You are on page 1of 5

Moment-gradient factor for lateral torsion-flexure buckling of steel I-

beams
E.Y. Sayed-Ahmed
Ain Shams University, Structural Engineering Dept., Cairo, Egypt. (on leave to University of Qatar, Civil
Engineering Dept., P.O Box 2713, Doha, Qatar)

ABSTRACT: Lateral torsion-flexure buckling of thin-walled steel I-beams subject to flexure is one of the most impor-
tant aspects which should be considered in design. Codes of practice deals with this matter by using the critical mo-
ment of a simply supported beam subject to two equal and opposite end moments and relate it to the critical moment
of other loading cases using a moment-gradient factor which is also referred to as “the equivalent moment factor”. A
numerical model based on the finite element technique is adopted to investigate the values of the moment-gradient fac-
tor for different loading configuration. The model is extended to investigate the effect of the load eccentricity, meas-
ured from the shear centre of the beam’s cross-section, on the critical moment which initiates lateral torsion-flexure
buckling. Equations which consider the effect of load eccentricity from the shear centre are introduced and checked
using the numerical analysis results.

with respect to the shear centre of the cross-section.


The latter (Fig. 2) is ignored by codes of practice al-
1 INTRODUTION though loads are commonly applied either above or be-
low the shear centre of the cross section.
When a thin-walled steel I-beam is subjected to flexure a) P b)
P
about its axis of greatest flexural rigidity with insufficient
w w
lateral bracing, out-of-plane bending and twisting may Shear centre
occur when the applied load reaches its critical value h
(Fig. 1). At this critical load, the in-plane bending de- h
γ γ P
formation ceases to be a stable configuration for the P
beam and lateral torsion-flexure buckling occurs
(Galambos 1998, Chen & Lui 1987). Figure 2. Location of the load with respect to the shear centre of
y the cross section: a) above, and b) below the shear centre.
Mo Mo For design purposes, codes of practice allow design-
ers to evaluate the critical moment causing lateral buck-
ling for a simply supported beam subjected to uniform
z
C.L. bending and modify it using a moment-gradient factor
which depends on the type of loading.
A numerical analysis is performed on steel I-beams
x
subjected to different types of loads to evaluate the criti-
cal moment causing lateral buckling and the moment-
gradient factor for each load case. The effect of the load
w location with respect to the shear centre on the moment-
y gradient factor is scrutinized.
γ z
2 LATERAL STABILITY OF STEEL I-BEAMS

Lateral torsion-flexure buckling has been approached by


Figure 1. Lateral torsion-flexure buckling of an I-beam. assuming that the beam is geometrically perfect and un-
The critical load depends on the laterally unbraced dergoes a small deflection when subject to in-plane
length, the mechanical properties of the cross section, bending and its cross-section does not distort during
the end support, the loading type, and the load location buckling. If two equal and opposite torques are applied
to the ends of a simply supported thin walled I-beam un- y Warping is prevented: applied torque is
restrained against warping, uniform torsion is generated resisted by the warping restraint torsion
along the beam’s length accompanied by warping of the
cross section (Fig. 3). This torsion (St. Venant torsion
Τsv) is solely resisted by a shear stress (St. Venant shear z
stress) developed in the cross section which is given by:
dγ n
Tsv = G J ⋅ J = Σ bi ⋅ t i3 (1) T
dx i =1 x

where G is the shear modulus, J is the torsional constant


and γ is the angle of twist. Free to warp: Applied torque is
resisted by St. Venant torsion

Figure 4. Beams restraint against warping at one end.


T T Combining the last two equations which describe the
out-of-plane equilibrium results-in:
d 4γ d 2γ M o
2
ECw − GJ − γ =0 (5)
Figure 3. Simply supported beam unrestrained against warping dx 4 dx 2 EI y
and subjected to uniform torsion.
When the section has one end restrained against Solution of Equation 5 yields the critical moment for a
warping (Fig. 4), the applied torque at this end would be simply supported beam subjected to two equal and op-
resisted by a “warping restraint torsion” given by: posite end moments which is given by:

d 3γ I f h2 π π ECw
Tw = − E C w CW = (2) M ocr = EI y GJ ( 1 + Wr2 ) WR = ( ) (6)
3 L L GJ
dx 2
where E is the Young’s modulus, Cw is the warping con- where WR represents the warping restraint contribution
stant of the section, h is the distance between the to the beam resistance.
flanges’ centroid, and If is the second moment of area of For beams subjected to other types of loading, the
the compression flange. Generally, the applied twisting effect of the moment gradient on the critical moment can
moment is resisted by both the St. Venant torsion and be accounted for by the use of a moment-gradient factor
the warping restraint torsion where: Cb (Salvadori 1955). Thus:

dγ d 3γ Cbπ πE 2
T = Tsv + Tw = G J ⋅ − E Cw 3 (3) M cr = Cb M ocr = EI y GJ + ( ) I y Cw (7)
dx dx L L
The first term of Equation 3 represents the resistance of For beams subjected to unequal end moments (MA and
the section to twisting while the second is the resistance MB), the moment-gradient factor may be given by
of the section to warping. (Galambos 1998, Chen & Lui 1987):
For a simply supported I-beam subject to uniform MA M
bending (Fig. 1), the conditions of equilibrium are: Cb = 1.75 + 1.05 + 0 .3 ( A ) 2 ≤ 2 .3 (8)
MB MB
d 2v
EI z 2 + M o = 0 where MA is the smaller moment and the ratio MA/MB is
dx
positive for beams bent in double curvature and negative
d 2w for single curvature. CAN/CSA-S16.1-94 (CISC
EI y 2 + γ M o = 0 (4)
dx 2000) adopted Equation 8 with a change in its limiting
value from 2.3 to 2.5. The same specification uses Cb of
dγ d 3γ dw
GJ − ECw 3 − Mo = 0 1.0 when the bending moment between the end supports
dx dx dx is greater than the end moment. The AISC-LRFD
where Mo is the applied end moment, Iz and Iy are the (AISC 2000) defines the following general equation for
cross-section’s second moment of area about the z-z the moment-gradient factor:
and the y-y axes respectively, γ is the angle of twist and
w is the lateral displacement of the beam.
12.5 ⋅ M max moment through its span is 294.5 kN.m. The dimensions
Cb = (9) and the span of the analyzed beams are chosen such that
3 ⋅ M 1 + 4 ⋅ M 2 + 3 ⋅ M 3 + 2.5 ⋅ M max
critical moment will be reached before yielding initiates in
where M1, M2, M3 are the absolute values of the mo- any part of the beam’s cross section: based on Fy = 300
ments at the quarter point, midpoint and three-quarter MPa, the yield moment My of the considered section is
point of the beam, respectively and Mmax is the maximum 1006.5 kN.m and the plastic moment Mp is 1110.6
moment acting on the beam. kN.m (Fig. 5).
In the previous Cb equations, the load was assumed The finite element package ANSYS 5.4 is used for
to act along the shear centre of the beam’s cross sec- preprocessing the model, solution of the equations and
tion. Thus, the effect of the applied load location with re- post-processing the results. Figure 6 shows a typical fi-
spect to the shear centre is ignored. nite element mesh adopted in the analysis.

3 THE FINITE ELEMENT MODEL

Numerical analysis based on the finite element technique


is used to evaluate the critical moment of I-beams sub-
jected to different types of loading. The results are re-
lated to the critical moment of a simply supported beam
subject to a constant moment to estimate the value of Cb
for each analyzed case of loading. The analysis also in-
vestigates the effect of the load location with respect to
the shear centre of the beam’s cross-section on the criti-
cal moment. Figure 6. Typical finite element mesh adopted in the analysis.

3.1 Model description


3.2 Beams subjected to end moments
Shell elements with 8 nodes and 6 degrees of freedom
per node are used to model the I-beams. The beams are The first group analyzed is composed of beams sub-
considered to be simply supported in flexure and in tor- jected to end moments (Fig. 7). Table 2 lists a compari-
sion: at the beam’s ends, rotation and warping about the son between the maximum vertical deflections resulting
weak axis are free while rotation about the centroidal from the numerical analysis and their theoretical values.
axis is restrained. The beams have 12.0 m span and the A lateral torsion-flexure buckling mode for Beam B1 is
cross-section shown in Figure 5. The mechanical prop- shown in Figure 8.
erties of the cross section considered in the analysis are M M M M/2
listed in Table 1.
Fy =300 MPa Fy =300 MPa 229x20 Beam B1 Beam B2
M
- M M/2
- Beam B3 Beam B4
M M
612
572.8x12 Beam B5
+ Figure 7. First group (Group 1) of the analyzed I-beams.
+
Fy =300 MPa Fy =300 MPa 229x20
Figure 5. Cross section of the analyzed beam and stress distri-
bution through the section at the yield and plastic moments.
M M
Table 1. Properties of the I-section adopted in the analysis . Beam B1
A Ix Iy rx ry Sx Sy
Prop-
(103) (106) (106) mm (106) (103) (103)
erty
mm2 mm4 mm4 mm mm3 mm3
Value 16.03 993.7 39.2 249 49.5 3355 343
Zx Zy Cw J E G Fy
Prop-
(106) (106) (109) (103) (GPa) (GPa) (MPa)
erty
mm3 mm3 mm6 mm3
Value 3703 513.9 3448 1482 200 76.9 300 Figure 8. A typical lateral buckling mode (Beam B1).
The moment-gradient factor is evaluated for each
Using Equation 6, the critical moment Mocr of the load case by relating the critical moment resulting from
analyzed steel beam when it is subjected to a constant
the finite element analysis to the critical moment calcu- P P P
lated using Equation 6. The values of Cb obtained based
on the finite element results are listed in Table 2 and Beam B6 Beam B7 P
plotted in Figure 9. These values are compared to the
values of Cb which are calculated using Equations 8 and P
9. It is evident from Figure 9 that Equations 8 and 9 re- Beam B8
sults-in an excellent estimate for Cb up to a ratio of
MA/MB of +0.5 (beams bent in double curvature) where P
they become conservative compared to the results of the Figure 10. Second group (Group 2) of the analyzed I-beams.
finite element analysis. Imposing the CAN/CSA-S16.1-
94 upper limit on Equation 8 produces better results Table 3. Results of the numerical analysis for Group 1.
which almost match the numerical analysis results. δmax-FE δtheo Mcr-FE Cb
Beam Cb-FE
(mm) (mm) kN.m CSA Eq. 9
Cb-F.E . B6 18.802 288.1 0.978
2.5 B7 18.79 18.11 385.44 1.309 1.0 1.32
Cb-Eq. 8 & CISC limits Cb-Eq. 8
B8 18.802 530.49 1.801

2.0 MA To include the effect of the load location with respect


MB Cb-Eq. 9 - AISC to the shear centre, the following definition for Cb of
Cb

beams subjected to concentrated loads is proposed


1.5 (Nethercot & Rocky 1971, Nethercot 1983):
 A / B load acts at the top flange
1.0 
Cb =  A load acts at the shear centre (10a)
-1.00 -0.50 0.00 0.50 1.00
MA/MB  AB load acts at the bottom flange

Figure 9. The moment-gradient factor for Group 1 beams.
A = 1 .35 ; B = 1 + 0 .649 W − 0 .180W 2
Table 2. Results of the numerical analysis for Group 1. (10b)
Cb and W = ( π / L ) ( EC w ) /( GJ )
δmax-FE δtheo Mcr-FE
Beam Cb-FE Eq. 8 Eq. 9
(mm) (mm) kN.m The values of the moment-gradient factor determined
CSA AISC
B1 12.303 12.283 287.7 0.98 1.0 1.0 using the numerical analysis and Equation 10 are plotted
B2 9.255 9.212 378.7 1.286 1.3 1.25 in Figure 11 together with the values adopted by the
B3 6.312 6.305 520.5 1.767 1.75 1.67 CAN/CSA-S16.1-94 (CISC 2000) and the AISC-
B4 3.639 3.617 707.6 2.403 2.3 2.17 LRFD (AISC 2000) provisions. It is evident from this
B5 1.565 1.573 770.4 2.616 2.3 2.27 figure that Equation 10 significantly matches the results
of the numerical analysis.

3.3 Beams subjected to central concentrated loads 1.8


P
Cb-Eq. 10
Beams subjected to central concentrated loads repre- 1.6 Cb-F.E.
sent the second group analyzed using the numerical 1.4
model. The load is applied on the top flange, at the shear Cb - Eq. 9 & AISC
Cb

centre of the cross section and on the bottom flange of 1.2


the beam (Fig. 10). Table 3 lists a comparison between Cb - CISC (2000)
the maximum vertical deflections resulting from the nu- 1.0
merical analysis and their theoretical values. In this table, Below the shear centre Above the shear centre
the moment-gradient factor is evaluated by relating the 0.8
critical moment resulting from the numerical analysis to -0.50 0.00 0.50
the critical moment of a beam subjected to constant hP / h
bending (Eq. 6). Figure 11. The moment-gradient factor for Group 2 beams (h p is
The values of Cb obtained numerically are compared the load distance measured from the shear centre and h is the
to those calculated using Equation 9 (AISC 2000). It is cross section height).
evident from Table 3 that Equation 9 overestimates the
value of Cb when the load acts above the shear centre
(Beam B6) and underestimates it when the load acts be- 3.4 Beams subjected to a uniformly distributed load
low the shear centre (Beam B8). Thus, the current
method of estimating the critical moment is significantly The third group analyzed using the numerical model is
unsafe when the loads act above the shear centre and composed of beams subjected to a uniformly distributed
over-conservative for loads acting below the shear cen- load. The load is applied on the top flange, at the shear
tre. centre of the cross section and on the bottom flange
(Fig. 12). Table 4 lists a comparison between the maxi- 1.5
mum vertical deflections resulting from the analysis and w
their theoretical values. 1.4 Cb-Eq. 11
w w w 1.3 Cb-F.E.
1.2
Beam B9 Beam B10 w

Cb
1.1 Cb - Eq. 9 & AISC
w 1.0 Cb - CISC (2000)
0.9
Beam B11 0.8 Below the shear centre Above the shear centre
w
0.7
Figure 12. Third group (Group 3) of the analyzed beams. -0.50 0.00 0.50
hP / h
Table 4. Results of the numerical analysis for Group 3.
δmax-FE δtheo Mcr-FE Cb Figure 13. The moment-gradient factor for Group 2 beams (h p is
Beam Cb-FE the load distance measured from the shear centre and h is the
(mm) (mm) kN.m CSA Eq. 9
cross section height).
B9 22.947 253.92 0.86 A finite element analysis was performed to scrutinize
B10 22947 22.23 326.9 1.11 1.0 1.136
the values of the moment-gradient factor for different
B11 22.945 417.75 1.42
cases of loading. The model was used to establish the
The values of Cb obtained numerically are listed in effect of the load location with respect to the cross-
Table 4 and compared to those calculated using Equa- section’s shear centre on the critical moment. The analy-
tion 9. It is evident that Equation 9 overestimates the sis revealed that the moment-gradient factor currently
value of Cb when the load acts above the shear centre used for the design of steel I-beams subjected to end
(Beam B9) and underestimates it when the load acts be- moments is conservative and gives acceptable results.
low the shear centre (Beam B11). Thus, the current However, it is recommended to increase the limiting
method of estimating the critical moment is again signifi- value of the commonly used Equation for this loading
cantly unsafe for uniformly distributed load acting above case from 2.3 to 2.5 which matches the CAN/CSA-
the shear centre and conservative for loads acting below S16.1-94 provisions.
the shear centre. On the other hand, for beams subjected to either a
To include the effect of the load location with respect concentrated or uniformly distributed load acting away
to the shear centre, the following definition for Cb of from the section’s shear centre, the currently used mo-
beams subjected to uniform loads is proposed (Nether- ment-gradient factor is inefficient. For loads acting be-
cot & Rocky 1971, Nethercot 1983): low the shear centre, the currently used factor underes-
A / B load acts at the top flange timates the value of the critical moment. For loads acting
 above the shear centre the currently used factor overes-
Cb =  A load acts at the shear centre (11a) timates the critical moment and thus, produces an unsafe
 AB
 load acts at the bottom flange design. Equations expressing the moment-gradient factor
for such cases have been investigated and found to be
competent.
A = 1 .35 ; B = 1 + 0 .535W − 0 .154W 2
(11b)
and W = ( π / L ) ( EC w ) /( GJ )
5 PREFERENCES
The Cb values determined numerically and using Equa-
tion 11 are plotted in Figure 13 together with the values American Institute of Steel Construction (AISC). 2000. Load
adopted by the CAN/CSA-S16.1-94 (CISC 2000) and Resistance Factor Design. 2nd edition. American Insti-
provisions and the AISC-LRFD provisions (AISC tute of Steel Construction, Illinois, USA.
2000). It is evident from this figure that Equation 11 sig- Canadian Institute of Steel Construction (CISC). 2000. Hand-
nificantly matches the results of the finite element analy- book of Steel Construction. 7th edition, Ontario, Canada.
Chen, W.F. & Lui, E. M. 1987. Structural Stability: theory and
sis. Implementation. Elsevier Sc. Publishing, NY, USA.
Galambos, T. V. 1998. A guide to Stability design criteria for
metal structures. John Wiley & Sons Inc., NY. USA.
4 SUMMARY AND CONCLUSIONS Nethercot, D.A. 1983. Elastic lateral buckling of beams. Ch 1 –
Beams and beam-columns, stability and strength, R. Nara-
Lateral torsion-flexure buckling is an important design yanan (ed), Elsevier Sc. Publishers, London, UK.
aspect for I-beams. Codes of practice use the critical Nethercot, D.A. & Rocky, K.C. 1971. A unified approach to the
elastic lateral buckling of beams. The Structural Engineer
moment of a simply supported beam subject to a con- 49(7): 321-330.
stant moment and relate it to the critical moment of other Salvadori, M. G. 1955. Lateral buckling of I-beams. ASCE
cases of loading via a “moment-gradient factor”. Transaction 120: 1165-1177.

You might also like