Documente Academic
Documente Profesional
Documente Cultură
www.elsevier.com/locate/jcsr
Abstract
Using a commercially available finite element analysis package ABAQUS, this paper
presents the results of a numerical parametric study to investigate the effects of a number of
design factors on the lateral torsional buckling bending moment resistance of a steel beam
with non-uniform temperature distributions. The results of these numerical studies are then
used to check the predictions of current design methods in BS 5950 Part 8 and ENV 1993-
1-2. It is found that both methods predict much lower critical temperatures than ABAQUS
and neither method is suitable to deal with the effects of non-uniform temperature distributions.
To enable ENV 1993-1-2 to be used, it is necessary to modify the lateral torsional buckling
slenderness of a steel beam to consider the effects of non-uniform temperature distributions
on the elastic critical buckling resistance and the effective cross-sectional shape of the steel
beam. An example is provided to show how the modified effective length may be calculated.
2003 Elsevier Science Ltd. All rights reserved.
Keywords: Lateral torsional buckling; Non-uniform temperature distribution; Steel beam; Fire; Slender-
ness; Design method
1. Introduction
Lateral torsional buckling is one of the most important modes of failure of a steel
beam at ambient temperature. Consequently, there have been many research studies
Corresponding author. Tel.: +44-161-200-8968; fax: +44-161-200-4646.
E-mail address: yong.wang@umist.ac.uk (Y.C. Wang).
0143-974X/03/$ - see front matter 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0143-974X(03)00003-8
1010 Y.Z. Yin, Y.C. Wang / Journal of Constructional Steel Research 59 (2003) 10091033
Nomenclature
Cw warping constant
E modulus of elasticity
(E)T,(E)B modulus of elasticity at the top and bottom flanges at elevated
temperatures
ECw warping rigidity of cross-section
(ECw)B warping rigidity of cross-section at the bottom flange temperature
EIz flexural rigidity of cross-section
(EIz)B flexural rigidity of cross-section at the bottom flange temperature
fy yield strength at ambient temperature
G shear modulus at ambient temperature
GJ torsional rigidity of cross-section
(GJ)B torsional rigidity of cross-section at the bottom flange temperature
Iz second moment of area of cross-section about the minor axis
J torsional constant
kE,fi retention factor for stiffness of steel at elevated temperatures
ky,fi retention factor for strength of steel at elevated temperatures
L effective length of beam
Mb,uniform,Mb,nonuniform elastic lateral torsional buckling resistance under
uniform or non-uniform temperature distributions
Mb elastic lateral torsional buckling resistance at ambient temperature
for beams without warping restraint
Mb elastic lateral torsional buckling resistance at ambient temperature
for beams with warping restraints
Mc plastic bending moment capacity at ambient temperature
lLT lateral torsional buckling slenderness at ambient temperature
lLT,fi lateral torsional buckling slenderness at elevated temperatures
lLT,uniform,lLT,nonuniform lateral torsional buckling slenderness under uniform
and non-uniform temperature distributions
of this topic and it is also comprehensively dealt with in design codes of practice
for the ambient temperature condition. In contrast, there are only a few studies of
this topic under fire conditions and current design methods for the lateral torsional
buckling resistance of a steel beam are relatively brief compared to many other
aspects of steel structural fire engineering design.
This problem comes about because lateral torsional buckling in fire is a relatively
minor problem. At ambient temperature, occurrence of lateral torsional buckling of
a steel beam is often associated with the failure of roof beams where there is not
sufficient lateral restraint to the compression flange of the beam. Since roof beams
do not usually require fire resistance, they do not need be checked for lateral torsional
buckling. Beams forming part of a fire resistant floor are required to have sufficient
Y.Z. Yin, Y.C. Wang / Journal of Constructional Steel Research 59 (2003) 10091033 1011
fire resistance, however, such beams will usually have floor slabs on top. Since these
are normally designed as simply supported, their compression flanges are continu-
ously restrained under normal gravity loading and lateral torsional buckling does
not occur.
Nevertheless, under certain circumstances, it may still be necessary to evaluate
the lateral torsional buckling capacity of a steel element. Such circumstances include
continuous beams in the hogging bending moment region, in parallel beam systems
where there may not be sufficient lateral and torsional restraints or in columns under
combined compression and bending.
Among a handful of relevant research studies of lateral torsional buckling of steel
beams, Bailey et al. [2] performed a numerical analysis for beams with a uniform
temperature distribution and found that the current design method in BS 5950 Part
8 [4] over predicts the lateral torsional buckling resistance of steel beams within the
practical range of beam slenderness. Piloto and Vila Real [10] conducted a few
experiments of slender beams at uniform elevated temperatures. Vila Real and
Franssen [13] carried out a numerical study and proposed a design method for lateral
torsional buckling of steel beams with uniform temperature distribution. This design
method has been adopted in the current version of ENV 1993-1-2 [6].
It appears that there has not been any study of the effects of non-uniform tempera-
ture distributions in the cross-section of a steel beam on its lateral torsional buckling
resistance. This is the objective of the present study. This paper will present the
results of a numerical study using a general finite element analysis package
ABAQUS [1]. This program is used to carry out a comprehensive parametric study
to investigate the effects of various design factors on limiting temperatures of steel
beams with non-uniform temperature distributions in the cross-section. The factors
under investigation include beam slenderness, type of non-uniform temperature dis-
tribution, load type, position and magnitude. The results of these numerical simula-
tions are then used to check the design methods in BS 5950 Part 8 [4] and ENV
1993-1-2 [6]. Both methods only consider the effect of non-uniform temperature
distributions in a steel beam on the beams plastic bending moment capacity. This
paper suggests that the beams slenderness is also affected. A method is proposed
to deal with this change and a design example is given to demonstrate how to use
the proposed method.
In BS 5950 Part 8 [4], two methods may be used: the limiting temperature method
and the bending moment capacity method. In the limiting temperature method, a
1012 Y.Z. Yin, Y.C. Wang / Journal of Constructional Steel Research 59 (2003) 10091033
load ratio is calculated, which gives the ratio of the applied bending moment on a
beam in fire to the beams plastic bending moment capacity at ambient temperature.
BS 5950 Part 8 gives two sets of limiting temperatureload ratio relationships, one
for uniform heating and one for non-uniform heating. These relationships are plotted
in Fig. 1. It can be seen that for uniform heating, the limiting temperature-load
ratio relationship is close to the steel stress at 2% straintemperature relationship.
The limiting temperatureload ratio relationship for non-uniform heating is close to
that obtained by dividing the load ratio of uniform heating by 0.7.
The corresponding method in ENV 1993-1-2 [6] uses the utilisation factor and
gives the critical temperature of a beam as a function of the utilisation factor. For
uniform heating, this utilisation factor is equivalent to the load ratio in BS 5950 Part
8. The ENV utilisation factorcritical temperature relationship is plotted in Fig. 1.
It can be seen that this relationship is almost identical to the BS 5950 load ratio
limiting temperature relationship. Indeed, the ENV method is the result of a curve
fitting exercise to the BS 5950 results. For non-uniform heating, the simple method
in ENV 1993-1-2 allows the plastic bending moment capacity of a beam to be
obtained by dividing that for uniform heating by 0.7.
A more accurate method may be used in both BS 5950 and ENV 1993-1-2. Here,
the plastic bending moment capacity of a steel beam cross-section is numerically
evaluated by dividing the cross-section with non-uniform heating into many slices.
The plastic neutral axis is found to divide the cross-section into two parts with equal
force resultants, one in tension and the other in compression. The resultant bending
moment of these forces is the beams plastic bending moment capacity.
cLT,fi 1 / f 2
l2LT,fi,
LT,fi
where
fLT,fi
1
2
1 0.65235 / fy lLT,fi l2LT,fi .
This reduction factor is based on a study of Vila Real and Franssen [13] and has
been obtained after calibrating against the numerical results obtained using the finite
element analysis software SAFIR [7].
If a steel beam is non-uniformly heated, the present version of ENV 1993-1-2
treats the beam as uniformly heated at the temperature of the compression flange of
the steel beam. Since the lower flange of a beam is usually heated to higher tempera-
tures, this recommendation is only applicable when the lower flange is under com-
pression, i.e. when the beam is under a hogging bending moment. Under this circum-
stance, the beams plastic bending moment capacity will be underestimated. On the
other hand, if the compression flange of a beam is at lower temperatures, the design
method in the current version of ENV 1993-1-2 may not be safe.
Thus, neither BS 5950 Part 8 nor ENV 1993-1-2 is able to deal with lateral tor-
sional buckling of a steel beam with non-uniform temperature distributions because
they do not consider two important effects of non-uniform heating on the lateral
torsional buckling slenderness of a beam. Firstly, the beams elastic lateral torsional
buckling resistance is much higher. Secondly, the cross-section has become unsym-
metrical about the major axis. For the latter, the beam slenderness based on a sym-
1014 Y.Z. Yin, Y.C. Wang / Journal of Constructional Steel Research 59 (2003) 10091033
metrical I-section may not be suitable. Both effects of non-uniform heating on the
lateral torsional buckling slenderness of a beam will be investigated in more detail
in a later section on design calculations.
3. Numerical simulations
There are very few experimental studies of lateral torsional buckling of steel beams
under fire conditions. To check design methods, the present study will use the
numerical results of a general finite element analysis package ABAQUS [1].
3.1. Procedure
only, e.g. fin plate connections or web cleat connections. The second type (WR) is
applicable to cases where connections are made to the entire cross-section, e.g. flush
or extended end plate connections.
The finite element simulation is divided into three steps. The first step is eigenvalue
buckling analysis. The result of this analysis is used to specify initial imperfections
of the beam according to its critical buckling mode. In the second and third steps,
loads and temperatures are applied consecutively. During the elevated temperature
analysis, an unrestrained beam will buckle laterally and the numerical solution pro-
cess may become unstable. To obtain a stable solution a pseudo-dynamic analysis
with a dissipated energy fraction of 1E-10 is used. This energy fraction is chosen
by trial and error to be small enough so as to have almost no effect on the behaviour
of the beam. The critical temperature of a beam is taken as that at runaway deflection.
If runaway deflection does not happen, a beam is assumed to have failed when its
maximum in-plane deflection reaches L/20.
Steel has a yield strength and Youngs modulus of 275 and 200 000 N/mm2,
respectively, at ambient temperature. Elevated temperature properties of steel are
according to ENV 1993-1-2. A constant thermal expansion coefficient of 1.4
105 (mm/mm C) is used.
To determine the optimal mesh of shell elements for numerical analysis, Table 1
compares the results for a steel beam UB 457 152 60 under uniform bending
and heating. From this comparison, it is noticed that the flange element size has a
large effect. For combined accuracy and computational economy, the beam flange
is divided into four shell elements and the web into eight in later parametric studies.
The number of elements along the beam length is to ensure that the aspect ratio of
shell elements is about 35.
Table 1
Sensitivity of beam buckling moment to different finite element meshes
WR-S1 2 4 40 90.7
WR-S2 2 8 40 95.8
WR-S3 4 4 40 116.1
WR-S4 4 8 40 115.8
WR-S5 4 8 80 110.2
WR-S6 8 8 80 110.7
Elastic theory (WR) 120.1
NWR-S7 4 8 40 91.0
Elastic theory (NWR) 96.4
NWR, S7, beams without warping restraint; WR, S1~S6, beams with warping restraint.
1016 Y.Z. Yin, Y.C. Wang / Journal of Constructional Steel Research 59 (2003) 10091033
p4E2CwIz p2EIzGJ
Mb (1)
L4 L2
in which, Iz is the second moment of area of the cross-section about the minor axis;
J is the torsional constant; Cw is the warping constant; and L is the beam effective
length. The design values are calculated according to BS 5950 Part 1 [3]. For a
beam without warping restraint (NWR), L equals to the beam span. For a beam with
warping restraint (WR), L is a variable. Some research studies have been carried out
on the effect of end warping restraint on lateral torsional buckling resistance of
beams [8,9,11]. It has been found that the lateral torsional buckling bending moment
resistance will increase by about 30100% compared to no warping restraint, i.e.
Fig. 4. Comparison of lateral torsional buckling bending moment resistance between theoretical and
ABAQUS results.
the effective length is about 0.70.9 of that of beams without warping restraint. BS
5950 Part 1 recommends a value of 0.85L as the effective length of a steel beam
with both flanges partially restrained against rotation on plan. This value is adopted
in this study. Fig. 4 shows that the numerical simulations are accurate and also the
BS 5950 recommended effective length for restrained warping is suitable.
4. Parametric study
To study the lateral torsional buckling behaviour of steel beams with non-uniform
heating and to provide information to amend design methods, a parametric study has
been performed. In the present study, an I-section of 457 152 60 UB is used
with a span L of 3 or 8 m, giving an out-of-plane slenderness ly of 70 or 300.
1018 Y.Z. Yin, Y.C. Wang / Journal of Constructional Steel Research 59 (2003) 10091033
Three types of load are applied as shown in Fig. 5, with the transverse loads
applied at the top flange or centroid of the cross-section, respectively. Table 2 gives
the lateral torsional buckling bending moment resistance according to BS 5950 Part
1 for different spans and load cases.
Two levels of applied load, producing a maximum bending moment in a beam of
0.4 and 0.8 Mb, respectively have been considered. These equal to load ratios 0.4
and 0.8 for beams without warping restraint. For beams with warping restraint, the
load ratios are about 0.34 and 0.68, respectively.
Four different temperature profiles in the beam cross-section, one uniform and
three non-uniform, are analysed. These temperature profiles are shown in Fig. 6.
Profile N3 is similar to that of a beam with three sides heated and after a long fire
exposure. Profile N1 is linear and has the same upper and lower flange temperatures
as profile N3. Profile N2 has a similar average temperature as N1, but a lower top
flange temperature and represents the temperature profile of a beam after a short
fire exposure.
The parametric study is in two parts: laterally restrained beams and laterally unre-
strained beams. The first part is to check various calculations for the plastic bending
moment resistance of a beam and the second part to consider the slenderness effect.
Table 2
Buckling resistance at ambient temperature, according to BS 5950 Part 1
L Beams with no warping restraint (NWR) Beams with warping restraints (WR)
ly Mb / Mc ly Mb / Mc ly Mb / Mc ly Mb / Mc
P, point load; D, uniformly distributed load; M, uniform bending moment; m, load at centroid; t, load at
top flange; Mc, plastic bending moment resistance; Mb, buckling resistance without warping restraint; Mb,
buckling resistance with warping restraints.
Y.Z. Yin, Y.C. Wang / Journal of Constructional Steel Research 59 (2003) 10091033 1019
In all cases, the beam is simply supported at both ends and there is no restraint to
its thermal expansion.
In this study, beams under point loads or uniformly distributed loads are analysed.
The boundary conditions allow the end sections of a beam to rotate as a plane about
the major axis, but prevent rotation about the minor axis and twist. Lateral restraints
are applied at the web of a beam along its length. Table 3 lists the failure tempera-
tures of laterally restrained beams under loads of 0.8 Mc.
As expected, beam failure temperatures for different loading cases and spans are
almost identical. They give the maximum lower flange temperatures at which the
plastic bending moment capacity of a beam cross-section is reduced to 0.8 Mc.
Compare results of different temperature profiles, Table 3 shows that the maximum
failure temperatures for non-uniform heating are much higher than for uniform heat-
ing. Since profiles N1 and N2 have similar average temperatures, their failure tem-
peratures are very close. Profile N3 has a higher average temperature than the other
two and the resulting failure temperatures are lower. Therefore, non-uniform tem-
Table 3
Maximum failure temperatures of laterally restrained beams under 0.8Mc loading
Load type Span L (m) Failure temperature (oC) for temperature profile
U N1 N-2 N3
perature distribution has some effect on plastic bending moment capacities of beams.
Nevertheless, the difference between non-uniform heating and uniform heating is
higher than between different profiles of non-uniform heating.
Fig. 7 compares the load ratiofailure temperature curves from numerical simula-
tions and BS 5950 Part 8 calculations for an 8-m beam under point loads at centroid.
Since the beams have reached their plastic bending moment capacity at failure, Fig.
7a shows that the ABAQUS results are almost identical with those obtained using
the BS 5950 bending moment capacity method. Fig. 7b compares results of numerical
simulations and BS 5950 limiting temperature calculations. Clearly, using only one
curve to represent different non-uniform heating profiles is not accurate. Neverthe-
less, the BS 5950 design curve is in between results of different non-uniform tem-
Fig. 7. Temperature-load ratio curves for a 8-m beam under point loads at centroid.
Y.Z. Yin, Y.C. Wang / Journal of Constructional Steel Research 59 (2003) 10091033 1021
perature distributions. In addition, within the practical load ratio range of 0.30.7,
the BS 5950 curve is close to different numerical result curves which have been
generated using a wide range of non-uniform temperature profiles. Overall, the BS
5950 limiting temperature design curve may be considered to be acceptable for prac-
tical prediction of the plastic bending moment capacity of a beam with non-uniform
temperature distributions.
P-m 3 0.73 0.4 551 691 0.81 0.36 586 791 795 708
0.8 236 462 0.72 342 507 592 526
8 0.28 0.4 569 850 0.33 0.34 606 971 954 851
0.8 285 573 0.68 435 648 822 663
P-t 3 0.54 0.4 556 736 0.64 0.34 621 851 861 754
0.8 236 519 0.68 456 660 671 596
8 0.18 0.4 597 911 0.22 0.33 652 1000 1000 911
0.8 424 599 0.65 537 821 904 801
D-m 3 0.67 0.4 561 701 0.77 0.35 594 804 808 721
0.8 274 518 0.70 383 537 615 552
8 0.24 0.4 585 867 0.28 0.35 626 992 973 867
0.8 367 641 0.69 505 708 844 724
D-t 3 0.54 0.4 552 700 0.64 0.34 615 841 852 746
0.8 220 507 0.68 440 585 659 584
8 0.18 0.4 592 902 0.22 0.33 647 1000 1000 902
0.8 400 683 0.64 529 795 894 794
M1 3 0.64 0.4 553 746 0.73 0.35 586 847 851 741
0.8 246 516 0.70 353 508 628 548
8 0.22 0.4 584 875 0.26 0.34 630 994 972 871
0.8 360 622 0.68 509 676 847 686
M2 3 0.64 0.4 553 604 0.73 0.35 586 660 681 640
0.8 246 341 0.70 353 421 504 435
8 0.22 0.4 584 664 0.26 0.34 630 716 826 715
0.8 360 511 0.68 509 527 588 570
Y.Z. Yin, Y.C. Wang / Journal of Constructional Steel Research 59 (2003) 10091033
Y.Z. Yin, Y.C. Wang / Journal of Constructional Steel Research 59 (2003) 10091033 1023
Fig. 9. Temperaturedeflection curves of an 8-m beam under point loads at centroid, with warping
restraints.
average temperatures, are very close. At a high load ratio (0.8 for no warping restraint
and about 0.68 for warping restraints), the applied stress is high so that the lower
flange stiffness almost diminishes with the top flange contributing most to the beam
stiffness. Therefore, the beam failure temperatures are strongly related to the top
flange temperature. As a result, the failure temperatures of profiles N1 and N3 are
close and the failure temperatures of profile N2 are much higher. Fig. 10 shows the
effects of load ratio on beam stiffness.
1024 Y.Z. Yin, Y.C. Wang / Journal of Constructional Steel Research 59 (2003) 10091033
Fig. 11. Comparison of failure temperatures between BS 5950 and ABAQUS for the 3-m beam with
warping restraints. (a) Load ratio = 0.34; (b) load ratio = 0.68.
flange point loads and negative uniform bending moment, respectively. Results for
beams with or without warping restraint show similar trend so that only results for
beams with warping restraints are presented. BS-U and BS-NU are for BS 5950 Part
8 calculations under uniform and non-uniform temperature distribution, respectively.
From these figures, it is observed that ABAQUS predicts lower failure tempera-
tures for beams under uniform temperature distribution than does BS 5950 Part 8.
This has been pointed out by Bailey et al. [2] and is a result of the design method
not being able to consider the non-linear se relationships of steel at elevated tem-
peratures. However, for beams with non-uniform temperature distributions, the
numerically predicted failure temperatures are generally much higher than the limit-
ing temperatures calculated by BS 5950 Part 8. In particular, when temperature dif-
ferences between the top and bottom flanges are high, the numerically predicted
1026 Y.Z. Yin, Y.C. Wang / Journal of Constructional Steel Research 59 (2003) 10091033
Fig. 12. Comparison of failure temperatures between BS 5950 and ABAQUS for the 8-m beam with
warping restraints.
beam failure temperatures are well above the design limiting temperatures given by
BS 5950 Part 8.
ECw EIz EIz GJ
p4 (ECw)B(EIz)B p2 (EI ) (GJ)B
(ECw)B (EIz)B (EIz)B (GJ)B z B
L4 L2
where (ECw)B, (EIz)B and (GJ)B are cross-sectional stiffness of the steel beam at a
uniform temperature of the lower flange.
The warping rigidity (ECw) of a cross-section is mainly affected by the flexural
rigidity (EIz) about the minor axis. The torsional rigidity (GJ) changes according to
the average stiffness of the cross-section, which is similar to that of flexural rigidity
about the minor axis. For simplicity, it is acceptable to assume that all cross-sectional
properties change at the same rate. Thus,
ECw EIz GJ (E)T (E)B
(3)
(ECw)B (EIz)B (GJ)B 2(E)B
where (E)T and (E)B are the reduced values of Youngs modulus of steel of the top
and bottom flanges, respectively.
1028 Y.Z. Yin, Y.C. Wang / Journal of Constructional Steel Research 59 (2003) 10091033
(E)T (E)B
Mb,nonuniform Mb,uniform . (4)
2(E)B
M .
Mc
lLT (5)
b
0.7[(E)T (E)B]
lLT,nonuniform lLT,uniform / 2(E)B
(6)
where the coefficient 0.7 relates the plastic bending moment capacity under non-
uniform temperature distributions to that under a uniform temperature distribution.
5.3. Comparisons with ENV 1993-1-2 with modified lateral torsional buckling
slenderness
The beam slenderness modification factors described in the previous section should
be applied in fire. Since BS 5950 Part 8 does not use beam slenderness for fire design,
comparisons will be made with ENV 1993-1-2 calculations using the modified beam
slenderness. The results are shown in Figs. 13 and 14. In order to make the figures
clear, only beams with warping restraints are included. Results for beams without
warping restraint are similar. In these figures, MEC3 refers to calculations that fol-
lows the ENV procedure, but using the modified beam slenderness and in which the
plastic bending moment capacity is obtained by dividing that of uniform temperature
distribution by 0.7. In results indicated by MEC3Mp, the beams plastic bending
moment is calculated using the full bending moment capacity method.
It can be seen that using the aforementioned modifications to the beam slenderness
can drastically improve accuracy of the prediction method in ENV 1993-1-2. The
improvement is especially impressive for slender beams or steep temperature gradi-
ents in the cross-section (e.g. temperature profile N2 in this study).
Fig. 15 shows that if the real plastic bending moment capacity of the cross-section
is used (MEC3Mp), the agreement between ENV 1993-1-2 calculations and numeri-
cal results is better than using the simplification method of calculating the beams
plastic bending moment capacity (MEC3). Nevertheless, for practical design, the
simplified method may still be used.
In all cases, using the ENV calculation method with modified beam lateral tor-
sional buckling slenderness still gives results on the safe side.
Appendix A provides an example of using the modified ENV method.
Y.Z. Yin, Y.C. Wang / Journal of Constructional Steel Research 59 (2003) 10091033 1029
Fig. 13. Comparison of failure temperatures between ENV 1993-1-2 and ABAQUS for the 3-m beam
with warping restraints.
6. Conclusions
This paper has presented the results of a numerical study, using ABAQUS, of the
lateral torsional buckling behaviour and strengths of steel beams with non-uniform
temperature distributions. The results of these numerical simulations are also com-
pared with calculations using design methods in BS 5950 Part 8 and ENV 1993-1-
2. From these results, the following conclusions may be drawn:
1. At the same load ratio, defined as the ratio of the applied load on a steel beam
in fire to the capacity of the beam at ambient temperature, the critical temperature
of the beam increases with decreasing ratio of the beams elastic critical moment
resistance to its plastic bending moment capacity.
2. Under non-uniform temperature distributions, the critical temperature of a beam,
1030 Y.Z. Yin, Y.C. Wang / Journal of Constructional Steel Research 59 (2003) 10091033
Fig. 14. Comparison of failure temperatures between ENV 1993-1-2 and ABAQUS for the 8-m beam
with warping restraints.
taken as the maximum temperature of the flanges, will depend on both the average
temperature and the temperature gradient of the beam. At low load ratios, the
lateral torsional buckling behaviour of a beam is controlled by the average elastic
modulus of the cross-section so that the average temperature of the cross-section
has noticeable influence on the beams behaviour. At high load ratios, temperature
of the less heated flange dominates the beam behaviour.
3. For uniformly heated beams, BS 5950 Part 8 overpredicts and ENV 1993-1-2
underpredicts beam failure temperatures. However, both methods predict much
lower beam failure temperatures for beams with non-uniform temperature distri-
butions.
4. Non-uniform temperature distributions will result in two important changes to the
slenderness of a steel beam. Firstly, a symmetrical section at ambient temperature
will become effectively unsymmetrical at non-uniform elevated temperatures due
Y.Z. Yin, Y.C. Wang / Journal of Constructional Steel Research 59 (2003) 10091033 1031
Fig. 15. Comparison of failure temperatures between ENV 1993-1-2 simplified and full bending moment
capacity for the 8-m beam with warping restraints.
Calculate the limiting temperature of an 8-m UB 457 152 60 beam with warping
restraints under sagging uniform bending moment under temperature profile N1, with
a load ratio of 0.68.
At ambient temperature
L 0.85 8000 l 211
l 211, 5.6v 0.7849
iz 32.3 c 37.5
uvl uvl
lLT 1.656cLT 0.246
l1 93.3235 / fy
Mb cLTWfy
At elevated temperatures
Using shape factor N 1.0v 0.6575lLT 1.387
lLTkyB / kEB
lLT,f
0.7(kET kEB)
2(kEB)
Mb,f cLT,fWfykyB / 0.7
L.R. cLT,fkyB / (0.7cLT)
At a lower flange maximum temperature of 600 C, the upper flange temperature
is 300 C
kET 0.80, kEB 0.31, kyB 0.47
lLT,f 1.526cLT,f 0.278 .
L.R. 0.76
At a lower flange maximum temperature of 700 C, the upper flange temperature
is 350 C
kET 0.75, kEB 0.13, kyB 0.23
lLT,f 1.199cLT,f 0.384 .
L.R. 0.51
If L.R. = 0.68 Limiting temperature T = 632 oC.
References
[1] ABAQUS (2001), Standard users manual. Hibbitt, Karlsson and Sorensen, Inc.
[2] Bailey CG, Burgess IW, Plank RJ. The lateral-torsional buckling of unrestrained steel beams in fires.
Journal of Constructional Steel Research 1996;36(2):10119.
Y.Z. Yin, Y.C. Wang / Journal of Constructional Steel Research 59 (2003) 10091033 1033
[3] BSI, BS 5950, Structural use of steelwork in building. Part 1, Code of practice for design in simple
and continuous construction. British Standards Institution, London, 1985.
[4] BSI, BS 5950, Structural use of steelwork in building. Part 8, Code of practice for fire resistant
design. British Standards Institution, London, 1990.
[6] European Committee for Standardisation (CEN). ENV 1993-1-2, Eurocode 3 Design of steel struc-
tures. Part 1.2 General rules/structural fire design. British Standards Institution, London, 2001.
[7] Franssen JM, Kodur VKR, Mason J. Users manual for SAFIR2001. A computer program for analy-
sis of structures submitted to fire. University of Liege, Belgium, 2000.
[8] Galambos TV. Guide to stability design criteria for metal structures. John Wiley & Sons, 1998.
[9] Ojalva M, Chambers RS. Effect of warping restraints on I-beam buckling. Journal of the Structural
Division, ASCE 1997;103(ST12):235160.
[10] Piloto PAG, Vila Real PMM. Lateral torsional buckling of steel I-beams in case of fire experi-
mental evaluation. In: Proceedings of the First International Workshop on Structures in Fire, Copen-
hagen, Denmark. 2000. p. 99105.
[11] Salmon CG, Johnson JE. Steel structures: design and behaviour. Harper & Row, 1980 2nd ed.
[13] Vila Real PMM, Franssen JM. Lateral torsional buckling of steel I-beams in case of fire numerical
modeling. In: Proceedings of the First International Workshop on Structures in Fire, Copenhagen,
Denmark. 2000. p. 7193.