Bocno Izvijanje Sacastih Nosaca
Bocno Izvijanje Sacastih Nosaca
Bocno Izvijanje Sacastih Nosaca
l
t
t
b
w
w
wc
(2)
- Ior Class 3:
w
t l 36
0
or
2
0
36
1
14
(
(
,
\
,
,
(
j
l
t
t
b
w
w
wc
(3)
I. Radi, D. Markulak Bono izvijanje saastih nosaa
TEHNIKI VJESNIK 14(2007)(1,2)25-35 27
where:
b
wc
- is the tee steam outstand b
w1
or b
w2
indicated in
Iigure 2.
l
0
- is the length deIined in Iigure 3.
This paper investigates the possibility oI stability
loss due to lateral buckling oI castellated beams, be-
cause they are relatively slender and they have web
openings, which have an inIluence on their resistance.
Figure 3. Length l0
Slika 3. Dulfina l0
2
Formulation of research objective
Formulacija cilja istraivanja
Lateral buckling is the problem oI stability loss and
bearing capacity Iailure beIore plastiIication oI cross-
section. Beams primarily constructed Ior taking Iorces
in their major plane are sensitive to lateral buckling, un-
til they have smaller bending stiIIness around minor
axis (I and U cross-sections).
Elastic critical moment Ior lateral buckling M
cr
ac-
cording to elastic theory represents minimum moment
which causes lateral buckling and according to |2| Ior
biaxial symmetrical cross-section can be calculated
Irom the Iollowing expression:
( )
( )
( )
]
]
]
]
]
,
,
,
,
+
+
(
(
,
\
,
,
(
j
g g
:
t
:
w
w
:
cr
: C : C
I E
I G L k
I
I
k
k
L k
I E
C M
2
2
2
2
2
2
2
2
1
(4)
where:
C
1
, C
2
- coeIIicients depending on the loading and end
restraint conditions
Iactor k reIers to end rotation on plan.
Iactor k
w
reIers to end warping.
I
:
- is the second moment oI area
I
w
- is the warping constant
I
t
- is the torsion constant
G - is the shear modulus
:
g
- is the distance between the point oI load application
and the shear center.
Results oI the expression (4) are very sensitive Ior
the Iactors k and k
w
whose geometrical interpretation is
shown in Figure 4.
Figure 4. Plan view illustrating idealised end restraints
Slika 4. Graficka interpretacifa faktora k i kw
Lateral buckling resistance M
b,Rd
must be bigger
than calculated bending moment M
v,Ed
, is given by mul-
tiplying reduction Iactor
LT
with cross-section bend-
ing resistance:
1
,
M
v v
LT Rd b
f W
M
(5)
Reduction Iactor Ior lateral buckling
LT
can be
determined according to expression:
2 2
1
LT LT LT
LT
+
but 0 , 1
LT
(6)
where:
cr
v v
LT
M
f W
, (7)
Lateral Buckling of Castellated Beams I. Radi, D. Markulak
TECHNICAL GAZETTE 14(2007)(1,2)25-35 28
( ) [ ]
2
2 , 0 1 5 , 0
LT LT LT LT
+ + , (8)
LT
- imperIection Iactor corresponding to the
appropriate buckling curve which depends on the type
and geometrical characteristics oI cross-section (W
v
depends on the class oI cross-section).
Because oI the relatively high slenderness,
castellated beams, applied on greater spans and
signiIicant web openings (even to 75 in relation to
the Iull web), can be considered as the beams sensitive
to lateral buckling. According to |1| the lateral-
torsional buckling resistance should be checked using
the cross-sectional properties relevant to the mid-length
oI openings, i.e. netto cross-section in the middle oI the
opening where the smallest geometrical characteristics
should be taken into lateral buckling resistance
checking. It means that validation oI lateral buckling
resistance is the same Ior castellated beams and
standard plate girders, but with reduction oI
geometrical characteristics castellated beam cross-
section.
So the research objective is estimation oI accuracy
Ior this simpliIied approach to lateral buckling
resistance oI castellated beams, because that is still a
stability problem oI the beam which doesn't have a
uniIorm cross-sectional properties along the whole
length.
3
Methods of problemresearch
Metode istraivanja problema
From the research objective it arises that it is nec-
essary to determine the inIluence oI reducing lateral
buckling resistance oI castellated beams by web open-
ings in relation to plate girders with the same geometri-
cal dimensions (same height, width and thickness oI
Ilange, and web thickness). Taking this into considera-
tion calculations were perIormed Ior the elastic critical
moment Ior lateral buckling according to |2|, (
EC
cr
M ),
and applying Iinite element method using linear buck-
ling analysis, (
FEM
cr
M ).
Generally, linear buckling analysis starts with the
equilibrium equation Ior the element subjected to a
constant Iorce:
[ ] K u P
(9)
where:
[ ] K - stiIIness matrix
u - displacement vector
P - applied load vector.
In order to include the diIIerential stiIIness eIIects,
the diIIerential stiIIness matrix, |K
D
|, is introduced and
this matrix Ior a typical element is a Iunction oI the
state oI stress in the element.
Hence, by introducing k
cr
as an arbitrary scalar
multiplier Ior another "intensity" oI load, the
equilibrium equation becomes:
[ ] [ ] ( ) +
*
cr D cr
K k K u k P
, (10)
where
*
u
is the modiIied displacement vector result-
ing Irom displacements under the intensity oI load, and
Irom diIIerential stiIIness eIIects.
By perturbing the structure slightly at a variety oI
load intensities, the "intensity" Iactor k
cr
to create
unstable equilibrium conditions, will be the Iactor to
cause buckling:
[ ] [ ] ( )
0 +
*
cr D
K k K u
. (11)
This requires the solution oI an eigenvalue problem:
[ ] 0
cr D
K k K
. (12)
The solution is nontrivial, (k
cr
diIIerent Irom 0)
only Ior speciIic values oI k
cr
that would make the ma-
trix
[ ]
cr D
K k K singular.
The product oI the Iirst load intensity Iactor or the
Iirst eigenvalue k
cr
with the applied load would give the
Iirst buckling load oI the model, and the eigenvector ,
the buckled shape.
The described method applied Ior elastic linear
buckling analysis to determine a critical load intensity
Iactor k
cr
in this paper is perIormed using soItware
package Robot Millennium. So, linear buckling analysis
results are critical intensity Iactors k
cr
Ior each shape
(mode) oI buckling.
The product oI calculated load intensity Iactor and
the load which the model is subjected to, gives critical
load, which leads to stability loss (or buckling). When
a critical load is known it is not diIIicult to calculate
the value oI critical bending moment. It depends on
static system oI the analyzed structural element, type
and distribution oI load.
This paper analyzes simply supported beams with
various spans subjected to uniIorm load, and critical
bending moments can be calculated using the Iollowing
expression:
2
8
FEM
cr cr
q L
M k
, (13)
where:
k
cr
- critical load intensity Iactor
q - uniIorm load
L - span oI castellated beam.
I. Radi, D. Markulak Bono izvijanje saastih nosaa
TEHNIKI VJESNIK 14(2007)(1,2)25-35 29
For this research two groups oI rolled proIiles are
chosen - IPE and HE-A, considering diIIerent geomet-
rical characteristics oI cross-section. It is assumed that
castellated beams will be used like rooI beams so they
are designed taking into account the loads oI rooI cover
("sandwich" panel) and snow.
Distance between the beams made oI IPE proIiles
is 2,20 m, and those made oI HE-A proIiles is 2,50 m.
Analysis is perIormed on castellated beams whose
height is 50 higher than the original IPE and HE-A
proIiles. Spans and chosen cross-sections oI the ana-
lyzed beams are shown in Table 1.
Table 1. Castellated beams
Tablica 1. Sacasti nosaci
CROSS
SECTION
Original
profile
height
Height of
castellated
beam
Span
of
beam
Distance
between
beams
Width
(a0) and
height
(h0) of
opening
Distance
of
openings
(w)
mm m mm
IPE 240 240 360 8,0 2,20 253,5 126,75
IPE 300 300 450 10,0 2,20 317,0 158,5
IPE 360 360 540 12,0 2,20 380,0 190,0
HE 220A 210 315 8,0 2,50 213,0 106,5
HE 260A 250 375 10,0 2,50 246,0 123,0
HE 300A 290 435 12,0 2,50 296,0 148,0
All beams are subjected to uniIorm load on whole
length, and the load is located on top (compressed)
Ilange. Each oI these six models is analyzed Ior the
three diIIerent cases oI lateral supporting i.e. Ior the
cases:
- k 1,0 and k
w
1,0,
- k 0,5 and k
w
1,0,
k 1,0 and k
w
0,5.
Total oI 18 3D models are created in Robot Mil-
lennium using Iinite elements type SHELL. Only halI
oI beam was modeled and on one end were used sym-
metry boundary conditions, and on the other end were
boundary conditions oI the case, which we want to
model.
It was needed to pay special attention to boundary
conditions modeling Ior diIIerent cases oI lateral sup-
porting because models are very sensitive to this prob-
lem. Even relatively small corrections oI boundary
conditions can give signiIicant diIIerences in calcula-
tion results. Plate girders were modeled as "control
models", with the same geometrical dimensions as ana-
lyzed castellated beams. For these beams it was possi-
ble to apply conventional "manual" lateral buckling re-
sistance veriIication according to |2|. It is important to
accentuate that buckling length oI analyzed beams in
load plane is equal to beam length, but lateral buckling
length out oI load plane depends on Iactor k.
4
Calculation results
Rezultati prorauna
In Figures 5, 6 and 7 are shown boundary condi-
tions oI models and buckling shapes Ior all three cases
oI lateral supporting which were analyzed. In Table 2
are shown results oI perIormed research - elastic criti-
cal moments oI lateral buckling: using expression (13)
-
FEM
cr
M , and according to |2| -
C
cr
M (expression (4)).
Figure 5. Boundarv conditions and buckling shape for case k 1,0 and kw 1,0
Slika 5. Rubni uvfeti i oblik i:vifanfa :a slucaf k 1,0 i kw 1,0
Lateral Buckling of Castellated Beams I. Radi, D. Markulak
TECHNICAL GAZETTE 14(2007)(1,2)25-35 30
Figure 6. Boundarv conditions and buckling shape for case k 0,5 and kw 1,0
Slika 6. Rubni uvfeti i oblik i:vifanfa :a slucaf k 0,5 i kw 1,0
Figure 7. Boundarv conditions and buckling shape for case k 1,0 and kw 0,5
Slika 7. Rubni uvfeti i oblik i:vifanfa :a slucaf k 1,0 i kw 0,5
For the interpretations oI the given results it is nec-
essary to comment the results oI lateral buckling criti-
cal moments only Ior castellated beams Iirst, and then
compare them with the calculated values oI critical
moments Ior the castellated and appropriate plate gird-
ers.
Graphical illustrations oI the appropriate values oI
critical moments Ior castellated beams are shown in
Figures 8, 9 and 10. It can be noticed that "manual"
procedure according to |2| gives mostly conservative
estimation oI the critical moment, except Ior simply
supporting case (k 1,0 and k
w
1,0) when values are
practically equal (but it is needed to say that these
boundary conditions are very rare in practice). For all
I. Radi, D. Markulak Bono izvijanje saastih nosaa
TEHNIKI VJESNIK 14(2007)(1,2)25-35 31
other lateral supporting cases FEM 3D models give lar-
ger values, whereas Ior (k 0,5; k
w
1,0) case diIIer-
ences between critical moments are smaller and they
are in the range oI 6 , and Ior (k 0,5; k
w
1,0) case
these diIIerences are greater (in the range oI 10 ).
This last lateral supporting case (k 0,5; k
w
1,0) is
hard to model precisely, and in this paper two ways oI
modeling are used which gave almost the same results.
The Iirst way was by using the option "rigid links"
which gives a possibility to set the same values oI dis-
placement and rotations Ior all points in speciIied cross-
section (Figure 7). The second way is to replace the
relatively short area oI supporting edge with the BEAM
element, which is a more precise way to control bound-
ary conditions (regarding 3D SHELL model). Gener-
ally, greater deviations are noticed on IPE cross-
sections with regard to HE-A cross-sections. That can
be associated with relative diIIerences between geo-
metrical characteristics oI cross-sections (elastic and
plastic) oI those section types through a web opening
("netto") and out oI opening ("brutto"), where that rela-
tion on IPE section is twice as high.
Table 2. Elastic critical moments of lateral buckling for plate girders and castellated beams calculated using Robot Millennium and Eurocode 3
Tablica 2. Elasticni kriticni momenti bocnog i:vifanfa :a punostfene i sacaste nosace dobivenih i: Robot Milleniuma i prema Eurokod 3
propisima
PLATE GIRDERS CASTELLATED BEAMS
k = 1,0
kw= 1,0
k = 0,5
kw= 1,0
k = 1,0
kw= 0,5
k = 1,0
kw= 1,0
k = 0,5
kw= 1,0
k = 1,0
kw = 0,5
CROSS SECTION
kNm
EC
cr
M 27,92 44,56 39,55 25,72 40,87 37,87
IPE 240
FEM
cr
M 27,57 46,27 44,17 26,12 44,37 42,85
EC
cr
M 43,18 68,55 64,27 39,43 62,25 61,51
IPE 300
FEM
cr
M 42,55 69,03 71,52 40,25 66,12 69,48
EC
cr
M 63,89 101,58 93,82 58,78 93,00 90,02
IPE 360
FEM
cr
M 63,06 103,05 104,86 59,77 98,85 101,93
EC
cr
M 118,44 185,06 201,82 114,89 179,17 199,42
HE 220A
FEM
cr
M 115,33 189,08 225,11 111,93 184,50 221,94
EC
cr
M 169,23 264,42 288,72 165,26 257,82 286,02
HE 260A
FEM
cr
M 164,29 268,95 322,26 159,08 262,88 318,09
EC
cr
M 235,96 368,83 401,43 229,73 358,47 397,21
HE 300A
FEM
cr
M 229,68 377,95 448,57 223,55 369,42 442,71
Figure 8. Critical moments of castellated beams for boundarv conditions k1,0 and kw1,0
Slika 8. Kriticni momenti :a sacaste nosace :a rubne uvfete k 1,0 i kw 1,0
Lateral Buckling of Castellated Beams I. Radi, D. Markulak
TECHNICAL GAZETTE 14(2007)(1,2)25-35 32
Figure 9. Critical moments of castellated beams for boundarv conditions k0,5 and kw1,0
Slika 9. Kriticni momenti :a sacaste nosace :a rubne uvfete k 0,5 i kw 1,0
Figure 10. Critical moments of castellated beams for boundarv conditions k1,0 and kw0,5
Slika 10. Kriticni momenti :a sacaste nosace :a rubne uvfete k 1,0 i kw 0,5
Next step is comparison oI the critical moments Ior
the castellated and appropriate plate girders, (Figure
11, 12 and 13). Generally, we can conclude that there is
a very good correspondence between the results Ior
castellated and plate girders in relation to the applied
method oI critical moments calculations - "manual" ac-
cording to |2| or Irom FEM models. Also it is shown
that diIIerences in moments Ior castellated and appro-
priate plate girders are relatively small, and they are Ior
the beams with IPE sections in the range oI 5 , while
Ior the beams with HE-A proIiles (stiIIener about mi-
nor axis) they are even smaller and they are max. 3 .
The research results show that inIluence oI the web
openings is relatively small Irom the aspect oI the ana-
lyzed problem. The research results Iit well into
previously published researches |4|.
A good illustration is example oI Iictive conIigura-
tion castellated beams shown in Figure 14. This Iigure
shows variation oI load intensity critical Iactor k
cr
Ior
various castellated beams made oI IPE 360 proIile Ior
which web openings are progressively increased. Start-
ing model is plate girders Ior which web opening is
equal to zero, and Iinal case is the model oI Iictive
beam with two openings which are stretched along the
whole halIspan. Beside the beams with polygonal (hex-
agonal) web openings the beams with rectangular web
openings are also analayzed. Even here illustrated exact
example oI beam shows the general trend. In the dia-
gram the area which EC deIines as maximum allowed
web opening the width is shadowed (expression (1)).
I. Radi, D. Markulak Bono izvijanje saastih nosaa
TEHNIKI VJESNIK 14(2007)(1,2)25-35 33
In the diagram it can be seen that in this area diIIer-
ences between critical moments are negligible, espe-
cially Ior the beams with polygonal web openings,
which have greater lateral buckling resistance than
those with rectangular web openings (better "restrains"
compressed tee).
Figure 11. Critical moments of lateral buckling for plate girders and castellated beams calculated using Robot Millennium k 1,0 and kw 1,0
Slika 11. Kriticni momenti :a sacaste i plocaste nosace dobiveni na modelima u Robotu Millennium :a rubne uvfete k 1,0 i kw 1,0
Figure 12. Critical moments of lateral buckling for plate girders and castellated beams calculated using Robot Millennium k 0,5 and kw 1,0
Slika 12. Kriticni momenti :a sacaste i plocaste nosace dobiveni na modelima u Robotu Millennium :a rubne uvfete k 0,5 i kw 1,0
Lateral Buckling of Castellated Beams I. Radi, D. Markulak
TECHNICAL GAZETTE 14(2007)(1,2)25-35 34
Figure 13. Critical moments of lateral buckling for plate girders and castellated beams calculated using Robot Millennium k 1,0 and kw 0,5
Slika 13. Kriticni momenti :a sacaste i plocaste nosace dobiveni na modelima u Robotu Millennium :a rubne uvfete k 1,0 i kw 0,5
Figure 14. Alteration of critical factors kcr for different configuration "castellted" beams
Slika 14. Promfena kriticnih koeficifenta kcr :a ra:licite konfiguracife "sacastih" nosaca
5
Conclusion
Zakljuak
In this paper the research into the castellated beams
stability Irom the aspect oI the lateral buckling resis-
tance is described, Ior which new Eurocode 3 suggests
simpliIied lateral buckling resistance veriIication. The
suggested method Ior lateral buckling veriIication can
be perIormed as well as the method Ior standard plate
girders but with reduction oI geometrical
characteristics oI cross-section so that geometrical
characteristics oI cross-section through openings
should be used in veriIication.
I. Radi, D. Markulak Bono izvijanje saastih nosaa
TEHNIKI VJESNIK 14(2007)(1,2)25-35 35
The research results show that web openings oI the
castellated beams have little inIluence on determining
oI M
cr
. Values oI lateral buckling critical moments ob-
tained on 3D FEM models, depending on lateral
boundary conditions, give practically the same or a lit-
tle higher values (up to 10 ) in relation to "manual"
calculation method.
It is very indicative that the parallel FEM analysis
oI the castellated and appropriate plate girders shows
that diIIerences in obtained lateral buckling critical
moments are practically negligible (up to max. 5 ), so
it can be concluded that the simpliIication oI critical
moments which can be calculated is the same as that oI
the critical moments Ior standard plate girders but tak-
ing into consideration geometrical characteristics oI the
cross-section through openings. This method gives
good estimation oI critical moment. The IPE or slender
beams, where the relation "netto" and "brutto" oI cross-
section geometrical characteristics is somewhat higher,
give some greater deviation, while deviation on HE-A
proIiles is practically negligible.
Using complex 3D FEM models oI castellated
beams it is possible to determine lateral buckling resis-
tance oI castellated beams more precisely. It can be ac-
complished by variation oI relevant parameters: stan-
dard supporting conditions, lateral supporting condi-
tions, detailed model oI beam geometry with possible
transverse stiIIeners and intermediate supporting along
span, introducing possible geometry and material im-
perIections, various load conditions etc. So in the next
researches FEM models oI continuous beams can be
made with various lateral support conditions and load
type.
6
References
Literetura
|1| CEN - European Committee Ior Standardization (1992):
Eurocode3 - Design oI steel structures - Part 1-1: Gen-
eral rules and rules Ior buildings, ENV 1993-1-
1:1992/A2:1998, Annex N
|2| CEN - European Committee Ior Standardization (2003):
Eurocode3 - Design oI steel structures - Part 1-1: Gen-
eral rules and rules Ior buildings, prEN 1993-1-1:2003
(E)
|3| Robot Millennium v. 17.0, User's Manual, UPDATE
November 2003
|4| Demirdjian, S. Stability oI Castellated beam webs, Mas-
ter Thesis, 1999., Department oI Civil Engineering and
Applied Mechanics, McGill University Montreal, Can-
ada.
|5| Fuller, I. J. Development oI an Object Oriented Hand-
book and a Numerical Methods Application Ior Steel
Structures, Masther Thesis, Department oI Civil and En-
vironmental Engineering, Carleton University, Ottawa,
Ontario, Canada, September 1999.
|6| Androic, B.; Dujmovic, D.; Dzeba, I. Metalne konstruk-
cije 1, Institut gradevinarstva Hrvatske, Zagreb, 1994.
|7| Tkalcevic, V.; Dzeba, I.; Androic, B. Proracun sacastih
nosaca prema Eurokodu 3, Gradevinar 58 (2006) 9, str.
709-716.
|8| Deljak, M. Bocno izvijanje prosupljenih nosaca, Dip-
lomski rad, Sveuciliste J. J. Strossmayer u Osijeku,
Gradevinski Iakultet, 2006.
|9| Markulak, D. Celicne konstrukcije interna skripta dio
1, Sveuciliste J. J. Strossmayer u Osijeku, Gradevinski
Iakultet, Osijek, 2004.
Author`s Address (Adresa autora):
Ivan Radic, dipl. ing. grad.
Sveuciliste J. J. Strossmayera u Osijeku
Gradevinski Iakultet
Drinska 16a, 31000 Osijek, Croatia
e-mail: radicgIos.hr
prof.dr.sc. Damir Markulak, dipl. ing. grad.
Sveuciliste J. J. Strossmayera u Osijeku
Gradevinski Iakultet
Drinska 16a, 31000 Osijek, Croatia
e-mail: markulakgIos.hr