Professional Documents
Culture Documents
Materials
0. A. Bauchau Beam theory plays an important role in structural analysis. The basic assumption is
that initially plane sections remain plane after deformation, neglecting out-of-plane
Assistant Professor,
Department of Mechanical Engineering,
warpings. Predictions based on these assumptions are accurate for slender, solid,
Aeronautical Engineering and Mechanics,
cross-sectional beams made out of isotropic materials. The beam theory derived in
Rensselaer Polytechnic Institute, this paper from variational principles is based on the sole kinematic assumption that
Troy, N.Y. 12181 each section is infinitely rigid in its own plane, but free to warp out of plane. After a
short review of the Bernoulli and Saint- Venant approaches to beam theory, a set of
orthonormal eigenwarpings is derived. Improved solutions can be obtained by
expanding the axial displacements or axial stress distribution in series of eigen-
warpings and using energy principles to derive the governing equations. The im-
proved Saint-Venant approach leads to fast converging solutions and accurate
results are obtained considering only a few eigenwarping terms.
1 Introduction
Timoshenko beam theory is well known by structural More recently, with the increasing use of composite
designers and is widely used as a first approximation in materials, Saint-Venant's principle has been investigated
numerous structural applications. For solid cross-sectional theoretically for anisotropic elasticity [3, 4], The decay length
beams made out of isotropic materials, this theory gives 5 was found to be proportional to V£/G {E is the longi-
accurate predictions for aspect ratios L/h a 5 (L is the span of tudinal Young's modulus and G the shear modulus). This
the beam and h its height). Because it is based on the means that for highly anisotropic materials the decay length
assumption that cross sections remain plane after defor- can be much larger than for isotropic materials. Goetschel [5]
mation, the theory predicts a linear distribution of axial calculated these decay length for different beams made out of
strains. However, additional axial strains are induced by composite materials. For a thin-walled rectangular cross-
warping incompatibilities generated either by specific loading section 8 is found to be Ih, and for an / cross section 5= I5h.
or boundary conditions, or by the occurrence of nonuniform Beam theory can still be applied if L/5&5, i.e., L/h^35 and
torsion or bending. According to Saint-Venant's Principle 75, respectively. Needless to say, such a restriction makes
these additional strains tend to decay away from the per- beam theory inapplicable to practical structures.
turbation that created them. The decay length 8 is defined as The purpose of this paper is to derive a beam theory that
the distance it takes for these strains to decay to a neglectible can be used when only the geometric condition L/h>5 is
value. For solid cross-sectional beams made out of isotropic satisfied.
materials, this decay length is of the order of the height of the
beam, i.e., 8 — h. A physical meaning of the validity range of 2 Geometry Assumptions
the theory is found by rewriting it as L/55:5, stating that the
span of the beam must be large compared to the perturbed Thin-walled box beams with closed cross sections will be
zone 8. considered (see Fig. 1). The contour of the section (denoted f)
In the case of thin-walled box beams, it was recognized is a parametric function of the variable s, while the variable z
early that beam theory was only a poor approximation,
because warping and shear lag effects are significant for those
structures. Von Karman and Chien [1] gave a solution to this
problem using a system of orthogonal states of stress. Argyris
and Dunne [2] derived solutions for more general problems
and showed the existence of fast converging series based on a
similar system of orthogonal stress states.
I a
dx
ds
ds is the shear force (similar definition for Ty), W=
\ [\o ^dz] [-^2\fAnnW2ds+ \(AqqT2ds] (25)
This results in a pair of linear equations for the coordinates of \ AmW,Wjds = 6Ut \ Awrtrjds=n26„ (27)
the shear center xk, yk. The Saint-Venant solution is now
summarized with a subscript "sv":
yuz = 99.5
U = 0. V = 0.007 9=0.
n = nb + YlAn„WiF; v = vb + '£viFi (32)
When infinite series are used, equations (32) give the exact
solution of the problem under the sole assumption of infinite
inplane rigidity of the section.
The Reissener principle (14) is used here again to derive the For infinite series, equation (36) give the exact solution to the
solution. The system of two differential equations resulting problem, again under the assumption of infinite inplane
from the variations on w and q is integrated as previously for rigidity of the section. Solution (32) is based on an assumed
the terms pertaining to the basic Saint-Venant solution (see displacement approach and yield a lower bound of the strain
Appendix A). The remaining terms of the system can be energy if a truncated series expansion is used. On the other
identified as the Euler equations of the functional (25) which hand, solution (36) is based on an assumed equilibrating stress
are satisfied by all eigenwarpings. The complete solution to field yielding an upper bound of the strain energy for trun-
this system is then placed into the functional (15) which cated series expansion.
becomes
\K?_ 8
Numerical Examples
Tfl - TTRs, dz (34)
•HJo V 2 M 2 ' A specific example is treated here, using the different
where -KRSU is the Reissner functional for the Saint-Venant approaches described in the previous sections. Figure 2
solution (15). Here again the basic solution is decoupled from depicts the thin-walled, rectangular cross-sectional beam to be
the corrective terms, as are the different corrective terms from analyzed. The aspect ratio L/h=4. The first step is to
each other. Minimization of -KRS„ will render the Saint-Venant calculate the eigenwarpings W,- and the associated eigenvalues
solution (19) and minimization of the corrective terms with IXj. The eigenvalue problem (26) can be solved using a finite
respect to K, gives element technique where the function Wis discretized over the
section. Then, a standard subspace iteration method allows
Kr-riK,= -d! (35) the calculation of the eigenvalues and eigenwarpings. For this
The solution of this simple differential equation gives the example, 36 nodal points were used to model the section and
solution of the problem as 24 eigenwarpings were extracted. The first few modes are
shown in Fig. 3. If the beam is subjected to a uniform loadp 0
W, U, in the y direction, the differential equation (31) to be solved
w=w„,+ . K\
for the improved Bernoulli approach is
V; Fr-riF,= -p0Vi(L-z)
1 Boundary conditions at z = 0 F,=Q; atz=LFf = 0
/ / A /
This equation is readily integrated to give the final solution
(32). In particular the tip deflection of the beam is found to
/ ri i Pi be:
y
tip _
+ +
L
P0L« 8/ v 2LM, YW^ L ^L2
Po ^ - '
1 + iXjLshjXjL
(37)
lx}L2chiAjL }
ST-VENANT SOL. The first term represents the bending deflection, the second
IMPROVED SOL. term is the shearing deflection, and the summation represents
the corrective deflections due to the different eigenwarping
-40 terms. Alternatively the improved Saint-Venant approach
requires to solve equation (35),
Ki-riK,=p0V,
Boundary conditions at z = 0KI= —paLVj\ dAz=L K,= 0
-80 From the general solution (36) the tip deflection is now
1
PoL* 871 + 2L2
1 + iXjLsYijXjL '
1 (38)
-120 +Y W 2
' L ch/j,Z, J
Table 1 lists the numerical results for the two approaches. A
Fig. 4 Stress distributions in the upper skin of the beam (root section) comparison between the basic Bernoulli and Saint-Venant
under uniform transverse load p 0 solutions shows a discrepancy of 20 percent. After adding
three corrective terms this discrepancy is reduced to about 1
percent, showing that both solutions have converged. It is
important to note the excellent convergence rate of the Saint-
Venant approach in (l/j«.,L)4 versus (l//x,L) 2 for the Bernoulli
approach. Considering one term only in the improved Saint-
Venant approach leads to a solution acceptable for
600- engineering purposes. It should be noted here that according
to (26), the eigenvalues ^, are proportional to •4Aqq/A„„
Thus the magnitude of the corrective terms in the improved
400- Saint-Venant approach is proportional to (A„„/Aqq)2
resulting in large corrections for highly anisotropic materials.
If the beam is subjected to a uniform torque m0, the tip
200-
twist 0 tip /m o Z, 4 can be calculated in a similar fashion. The tip
q/m
oL'i. twist are \/2IpL2 and 1/2JL2 for the basic Bernoulli and
Saint-Venant approaches, respectively. The corrective terms
are identical to those in (37) and (38) except that K, is replaced
\
by E,. The results are summarized in Table 1 and show a 44
percent discrepancy for the basic solutions. The corrective
-200- terms reduce this discrepancy to about 3 percent. Again the
Saint-Venant approach shows a better convergence rate; two
corrective terms are necessary here to obtain an accurate
ST-VENANT SOL.
-400- solution.
IMPROVED SOL.
It is also interesting to compute the stress distribution at the
root of the beam. Figure 4 shows the distribution of axial
-600- stresses n/p0L2 and shear stess q/p0L2 in the upper face of
the beam under a uniform loading p0, using the improved
Fig. 5 Stress distributions in the upper skin of the beam (root section) Saint-Venant approach (36). Instead of the uniform axial
under uniform torque m 0 stress predicted by the basic theory, a large shear lag effect is
(B2)
Ar
APPENDIX B
where
The inplane stiffness of an orthotropic laminate can be
A„„ =A i -A2l2/A22, A'qq
nn=A
—Ml -A\/A
-*66 _ 71
^ 26/^122>
written in matrix form as [7]:
A„Q=Ai6-Al2A26/A22 (53)
A 12 Ai6~ 6
when the material is isotropic, an orthotropic with the axes of
A2\ -422 A 26 e< (M) orthotropy parallel to the axis of the beam and the tangent to
the cross-sectional curve, we havey4 16 =A26 = 0 , thus ,4,^ = 0
A*, A 62 A66_ X and equations (32) are decoupled as (1) and (2).
.Readers Of
The Journal Of Applied Mechanics
Will Be Interested In:
MPC-Volume 25
Presented at The Winter Annual Meeting, New Orleans, LA, December 9-14
Fatigue is an important design parameter in the development of equipment for the power and transportation industries. The pur-
pose of this Volume is to present to industry some of the data that has been developed under the guidance of Subcommittee 3 of
The Metal Properties Council and to provide a forum for the presentation of material property data on some materials that have
been exposed to service conditions at high temperature.
Specific topics include: Behavior of Notched Specimens in Fatigue and Creep-Fatigue at Elevated Temperatures, A Program for
Determining the Strain Control Fatigue Properties of Aluminum Alloys Used in the Transportation Industry, and Failure of 316
Stainless Turbine Valves in Long Term Elevated Temperature Service.