Construction and Building Materials 54 (2014) 512–519

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

Experimental investigation of residual stresses in steel cellular

and castellated members
Delphine Sonck ⇑, Rudy Van Impe, Jan Belis
Department of Structural Engineering – LMO, Ghent University, Technologiepark-Zwijnaarde 904, 9052 Ghent, Belgium

h i g h l i g h t s

 Cellular or castellated members are fabricated from hot-rolled parent sections.

 Residual stresses lower the global buckling resistance of the parent sections.
 The production process influence was determined by measuring the residual stress.
 During the production process, the residual stresses become more detrimental.

a r t i c l e i n f o a b s t r a c t

Article history: The global buckling resistance of hot-rolled steel I-section members is adversely influenced by the pres-
Received 26 August 2013 ence of residual stresses. It is expected that thermal effects during the production of cellular and castel-
Received in revised form 29 November 2013 lated members will influence the already present stresses in the hot-rolled parent sections, but it is yet
Accepted 16 December 2013
unknown to what extent. In this paper, the experimental investigation of the residual stresses in these
Available online 23 January 2014
members is described, and it is shown that the production process increases the already present com-
pressive flange stresses, which will be detrimental for their buckling resistance. This effect will be even
Keywords:
more pronounced for deviating production procedures.
Residual stresses
Cellular members
Ó 2013 Elsevier Ltd. All rights reserved.
Castellated members
Hot-rolled I-members
Imperfection
Flexural buckling
Lateral–torsional buckling

1. Introduction
Cellular or castellated members are I-section members with cir-
cular or hexagonal web openings regularly spaced along the length
of the member. Usually, they are manufactured starting from steel
hot-rolled I-section members of which the web is thermally cut
according to a certain pattern, after which the obtained halves
are shifted and rewelded (Fig. 1). Variations in the cutting pattern
and the properties of the top and bottom halves of the beam result
in a large variety of final member geometries, in which tapered
beams, different web opening shapes or asymmetric beams are
possible.
The main advantage of cellular or castellated members com-
pared to normal I-section members is their increased strong-axis
bending resistance. Furthermore, service ducts can be guided
⇑ Corresponding author. Tel.: +32 9 2645469.
E-mail addresses: Delphine.Sonck@UGent.be (D. Sonck), Rudy.VanImpe@
UGent.be (R. Van Impe), Jan.Belis@UGent.be (J. Belis).
through the web openings instead of under the member, thus
diminishing the necessary total story height. Additionally, they
can also be used out of aesthetic considerations because of their
lighter appearance. The members are mainly used for applications
in which they are loaded in bending, both in steel and steel–
concrete construction. However, they are also increasingly used
for columns and beam-columns, where the column is subjected
to both bending and compression.
Clearly the presence of the web openings will influence the fail-
ure behaviour of cellular or castellated members [1]. For instance,
new localized failure modes can arise, such as Vierendeel bending
around the web openings or local buckling of the web surrounding
the openings. Specific rules for these failure modes are currently
being developed and optimized [2–8]. However, the already exist-
ing failure modes for I-section members will also be modified,
because of the change in geometry and the influence of the produc-
tion process on imperfections such as geometric imperfections and
residual stresses. Examples of this are the global instability failure
modes for members loaded in compression and/or bending, such as

0950-0618/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.conbuildmat.2013.12.045
 D. Sonck et al. / Construction and Building Materials 54 (2014) 512–519
Fig. 1. Standard fabrication method of castellated and cellular members starting from a plain-webbed parent section.
flexural buckling or lateral–torsional buckling. Unfortunately, the
existing research for these global instability modes is incomplete,
conflicting or lacking altogether [9–11]. An important aspect of
the global instability behaviour of cellular and castellated mem-
bers that has not been investigated yet is the influence of residual
stresses.
Residual stresses are internal stresses which can exist in a
member not subjected to external loads. Hence, residual stresses
are always in static equilibrium. In hot-rolled I-section members,
thermal residual stresses originate during the cooling after the
hot-rolling process because of the non-uniform cooling and corre-
sponding thermal contraction, coupled with a temperature-depen-
dent yield stress. Typically, the flange tips will be in compression
and the flange centres in tension, while the sign and variation of
the residual stresses in the web varies depending on the cooling
parameters and cross-section geometry [12–17]. Examples of typ-
ical thermal residual stress shapes are shown in Fig. 2 [17,18]. If
the straightness of the beams after cooling is not sufficient, the
beams will undergo a cold-straightening process which redistrib-
utes the residual stresses in the flanges [19,20]. As a result, the pos-
sible residual stress patterns in I-section members are very diverse.
The influence of residual stresses on the global buckling failure
of I-section members without web openings has been studied
extensively for the cases of flexural buckling [14,15] and lateral–
torsional buckling [21], which are the most important global fail-
ure modes for I-section members. It was shown that the residual
stresses have a detrimental effect on the global buckling resistance
of the I-section members, caused by the compressive residual
stresses at the flange tips. Due to these stresses, the onset of yield-
ing will occur quicker at the flange tips, increasing the members’
flexibility and reducing the corresponding buckling resistance. Fur-
thermore, it was shown that the redistribution of the flange stres-
ses due to cold-straightening is always advantageous for the failure
load [16,19,20,22]. Therefore, it was decided not to take this into
account in the proposed standard residual stress patterns and to
consider only the more detrimental thermal residual stress pattern
in finite element calculations to determine the buckling resistance.
Likewise, only non-straightened members were used in the flex-
ural buckling experiments to determine the buckling resistance
of columns [23].
During the cutting and welding processes in the production of
cellular or castellated members, heat is introduced at the cutting
(a)
Fig. 2. Residual stress pattern in hot-rolled I-section members: (a) according to ECCS [18]; (b) according to Young [17].
513
and welding locations, which will influence the already present
residual stresses. At the locally heated zone, very high tensile
residual stresses (up to the material’s yield stress) will be present,
balanced by compressive (and possibly also tensile) stresses far-
ther away from the location of heat introduction [15,19,23–28].
Typical schematized residual stress distributions for plates cut at
one side and plates welded in their middle are depicted in Fig. 3.
As a result, it is very well possible that the production process in-
creases the magnitude of the compressive residual stresses in the
flanges, hence decreasing the global buckling failure load. Never-
theless, the change in residual stresses of cellular and castellated
members during their production process has not been investi-
gated yet and it is currently not known how significant the possible
detrimental effect on the members’ stability will be.
In this paper, the effect of the production process of cellular and
castellated beams on the residual stresses is investigated by mea-
suring the residual stresses in a series of castellated and cellular
members. Measurements were done both for the parent sections
and the resulting castellated and cellular members. By comparing
the measured stresses before, during, and after production, the ef-
fect of the production procedure could be studied. In this paper,
tensile stresses will be taken as positive stresses and compressive
stresses as negative.
2. Residual stress measurement method and specimens
In this section, an overview will be given of the used measure-
ment method and the specimens in which the residual stresses
were measured. All measurements were done at the Laboratory
for Research on Structural Models (LMO, Ghent University).
2.1. Measurement method
The residual stresses were measured using the sectioning meth-
od, which is a destructive relaxation method. The longitudinal
residual stresses in each segment can be calculated using the strain
difference before and after the relaxation of each segment. The
relaxation of each segment is accomplished by the sectioning of
the member. First, the member is cut transversally at the location
of measurement, after which a series of longitudinal cuts is made
to separate all segments in which the strains are measured
(b)
514 D. Sonck et al. / Construction and Building Materials 54 (2014) 512–519

(a) (b)

Fig. 3. Schematized residual stress pattern in: (a) plate welded in middle; (b) plate cut at one end.

were permanently damaged during the measurements: from the

total of 360 used strain gages, 31 were found faulty. The measured
strains in these gages (if any) were not included in the results.

2.2. Used specimens

Four castellated and two cellular members were made starting

Fig. 4. Overview of the sectioning method. from six IPE160 parent sections: three geometries were used and
two members were made for each geometry. Geometries CS1 and
CS3 had hexagonal openings, while the CS2 beams had circular
(Fig. 4). Because of its straightforwardness, this method is the pri- openings (Fig. 5).
mary method for the longitudinal residual stress measurement in The steel grade of the members was S275, with a nominal yield
steel I-section members, having been extensively used in the past stress of 275 MPa. Tensile tests to determine the yield stress fy
as well as nowadays [12–15,17,20,22,24–29]. were executed on specimens from the flanges and the web of each
At each measurement location, the residual stress magnitude member (Table 1). The Young’s modulus E could also be derived
rres was calculated according to Eq. (1) using the measured relax- from the tensile test measurements. Unfortunately, some of the
ation strains emeas and the modulus of elasticity E. It was assumed obtained values for E were too aberrant from the mean values gi-
that the specimens behaved elastically during the measurements, ven in literature [30], probably due to a faulty displacement mea-
that the transverse residual stresses could be neglected, and that surement during the test (these values are not included in Table 1).
the thermal influence of the cutting process on the residual stres- As a result, it was chosen not to use the measured values of E for
ses was negligible. the residual stress calculations, but to use E = 210 GPa as a safe
mean value.
rres ¼ E  emeas ð1Þ Taking the variability of the values of the Young’s modulus E
into account, the possible error on the residual stress measure-
The residual stresses were expected to vary along the length of
ments increases. This can be illustrated for the flange stresses,
the non-uniform castellated and cellular members. To study this
using values of Emean ± 2 Est. dev., based on the measured values of
effect, the residual stresses in these members were measured be-
E in the flanges. If the real value of E would be 179 GPa instead
tween the openings (web post in Fig. 1) and at the centre of an
of the used 210 GPa, the calculated residual stress is 15% larger
opening (tee section in Fig. 1). Adhesively bonded electrical strain
than its real value. On the other hand, if E = 223 GPa, the calculated
gages were used to cope with the necessary short gage length. An-
residual stress is 6% smaller than the real residual stress value.
other advantage of using electrical gages was that the strain varia-
Six IPE160 beams of 12 m long were used as the parent section.
tion during the measurement (and cutting) process could be easily
Before the production of the castellated and cellular members, a
monitored continuously. The resistance changes corresponding
specimen of 1 m long was cut from each parent section beam, to
with the strain variations in the strain gages were measured using
determine the original residual stress pattern in the parent sec-
a Wheatstone bridge. Temperature variations during the sectioning
tions. Geometries CS1 and CS3 were made following the standard
process were accounted for by using self-temperature-compen-
sated strain gages specifically designed for mild steel. The error
on the strain measurement is estimated to be ±1.5%. Table 1
The cutting process was executed with an electrical band saw, Yield stresses obtained from the tensile tests, comparison with literature [30].
combined with a handsaw for some of the transverse cuts. Cutting Yield stress, fy [MPa] Young’s modulus, E [GPa]
oil was used to limit local heating of the specimens. A protective
# Samples fy,mean fy,st. dev. # Samples Emean Est. dev.
coating was used to protect the strain gages from this oil. During
Web 8 332 7 6 195 12
the cutting, the measured strains were continuously monitored
Flanges 12 346 5 11 201 11
to be sure that there was no malfunctioning of the strain gages Literature – 327.93 18.96 – 205.47 13.18
due to damage suffered during the multiple manipulations
throughout the sectioning process. However, some strain gages

Fig. 5. Dimensions [mm] of the parent section PS and the castellated and cellular member geometries CS .
 D. Sonck et al. / Construction and Building Materials 54 (2014) 512–519 515

Fig. 6. Production process: (a) cutting of the web; (b) welding of the castellated members; (c) additional cutting of the circular openings for the CS2 geometries.

procedure depicted in Figs. 1 and 6a and b. First, the parent section length: the residual stresses were determined between two open-
webs were cut according to the corresponding pattern for hexago- ings (web post WP) and at the centre of an opening (tee section TS).
nal openings using an oxyfuel cutting process. After the cutting of The measurements in the unwelded castellated members speci-
the web, both halves of the beam were shifted and welded together mens occurred for both the top and bottom halves of the cut
again using a semiautomatic MAG welding process. A but joint IPE160 section. All specimens were taken from the same end of
existing of two weld passes – one on each side of the web – was the members. The length of each specimen was chosen such that
used. For each of both geometries, a section was left unwelded so the distance between the measurement locations and the speci-
that the influence of the cutting process alone on the residual men edge was at least twice the height of the specimen. Thus, pos-
stresses could be studied in the obtained castellated member sible disturbances of the residual stresses by edge effects were
halves. avoided. The remainder of the castellated and cellular members
The CS2 geometries had circular openings. Normally, cellular is being used for buckling experiments which are beyond the scope
members are fabricated following a similar procedure as for castel- of this contribution.
lated members, with two cut lines in the web instead of one to ob-
tain the circular opening shapes (Fig. 1). However, the production
3. Measurement results and discussion
of the two CS2 members deviated from this procedure: the mem-
bers were made by cutting circular openings around the hexagonal
3.1. Measurement results
openings of two completed castellated members with a CS1 geom-
etry (Fig. 6c). Thus, the effect of deviating production procedures
The results of the residual stress measurements in the parent
on the residual stresses could be examined as well. The heat input
section are shown in Fig. 7. The effect of the cutting of the parent
using this deviating procedure was larger than for the standard
sections of the CS1 and CS3 geometries can be seen in Fig. 8, both
production procedure of CS1 and CS3 because the cutting process
for the tee section and the web post. The final residual stress dis-
had to be restarted for each separate web opening.
tribution in the geometries CS1 and CS3 is shown in Fig. 9, while
The strains were measured at five locations in each flange. In
the residual stress distribution for CS2 is shown in Fig. 10. The
the web, five strain gages were used for the parent sections, while
measurements for geometries CS1 and CS3 are shown together,
two gages were used at the tee section and eight at the web post
since they were found to be very similar.
for the cellular or castellated members. It was expected that mea-
In Figs. 7–9, the measured values of the residual stresses are
surements on one side of the flanges or the web would suffice,
shown for each specimen, together with the corresponding least-
since it had been shown that the through-thickness stress varia-
square fit (LSF) curves for the residual stresses in the flanges. The
tions in thin-walled hot-rolled members such as the considered
LSF curves were obtained by mirroring the results about the hori-
specimens is limited [12,13]. Consequently, all measurements
zontal and vertical symmetry axes of the cross-section, after which
were done on the outer side of the flanges and on the same side
these values were fitted to a linear function. This function shape is
of the web. The exact location of the strain gages is given in
based on the perceived variation of the residual stresses across the
Appendix A.
flanges.
An overview of the specimens in which the residual stresses
Since the compressive residual flange stresses have a dominant
were measured is given in Table 2. In total, the residual stresses
detrimental influence on the global buckling resistance, an over-
were measured in six parent section specimens (PS), two unwelded
view of the variation of the flange stresses during the production
castellated member specimens (AC) and seven castellated or
process is depicted in Fig. 11. Here, the least-square fit curves of
cellular member specimens (AW). For both the AC and AW speci-
the measured residual stresses in the flanges are shown.
men types, measurements were made at two locations along the

3.2. Discussion of residual stress measurements

Table 2
An overview of the test specimens. Measurements were done in three different 3.2.1. Influence of cold-straightening
sections: parent section (PS), sections after cutting (AC) and after welding (AW). In the It was expected that the plastic deformation corresponding
latter two, measurement occurred at the web post between two openings (WP) and at with the cold-straightening of a member around its weak axis is
the tee section at an opening (TS).
perceivable through antisymmetric residual stress deviations in
Member name # Locations # Locations after # Locations after the flanges.
parent section (PS) cutting (AC) welding (AW) In the parent sections (Fig. 7), no influence of the cold-straight-
PS WP TS WP TS ening of the beam was observed in the measured residual stresses
CS1L1 1 2  0.5 2  0.5 1 1 in the flanges. However, in the flanges of the AC and AW speci-
CS1L2 1 0 0 1 1 mens, the influence of cold straightening on the residual flange
CS2L1 1 0 0 2 2 stresses is visible (Figs. 8–10).
CS2L2 1 0 0 1 1
CS3L1 1 2  0.5 2  0.5 1 1
CS3L2 1 0 0 1 1 3.2.2. Parent sections (PS)
Total 6 2 2 7 7 The measured residual stresses in the parent section (Fig. 7) are
consistent with earlier results from literature and lie between the
516 D. Sonck et al. / Construction and Building Materials 54 (2014) 512–519

(a) (b)

Fig. 7. Measured residual stresses in the parent section PS (a); comparison of the measured PS residual stresses with results from literature (b).

(a) (b)

Fig. 8. Measured residual stresses in CS1 and CS3 in both halves after the cutting of the web during production: (a) in the web post; (b) in the tee section.

residual stress patterns proposed by Young and the ECCS [17,18],
shown in Fig. 7. Furthermore, it can be seen that the measured
residual stresses were very similar for the six parent section spec-
imens, suggesting that all parent sections came from the same
batch.
3.2.3. Member halves after cutting of the parent sections (AC)
In Figs. 8 and 11, the residual stresses in the parent section
halves obtained after the oxycutting of the web are depicted. It
was expected that this cutting of the parent sections would have
a dual influence on the stresses in the obtained halves. Firstly,
the thermal influence of the cutting process will induce tensile
stresses at the cut, balanced by both compressive and tensile stres-
ses further away from the cut. Secondly, an elastic rebound of the
isolated halve of the member will occur when the forces acting on
it by the other halve are removed. For the considered specimens,
the elastic rearrangement of the residual stresses due to the elastic
rebound was expected to increase the compressive flange stresses.
 D. Sonck et al. / Construction and Building Materials 54 (2014) 512–519 517

(a) (b)

Fig. 9. Measured residual stresses in completed CS1 and CS3: (a) in the web post; (b) in the tee section.

(a) (b)

Fig. 10. Measured residual stresses in completed CS2: (a) in the web post; (b) in the tee section.

However, this compressive stress increase was expected to be
counteracted by the increase in tensile flange stresses due to ther-
mal effects.
The total influence of the oxycutting process on the flange
residual stresses in the beam halves was seen to be rather lim-
ited (Figs. 8 and 11). However, the cutting process did influence
the residual stresses in the web, with high tensile stresses near
the location of the cut. The perceived small influence of the cut-
ting process on the flange stresses is likely due to the counter-
acting effects of the local heat introduction and the elastic
rebound, which are expected to be of the same order of magni-
tude. Nevertheless, it is difficult to draw definitive conclusions
518 D. Sonck et al. / Construction and Building Materials 54 (2014) 512–519
(a)
Fig. 11. Evolution of the residual stresses in the flanges during the production process for geometries CS1&CS3 (a) and CS2 (b). The residual stresses in the parent section are
given by PS. The stresses in the completed members (AW) are given at the web post (WP) and tee section (TS). For geometries CS1 and CS3 (a), the stresses in the separate
halves after the cutting of the web (AC) are given as well for the tee section (TS) and the web post (WP).
from these measurements, since there was only a limited num-
ber of tested specimens for this case and no related results are
available in literature. It could be further investigated by using
finite element software, but since the focus of this work was
on the residual stresses in the completed cellular or castellated
members, this was not further examined.
In the completed cellular or castellated members, the effect of
the elastic rebound will only occur partially because both halves
will be pushed together in a straight line before welding. Conse-
quently, the equilibrating bending moment is delivered externally
to the member halves. However, a rearrangement of residual stresses
is still expected because of the normal equilibrium requirements.
3.2.4. Completed castellated or cellular members (AW)
Due to the local heat introduction during welding, very high
tensile stresses were expected at the weld. It is unclear to what ex-
tent the tensile zone width in the completed member will still be
influenced by the previous effect of the cutting of the web. Accord-
ing to [26], the effect of welding at the same location as the cut al-
most completely relieves the already present residual stresses due
to cutting, while according to [23] the tensile stress zone size will
depend on both the cutting and welding process.
The effect of the welding process on the residual stresses in the
completed castellated sections CS1 and CS3 is clearly noticeable in
Figs. 9 and 11. At the flange tips, a significant residual stress de-
crease of about 50 MPa is noticeable, while the residual stresses
at the flange centres decrease as well. This increase in compressive
flange stresses is caused by an increase of the web (tensile) resid-
ual stresses at the location of the weld and the cut, as expected.
Taking into account the variation of the modulus of elasticity E
and the error on the strain measurement, it is possible that the
flange stress decrease due to the production process is less extreme
(42 MPa). Nevertheless, the magnitude of the residual stress de-
crease at the flange tips remains significant.
The additional effect of the cutting of the circular openings in
CS2 can be seen in Figs. 10 and 11. The effect of the additional heat
input close to the flanges is noticeable in the tee section, where the
tensile residual stresses in the web are very high. These stresses are
balanced by a large increase of compressive residual stresses in the
flanges. At the web post, the residual stresses in the flanges de-
crease less drastically to the same level as for geometries CS1
and CS2. The measured residual stresses in the web at the tee sec-
tion were very high (up to 371 MPa), even larger than the mea-
sured yield stress, which is theoretically impossible. This could
be caused by the high assumed value of E: if the real Young’s mod-
(b)
ulus would only be 10% lower, a value of 334 MPa would be ob-
tained, which is far more realistic. Other, but less likely
explanations could be the hardening of the steel, three-dimen-
sional stress fields close to the fillet between web and flange, or lo-
cal deviations of the yield stress.
The results for the completed geometries CS1 and CS3 show
an increase of the compressive residual stresses in the flange be-
cause of the production process. The deviating production pro-
cess of the CS2 geometry causes even higher compressive
residual stresses at the flange. As a result, it is expected that
the effect of residual stresses is more detrimental for the global
buckling resistance of cellular or castellated members than it is
for their parent sections.
4. Conclusions
It is expected that the global buckling behaviour of cellular and
castellated members is different than that of plain-webbed mem-
bers, due to the effect of the modified geometry and the influence
of the production process on the present imperfections. Regarding
the latter effect, the largest influence is expected for the residual
stresses.
In this paper, it was examined to what extent the cellular and
castellated member production processes modify the residual
stresses present in these members. This was done by measuring
the residual stresses in the parent sections and completed castel-
lated members, as well as in the cut section halves. Additionally,
the effect of a deviating production procedure for the cellular
members was studied. The focus was on the flange residual stres-
ses because the compressive residual stresses in the flanges will
have a dominant detrimental effect on the global buckling resistance.
The residual stresses in the completed castellated and cellular
members are the most important, because these will determine
the buckling resistance of these members. In the completed castel-
lated members, the influence of the production process on the
flange residual stresses could be seen clearly: the compressive
residual flange stresses increased both at the web post and at the
tee section. The modified residual stresses will have a detrimental
effect on the castellated members’ stability. It is expected that the
residual stresses in cellular members made according to the stan-
dard production procedure will correspond well to the values mea-
sured for the castellated members.
The detrimental effect of the production process was even lar-
ger for the cellular members made according to the deviating pro-
duction procedure. Because of the local heat input close to the
 D. Sonck et al. / Construction and Building Materials 54 (2014) 512–519 519

Table 3
Location of strain gages in flanges and in web (y and z are given in mm).

Flanges CS1–CS3 y1 y2 y3 y4 y5
30 15 0 15 30
Web CS1 and CS2 z1 z2 z3 z4 z5 z6 z7 z8
PS 48.6 18.6 0 18.6 48.6   
AC and AW: TS 78.9 78.9 – – – – – –
AC and AW: WP 78.9 48.9 30.3 11.7 11.7 30.3 48.9 78.9
Web CS3 z1 z2 z3 z4 z5 z6 z7 z8
PS 53.5 .26.5 0 26.5 53.5 – – –
AC and AW: TS 93.1 93.1 – – – – – –
AC and AW: WP 93.1 66.1 39.6 13.1 13.1 39.6 66.1 93.1

flanges during the cutting of the circular openings in the completed
castellated sections, the measured flange residual stresses at the
tee section were totally in compression. At the web post, the effect
of the supplementary cutting procedure was smaller: the residual
stresses at the flanges corresponded well with the residual stresses
measured in the castellated sections. However, care should be ta-
ken not to introduce local heat close to the flanges in the com-
pleted cellular or castellated members, as this could have
dramatic consequences for the compressive residual stresses in
the flanges and the corresponding global buckling resistance of
the concerned members.
The effect of the adversely modified residual flange stresses
should be taken into account when calculating the resistance of
cellular or castellated members. In future research, these results
could be used to construct a valid finite element model to study
the global buckling resistance of cellular and castellated members,
based on which accurate expressions for the buckling resistance
could be developed.
Acknowledgement
The authors would like to thank Huys-Liggers (Venlo, The
Netherlands) for supplying the specimens used in the experiments.
Appendix A
The exact location of the strain gages, according to Fig. 12, is
given in Table 3.
Fig. 12. Location of strain gages.
References
[1] Kerdal D, Nethercot DA. Failure modes for castellated beams. J Constr Steel Res
1984;4(4):295–315.
[2] Chung K. Investigation on Vierendeel mechanism in steel beams with circular
web openings. J Constr Steel Res 2001;57(5):467–90.
[3] Tsavdaridis KD, D’Mello C. Web buckling study of the behaviour and strength
of perforated steel beams with different novel web opening shapes. J Constr
Steel Res 2011;67(10):1605–20.
[4] Bitar D, Martin P-O, Galéa Y, Demarco T. Poutres Cellulaires acier et mixte:
partie 1 proposition d’un modèle pour la resistance des montants.
Construction Métallique 2006;1.
[5] Durif S, Bouchaïr A. Behavior of cellular beams with sinusoidal openings. Proc
Eng 2012;40:108–13.
[6] Liu T. Steel beams with large web openings of various shapes and sizes: finite
element investigation. J Constr Steel Res 2003;59(9):1159–76.
[7] Tsavdaridis KD, D’Mello C. Vierendeel bending study of perforated steel beams
with various novel web opening shapes, through non-linear finite element
analyses. J Struct Eng 2012;138(10):1214–30.
[8] Vassart O. Analytical model for cellular beams made of hot rolled sections in
case of fire [PhD dissertation]. Clermont-Ferrand: Université Blaise-Pascal –
Clermont II; 2009.
[9] Sonck D, Vanlaere W, Van Impe R. Influence of plasticity on lateral–torsional
buckling behaviour of cellular beams. Mat Res Innov 2011;15(Suppl. 1):158–61.
[10] Sonck D, Boissonnade N, Van Impe R. Instabilities of cellular members loaded
in bending or compression. In: Proceedings of the annual stability conference
Structural Stability Research Council (SSRC2012) Grapevine, Texas, 2012.
[11] Sonck D, Vanlaere W, Van Impe R. Elastic lateral–torsional buckling of cellular
beams. In: Proceedings of International symposium ‘‘steel structures: culture
and sustainability 2010’’ (SSCS2010); 2010, Turkish Constructional Steelwork
Associations (TUCSA) Istanbul, p. 573–82
[12] Huber AW, Beedle LS. Residual stress and the compressive strength of steel.
Weld J 1954;33(12):589–614 [Res. supl.].
[13] Huber AW. Residual stresses in wide-flange beams and columns. Fritz
Laboratory Reports, Paper 1498, 1956.
[14] Beedle LS, Huber AW. Residual stress and the compressive properties of steel –
a summary report. Fritz Laboratory Reports, paper 46, 1957.
[15] Beedle LS, Tall L. Basic column strength. J Struct Div 1960;86(7):139–73.
[16] Frey F. Effet du dressage à froid des profilés laminés en double té sur leur force
portante. IABSE publications 1969;29(II):101–23.
[17] Young BW. Residual stresses in hot rolled members. In: Proceedings IABSE
colloquium: on column strength, Paris, 1975. pp. 25–38.
[18] ECCS. Ultimate Limit State Calculations of Sway Frames with Rigid Joints
(Publication no. 33), 1984.
[19] Lay MG, Ward R. Residual stresses in steel sections. J Aust Inst Steel Constr
1969;3(3):2–21.
[20] Alpsten GA. Residual stresses, yield stress and column strength of hot-rolled
and roller-straightened steel shapes. In: Proceedings IABSE colloquium: on
column strength, Paris, 1975.
[21] Galambos TV. Inelastic lateral buckling of beams. J Struct Div ASCE 1963;89(236).
[22] Jez-Gala C. Residual stresses in rolled I-sections. Proc Inst Civ Eng (ICE)
1962;23(3):361–78.
[23] ECCS. Manual on stability of steel structures (Publication no. 22), 1976.
[24] Alpsten GA, Tall L. Residual stresses in heavy welded shapes. Fritz Laboratory
Reports, paper 334, 1969.
[25] Nagaraja Rao NR, Tall L. Residual stresses in welded plates. Fritz Laboratory
Reports, no. 249.7, 1960.
[26] Nagaraja Rao, NR, Estuar FR, Tall L. Residual stresses in welded shapes. Fritz
Laboratory Reports, no. 249.18, 1963.
[27] TallL. Residual stresses in welded plates. A theoretical study. Fritz Laboratory
Reports, no. 249.11, 1961.
[28] Tall L. Welded built-up columns. Fritz Laboratory Reports, paper 65, 1966.
[29] Spoorenberg RC, Snijder HH, Hoenderkamp JCD. Experimental investigation of
residual stresses in roller bent wide flange steel sections. J Constr Steel Res
2010;66(6):737–47.
[30] Simões da Silva L, Rebelo C, Nethercot DA, Marques L, Simões R, Vila Real PMM.
Statistical evaluation of the lateral–torsional buckling resistance of steel I-
beams, Part 2: variability of steel properties. J Constr Steel Res
2009;65(4):832–49.