher than S235, the constant values mate loads because representative initial imperfections
in Table 5.1 of EC3, Part 1, may nevertheless be retained were used instead of residual stresses and the effects of
for rolled I- and H-sections only based on the results of plasticizing. Therefore, the results of [20] are too unfa-
refined FEM calculations. Exceptions were identified vourable.
for grade S460 sections in buckling direction z-z. In the following, the bending moment according to
–– In other cases, for all sections with steel grades higher second-order theory, caused by a compression force N, a
than S235, the j values given in Table 5.1 of EN 1993-1-1 bow imperfection e0 and an external bending moment Mex,
have to be multiplied by ε, with ε = 0.924 (S275), 0.814 is added to Mpl,N. This Mpl,N value describes the basis for
(S355) and 0.748 (S420). further investigations, indicated by the interaction factor in
Eq. (8). The analysis according to second-order theory may
be carried out by a commercial computer program. To sim-
7 Axial compression and bending plify utilization, it is helpful to apply a formula. Therefore,
7.1 General Mpl,N is given by Eq. (13), which consists of two parts: one
for the bending moment due to compression force N and
All results shown in the previous sections are based on its bow imperfection as in Eq. (2), and one for Mex result-
investigations involving centrally loaded members. If bend- ing from external loads. This formulation is used for load
ing moments are present, the results may be less favoura- cases 1 to 7 according to Table 2.
ble, because a larger part of the cross-section in the longi-
tudinal direction may be plasticized, thus leading to a stiff- N
1 + δ·
ness reduction and therefore higher necessary bow L 1 Ncr
Mpl,N = N· ·
(13) + Mex ·
imperfections for simplified calculations. If this stiffness j N N
1− 1−
reduction is neglected, the calculated load-carrying capac- Ncr Ncr
ity might be too high.
Intensive additional investigations were carried out by where:
Lindner [10], [13] for the German code DIN 18800-2 [4], d correction factor, also called “Dischinger factor”, see Ta-
leading to a maximum deviation of up to approx. 5 % for ble 1
constant moment distribution and very slender members.
At the time this was also discussed in great detail in TC8 Introducing the relative capacities nu with Eq. (14a) and
of ECCS [11]. In the end it was decided to neglect this ef- mu with Eq. (14b):
fect of axial compression and bending because bigger er-
rors seem to occur for very special cases only. But it has to N
nu =
(14a)
be stated clearly that these moderate differences were only χ ·Npl
possible and acceptable because…
–– the plastic shape factor of I-sections subjected to bend- M
m u = ex
(14b)
ing about the weak axis z-z was limited in DIN 18800 Mpl
[4] to apl,z = 1.25,
–– in the original version of the new DIN 18800-2 (1990), and N/Ncr with Eq. (15):
no steel grade S460 was included and therefore the less
favourable results of S460 compared with S235 have no N
= nu · χ · λ2
(15)
effect, and Ncr
–– Eurocode 3 does not limit the plastic shape factor and
has not taken consequences from the fact that the re- leads to Eq. (16):
sults for S460 are less favourable than those for S235.
Mpl,N Npl L 1
(16) = nu · χ · · ·
Since those discussions, many more numerical results for Mpl Mpl j 1 − n u · χ · λ 2
compression and bending have become available. The in-
teraction equations for compression and biaxial bending in 1 + δ · nu · χ · λ 2
+m u ·
EN 1993-1-1, Annexes A and B [5], are mostly based on 1 − nu · χ · λ 2
numerical FEM results that were calculated by Lindner in
Berlin [13] and by Ofner in Graz [15]. Finally, the following Eq. (17) must be fulfilled:
During the elaboration of the German NA to EN
1993-1-1, a paper by Wagenknecht [20] dealt with the Npl L 1 1 + δ · nu · χ · λ 2
problem of bow imperfections based on diagrams first nu · χ ·
(17) · · + mu ·
Mpl j 1 − n u · χ · λ 2 1 − nu · χ · λ 2
published by Roik and Wagenknecht [16]. Unfortunately, = 1.0
ψ
the results of charts in [16], [20] do not present real ulti-

356 Steel Construction 9 (2016), No. 4


J. Lindner/U. Kuhlmann/A. Just · Verification of flexural buckling according to Eurocode 3 part 1-1 using bow imperfections

The value 0.9114 means that when using the interaction


equation from EN 1993-1-1, section 6.2.9.1, in Eq. (17), the
load may still be increased and therefore this result is on
the unsafe side compared with the load combination of the
FEM calculation.
Iteration results in hul = 1.060 > 1.0, where hul is de-
fined by Eq. (18):

U.L.[interaction]
ηul =
(18)
U.L.[FEM]
Fig. 18. Structural system and loading for example 2
U.L.[interaction] ultimate load calculated by the
It should be mentioned that contrary to sections 3 to 5, in relevant interaction formula (19a)
this case the interaction factor y must be multiplied by the
value of the relative axial force nu. U.L.[FEM] ultimate load calculated by FEM
analysis(19b)
7.2 Example 2
7.3 Systematic investigations
Compression force N and bending moment My caused by a 7.3.1 Steel grade S235
distributed load (load case 2, Table 2), are given in Fig. 18
(IPE 500, buckling direction y-y, S235, buckling curve “a”). For detailed investigations, the results of Ofner [15] in the
The load calculations of Ofner [15], [1], neglecting the presentation of [1] were used here. The following sections

radii r, lead to the following ultimate loads: for slenderness values l = 0.5, 1.0 and 1.5 were taken into
account:
nu = 0.3195 –– IPE 200 and IPE 500, buckling direction y-y, buckling
curve “a”
mu = 0.6190 –– IPE 200 and IPE 500, buckling direction z-z, buckling
curve “b”
χ = 0.6783 –– HEB 300, buckling direction y-y, buckling curve “b”
–– HEB 300, buckling direction z-z, buckling curve “c”

l = 1.0
Additional results for RHS 200 × 10 from [15], [1] are not
The basic data are similar to those of example 1 in section evaluated here.
5: Ofner [15] considered six different load cases (Nos. 1,
2, 3, 6, 8 and 9 from Table 2).
A = 2 · 20 · 1.6 + (50 – 2 · 1.6) · 1.02 = 111.74 cm2 Additionally, results from Lindner [13] are taken into
account, which are calculated for different sections and all

64 load cases shown in Table 2 for slenderness values l = 0.6,
a = 1.0 − = 0.4272
111.74 0.8, 1.0, 1.5, 2.0, 2.5 and 3.0:
–– IPE 300, buckling direction y-y, buckling curve “a”, load
Mpl = 49521.8 kNcm cases 2, 4
–– IPE 300, buckling direction z-z, buckling curve “b”, load
Npl = 2625.8 kN cases 1, 2, 8
–– HEM 340, buckling direction y-y, buckling curve “a”,
L = 1909.8 cm load case 1
–– HEB 200, buckling direction y-y, buckling curve “b”,
Interaction according to EN 1993-1-1, section 6.2.9.1 (see load cases 1, 2, 3, 8, 9, 10, 11
Eq. (10b)): –– HEB 200, buckling direction z-z, buckling curve “c”,
load cases 1, 9, 10, 11
1 − 0.6783·0.3195 0.7833 –– HEM 140, buckling direction y-y, buckling curve “b”,
ψ= = = 0.9961
1 − 0.5·0.4272 0.7864 load case 1
–– HEM 200, buckling direction z-z, buckling curve “c”,
Correction factor d = 0.032 (Table 1): load case 1

2625.8 1 There are two restrictions in the investigations mentioned:


0.3195·0.6783· ·(1909.8 / 250)·
49, 251.8 1 − 0.3195·0.6783 the work of Ofner [15] was limited to a non-dimensional

slenderness l = 1.5 and the calculations by Lindner [13] were
1 + 0.032·0.3195·0.6783
+0.6190· mainly calculated for nu = 0.4. Therefore, additional ultimate
1 − 0.3195·0.6783
load calculations (taking into account residual stresses, ini-
0.9961
tial bow imperfection L/1000, plasticizing within the
0.1121 + 0.7957 cross-section and along the length of the member) were car-
= = 0.9114 < 1.0
0.9961 ried out similarly to those in section 7.3.2, see Lindner [7]:

Steel Construction 9 (2016), No. 4 357


J. Lindner/U. Kuhlmann/A. Just · Verification of flexural buckling according to Eurocode 3 part 1-1 using bow imperfections

Table 2. Load cases investigated

LC 1 LC 2

LC 3 LC 4

LC 5 LC 6

LC 7 LC 8
Fig. 19. Utilization rate hul for different I-sections, S235, de-

pending on the non-dimensional slenderness l , axial com-
LC 9 LC 10 pression N (nu = 0.4) and bending moment My (not men-
tioned in detail), buckling direction y-y [13]

LC 11 LC = load case

–– IPE 140, buckling direction y-y, buckling curve “a”


–– IPE 140, buckling direction z-z, buckling curve “b”
–– HEA 360, buckling direction y-y, buckling curve “b”
–– HEA 360, buckling direction z-z, buckling curve “c”

These investigations were executed within a slenderness



range l = 0.5 to 3.0 and for an axial force proportion nu ≈
0.15 to 0.85
In the previous sections it was mentioned that five Fig. 20. Utilization rate hul for HEB 300, S235, depending

types of interaction nearly reach comparable results each on the non-dimensional slenderness l , axial compression N,
time, see Fig. 3: considering different nu values and bending moment My (not
–– Interaction according to EN 1993-1-1 (EC3-1-1), section mentioned in detail), buckling direction y-y [13]
6.2.9.1, see Eqs. (10b) and (10c)
–– Interaction proposal by Kindmann/Stroetmann, see
Eqs. (10h) and (10j) The following two diagrams again show results for the
–– Interaction proposal by Bureau, see Eqs. (10n) to (10q) basic case of buckling direction y-y (strong axis), steel
–– Interaction proposal by Lindner, see Eqs. (10r) to (10t) grade S235 and depending on the proportion of the axial
–– Interaction according to DIN 18800-1, see Eqs. (10f) load nu. Using FEM data from Ofner [15], the tendency
and (10g) could be mentioned that in nearly all cases the non-dimen-

sional slenderness l = 1.0 was less favourable than the
In order to minimize the number of results, normally only other two slenderness values (0.5, 1.5). Therefore, in the

the interaction from EC3-1-1 is used for the following eval- following, only the results for “l” = 1.0 are commented in
uations. The following diagrams give the ratio ηul of the detail.
ultimate load calculated by the relevant interaction for- Fig. 21 shows that for an IPE 200 section and buck-
mula of EC3-1-1 to the ultimate load calculated by FEM ling direction y-y, different load cases LC 1, 2, 3 and 8 ac-
analysis, see Eq. (18). A utilization ratio value ηul < 1.0 cording to Table 1 have only a small influence on the most
shows that the verification according to second-order the- unfavourable value of the utilization rate hul, which
ory is on the safe side. With a utilization ratio ηul > 1.0, the reaches a maximum in the range nu = 0.3 to 0.4 with hul =
verification may lead to load combinations exceeding the 1.056 > 1.0. In addition to the h1 results (representing the
FEM results, which means they are on the unsafe side. The plastic interaction according to EN 1993-1-1 (EC3-1-1),
results were calculated assuming that for second-order the- section 6.2.9.1, see Eq. (10b)), h2 results for the linear plas-
ory, an initial bow imperfection is based on j values given tic interaction (see Eq. (1)) are also given as dashed lines
in EC3-1-1, Table 5.1. in the diagram. All of them fall well below 1.0, indicating
In Fig. 19 the results for constant bending moment My that the linear plastic interaction leads here to rather con-
distribution utilization rates > 1 occur in all cases for the servative results.
highest investigated value of the non-dimensional slender- As load case 1 (constant moment distribution) does

ness l = 3.0, although for IPE 300, hul also lies on the un- not dominate in practical design, load case 2 has been cho-

safe side for the smallest value l = 0.6. sen for all further comparisons.
Results for other moment distributions are more satis- Therefore, the results for different sections are com-
factory, see Fig. 20. The figure shows that small values of pared in Fig. 22 for load case 2. For buckling curve “a”, IPE
nu, e.g. nu = 0.2, 0.4, are less favourable than values such 200 and IPE 500 nearly lead to the same most unfavoura-
as nu = 0.6, 0.8. ble value; however, the result for IPE 140 is worse, leading

358 Steel Construction 9 (2016), No. 4


J. Lindner/U. Kuhlmann/A. Just · Verification of flexural buckling according to Eurocode 3 part 1-1 using bow imperfections

Fig. 21. The hl values for IPE 200, S235, different nu values, Fig. 23. Utilization rate hul for IPE 140 (buckling curve “a”)

depending on the non-dimensional slenderness l , axial and HEA 360 (buckling curve “b”), S235, depending on

compression N (plotted as nu) and bending moment My, dif- non-dimensional slenderness l , different values of axial
ferent load cases (numerical data mu not mentioned in de- compression N (given as nu) and bending moment My, load
tail), buckling direction y-y, Ofner [15], [1] case 2 (numerical data not mentioned in detail), buckling
direction y-y, Lindner [7]

Fig. 22. Utilization rate hul for different sections, S235, de- Fig. 24. Utilization rate hul for IPE 200, S235, depending on
pending on different values of axial compression N (plotted different values of axial compression N (plotted as nu) for
– –
as nu), non-dimensional slenderness l , = 1.0 and bending non-dimensional slenderness l = 1.0 and bending moment
moment My, load case 2 (numerical data not mentioned in Mz for different load cases (numerical data mu not men-
detail), buckling direction y-y, based on Ofner [15], [1] and tioned in detail), buckling direction z-z, Ofner [15], [1]
Lindner [7]

to a value h1 = 1.074 > 1.0, which is on the unsafe side. 3 and 8 according to Table 1 only have a small influence
HEB 300 and HEA 360, which are associated with buck- on the most unfavourable value of the utilization hul,
ling curve “b”, are less favourable. All results for h2 (linear which lies in the range nu = 0.6 to 0.7 with a maximum hul
plastic interaction) are on the safe side. = 1.094 > 1.0. Load case 1 (constant moment distribution)
In addition, Fig. 23 shows the dependency of the re- is not considered applicable because it seldom appears in

sults concerning the non-dimensional slenderness l for the practical design. Again, load case 2 (distributed load) was
same critical sections as in Fig. 22. For the plastic interac- chosen to be used for all further comparisons.
tion according to EN 1993-1-1 (EC3-1-1), section 6.2.9.1, Therefore, the results for different sections are com-
Eq. (10b) utilization rate h1 (solid lines), the most unfa- pared in Fig. 25 for load case 2. IPE 200 and IPE 500

vourable result occurs for l = 1.0. In contrast to this, for h2 (buckling curve “b”) and HEB 300 (buckling curve “c”)
(dashed lines) with linear plastic interaction and j = 250, result in an almost identical unfavourable value h = 1.081

the high slenderness values l > 2.15 result in values on the > 1.0, which is on the unsafe side. The results for sections
unsafe side, with h2 ≈ 1.03 > 1.0. The attempt to avoid IPE 140 and HEA 360 (investigations by Lindner [7]) are
these small unsafe results by increasing the initial bow im- not plotted here but give comparable results.
perfection requires very small j factors (or high imperfec- It can be observed that the values relating to the inter-

tion values) up to j = 100 for l = 3.0, whereas j = 500 would action of EN 1993-1-1 (EC 3-1-1), 6.2.9.1, Eq. (10c), utili-

be sufficient for l = 1.5 and linear plastic interaction. This zation rate η1 (solid lines), exceed the border of 1.0, which
also indicates that the modification of the initial bow im- leads to the question of how far the results may be changed
perfection is not the way to solve the exceedance in the through the choice of another type of interaction. Fig. 26
case of bending and compression. shows results for an IPE 140 (buckling curve “b”) depend-

The following diagrams summarize the investigations ing on the non-dimensional slenderness l for different
for buckling direction z-z for sections in steel grade S235. ­values of the proportional axial force nu. For interaction
Fig. 24 shows that for an IPE 200 section and buckling due to EN 1993-1-1 (EC3-1-1), section 6.2.9.1, Eq. (10c),
direction z-z (buckling curve “b”), different load cases 1, 2, the maximum value is max h1 = 1.111, but is reduced to

Steel Construction 9 (2016), No. 4 359


J. Lindner/U. Kuhlmann/A. Just · Verification of flexural buckling according to Eurocode 3 part 1-1 using bow imperfections

–– IPE 140, buckling direction y-y, buckling curve “a0”


–– IPE 140, buckling direction z-z, buckling curve “a”
–– HEA 360, buckling direction y-y, buckling curve “a”
–– HEA 360, buckling direction z-z, buckling curve “b”

Some of the results are shown in Fig. 25 and Fig. 26.


Fig. 27 shows the results for the non-dimensional slen-

derness l = 1.0 for IPE 140 and HEA 360 sections, both
buckling directions y-y and z-z, and for the interaction
from EC3-1-1 (solid lines) and linear plastic interaction
(dashed lines). It shows that for buckling direction y-y, the
Fig. 25. Utilization rate hul for different sections, S235, de- most unfavourable result is h = 1.103, whereas for buckling
pending on different values of axial compression N (plotted direction z-z, the maximum value is η = 1.156, both values
– being remarkably higher than the results for S235 in sec-
as nu), non-dimensional slenderness l , = 1.0 and bending
moment Mz, load case 2 (numerical data not mentioned in tion 7.3.1. In order to avoid these excessive values, a solu-
detail), buckling direction z-z, Ofner [15], [1] tion seems to be the linear plastic interaction given by the
dashed lines. However, these sufficient results are given for

a non-dimensional slenderness l = 1.0. Whether other slen-

derness values, e.g. high ones such as l = 3.0, show a less
favourable effect is investigated in the following, with the
results given in Fig. 28.
Fig. 28 shows the results for IPE 140, buckling direc-
tion z-z, buckling curve “a” and load case 2 (distributed

load) depending on different slenderness values l = 0.5,
1.0, 1.5 and 3.0 for the relative axial forces nu ~ 0.25, ~ 0.4,
~ 0.53 and ~ 0.65. Furthermore, three different interaction
equations are taken into account:
–– Interaction according to EC3-1-1, section 6.2.9.1, see
Eq. (10c) (solid lines)
Fig. 26. The hl values for IPE 140, S235, different values of
–– Linear plastic interaction, see Eqs. (1) and (10a) (dashed
axial compression N given as nu, depending on non-dimen-
– lines)
sional slenderness l and bending moment Mz, load case 2
(numerical data mu not mentioned in detail), buckling direc- –– Interaction according to DIN 18800-1 [4] with a re-
tion z-z, Lindner [7] duced shape factor apl = 1.25, the interaction according
to EN 1993-1-1 gives similar results (chain-dot lines)

h2 = 1.094 > 1.0 for linear plastic interaction (dashed lines). The most unfavourable result occurs for l = 1.0, nu =
If similar to the restriction in DIN 18800 [4], the plastic 0.67 (~ 0.65), with h = 1.156 > 1.0 on the unsafe side, but
shape factor is limited to apl,z = 1.25, but for the interaction the results for other nu values are similar. This finding for
according to EN 1993-1-1 (see chain-dot lines), the most nu is similar to S235 in Fig. 25. It is important to recognize
unfavourable value max h3 = 0.953 < 1.0 is sufficient. that the linear plastic interaction gives significantly smaller

All results for second-order theory were calculated values within the range l = 0.5 to 1.0. Unfortunately, this

with the initial bow imperfections, i.e. j-values, contained does not apply within the range l = 1.5 to 3.0, where the
in EC3-1-1, Table 5.1. Further investigations show that worst result is h = 1.124 and thus only marginally smaller
greater initial bow imperfections, i.e. smaller j values, only than h = 1.156. When a reduced plastic shape factor of
have a small influence on the unfavourable results, espe- 1.25 similar to DIN 18800-1 [4] is taken into account, then
cially in the range of small axial forces nu. the worst value diminishes to h = 0.990 < 1.0!
Two main results can be summarized:
–– The exceedance on the unsafe side cannot be corrected 8 Summary and conclusions
effectively by smaller j values because these values
would be extremely small. Investigations were carried out concerning the necessary
–– The exceedance on the unsafe side of around 10 % is so numerical values for initial bow imperfections e0, given as
high that special measures should be taken. the non-dimensional, representative bow imperfection j =
L/e0. The influences of different rolled I-sections, different
7.3.2 Steel grade S460 interaction equations and different steel grades were con-
sidered. Only for high axial forces are these imperfections
No detailed data from earlier investigations are available important. For steel grades S235 and higher, the constant
for steel grade S460. Therefore, FEM calculations were values in Table 5.1 of EC3, Part 1-1, may be retained for
carried out [14]. It was shown earlier in this paper, see Fig. rolled I- and H-sections, partly based on the results of re-
11 to Fig. 13, that within one group of sections, the results fined FEM calculations. Exceptions were identified for
are very close together. Therefore, two typical sections sections in buckling direction z-z and for steel grade S460.
were chosen for this calculation and subjected to y-y and Special attention was paid to loading by compression
z-z buckling: and bending moments assuming the j values as bow imper-

360 Steel Construction 9 (2016), No. 4


J. Lindner/U. Kuhlmann/A. Just · Verification of flexural buckling according to Eurocode 3 part 1-1 using bow imperfections

Fig. 27. Utilization rate hul for IPE 140 and HEA 360, S460, Fig. 28. Utilization rate hul for IPE 140, S460, depending on

different values of axial compression N given as nu, non-di- the non-dimensional slenderness l , different values of axial

mensional slenderness l = 1.0 and bending moment Mz, compression N given as nu and bending moment Mz, load
load case 2 (numerical data mu not mentioned in detail), case 2 (numerical data mu not mentioned in detail), buck-
buckling directions y-y and z-z, Lindner [7] ling direction z-z, Lindner [7]

Table 3. Options for the design of members loaded in compression and bending

Buckling Option 1 – ULRF Option 2 – type of interaction


Buckling curve ­perpendicular S235, S275, S355, S235, S275, S355,
to axis for I-sections S460 S460
S420 S420
1 2 3 4 5 6
a0 y-y - 1.10 - linear plastic
y-y 1.08 1.05 linear plastic linear plastic
a
z-z - 1.15 - any, but apl,z = 1.25
y-y 1.05 - linear plastic -
b
z-z 1.10 1.15 any, but apl,z = 1.25 any, but apl,z = 1.25
c z-z 1.10 - any, but apl,z = 1.25 -

fections given so far in Table 5.1 of EC3-1-1. In nearly all discrepancies occur for buckling direction z-z and grade
load cases the results for a combined loading of axial force S460. These discrepancies are higher than for mild steels
plus bending moment are less favourable than for mem- (S235 etc.), see Table 3. This may be explained by the large
bers loaded by axial force only and, furthermore, lie on the plasticized zones when for buckling direction z-z the plas-
unsafe side. The discrepancies can be observed for both tic shape factor apl,z is high and there is a large difference
buckling directions, different buckling curves, various slen- between the deformations of plastic hinge theory, on which
derness ranges and different proportions of axial forces. the interactions are based, and the deformations of FEM
The highest h values occur for buckling direction z-z and calculations considering plastic zones. In these cases lim-
steel grade S460. iting the plastic shape factor to, maybe, apl ≤ 1.25, similarly
The following choices can be offered: to DIN 18800-1 [4], can be of help, see Fig. 26 and Fig. 28.
Then any kind of plastic interaction may be used.
Option 1: Utilization rates hul > 1 are accounted for by This method is not efficient for buckling direction y-y,
dividing the right-hand side of the interaction equation by as the shape factors are usually in the range apl,y = 1.14 to
a factor called ULRF > 1.0 (ultimate limit reduction factor) 1.18. Here, the application of a linear plastic interaction
which covers the maximum discrepancies on the safe side. instead of the use of the interaction equation in EN 1993-
For linear interaction, Eq. (1) changes to Eq. (1*): 1-1, section 6.2.9.1, may solve the discrepancies, see Fig. 22
to Fig. 25 and Fig. 27.
N M 1 As a conservative solution (Option 3) it is always suf-
+
(1*) ≤
Npl Mpl ULRF ficient to use a linear elastic interaction equation.
These two options are shown in detail in Table 3. For
The same effect is reached by dividing the interaction fac- reasons of economy, a distinction should be made between
tor y by ULRF, leading to y* = y/ULRF. buckling directions y-y and z-z and two groups of steel
Table 3 gives values for ULRF derived from the inves- grades: S235 (up to S420) und S460. This table may be
tigations shown in section 7. Using the given values, small shortened by not differentiating between buckling direc-
deviations up to 3 % in special cases nevertheless occur tions y-y and z-z and/or by not differentiating between
and are not accounted for. steel grades. But in these cases the results are less eco-
nomic.
Option 2: Utilization rates hul > 1 are accounted for by These results may be used for all rolled I- and H-sec-
using a variation of the type of interaction. The greatest tions. In other cases, for all welded profiles with steel

Steel Construction 9 (2016), No. 4 361


J. Lindner/U. Kuhlmann/A. Just · Verification of flexural buckling according to Eurocode 3 part 1-1 using bow imperfections

grades higher than S235, the j values given in Table 5.1 of [13] Lindner, J.: Vergleiche verschiedener Nachweise für Druck
Eurocode 3, Part 1-1, have to be multiplied by ε, with ε = und einachsige Biegung (Comparison of different design
0.924 (S275), 0.814 (S355) and 0.748 (S420). checks for beams subjected to compression and uniaxial bend-
Additional investigations for welded sections would ing). Report 2059A, Fachgebiet Stahlbau, TU Berlin, 1984,
documented in [1] as ref. [45].
allow the approach to be confirmed and improved.
[14] Lindner, J.: LIDUR. EDV-Programm zur Berechnung der
Traglasten von beliebig gelagerten geraden Stabsystemen.
Keywords: stability; initial bow imperfection; flexural buckling;
(computer program for ultimate load calculations concerning
Eurocode 3; second-order theory
straight systems with any boundary condition). TU Berlin,
Fachgebiet Stahlbau, ver. 07/14 “LIDALL4” (unpublished).
References
[15] Ofner, R.: Traglast von Stäben aus Stahl bei Druck und
Biegung. (Ultimate load of steel members subjected to axial
[1] Boissonnade, N.; Greiner, R.; Jaspart, J. P.; Lindner, J.: ECCS
compression and bending). Diss., Institut für Stahlbau, Holz-
TC 8, No. 119: Rules for member stability in EN 1993-1-1,
bau & Flächentragwerke der TU Graz, No. 9, 1997, docu-
Background documentation and design guidelines. ECCS
mented as ref. [60] on a CD in [1].
pub., Brussels, 2006.
[16] Roik, K.; Wagenknecht, G.: Traglastdiagramme zur Bemes-
[2] Bureau, A., et al: N+My interaction for cross section resis­
sung von Druckstäben mit doppeltsymmetrischem Quer-
tance of doubly symmetric I-profiles. Working doc. No. [24a]
schnitt aus Baustahl (ultimate load diagrams for design of
Working Group 1 of CEN/TC 250/SC 3, Zurich, 14 Oct 2013.
compression members of doubly symmetrical steel cross sec-
[3] CEN/TC 250/SC 3, Doc. N2164: Classification of rolled
tion and uniaxial bending). Konstruktiver Ingenieurbau Be­
I-Profiles fabricated in steel grade S460 within Table 6.2 of EN
richte No. 27, Vulkan-Verlag, Essen, 1977.
1993-1-1, Decision 19/2015 on Amendment AM-1-1-2015-
[17] Sedlacek, G.: Consistency of the equivalent geometric im-
001, 2015.
perfection used in design and in the tolerances for geometric
[4] DIN 18800:2008-11: Stahlbauten, – Teil 1 Bemessung und
imperfection used in execution. Doc. ECCS-TC8-2010-06-001,
Konstruktion, – Teil 2: Stabilitätsfälle – Knicken von Stäben
Feb 2010.
und Stabwerken (Steel structures, – Part 1 Design and Con-
[18] Stroetmann, R.; Faßl, T.: Influence of the interaction formu-
struction, – Part 2: Stability – Buckling of bars and skeletal
las for the plastic cross section resistance to the assumption
structures.
of bow imperfection for flexural buckling. Working doc.
[5] EN 1993-1-1:2009, Eurocode 3: Design of steel structures –
N0019, Working Group 1-1 of CEN/TC 250/SC 3, Oct. 2014.
Part 1-1: General rules and rules for buildings.
[19] Taras, A.; Kuhlmann, U.; Just, A.: Design of Compression
[6] Kuhlmann, U.; Feldmann, M.; Lindner, J.; Müller, C.; Stroet-
Members by 2nd Order Analysis – Imperfection Amplitudes,
mann, R., (assisted by Just, A.): Eurocode 3: Bemessung und
Material Dependency, Influence of gM1. Doc. ECCS – TC8,
Konstruktion von Stahlbauten, Band 1: Allgemeine Regeln
TC8-2013-06-005, June 2013.
und Hochbau. DIN EN 1993-1-1 mit Nationalem Anhang.
[20] Wagenknecht, G.: Imperfektionsannahmen im Stahlbau für
Kommentar und Beispiele. Pub.: bauforumstahl, Beuth Ver-
Biegeknicken (Assumptions for imperfections in steel struc-
lag/Ernst & Sohn, Berlin, 2014.
tures for flexural buckling). Berichte FH Gießen-Friedberg,
[7] Lindner, J.: Ultimate load calculations concerning compres-
No. 1, 2014, Working doc. DIN NA 005-08-16 AA N 1108.
sion and uniaxial bending, IPE 140 and HEA 360 of steel
grade S235 and S460, Background document, Berlin, 2016
(unpublished). Authors
[8] Lindner, J.; Just, A.; Kuhlmann, U.: Bow imperfections for Univ.-Prof. Dr.-Ing. Joachim Lindner
flexural buckling according to Eurocode 3 part 1-1. Proc. of j.lindner-berlin@t-online.de
SDDS 2016: Intl. Coll. on Stability & Ductility of Steel Struc- Consulting engineer
tures, Timisoara, 2016, pp. 525–532. Formerly: Technische Universität Berlin, Chair of steel structures
[9] Lindner, J.: Classification of rolled I-profiles fabricated in Wallot-Str. 3
steel grade S460 within table 6.2 of EN 1993-1-1 – rev. 3, Mar 14193 Berlin
2015, Working doc. N0037, Working Group 1-1 of CEN/ Germany
TC 250/SC 3.
[10] Lindner, J.; Scheer, J.; Schmidt, H.: Stahlbauten, Erläuter- Prof. Dr.-Ing. Ulrike Kuhlmann
ungen zu DIN 18800 Teil 1 bis Teil 4 (Steel structures, Com- sekretariat@ke.uni-stuttgart.de
mentary to DIN 18800 Part 1 to Part 4). Beuth Verlag/Ernst
& Sohn, 3rd ed., Berlin, 1998. Dipl.-Ing. Adrian Just
[11] Lindner, J.; Lagae, G.: Minutes of TC 8 meeting in Luxem- adrian.just@ke.uni-stuttgart.de
bourg, 22 May 1998, paper TC8-1998-001. University of Stuttgart
[12] Lindner, J.: Equivalent initial bow imperfections for hollow Institute of Structural Design
sections. Tubular Structures VI, Proc. of 6th Intl. Symp. on Pfaffenwaldring 7
Tubular Structures, Melbourne, Australia, 1994, Balkema Rot- 70569 Stuttgart
terdam, pp. 159–162. Germany

362 Steel Construction 9 (2016), No. 4

You might also like