Lateral-torsional buckling of orthotropic rectangular section beams

Bambang SURYOATMONO
Professor
Department of Civil Engineering, Parahyangan Catholic University
Bandung, Indonesia
Adhijoso TJONDRO
Associate Professor
Department of Civil Engineering, Parahyangan Catholic University
Bandung, Indonesia

Summary
The elastic lateral-torsional buckling of rectangular section beams made of wood under various
loading conditions are investigated in this paper. Three dimensional finite element models
considering the orthotropic behavior of wood are utilized to predict the elastic buckling load. The
effects of depth/width ratios of the cross section, wood species, and cross grain of the beams on the
predicted buckling loads are investigated. For all the cases considered, the critical loads predicted
using finite element models assuming wood as orthotropic material are much smaller than those
predicted assuming wood as isotropic material. The predicted critical loads of beams with cross grain
of 45o are approximately a third of the predicted critical loads of beams without any cross grain. The
finite element results agree very well with the beam tests on several species of the hardwood
category.
1. Introduction
Lateral-torsional buckling (LTB) is a limit state where beam deformation includes in-plane
deformation, out-of-plane deformation, and twisting [1]. Consider, for example, a simply supported
rectangular beam loaded by uniform moment M as seen in Figure 1. The beam is laterally supported
at both ends so that the cross sections at both ends cannot rotate but free to warp. Assuming that the
material is isotropic and linearly elastic, the critical moment, M0cr, that causes lateral-torsional
buckling to occur is
π
M cr = EI y GJ (1)
Lu
where E = modulus of elasticity, Iy =
moment of inertia with respect to the
weak (y) axis, G = shear modulus, and J
= torsional constant, and Lu is the
unbraced length of the beam. For
rectangular sections with depth d and
width b,
1 3
Iy = bd (2)
12
Fig. 1 Lateral Torsional Buckling (LTB) of a rectangular J = k1b 3d (3)
section beam.
where k1 is a cross section constant and
can be obtained from Table 1 [2].

Note that warping, a deformation that occurs in a thin walled open section, is neglected in Eq.
(1).The derivation of Eq. (1) can be found in literature [3].
For nonuniform moment diagram, the critical moment can be obtained by multiplying the critical
moment for uniform moment, Eq. (1), by an LTB modification factor for nonuniform moment Cb
[4]. For a simply supported beam laterally supported at its ends under uniform load and, Cb = 1.14
and if the beam is loaded by concentrated load at midspan, Cb = 1.32 [5]. Therefore, the critical
uniform load for a simply supported beam laterally supported at its ends is
8 π
wcr = 2
(1.14) EI y GJ (4)
Table 1 Constant k1 required for computing Lu Lu
torsional constant J of a rectangular section
[2] and the critical concentrated load at midspan for
a simply supported beam laterally supported at
d/b k1 its ends is
2 0.229
3 0.263 Pcr =
4
(1.32) π EI y GJ (5)
Lu Lu
4 0.281
It should be noted that Eq. (1), (4), and (5) were
5 0.291 derived by assuming that the beam material is
isotropic. In this paper, the applicability of these
equations for orthotropic material is investigated
by using Finite Element Method (FEM).

2. Finite Element Method

If wood is assumed as an orthotropic and elastic material with three mutually perpendicular
material principal axes (longitudinal, radial, and tangential), then the relationship between strain
components and stress components can be expressed as [6]
⎡ 1 μ RL μTL ⎤
⎢ E − − 0 0 0 ⎥
ER ET
⎢ L ⎥
μ LR 1 μTR
⎧ ε L ⎫ ⎢⎢ − − 0 0 0 ⎥ ⎧σ L ⎫
⎥⎪ ⎪
⎪ ε ⎪ ⎢ EL ER ET
⎥ σR
⎪ R ⎪ ⎢− μ LT −
μ RT 1
0 0 0 ⎥⎪ ⎪ (6)
⎪⎪ ε T ⎪⎪ ⎢ E L ER ET ⎥ ⎪⎪⎨ σ T ⎪⎪⎬
⎨ ⎬=⎢ ⎥
⎪γ LR ⎪ ⎢ 0 0 ⎥ ⎪τ LR ⎪
1
0 0 0
⎪γ LT ⎪ ⎢ GLR
⎥ ⎪⎪τ LT ⎪⎪
⎪ ⎪ ⎢ 1
⎩⎪γ RT ⎭⎪ ⎢ 0 0 0 0 0 ⎥ ⎩⎪τ RT ⎭⎪
GLT ⎥
⎢ 1 ⎥
⎢ 0 0 0 0 0 ⎥
⎣⎢ GRT ⎦⎥

where Ei are moduli of elasticity, μij are Poisson’s ratios, and Gij are shear moduli, where i = L, R,
and T. For an isotropic material, the relationship reduces to
⎡ 1 μ μ ⎤
⎢ E − − 0 0 0⎥
E E
⎢ ⎥
⎧ ε L ⎫ ⎢− μ 1
−
μ
0 0 0 ⎥ ⎧σ L ⎫
⎪ε ⎪ ⎢ E E E ⎥ ⎪σ R ⎪
⎪ R⎪ ⎢ μ μ ⎪ ⎪
0 ⎥ ⎪σ ⎪
1 (7)
⎪⎪ ε T ⎪⎪ ⎢− E −
E E
0 0
⎥⎪ T ⎪
⎨ ⎬=⎢ ⎥⎨ ⎬
⎪γ LR ⎪ ⎢ 0 0 ⎥ ⎪τ LR ⎪
1
0 0 0
⎪γ LT ⎪ ⎢ G ⎥ ⎪τ LT ⎪
⎪ ⎪ ⎢ 0 1
0 ⎥ ⎪⎪⎩τ RT ⎪⎪⎭
⎪⎩γ RT ⎪⎭ 0 0 0
⎢ G ⎥
⎢ 1⎥
⎢⎣ 0 0 0 0 0
G ⎥⎦

where G = E/{2(1+μ)}.
In the subsequent analyses using a finite element program (SAP2000 Advanced 11.0.0 ©), the beam
is modelled by using solid (3-D) elements. Boundary conditions of the finite element mesh are the
same as the boundary conditions used in the derivation of Eq.(1), i.e. no out-of-plane deformation
at both beam ends but both ends can rotate in the plane of the beam. By defining buckling analysis
in the program, the critical load along with the buckling modes can be obtained. The first mode is
the one that is associated with the critical (lowest) load. As seen in Fig. 2, the buckling mode shows
in-plane and out-of-plane deformations indicating that it is LTB mode.
Eq. (6) and (7) are incorporated in the
program for orthotropic and isotropic
materials, respectively. The hypothetical
beam is assumed to be made of walnut. The
static bending modulus of elasticity Esb of the
material is 11600 MPa [7]. The longitudinal
modulus of elasticity EL can be taken as ten
percent higher than the static bending
modulus of elasticity [7], i.e. EL = 1.1 x
11600 MPa = 12760 MPa. The other elastic
properties are taken from the literature [7],
namely ER = 1353 MPa, ET = 715 MPa, μLR =
0.495, μLT = 0.632, μRT= 0.718, GLR = 1085
MPa, GLT = 0.791, and GRT = 268 MPa. The
same beam is also analyzed assuming
isotropic behaviour by using the modulus of
Fig. 2 Typical lateral torsional buckling mode for elasticity E = 12760 MPa and Poisson’s ratio
simply supported beam laterally supported at its μ = 0.0525.
ends.

The FEM is used to analyze isotropic and orthotropic beams under three loading conditions, namely
uniform moment, distributed load, and concentrated load.

2.1 Uniform Moment

To investigate if FEM is appropriate in
predicting critical load, consider a simply
supported beam under uniform moment as seen
in Fig. 3. The beam length L = Lu = 800 mm. The
cross section is rectangular with the width b = 40
mm and depth d = 80 mm. Assuming the beam as
isotropic material, it can be seen in Table 2 that
Fig. 3 A simply supported beam under uniform FEM results in a critical moment Mcr = 24.46
bending moment diagram. Both ends of the kNm which is only 0.12% higher than the critical
beam are laterally supported. moment computed using Eq. (1). This result
shows that FEM can be used in buckling analysis
with a very high accuracy.
Table 2 Critical moment Mcr (kNm) computed To investigate if the isotropic equation (Eq. 1)
using isotropic equation (Eq.1), FEM assuming can be used in predicting the critical load of an
isotropic material, and FEM assuming orthotropic material, the same beam is further
orthotropic material. analyzed using FEM assuming orthotropic
material. The beam plane is L-T plane. The
Method Mcr (kNm)
analysis is repeated for several angles of cross
Isotropic Equation (Eq. 1) 24.43 grain, namely 0o, 15o, 30o, and 45o. Cross grain is
FEM, isotropic 24.46 defined as the angle between the beam
FEM, orthotropic, cross grain 0o 9.07 longitudinal axis and the material L axis. As seen
FEM, orthotropic, cross grain 15o 7.51 in Table 2 and Fig. 4, the critical moment for
FEM, orthotropic, cross grain 30o 4.88 orthotropic material is much lower than the
FEM, orthotropic, cross grain 45o 3.17 critical moment computed using Eq. (1). The
difference is higher for greater cross grain angle.
These results show that it is unsafe to use the
isotropic equation (Eq. 1) for orthotropic
material such as wood.
Fig. 4 The effect of cross grain on the critical end moments Mcr for a simply supported beam under
uniform bending moment diagram.

2.2 Distributed Load

The same beam as described before is now
loaded by uniformly distributed load w as
seen in Fig. 5. As before, the beam is
analyzed assuming as isotropic material
and orthotropic material with several
angles of cross grain. As seen in Table 3
and Fig. 6, the critical load wcr obtained
from FEM is only 5.54% lower than the
Fig. 5 A simply supported beam under uniform load. critical load computed using Eq. 4. This
Both ends of the beam are laterally supported. result shows that, once again, FEM can be
used in buckling analysis with a very high
accuracy.
Table 3 Critical load wcr (kN/m) computed using If the material is orthotropic, FEM shows
isotropic equation (Eq.4), FEM assuming isotropic that the critical load wcr is much lower than
material, and FEM assuming orthotropic material. the critical load for isotropic material. For
an orthotropic beam with cross grain of
Method wcr (kN/m)
45o, the critical load is only approximately
Isotropic Equation (Eq. 4) 348.09 12% of the critical load of isotropic beam.
FEM, isotropic 328.80 The conclusion for this loading case is the
FEM, orthotropic, cross grain 0o 109.30 same as in the case of uniform moment, i.e.
FEM, orthotropic, cross grain 15o 90.88 it is unsafe to use the isotropic equation
FEM, orthotropic, cross grain 30o 60.84 (Eq. 1) for orthotropic material such as
FEM, orthotropic, cross grain 45o 40.97 wood.
Fig. 6 The effect of cross grain on the critical load wcr for a simply supported beam under
uniformly distributed load.

2.3 Concentrated Load

To investigate the effect of cross section
aspect ratio, d/b, on the critical buckling load,
a hypothetical simply supported walnut beam
under a concentrated load P as seen in Fig. 7
is analyzed using FEM. The material
properties are the same as described before.
The beam length and width are still the same
as before, i.e. L = 800 mm and b = 40 mm.
Fig. 7 A simply supported beam loaded by The depth of the cross section varies to give
concentrated load at midspan. Both ends of the d/b ratios of 2, 3, 4, and 5.
beam are laterally supported.
Table 4 Comparison of critical loads Pcr (kN) As seen in Table 4 and Fig. 8 the difference
computed using isotropic equation (Eq. 5), FEM between FEM results for isotropic material
assuming isotropic material, and FEM assuming and Eq. 5 are quite small. The largest
orthotropic material. difference is 17.98% (for d/b = 5).
d/ Isotropic FEM, isotropic FEM, As in the loading cases analyzed before, the
b Equation (Eq. 5) orthotropic critical load obtained using FEM for
orthotropic material is much smaller than that
2 161.72 151.92 47.47
for isotropic material. The difference between
3 259.17 235.55 68.64
isotropic and orthotropic critical loads is
4 357.18 313.35 85.49
larger as the ratio d/b increases. This is due to
5 470.46 385.88 98.12
the fact that as d/b increases, the beam
behaves more like a deep beam and the
anisotropic nature of the material becomes
more pronounced.
Fig. 8 Critical load Pcr (kN) for various cross section aspect ratio d/b. The width b is kept constant
while depth d varies.

3. Static Bending Tests

To verify the results of FEM, static bending tests on simply supported beams were performed. The
experimental setup is shown in Fig. 9. As seen in the figure, both ends of the beam are laterally
supported. The loading given by the Universal Testing Machine (UTM) was displacement
controlled with a displacement rate of 2 mm/minute. A dial gauge was attached at midspan to
monitor the lateral deformation. Three species in hardwood category grown in Indonesia were
tested, namely albasia (albizzia falcata, batai), meranti = shorea spec. div.), and bangkirai (hopea
spec. div.). For each species, three specimens were tested. The specific gravity of each species were
0.90, 0.45, and 0.38 for bangkirai, meranti, and albasia, respectively. The moisture content of all
specimens were approximately 15%. The dimensions of all specimens were as follows: total length
Ltotal = 1260 mm, beam span L = 1160 mm, width b = 20 mm, and depth d = 100 mm.
By using the computer attached to the
UTM, the load versus midspan deflection
for each test was recorded. The typical
load - midspan deflection curve is shown
in Fig. 10. The static bending modulus of
elasticity Esb can be computed using the
elementary equation
PL3
E sb = (8)
48ΔI x
where load P and midspan deflection Δ are
the values during the linearly elastic stage
of the curve. Static bending moduli of
elasticity for each species computed using
Eq. (8) are 12233 MPa, 5026 MPa, and
3091 MPa, for bangkirai, meranti, and
Fig.9 Test setup for simply supported beam under a albasia, respectively.
concentrated load at midspan.
Fig. 10 Typical load (N) – deflection (mm) curve obtained from static bending test.
The critical load was taken as the load at which the beam started to deflect laterally as indicated by
the dial gauge. The results for each species are shown in Table 5. As expected, the critical load Pcr
for the species with the highest specific gravity is the largest. These experimental results are
compared with the FEM results.
Table 5 Comparison of critical loads Pcr (N) To perform buckling analysis using FEM,
obtained from FEM and static bending test. material properties for each species are needed.
As before, longitudinal modulus of elasticity EL
Species FEM, Static Difference
= 1.1Esb. Since there is no data of all other
orthotropic bending test
orthotropic elastic properties for the three
Bangkirai 2529.9 3933.0 -35.7% species tested, the properties were taken as
shown in Table 6. The critical buckling loads Pcr
Meranti 2160.0 1833.0 17.8%
obtained from FEM are shown in Table 5. It is
Albasia 1327.1 1317.0 0.8% interesting to note that although the elastic
properties in FEM are approximation, the
difference between experimental result and FEM
result is relatively small, especially for albasia.
Table 6 Elastic properties used in finite element analyses of three species, namely bangkirai,
meranti, and albasia.
EL (MPa) ER (MPa) ET (MPa) μLR μLT μRT GLT (MPa) GLR (MPa) GRT (MPa)
Bangkirai 13457 2072 1103 0.35 0.45 0.56 1090 1198 283
Meranti 5529 851 453 0.35 0.45 0.56 448 492 116
Albasia 3400 524 279 0.35 0.45 0.56 275 303 71

4. Discussion
As shown in the analyses of beams subject to various loading conditions using FEM, orthotropic
material always results in a lower critical load than isotropic material. The difference between the
two materials becomes larger if the beam has cross grain. While investigating various angles of
cross grain using real beam specimen is difficult, FEM is a versatile tool to perform such task.
Investigation on various depth/width ratio of the beam cross section shows that as the ratio
increases, the difference between isotropic and orthotropic materials become larger.

5. Conclusions
Since orthotropic assumption always results in lower critical load for the loading cases considered
in this paper, then isotropic equations for LTB should not be used for orthotropic material such as
wood. One possible approach is to use reduction factors in the isotropic equations. Such factor
should depend on beam dimensions, specific gravity, cross grain, and perhaps all other possible
factors that have not been considered in this paper.

6. Acknowledgements
The authors gratefully acknowledge financial support of Parahyangan Catholic University,
Indonesia. The authors also gratefully acknowledge the assistance of staffs at the Structural
Laboratory at the university.

7. References
[1] American Wood Council (AWC), Designing for lateral-torsional stability in wood members,
Technical Report 14, Washington, DC, American Forest & Paper Association (AF&PA),
2003, p. 1.
[2] Timoshenko, SP and Goodier, Theory of Elasticity, McGrawHill, New York, 1970, p. 312.
[3] Simitses G., Hodges, D.H., Fundamentals of Structural Stability, Butterworth-Heinemann,
Burlington, 2006, p. 253.
[4] American Institute of Steel Construction (AISC). Specification for structural buildings. AISC,
Chicago, 2005, p. 46.
[5] Segui, WT, LRFD Steel Design. 3rd Edition. Thomson Brooks/Cole, 2003, pp. 155.
[6] Bodig, J and Jayne, Mechanics of Wood and Wood Composites, Krieger Publishing Company,
Malabar, 1993, p. 98.
[7] Forest Product Society, Wood Handbook, Wood as an Engineering Material,. Madison, 1999,
pp. 4-2, 4-3.