WEB BUCKLING OF HIGH STRENGTH STEEL PLATE GIRDERS

INDUCED BY BENDING CURVATURE

Sérgio Nascimentoa, José Oliveira Pedroa,b e André Biscayaa

a
CERIS, Instituto Superior Técnico, Universidade de Lisboa, Portugal
b
GRID INTERNATIONAL, Consulting Engineers, Lisboa, Portugal

Resume. Simplified formula of the web slenderness limit given in EN 1993-1-5 to prevent the
flange-induced buckling does not usually govern the web design other than very slender Class
4 plate girder webs in steel S235 to S355. However, the same formula applied to plate girders
designed in HSS S690 gives lower slenderness limits likely to govern web’s design and re-
strains the possibilities of reducing their thickness. This paper reviews the background of the
simplified formula currently used and compares the web slenderness limits obtained with nu-
merical results from full nonlinear analysis of slender high strength steel girders.

1. Introduction
Standard hot-rolled I and H sections are frequently adopted in steel structures of buildings and
single-storey industrial halls. However, for long span steel frames and steel and steel-concrete
composite bridge decks with medium to long spans fabricated plate girders are usually adopt-
ed to fulfil the ultimate limit state requirements, such as strength and stability, or the defor-
mation and vibration serviceability limit states. Plate girders can be optimized to a high de-
gree by taking into account the distribution of bending moments and shear forces. Current au-
tomation of plate girders production much reduces labour costs and therefore material costs
become decisive. Moreover, the life cycle costs and environmental impact of structures are
nowadays also important issues, emphasising the need for material optimized designs.
The use of high strength steel (HSS) S690 is being proposed in bridge deck applications to
reduce steel quantity of plate girder, designed for bending and shear. According to recent in-
vestigations, for highway plate girders decks, the use of HSS S690 QL enables a reduction of
25-30% of the steel weight compared to designs using standard in S355 NL [1, 2].
For the optimal design of plate girders, the steel material should be moved as much as pos-
sible to the extreme fibers. By adopting slender webs – defined by the web height to thickness
ratio hw/tw – the level arm of the internal forces is maximised and thus the bending carrying
capacity. However, web plate buckling and ultimate resistance has always been a limitation to
this trend. So, considerable effort has been put to establish plate girder design resistances for
bending, shear, direct loads and combined stresses [3, 4, 5].

© 2019 Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin · ce/papers 3 (2019), Nos. 5 & 6
https://doi.org/10.1002/cepa.1191 wileyonlinelibrary.com/journal/cepa 160
161 |
XII Congresso de Construção Metálica e Mista Coimbra, Portugal

A plate girder under bending exhibits also flange-induced forces on web, due to the overall
bending curvature of the girder. It is normally assumed that the web provides a rigid linear
support to the compression flange against buckling in the plane of the web. However, when
the plate girder has large flanges and a very slender web, it was reported the web buckling
under pure bending, inducing also the compressed flange buckling in the plane of the web. To
prevent this event, a simple web slender upper limit ratio hw/tw is given in EN 3-1-5 [6], which
is primarily intended to be use for plate girders without stiffeners. The same formula is being
used during the last 50 years without much attention from designers, mainly because it usual-
ly do not govern the design of plate girders produced in standard steel grades S235 to S355.
However, when designing HSS plate girders, the trend to use slender webs is restricted by
this flange-induced slender formula. Therefore, the assumptions used to attain the simple web
slender limit should be reassessed and a comparison web slender limit obtained with numeri-
cal results from full nonlinear analysis of slender high strength steel girders is important.

2. Current design formula from EN 3-1-5

The web slender limit to prevent flange-induced buckling effect was initially proposed by
Basler [7]. This formula evolved into Eq. (1), given in clause 8(2) of EN 3-1-5 [6], that corre-
sponds to a simple web slender (i.e. web height to thickness ratio) upper limit for plate girders
curved in elevation, with the compression flange on the concave side, as shown in Fig. 1 a).
 w
 �
 yf f
≤ (1)
 ℎw 
�1 + 3 0 yf

f ∙σ
f σy
Ff = A
r0 , r

εf
σf σr σr
Af ∙ σf ∙ d
φ
Af ∙
σz εf ∙ dx
hw
2 dφ= hw/2
x dx ε y+ε r
hw σf εr
2 Af ∙
σz Ff =
εf

a) b)
Fig. 1: Basler simple column-like buckling mode for establishing the maximum web slenderness
(adapted from [8])

Eq. (1) assumes that the curved flange induces transverse compression in the web, which is
always likely to occur if the girder has both flanges with a constant radius in elevation. How-
ever, for a one side curved girder in elevation, if the curved flange induces transverse tension
in the web, it cannot cause buckling. It is also noted that for girders with only one flange
curved, Eq. (1) provides a conservative result, as the higher compressive force is then applied
to one edge of the web only, and the resulting critical buckling load is then higher.
 | 162

Steel and Composite Steel-Concrete Bridges

The slender web limit of EN 3-1-5 matches with Basler proposal for elastic section analy-
sis (k=0.55) of plate girders with constant height, and comprises the same assumptions. When
plastic section or a plastic global analyses are preformed k=0.40 or k=0.30 should be adopted.
Eq. (1) can be obtained adopting Basler model and assumptions.The curvature 1/r of the
curved plate girder, given by Eq. (2), comprises a planned initial curvature 1/r0, together with
a further curvature from deflection under load (limited to the yield strain of the flange divided
by half of the web depth) and an extra curvature from residual stresses σr = fyf / 2 (Fig. 1 b)).
Assuming that at the ULS the flange has installed the yielding force  =   , then the web
compressive stress σz is given at this stage by Eq. (3).
1⁄ =  (1 + 0.5)⁄( ∙  ⁄2) + 1⁄ = 3  ⁄( ∙  ) + 1⁄ (2)
z =  ⁄( ∙  ) =  ∙  ∙ 3  ⁄( ∙  ) + 1⁄( ∙  ) (3)
Modelling a one meter vertical strip of a web panel without longitudinal and transverse
stiffeners (and ignoring the effect of the longitudinal bending stresses in the web), as a perfect
simple supported column loaded with the compressive stress σz, the column-like buckling
mode should be the failure mode, and the Euler critical stress the stress limit, as in Eq. (4a).
Solving Eq. (4a) with restect hw/tw, the slender limit given by Eq. (4b) is obtained.
   (4a)
z ≤ cr →  ∙  3  ⁄( ∙  ) + 1⁄( ∙  ) ≤  
12(1 −   ) 
2  w  w
� � 0.55 �
 36(1 2 ) yf f yf f
∴ ≤ = (4b)
 ℎw ∙ ℎw ∙
�1 + 3  �1 + 3 
0 ∙yf 0 ∙yf

However, Basler and Thürlimann mentioned that the method proposed for attaining the
web slenderness limit was “somewhat crude” [8]. If fact, five main simplified assumptions
can be identified:
1) A symmetric plate girder is assumed loaded to pure bending, with the neutral axis located
at middle height of the cross-section;
2) For this symmetric plate girder loaded in bending, both flanges attain the yield strength
 , and generate the same compressive force on the web;
3) The web is supposed to be a perfect column hinged at both flange levels, with no longitu-
dinal stresses and no initial geometrical imperfections or residual stresses;
4) Residual stresses are considered at the flanges, with a peak of 0.5 in the region adja-
cent to the web-to-flange welded joint;
5) Transverse and longitudinal stiffeners are negleted.

Several comments can be made to the assumptions from the Basler model:
1) The curvature 1/ is based on the strain in the flange and the neutral axis placed at middle
height of a symmetric girder; steel plate girders can be symmetrical, but steel-concrete
composite plate sections are unlikely to be; still, for a simply supported girder with a
class 4 symmetric cross-section, the effective cross-section is asymmetric, and so the po-
sition of the neutral axis shifts downwards to the bottom tension flange; the curvature
based on the compressive flange and the ensuing web compressive stress σz will therefore
decrease, and the web slenderness limit may slightly increase;
2) When using HSS, the stability and fatigue issues often govern the design [2]; in such case
ULS combinations create stresses lower than the high steel strength; So, the compressive
deviation force applied to the web can be reduced from the maximum assumed value
163 |
XII Congresso de Construção Metálica e Mista Coimbra, Portugal

 =   , and the slender limit imposed by flange-induced buckling on the web will
therefore be increased;
3) Basler model assumed that the web is simply supported by the flanges and so the column
buckling length of the web is equal to the web depth w; however, the web edges are
welded to the flanges, and a top slab may exist, and so some rotational stiffness caused by
the flanges may exist. In that case, the buckling length is smaller than the web depth, in-
creasing the web slenderness limit;
4) Basler model neglects the longitudinal web stresses x due to bending, much higher than
the vertical stresses z . For a simply supported symmetric steel girder, Basler justified
this by assuming that the negative influence of the compressive stresses would be bal-
anced by the positive influence of the tensile stresses in the lower part of the web near the
tension flange. In fact, the web of a composite asymmetric girder is often almost all in
tension in the span regions, which avoids web buckling. To consider this effect the simple
column buckling checking of the web must be replaced by the plate bucking checking,
taking into account the actual stress state, and using the Reduced Stress Method proposed
in EN 3-1-5 [10, 11]. It is also interesting to notice that the UK National Annex to EN
1993-1-5, substituted Eq. (1) slender limit by a direct stress verification of the web,
namely for plate girders curved in elevation [12];
5) Basler model also neglects the geometric imperfections and residual stresses of the web;
even if critical loads of very slender columns are not much reduced by imperfections,
they tend to reduce the slender limit imposed by flange-induced buckling; moreover,
Basler model considers the flange residual stresses as being 0.5 for obtaining an addi-
tional curvature of the cross-section due to flange strains beyond first yield, and it is
known that for higher steel grades residual stresses are smaller in percentage of the yield
strength  [13];
6) Finally, Basler buckling column model clearly makes no allowance for transverse vertical
stiffeners on the webs; according to [14] these stiffeners will be of limited benefit on a
continuously curved flange unless closely spaced, so they can usually be ignored without
undue conservatism, but the possible benefits of transverse stiffeners on the flange-
induced buckling should be assessed.

In 2015, Abspoel modelled numerically the two G4 tests of Basler [9]. He also conduced
several experimental tests of I-shape plate girders with very slender webs. In all tests, vertical
buckling of the compressive flange occurred after reaching the peak of the P−δz diagram, and
only when the compression flange yields [9, 15]. Therefore, according with Abspoel, flange-
induced buckling is not the primary collapse mode and can be identified as a post-critical
phenomenon.
Even with all simplified assumptions, design practice shows that it is unlikely flange-
induced buckling governs the web design other than in very slender Class 4 plate girder webs
in S235 to S355. However, if applying the same flange-induced buckling formula to girders in
HSS S690, rather low slender limits are obtained for the webs. This effect can govern the
web’s design, limiting the prospect to reduce its thickness to profit from the HSS strength,
and thus justifying further investigation.
 | 164

Steel and Composite Steel-Concrete Bridges

3. Numerical analysis of slender plate girders in HSS

3.1 Calibration of the numerical model with Basler G4-T2 experimental test

For calibrating the numerical model, Basler’ G4-T2 experimental test was replicated numeri-
cally by means of general-purpose software ABAQUS [16]. Based on the setup, geometry,
material properties and procedures described in [8, 17, 18], a material and geometric nonline-
ar analysis is performed. Web’s initial imperfections measured by Basler were considered as
the web initial shape. After the first loading with two concentrated forced  at distance
2 =3810 mm from the supports, a 25 mm residual deflection was reported [17]. S4R ele-
ments type meshed with 50x50 mm2 and the arc-length method of convergence were adopted.
Boundary conditions are simulated to accurately meet with experimental test setup.
Table 1 compares the maximum applied load, the maximum mid-span flange displacement
and ultimate bending moment obtained in the numerical model, showing a very good agree-
ment with the Basler experimental test results [17]. Moreover, Fig. 2a) shows a global view of
the buckling obtained numerically, and Fig. 2b) gives a close look of the flange-induced buck-
ling similar to the experimental reported by Basler. Finally, Fig. 3 compares the complete
load-displacement diagram from Basler’ experimental test with the one obtained numerically,
showing the good concordance between the numerical and test results.

Table 1: Comparison of results between Basler’ G4-T2 experimental test and FE-model

Basler’ test FEM Difference

u [kN] 569.6 570.1 0.1%
δu [mm] 68.0 70.9 4.3%
u = 2  u [kNm] 2170.2 2172.1 0.1%
z
P
x y

a)

y= 89.5 mm

b)
Fig. 2: Basler’ G4-T2 FE-model failure mode: a) Global view (stresses in MPa); b) comparison of
failure modes at the experimental test G4-T2 and FE-model
165 |
XII Congresso de Construção Metálica e Mista Coimbra, Portugal

650
600
550
500
Loading force P [kN]

450
400
350 G4-T1 Load
300 G4-T1 Unload
250 G4-T1 Reload
200
150 G4-T1 Unload
100 G4-T2 Load
50 G4-T2 FEM
0
0 10 20 30 40 50 60 70 80 90 100
Vertical deflection δz of the bottom flange mid-span section [mm]
Fig. 3: Load-displacement diagram of the experimental test G4 and FE-model

3.2 Case study – High strength steel plate girder

To assess numerically the flange-induced buckling of a simply supported steel girder with
L=32000 mm span, the steel girder is submitted to a constant bending moment My, resulting
in constant curvature in the web panel. The compressive flange buckling into the web plane is
studied for different web thicknesses (10, 15 and 20 mm). Two cases are considered: a) the
web is unstiffened, or b) the web is transversally stiffened by plate stiffeners spaced =4000
mm between centres, as illustrated Fig. 4.
z

x y
u y=0
ous)
(c ontinu My
uy=uz=0

u x=0
a
a
a= a
0 m m
My 40 0 m
a 000 m
a 3 2
uy=uz=0 L=
a
a

Fig. 4: Case study FE-model scheme

Structural steel S690 is considered with the bilinear elastic-plastic behaviour of Fig. 5 [6,
19, 20]. The plate girder has a constant symmetrical cross-section, with equal top and bottom
flanges, as shown in Fig. 6. FE-models developed to evaluate numerically the case study con-
sider both geometric and material non-linear behaviour (GMNIA) along with a bow equiva-
lent initial imperfection of hw /200 [6] within each web panel and the residual stress patern
given in Fig. 7. Plates are meshed with 4-nodes shell elements with reduced integration
 | 166

Steel and Composite Steel-Concrete Bridges

(S4R), with mesh size approximately 150x150 mm2, optimized by a sensitivity analysis in or-
der to obtain a good balance between run time and accurate results. Girder’ boundary condi-
tions are presented in Fig. 4. An increase in bending moment is applied at both rigid trans-
verse end stiffeners, until the ultimate load and the failure mode are reached. The numerical
analysis used the arc-length convergence method.

700 700

σ [MPa] 50 50

770
690 ~E /250

tw = 10, 15 or 20 tw = 10, 15 or 20

3500
3500
E =210 GPa

εy =0.33% εu =10% ε [%] 50 a) 50 b)

700 700

Fig. 5: Stress-strain relationship adopted Fig. 6: Case study cross-sections: a) unstiffened;

b) Transversally stiffened (in mm)

=
=
w0 = hw/200 bf w0 = hw/200
hw =
=

0.5fy 0.5fy

a) b)
Fig. 7: Bow initial equivalent web imperfection and residual stresses for the case study:
a) unstiffened web; b) transversally stiffened web

Flange-induced buckling failure mode can be reported by the numerical models for the
case of a very slender web panel of tw=10 mm. Due to bending, the top flange and part of the
web adjacent to the flange start to yield as the load increases. When the web reaches yielding,
the top compressed flange loses the vertical support and buckles into the web. Fig. 8 a) dis-
plays a similar failure mode as reported by Basler’ test G4-T2 at the mid span of the unstiff-
ened girder only with a 10 mm thick web panel. Fig. 8 b) presents the same top flange buck-
ling into the web failure mode, symmetrically in the second panels from the support, for the
case of a stiffened 10 mm web girder.
167 |
XII Congresso de Construção Metálica e Mista Coimbra, Portugal

Consider that the girder cross-section is symmetrical, both top and bottom flanges were
expected to yield for the same load step, keeping the neutral axis constant and in mid-section.
However, as the web is a class 4 section under pure bending, local buckling of the web under
compression is detected in the early load stages, and causes the neutral axis to shift down-
wards. That explains why only the top flange under compression yields.

a) b)
Fig. 8: Case study FE-model failure mode for tw=10 mm web: a) unstiffened; b) transversely stiff-
ened (stresses in MPa)

Web behaviour is assessed by M-δz diagrams of Fig. 9 for the case of the unstiffened and
stiffened web, being δz the vertical displacement at the mid-span bottom flange. The effective
e
elastic bending moment el of each girder cross-section, evaluated applying EN 1993-1-5
e
rules and the safety factor γM1 =1.0, are also presented. After reaching el , the girder fails
due to the top flange and web local yielding and consequent the top flange buckling into the
web. This effect leads to a failure mode by a sudden fall of the bending resistance, when the
two flanges become closer. The same behaviour is attained for the unstiffened and stiffened
web, which shows that the transverse stiffeners spaced 4000 mm apart do not have any clear
benefit on the ultimate behaviour of the steel girder under a constant bending moment.Yet,
e
flange-induced buckling occurred always after reaching el and, therefore, for loads higher
than the design loads.

Fig. 9: M-δz diagram for the case study with an unstiffened and stiffened web panels
 | 168

Steel and Composite Steel-Concrete Bridges

4. Web thickness of HSS plate girders based on the EN 3-1-5 formula

Web thickness/slenderness of the case study analysed numerically (a typical symmetrical steel
plate girder) can be compared with the results given by EN 3-1-5 slender limit, assuming
yielding in one of the extreme fibers. Unstiffened plate girder webs are considered as well as
transversely stiffened webs at 4000 mm centres. To prevent flange-induced buckling accord-
ing to the EN 3-1-5, the web slenderness upper limit from Eq. (1) should be verified. Table 2
gives the web slenderness and the limit for each case, noticing that the web is always class 4.
For comparison purposes, the results are normalized, i.e. web thickness with a ratio over 1.0
(in bold) are prone to flange-induced buckling according to EN 3-1-5 formula.

Table 2: Web slenderness verification according to EN 3-1-5

tw fy  / Eq. (1) limit  ⁄
Case Study
[mm] [N/mm2] [-] [-] Eq. (1)
10 690 340 165 2.06
r 0= ∞
15 690 227 202 1.12
a = ∞ and 4 m
20 690 170 233 0.73

Some comments can be made based on Table 2 and FE-model results:

1) As the standard procedure does not take into account the transverse stiffeners, no differ-
ence in slenderness results is found between unstiffened and trasnversaly stiffened cases;
2) FE-model results also shows that vertical stiffeners spaced  ≃  have no relevance in
preventing flange-induced buckling failure mode and not increase the ultimate load;
3) For the web thickness’s investigated (i.e. 10, 15 and 20 mm) numerical models show no
e
evidence of web buckling up to the design bending moment el ; the local buckling of
the compressed flange into the web is reported after, and coupled with the yielding of the
compressed flange and web, for a load level about 10% over the design load;
4) For case study with tw = 10 and 15 mm, EN 1993-1-5 formula indicated the plate girders
are prone to flange-induced buckling, and a minimum tw = 17 mm is required;
5) Therefore, the standard simplified formula indicates that flange-induced buckling may
govern the design of 10 to 15 mm thick webs, which proves this formula is on the safe
side also for HSS plate girders.

5. Conclusions
From the study presented, the general conclusion can be presented about the flange-induced
buckling of HSS plate girders:
1) It was reported the web buckling of high strength steel plate girders, together with the
local buckling of the flange into the web, only for highly slender webs;
2) Results from FE-models put in evidence that flange-induced buckling is generally an is-
sue reported after the first flange yielding occurs;
3) Abspoel stated that only outstanding slender HSS S690 webs (with slenderness above
600) may present elastic web buckling before the first yielding occurs [9];
4) Thus, web slender limits given in EN 3-1-5 for class 4 sections are conservative and
should be reassessed for the design of HSS plate girders;
5) Vertical stiffeners spaced at  ≃  are ineffective to prevent flange buckling into the
web, namely when the local buckling mode is reported.
169 |
XII Congresso de Construção Metálica e Mista Coimbra, Portugal

Notation
Aw, Af Cross-sectional area of the web; Effective cross-sectional area of the flange
a Transverse stiffeners spacing
bf Flange width
E Young modulus
Ff Force in the flange
fy Yield stress (f for the flange, w for the web)
hw Depth of the web
k Factor which reduces the web slenderness with increasing expected strain in the
flanges (0.55 for elastic section analysis, 0.40 for plastic section analysis, 0.30
for plastic global analysis with hinge formation)
M Bending moment (y in y direction)
ele Effective elastic bending moment
P, Pu Load, Ultimate Load
r, r0 Radius or initial radius of curvature in elevation of the concave flange
t Thickness (f for the flange, w for the web)
u Degree of freedom (x in x direction, y in y direction, z in z direction)
δ Vertical displacement (z in z direction)
ε Strain (r residual, f flange, r residual, y yield, uultimate)
σ Stress (cr critical, f flange, r residual, x bending, y yield, z vertical)
υ Poisson coefficient
φ Angle variaton in one strip
ω0 Out-of-web initial imperfection
ω Out-of-web plane displacement (y in y direction)

References
[1] Reis, A., Oliveira Pedro, J., Baptista, C, Virtuoso, F, Vieira, C., Optimal use of High
Strength Steel grades within bridge. OPTIBRI. Report on Design C. European Commis-
sion. Research Programme of the Research Fund for Coal and Steel, Grand agreement
n° RFSR-CT-2014-00026. Lisbon, Portugal, 2018.
[2] J. Pedro; A. Reis; C. Baptista, High strength steel (HSS) S690 in highway bridges: Gen-
eral guidelines for design. Stahlbau vol. 87, Issue 6, pp. 555-564, 2018.
[3] Sinur, F., Behaviour of longitudinally stiffened plated girders subjected to bending–
shear interaction. Ph.D. Thesis, Faculty of Civil and Geodetic Engineering, University
of Ljubljana, 2011.
[4] Sinur, F., Zizza, A., Kuhlmann, U., Beg, D., Buckling interaction of slender plates–
Experimental and numerical investigations. Thin-Walled Structures, Volume 61, De-
cember 2012, pp. 121-131, 2012.
[5] Hasan, Q. et al., The state of the art of steel and steel-composite plate girder bridges.
Part I: Straight plate girders. Thin-Walled Structures, Volume 119, October 2017, pp.
988-1020, 2017.
[6] EN 1993-1-5, Eurocode 3: Design of steel structures – Part 1-5: Plated structural ele-
ments, CEN, Brussels, 2006.
 | 170

Steel and Composite Steel-Concrete Bridges

[7] Basler, K., Strength of plate girders. PhD Dissertation, Bethlehem, USA, 1959.
[8] Basler, K., Thürlimann, B., “Strength of plate girders in bending.” Fritz Engineering
Laboratory Reports No. 251-19, Lehigh University, USA, 1961.
[9] Abspoel R., Optimisation of plate girders. PhD Dissertation, University of Delft, Neth-
erlands, 2015.
[10] Zizza, A., Buckling Behaviour of Unstiffened and Stiffened Steel Plates under Multiax-
ial Stress States. Ph.D. Dissertation, Universität Stuttgart, Germany, 2016.
[11] Kuhlmann, U., Zizza, A., Pourostad, V., Einfluss der Zugspannungen auf das Beul-
verhalten Eines Beulfeldes Unter Mehraxialer Beanspruchung – Nachweisnach DIN EN
1993-1-5, Abschnitt 10. Stahlbau, 86(10), 932-944, 2017.
[12] NA+A1 to BS-EN 1993-1-5:2006, UK National Annex to Eurocode 3: Design of steel
structures – Part 1-5: Plated structural elements. BSI, London, 2016.
[13] Wang, Y., Li, G., & Chen, S., The assessment of residual stresses in welded high
strength steel box sections. Journal of Constructional Steel Research, 76, 93-99, 2012.
[14] Hendy, C., Murphy, C., Designers’ Guide to EN1993-2–Eurocode 3, Design of steel
structures, Part 2, Steel bridges. Published by Thomas Telford Publishing, 2007.
[15] Abspoel, R., Bijlaard, F., Optimization of plate girders. Steel construction, nº7(2),
pp. 116-125, 2014.
[16] Abaqus 2018 Documentation. Dassault Systèmes Simulia Corp, 2018.
[17] Basler, K., Yen, B., Mueller, J., Thürlimann, B., Web buckling test on welded plate
girders. Part 2: Test on plate girders subjected to bending, Fritz Engineering Laboratory
Reports No. 251-12, Lehigh University, USA, 1960.
[18] Basler, K., Yen, B., Mueller, J., Thürlimann, B., Web buckling test on welded plate
girders. Overall introduction and Part 1: The test girders”. Fritz Engineering Laboratory
Reports No. 251-11, Lehigh University, USA, 1960.
[19] Wang, J., Behaviour and design of high strength steel structures, PhD Dissertation, Im-
perial College, London, 2016.
[20] EN 1993-1-12, Eurocode 3: Design of steel structures – Part 1-12: Additional rules for
the extension of EN 1993 up to steel grades S700, CEN, Brussels, 2017.