Documente Academic
Documente Profesional
Documente Cultură
Engineering,
Printed in Great Britain.
1992.
G961-9526/92
$5.&l+ .CfJ
0 1992 Pergamon Press Ltd
STRUCTURAL
RESPONSE OF COMPOSITE
BEAMS
AND BLADES WITH ELASTIC COUPLINGS
RAMESH CHANDRA and INDERJIT CHOPRA
Center for Rotorcraft Education and Research, Department of Aerospace Engineering,
University of Maryland, College Park, MD 20742, U.S.A.
(Received
29 January
1992; final
version
accepted
20 February
1992)
Abstract-The
structural behavior of coupled, thin-walled, composite beams of open as well as
closed section was analyzed using Vlasov theory and then the results were validated by experiment.
The analysis modeled the walls of beams as general composite laminates and accounted for the
transverse shear deformation of the cross-section. The out-of-plane warping deformation of the
cross-section was included implicitly in this formulation. In order to validate the analysis,
graphite-epoxy beams of various cross-sections such as solid rectangular, I-section, single-cell
rectangular and two-cell airfoil were fabricated and tested for their structural response under tip
bending, torsional and extensional loads. Specialized bending-torsion and extension-torsion
couplings were introduced in these beams using proper ply lay-ups. Good correlation between
theoretical and experimental results was achieved. Transverse-shear-related couplings were found
to influence the structural response of open- as well as closed-section beams. For blades with
hygrothermally stable lay-ups, bending-transverse shear coupling increased the bending flexibility
by about 50%. The in-plane-bending coupling stiffness [B] of the walls of the beam generally
influenced the structural response of the beams quite significantly; this effect was expecially large
for I-beams. The influence of constraining the warping deformation was found to be substantial
on the structural response of open-section beams as compared to closed-section beams. A 630%
increase in the torsional stiffness due to constrained warping was noticed for graphite-epoxy
I-beams of slenderness ratio 30. The feasibility of achieving the desired levels of bending-torsion
and extension-torsion couplings in two-cell rotor blades was demonstrated.
NOTATION
chord and thickness of two-cell composite rotor blade
length of beam
coordinate system for plate segment
coordinate system for beam
displacements in the n, s, z directions, referring to the plate segment
displacements in the x, y, z directions, referring to the beam
membrane strains referring to the plate segment
bending curvature referring to the plate segment
rotations about the x, y, z axes, referring to the beam
transverse shear strains for the beam in the xz and yz planes, respectively
warping function
constrained warping parameter
stress field referring to the plate segment
stress resultants referring to the plate segment
moment results referring to the plate segment
axial force referring to the beam
bending moments referring to the beam
shear forces in the x, y directions, referring to the beam
torsion moment referring to the beam
bimoment (or warping moment) referring to the beam
stiffness matrix for the beam
applied torsion at the tip of the beam
applied force at the tip of the beam
axial force at the tip of the beam
Youngs moduli of the plies in the principal directions
Poissons ratio of the plies in the principal plane
shear modulus of the plies in the principal plane
differentiation with respect to the z coordinate of the beam
INTRODUCTION
Composite beams of open as well as closed sections form important structural elements of
aerospace systems, such as helicopter rotor blades and airplane wings. Although composite
COE2:5/7-o
347
348
R. CHANDRA
and I. CHOPRA
materials have been used by the aerospace industry for several years, bending-torsion and
extension-torsion couplings due to laminated composite have not been exploited at this
time to improve the aeroelastic characteristics of these systems. For this, it is necessary to
develop a simple, accurate model of composite beams which can be incorporated into
comprehensive aeroelastic analyses. It is now well established that nonclassical effects,
such as cross-section warping, transverse shear and warping constraint at the fixed end,
become important for the analysis of composite beams and therefore must be addressed
in the modeling of composite beams.
In order to study the aeroelastic stability of hingeless rotor blades, Hong and Chopra
(1985) developed a simple analysis for composite box beams with elastic couplings. That
analysis was based upon a solid-section approach similar to that used by Hodges and
Dowel1 (1974) for isotropic blades. Chandra et al. (1990) evaluated this structural model
using a detailed finite element analysis as well as experimental techniques, and identified
certain limitations. Smith and Chopra (1991) improved the model by introducing transverse shear deformation of the cross-section, a refined section warping and proper inplane elasticity. Chandra and Chopra (1992) used the improved model to predict the free
vibration characteristics of rotating box beams and satisfactorily correlated the calculated
results with experimental data obtained using an in wcuo rotor test facility.
Rehfield et al. (1990) formulated a composite beam theory for closed sections where
transverse shear deformation of the cross-section and constrained warping effects were
included. The local bending stiffness of the beam wall was, however, neglected. The
analysis was applied to extension-torsion coupled beams and a significant influence of
bending-transverse shear coupling on the static structural response of cantilevered beams
was shown. Hodges et al. (1991) used that analysis to predict the free vibration characteristics of nonrotating beams. Further, Rehfield et al. (1988) extended their formulation to
include multi-cell composite beams. Shear-flow calculations were based on an approach
very similar to that described by Megson (1974). The predictions of the static structural
response of two-cell composite beams were correlated with finite element results.
Kosmatka (1991) used an elasticity approach to investigate the influence of initial
twist on the structural behavior of composite beams. Single-cell D-section beams with
initial twist were analyzed. The importance of initial twist in the modeling of rotor blades
was highlighted. Bauchau (1987) presented a large-deflection finite element analysis of
curved and twisted composite beams. The analysis included transverse-shear- and torsionrelated warping. Giavotto et al. (1983) presented a finite element analysis for anisotropic
beams. Borri (1986) formulated a composite beam theory from a virtual work approach,
where large rotations and accurate cross-sectional warping were included. Minguet and
Dugundji (1990a, b) presented an analytical-experimental
study of solid-section composite beams undergoing arbitrarily large deformations. Good correlation between theory
and experiment was achieved.
Open-section composite beams of different cross-sections like rectangular, I, cruciform, etc. form structural elements of the flexbeam of bearingless rotors. Chandra and
Chopra (1991a, b) developed an analysis for a general open-section composite beam based
upon the Vlasov theory (Vlasov, 1961; Gjelsvik, 1981). The analysis was applied to predict
the static structural response and free-vibration characteristics of composite I-beams with
couplings, and results were validated by experiments. An in vucuo rotor test facility was
used to obtain experimental data to validate the free-vibration analysis of these beams
under rotation. These studies concluded that the segments of open-section composite
beams must be modeled as general composite laminates and contrained warping effects
must be included. Rehfield and Atilgan (1989) investigated the buckling of composite
open-section beams. Transverse shear deformation
of the beam cross-section was
included, but the local bending stiffness of the wall was neglected. Chandra and Chopra
(1991c, d) expanded the Vlasov theory to analyze multi-cell composite rotor blades with
elastic couplings under extension, bending and torsional loads. The predicted structural
response was successfully correlated with experimental data. These studies showed the
importance of multi-cell analysis, and also the feasibility of achieving a significant degree
of bending-torsion and extension-torsion couplings in blades. Most of the above studies
349
show the importance of nonclassical effects for composite beam analysis; however, these
have been restricted to a few simple beam sections.
The objective of this paper is to expand the existing composite beam analyses of the
authors (based on the Vlasov theory) to apply it to several different composite beams
involving open, single-cell and multi-cell closed sections, and to highlight the importance
of nonclassical effects for various beams. To validate the analyses, special bendingtorsion and extension-torsion
composite beams were built of graphite-epoxy composite
and tested under static bending, torsion and extensional loads. To carry out a comprehensive review of the modeling of different composite beams, existing predicted and
experimental results are complemented by many new results.
ANALYSIS
In this paper, the Vlasov theory is expanded to examine the structural response of
composite beams of open as well as closed sections (solid rectangular, I, thin-walled
rectangular and multi-cell airfoil). The flanges and webs of the beams are modeled as
general composite laminates and transverse shear deformation of the cross-section is
included. The essence of this theory is the reduction of two-dimensional
stress and
displacement fields (associated with plate segments of the beam) to one-dimensional
stresses and displacements identified with the beam. The six generalized beam displacements are determined from the plate displacements through geometric considerations,
whereas the generalized beam forces and their equilibrium
equations are obtained by
invoking the principle of virtual work.
The present analysis uses three coordinate systems: an orthogonal right-handed
Cartesian coordinate system (x, y, z) for the beam (Fig. la); an orthogonal coordinate
system (n, s, z), for any plate segment of the beam (Fig. lb) where the n axis is normal to
the mid-surface of any plate segment, the s axis is tangential to the mid-surface and is
along the contour line of the beam cross-section, and the z axis is along the longitudinal
axis of beam; and a contour coordinate system s, where s is measured along the
contour line of the cross-section from a judiciously selected origin (Fig. Id). The seven
generalized beam forces V,, I$, I$, M,, MY, T and h4, are shown in Fig. lc. The
torsional moment T consists of unconstrained warping torsion (St Venant torsion), and
constrained warping torsion (Vlasov torsion). As shown later, the Vlasov torsion and
bimoment M, are related to each other. The stress resultants, moment results and transverse shear forces acting on any general plate segment of the beam are shown in Fig. lb.
The plate stress and displacement fields are functions of s and z.
Fundamental assumptions
Three basic assumptions used in the present theory are:
(1) The contour (mid-line of the plate segments) of a cross-section does not deform
in its plane. This means that the in-plane warping of the cross-section is
neglected and the normal strain E, in the contour direction is neglected in comparison with the normal axial strain E,. This assumption was introduced by
Vlasov (1961).
(2) The normal stress o, is neglected in comparison with or.
(3) A general plate segment of the beam is governed by linear classical laminated
plate theory. This implies that the transverse shear deformation of the plate
segment is not accounted for, though the transverse shear deformation of the
beam is considered.
These assumptions imply that the nonzero membrane
the plate segment are E,, cSZ, k, and k,, .
R. CHANDRA
and LCHOPRA
I-Section
(b)
VY
VX
/J-VZ.
kf)
p
t
P,W
00
Fig. 1. (a) Coordinates of beams. (b) Stress and moment resultants acting on any general plate
segment of the beam. (c) Generalized beam forces. (d) Pictorial definitions of blade displacements
and rotations.
351
Kinematics
From geometric considerations Fig. Id, the plate displacements u(s, z) and u(s, z) are
related to the beam displacement U, V and & as:
u(s, 2) = U(z) sin &s) - V(z) cos O(s) - q(s)&(z)
(1)
(2)
shear
F(s) controls the variation of this strain along the contour of the beam cross-section. In
order to account for variation of shear modulus G along the contour, eqn (5) is rewritten as:
G,(s)
Eqs
SE 7 z) = - Gt 4;(z)
(6)
-h, 0
ccc
1,2.
where Ai = fi r ds.
9~dS=2Ai
03)
352
and 1. CHOPRA
R. CHANDRA
Equation (8) is used to compute G,(s). Gsl and GS2 are the value of G, associated with
circuits and G,r to G,, are the values of G, associated with branches. From Fig. 2,
Gsl = csl - es2
Gs2
Gs3
cs,
Gs4
&
(10)
where
c
a1
0.3%
d.5
s 0.3sc
Solving eqns (11) and (12),
Ar=2
&
P2 =
0
0.35c
ydx,
s 0.3%
4(-&+~)
A2=2
ydx.
s0
+A,&
(13)
($+-&i)(-&++)-(-j&G
G2
(14)
w = w + xd++ MJJ- HI
where the warping function,
(15)
p, is equal to:
(16)
4, = Exz - U
f#Jy= Eyz - Y.
(17)
(18)
(19)
Similarly
353
(20)
k,, = -26;.
(21)
Thus the nonzero membrane strains and bending curvatures in the plate segment are given
by relations (19), (4), (20) and (21).
Plate stress field
Using classical laminated
+ B&s
Nzs = Al&
+ &A,
+ B&s
Mz = 41~
+ &t&s + D,,kz
Mzs = 46~
+ %k,
+ B66w + Q,
(22)
+ h&
kz + Q&s
equations
- M&J,]
ds +
Using relations (l), (2) and (15) and taking the variation
SW, = N6W
+ v, 6U + v, 6V + T&,
+ M, 6#y + F, &,,
(M$?
- A&u').
(23)
branches
of W,,
+ Fy SE,,
(24)
where
N=
N,ds
ss
V, =
branches
branches
V, =
T =
Mw
ss
(25)
c
c
c
(26)
(-M&
(27)
(-M&qj
(28)
branches
(N,Y,
(29)
M,q)h
s s
Ad,
(N,y
e) ds
(30)
(N,x - Mz sin e) ds
ss
(31)
A4,
cos
s s
My =
F, =
is
my = -
A4, sin B ds
(32)
hf,c0seds.
(33)
ss
354
R.CHANDRA
and I.CHOPRA
It is difficult to compute the generalized blade forces V,, V, and T from relations
(25), (26) and (27) because of the contributions from different branches. These are simplified by using equilibrium equations of plate forces (Gjelsvik, 1981):
<=-My
(34)
v, = M;
(35)
T=T,+T,
(36)
where T, is the St Venant torsion (free warping) and T, is Vlasov torsion (constrained
warping). These are defined as:
(37)
It is to be noted that the second term in the equation of the St Venant torsion is zero for
an open section.
T, =
(Nzsr + M;q)ds.
s
equation, relation (37) is simplified
(38)
to:
T, = -ML.
(39
This gives the relationship between the Vlasov torsion and the warping moment
bimoment).
The external virtual work done by the applied loadings on the plate is:
ndW+
v,dU+
v,dV+
tfJ4, + m,&,
+ rn,6&
+ rn,d&
+fx&,,
+&BE~~
(or
(40)
where n, v,, v,,, t, m, , my, m,, f, and fY are generalized load intensities on the blade,
derived from the loadings on a shell (Gjelsvik, 1981).
The strain energy, l-I, is given as
l-I = +
ss
(N&
+ Nzs.szs + M,k,
+ M,,k,,)ds.
(41)
Using the relations between the blade forces and shell forces, the strain energy becomes
I-I = $[NW
+ F,E:, + Fy&
+ GxexZ + G,,E,,J.
(42)
The internal virtual work, w, is obtained from the strain energy as:
-Y
= NW
+ My+:
+ M,&
+ T& + Mad;
G, =
ss
+ F,E:, + Z$&
N,,cosOds
G,, =
(46)
v,+u,=o
(47)
N+n=O
(48)
T+t=O
(4%
MA-f-T-T,+m,=O
M; + V, + m,, = 0
(50)
(51)
355
By eliminating
(52)
&'-G,+f,=O
(53)
l$-G,,ffy=O.
(54)
(55)
(56)
(57)
(58)
M; + ml; - v, = 0
M:' + m; + vy = 0
M: - T, + rn: - t = 0
Beam force-displacement
F;-G,+f,=O
(5%
Fy)-gy+fy=o.
(60)
relations
There are nine generalized blade forces, namely N, MY, IV,, IV,, T, , F,, 4, G, and
G,, appearing in the above equations. The nine generalized forces are related to six generalized displacements. Using plate stress-strain relations (21) and plate strain-beam displacement relations (18), (4), (19) and (20), the following relations between the generalized
bar forces and displacement are obtained:
N
K,,
Mx
K12
KM
KM
KIS
Kl6
K17
KM
K19
K22
K23
K24
K25
K26
K27
K28
K29
6;
K33
K34
K35
K36
K37
K38
K3,
6:
K44
K45
K46
K47
K48
K49
6;
KS,
K56
KS,
K58
K,,
4:
K66
&7
&3
Km
GZ
Km
K7,
Km
&yz
Ku3 &g
&:,
-My
Mid
T$
Symmetric
GY
F,
.A-
(61)
49 _ _ cl, _
coupled
Extension-shear:
(62)
Bending-torsion:
K22
0
0
=
K,,
0
Jz,
K33 0
0 K44
K25
0
K38
0
0
0
0
K,,
0
K59
Kzg
K38 0
0
0
0 K,,
KB8 0
K99
(63)
and I.CHOPRA
R.CHANDRA
Extension-Torsion
Coupling in Solid
Rectangular Cross-Section Beam
Bending-Torsion
Coupling in Solid
Rectangular Cross-Section Beam
Bending-Torsion
Bending-Torsion
Coupling
in I-Beam
Coupled Beam
Bending-Torsion
Coupled Blade
Extension-Torsion
Coupling
Extension-Torsion
Extension-Torsion
in I-Beam
Coupled Beam
Coupled Blade
Fig. 3. (a) Lay-up details for coupled composite open-section beams. (b) Lay-up details for
coupled composite closed-section beams.
Following the procedure used by Chandra and Chopra (1991a), eqns (62) and (63) are
solved to give the bending slope and twist of bending-torsion
coupled beams under tip
loads.
Tip bending load:
= %
PI
coth&cosh
W&J,
(65)
cash AZ
(67)
357
where
(K,,),/
p=AI=
r
(68)
K44
It is to be noted that the influence of direct transverse shear manifests itself in the bending
slope relations (65) and (67) through the term L,,.
In order to evaluate the influence of constrained warping on the torsional behavior
of these beams, the torsional stiffness is represented as the relative torsional stiffness
(RTS) with respect to the St Venant torsional stiffness:
RTS =
P
,LI - 2 tanh(p/2)
(6%
It is interesting to note that the relative torsional stiffness is controlled by the constrained
warping parameter p. This parameter as defined by eqn (68) depends upon the coefficients
of ths stiffness matrix and length of the beam. Thus, the influence of geometry and
material on the torsional behavior of the beam is felt through this parameter.
Neglecting the constrained warping and direct transverse shear effects, the bending
slope and twist relations are simplified to:
Tip bending load:
= -AtWn),
K25
4~= K,,
Tip torsional
(212 - 22)
(70)
(71)
(212 - z2).
2K22)r
load:
(72)
(73)
where
W22)r
K22
K225
- -
(74)
K55
K22
VMr = KS, - ~22.
(75)
Note that for slender beams, the influence of direct transverse shear deformation on the
static structural response is negligible. Also for beams where the constrained warping
parameter is higher than 40, the constrained warping effects are negligible (Chandra and
Chopra, 1991a).
Extension-torsion
force-displacement
Extension-torsion:
(76)
Relations (76) control the extension-torsion
coupled behavior of these beams. Note
that the extension-torsion
coupling is via St Venant torsion.
358
R. CHANDRA
Bending-transverse
and I. CHOPRA
shear:
(77)
(78)
Relations (77) and (78) control the bending behavior of these beams and the bendingshear couplings are included in these relations.
In the present paper, the structural response of extension-torsion coupled beams is
confined to closed section beams. The influence of constrained warping on the structural
response of such beams is negligible as the constrained warping parameter is higher than
40 (Chandra and Chopra, 1991a).
Using relations (76), the twist due to the tip torsional load T and induced twist due
to the axial force F are given by:
(79)
t$z = -
K15
Fl.
K,IKSS - K:,
For beams subjected to a tip bending load P, the rotation $ is obtained from relation (77):
= K,,U
P
- W,2,/K,,K,,))
(81)
= L,,P
(82)
where
1
L77 = KT7(1 - (K$,/K,,K,,))
Using relations (Sl), (82) and (17) the bending slope is obtained as:
P
= L77p - KZ2(1 - (K;6/K22K66))
The first term in relation (84) represents the direct transverse shear effect whereas the
second term represents the influence of transverse shear-bending coupling on the bending
slope. Note that the bending-shear coupling reduces the bending stiffness. The ratio of
the first term to the second term at the tip can be used as a measure of assessing the direct
transverse shear effect. This ratio is termed the shear correction factor (SCF) and is
given by:
SCF=;L,,K,,(l
-&).
The analytical development presented above is generic for open- as well as closedsection composite beams and multi-cell rotor blades. The coefficients of the stiffness
matrix [K] are given in a general form. These can easily be specialized for open sections
by omitting the shear flow terms. The results in terms of bending slope and twist are also
generic.
FABRICATION
To validate the analysis, many different composite beams with specialized couplings
were built. These included solid rectangular section beams, thin-walled single-cell box
beams, thin-walled open-section I-beams and thin-walled,
multi-cell,
airfoil-section
blades. The solid rectangular-section
beams, thin-walled single-cell box beams and
359
Legend:
pJ!j
Aluminum
Molds
Composite
Lay-up
II
Peel Ply
Porous Bleeder
Bleeder Plies
Solid Beam
cl
Barrier Film
cl
Breather
Plies
I-Beam
Rotor Blade
Box Beam
Fig. 4. Details of composite beam fabrication.
beams
Figure 4 shows the fabrication details for the solid beam. Metal molds with spacers
were used to avoid slipping of plies during curing under pressure. Peel ply was wrapped
to provide the surface finish of the beam. In order to bleed out excess resin and to permit
the escape of volatiles during the curing process, a number of bleeder and breather layers
were then applied. A caul plate was used to facilitate the application of uniform pressure
on the laminate. The lay-up was cured in a microprocessor-controlled
autoclave according
to the curing cycle provided by the manufacturer. Thus, bending-torsion coupled composite solid beams with different ply orientations were built.
I-beams
Figure 4 shows the details of fabrication of I-beam specimens. Graphite-epoxy
prepreg layers were laid-up on a metal mold which consisted of two parts. Each of these
parts would yield beams of channel-section. For the fabrication of I-section beams, each
part of the mold was wrapped with the desired number of prepreg layers. These layers
were compacted by applying a vacuum between the mold and lay-up. The two parts of the
mold were placed back to back and the additional layers were introduced on the top and
bottom flanges. Thus, several bending-torsion coupled composite I-beams with different
ply orientations and slenderness ratios were built using an autoclave molding procedure,
as described in the previous section. Table 1 shows the details of these beams.
Box beams
Figure 4 shows the schematic of the split metal mold used to fabricate box beams.
Graphite-epoxy unidirectional prepreg layers were laid on the mold. For bending-torsion
coupled beams, each layer had two joints, which were intentionally staggered for better
strength. The layers in extension-torsion coupled beams had only one joint for each ply.
The lay-up was compacted using a vacuum pump and cured in a microprocessor-controlled autoclave, using the method described earlier. Thus, coupled beams of different
ply orientations and slenderness ratios were built. Table 2 shows the details of these
beams.
360
= 0.42
Geometric parameters:
Length = 36 in.; Clamped length = 6 in.; Effective length = 30 in.
Wall thickness = 0.04 in., Number of layers = 8, Ply thickness = 0.005 in.
Width and
height
(in.)
Cases
I-beam
I-beam
I-beam
I-beam
I-beam
I-beam
Flanges
2x2
1
2
3
4
5
6
1X1
1x
1x
1x
1x
0.5
0.5
0.5
0.5
Top
Bottom
wwl
w01,
Pow,
wol,m~l,
P~w,mol,
ww,
Web
10~901,
ww,
ww,([-
151,
[O/901,/[-301,
[O/90/0/-15],
[O/90/0/ 151,
ww,
ww,
W9%
w01,
ww,
ww,
Cases
Top flange
Bottom flange
Left web
Right web
Coupling
beam
beam
beam
beam
beam
beam
1
2
3
4
5
6
w9013
10/9%
w901,
IO/W,
[1516
i30i6
[-I516
k30i6
[15/-151,
[30/-301,
I--151151,
[-30/30],
f4516
[-4516
[45/-451,
[-45/45],
Ii516
Ii516
v5i6
b56i
[O/301,
[O/301,
[O/301,
[O/301,
None
Bend-Tor.
Bend-Tor .
Bend-Tor.
Ext.-Tor.
Ext.-Tor.
D-spar
Bottom flange
Top flange
Extension-torsion
Chord = 3 in.
Blade
Blade
Blade
Blade
Blade
Blade
Blade
1
2
3
4
5
6
7
Blade 8
Blade 9
Ku,
w51,
Hygrothermally
[O/151,
[O/451,
stable lay-up.
WI,
[O/151,
[O/301,
[O/451,
[20/-70],,
[20/-701,
v51,
[15/-151
[15/-151
[30/-301
[45/-451
[20/-701
[20/-701
[15/-151
Bending-torsion
Blade 10
Blade 11
coupled blades
[O/151,
[O/301,
[O/451,
[20/-70],,
[20/k701,
P5l.i
Chord = 6in.
[20/-70],,
Skin
mi
[O/151,
[O/301,
w/451,
[20/-70],,
(20/-701,
D51,
Web
[20/-70],,
[O/151,
[20/-701
[15/-151
coupled blades
[O/-15],
[O/k 15/O],
[15/-151
[O/-45],
[o/+45/0],
[45/-451
Composite
361
STRUCTURAL
TESTING
The coupled beams and blades were tested under bending, torsional and extensional
loads for their structural response. Tip loads were applied by means of dead weights and
pulleys, using a simple test set-up (Chandra et al., 1990). The bending slope and twist of
a generic point on the beam were measured using a 0.165 in. diameter mirror and a 2 mW
helium-neon laser. To reduce measurement error, laser-dot deflections of the order of
10 in. over a distance of 282 in. were used. Different load levels were used on various
beams to get laser dot-deflections of this order. Different clamping and loading fixtures
were used to simulate the proper clamping conditions for different beam specimens.
Figure 5 shows a shematic of the clamped ends of various beams.
In order to avoid buckling of the web of the I-beam while clamping, the web was
bolted between two metal inserts. At the loading end, a special fixture was used to ensure
Solid Beam
Box Beam
Legend:
a - Steel Clamps
b - Metal Insert
c - Specimen
Rotor Blade
I-Beam
Fig. 5. Clamped
ends of various
composite
beams.
362
that the bending load passes through the shear center. Warping displacements at the
clamped and loading ends were constrained in this test. For box beams, the clamped end
was reinforced by means of a metal insert to provide a clamped condition. While testing
extension-torsion
coupled beams in tension, the loading end was permitted to twist by
means of a special end fixture with a thrust bearing. Warping deformations at the
clamped end were constrained. In order to provide proper clamping for blades, a
mahogany insert was introduced in the blade at the fabrication stage. Also, the clamped
end of the blade was further reinforced by means of outside aluminum clamps, which
were machined accurately using numerically-controlled
machining. While mirrors and a
laser were used to measure the bending slope and twist of the blades, strain gages were
used to measure the response under extensional loads.
RESULTS AND DISCUSSION
The salient features of thin-walled composite beam analysis are: constrained warping,
modeling of walls of beams as general laminates, transverse shear deformation and elastic
couplings. Keeping these nonclassical effects in view, the following section is written.
Constrained
warping effect
Constraining the warping deformation of the beam either at the support or load point
changes its torsional behavior. This effect is more pronounced for open-section beams,
such as I-beams.
Figure 6 shows the influence of constrained warping on the torsional stiffness of
several cantilevered I-beams subjected to a tip torsional load. The torsional stiffness is
expressed in terms of relative torsional stiffness, which is the ratio of the torsional stiffness under a mixed condition of free and constrained warping to the torsional stiffness
corresponding to free warping (St Venant torsion). It is seen from this figure that the
relative torsional stiffness decreases with an increase in the constrained warping parameter,
,u, and reaches an asymptotic value of 1, which refers to the free-warping condition. The
parameter P, which is a function of the cross-sectional stiffnesses and length, determines
the influence of the constrained warping on the torsional stiffness of the beam. The
influence of the constrained warping on the relative torsional stiffness becomes very
significant for low values of P. For values of P larger than 40, the effect of constrained
warping on torsional stiffness is negligible and then the beam torsional behavior is
controlled by the St Venant torsion. For a cross-ply I-beam with a slenderness ratio of 15
(I-beam 1, p = 0.38), the relative torsional stiffness is 83, whereas for a cross-ply I-beam
with a slenderness of 30 (I-beam 2, p = 1.53), it is 6.37. Section details of these beams are
provided in Table 1. The influence of the composite material on the relative torsional
--
Relative
torsional
stiffness
loo 7
I-Beam
2 I-Beam
Aluminum
10 y
I-beam
1
f
0.1 ,
St. Venant
I
0.1
Torsion
Constrained
I
10
Warping
Parameter,
100
Composite
363
7
6
5
4
Relative
torsional
stiffness
3
2
1
0
I-Beam 2
Slenderness
ratio
I-Beam 3
= 30 Slenderness
ratio
= 60
Fig. 7. Torsional stiffnesses of graphite-epoxy I-beams subjected to tip torsional load (Chandra
and Chopra, 1991a).
stiffness is illustrated by comparing the results for a cross-ply I-beam with a slenderness
ratio of 60 (I-beam 3, p = 2.89) with an aluminum I-beam of identical slenderness ratio
(,u = 6.45). Note that the value of the relative torsional stiffness for the cross-ply, composite I-beam is 2.62, whereas for the aluminum I-beam, this value becomes 1.52. This
shows that the effect of constrained warping on the torsional stiffness is smaller for
isotropic materials than for composites. Figure 7 shows calculated and experimental
values of relative torsional stiffnesses of cross-ply I-beams for two slenderness ratios. A
good correlation between theory and experiment is seen only when constrained warping is
included in the analysis.
For the solid beams, thin-walled box beams and multi-cell thin-walled rotor blades
examined in this paper, the constrained warping parameter, p, is generally above 40.
Hence, the constrained warping effect is not important for these beams.
Flanges and webs as general laminates
Composite beams with flanges and webs made out of general composite laminates
may be required in many engineering applications. Neglecting their local bending stiffnesses may lead to a large error in the prediction of the structural response. This effect is
illustrated for I-beams and rotor blades.
The bending-torsion
coupling stiffness KZ5 for composite I-beam depends upon the
stiffness coefficients B,, and D,, of the flanges. BIG refers to extension-twist coupling and
D,, refers to bending-twist coupling of the plate segment of the I-beam. It is possible to
select the flanges such that B,, is zero, meaning that the flanges are symmetric with
respect to their own mid-surfaces. The 15 bending-torsion
coupled graphite-epoxy
I-beams (I-beam 4) are made out of flanges which are not symmetric with respect to their
own mid-planes. Hence, B,, for these beams is nonzero. In order to assessthe importance
of including B,, in the analysis, I-beam 6, whose flanges are symmetric with respect to
their own mid-planes, is examined analytically. Note that the number of plies and their
orientation in the flanges of I-beam 4 and I-beam 6 are identical, but the stacking
sequence is different (Table 1). Figure 8 shows the bending slope and induced twist of
these beams under unit tip bending load. Because of a nonzero B,, in I-beam 4, the
response of this beam is vastly different to that of I-beam 6.
Figure 9 shows the influence of the extension-twist coupling stiffness, B16, of the
branches of the blade spar on the blades structural response under tip bending and torsional loads. The general plate segment of the spar of Blade 5 is symmetric with respect
to its own mid-plane, and hence B16 equals zero. On the other hand, the spar of Blade 6
is made of laminates which are not symmetric with respect to their mid-plane and hence
results in a nonzero B,, . B16 influences the bending slope of the blade by increasing the
bending-transverse shear coupling Kz6. About a 25% increase in the bending slope due to
B16 is noticed for this blade. Note that the twist is virtually unaffected by B16. This is
CDE 2:5/7-E
364
R. CHANDRA
and
I. CHOPRA
0.025 +
0.020 j
Response
Rad.
0.015 {
0.010 I
I-Beam 6
B16 =0
I-Beam 4
0.025
0.020
Response
rad.
0.010
0.005
0.000
Bending slope
h-Blade
5
Spar lay-up:
[20/-7OJ,
B =0
0 Bl1:de 6
Spar lay-up
f20/-7014
B16*
Twist
because the torsional stiffness of the blade does not depend upon B,, . It is clear that this
effect will notbe captured if the bending stiffness of the wall of the blade is not included
in the analysis, as done by Rehfield et al. (1988).
Transverse shear effect
The influence of transverse shear deformation on the structural response occurs in
two ways: the direct transverse shear effect and the effect via transverse-shear-related
couplings. The direct transverse shear effect is controlled by the slenderness ratio of the
beam and its cross-sectional details, whereas the transverse shear related coupling effect
is controlled by the geometry and lay-up of the cross-section only. This effect is shown
below for box beams and rotor blades.
Figure 10 shows the influence of the direct transverse shear deformation on the tip
bending slope of cantilevered beams under unit tip bending load. The shear correction
factor is the ratio of the tip bending slope due to transverse shear deformation to the tip
bending slope due to bending. The slenderness ratio has a significant influence on the
Shear Correction Factor (SCF). For beams with a slenderness ratio of 10, the SCF value
is less than lo%, whereas for beams with a slenderness ratio of 5 this value can become
as high as 40%. Note that the lay-up of the beams has a small effect on the SCF. As the
365
0.3
Shear
Correction
Factor
Oq2
0.1
0.0
[1516
W3013
Aluminum
Fig. 10. Influence of transverse shear deformation on the tip bending slope of box beams.
ratio of shear modulus to Youngs modulus is much less for graphite-epoxy than for
aluminum, the values of the SCF for composite beams are much higher than those of
identical aluminum beams.
Figure 11 shows the influence of the bending-transverse shear coupling on the
predicted tip bending slope of extension-torsion
coupled beams under unit tip bending
loads. Measured values are also shown. Transverse shear has a significant influence on
response. A good correlation between analysis and experiment is achieved when the effect
of transverse shear is included in the analysis. For the [15], box beam (Box beam 5), the
bending-shear coupling increases the bending flexibility by about 40%. However, this
effect is less pronounced for the [O/30], box beam (Box beam 6).
Figure 12 shows the influence of the bending-shear coupling on the bending slope of
extension-torsion
coupled blades subjected to unit tip bending load. Again, bendingshear coupling appears important. For the blade with a hygrothermally
stable lay-up
(Blade 5), bending-shear coupling increases the bending slope by about 50%.
0 Analysis w/o bending-shear coupling
f3 Analysis with bending-shear coupling
m Experiment
Tip Bending
Slope
rad. 0.005
0.000
[1516
w3q
Fig. 11. Tip bending slope of extension-torsion coupled graphite-epoxy box beams under unit tip
bending load.
Influence
of ply orientation
on elastic couplings
The influence of ply orientation on the structural response is presented for I-beams,
box beams, rotor blades and solid beams.
Figure 13 shows the bending slope and induced twist of bending-torsion
coupled
I-beams subjected to a unit tip bending load. For uncoupled composite and metal beams,
there will be no twist under bending loads. As shown, the induced twist due to bending
R.
366
CHANDRA
and
I.CHOPRA
0.045
Bending Slope
rad.
Blade 5
1
1
Blade 6
Fig. 12. Influence of bending-shear coupling on the bending slope of extension-torsion coupled
blades (Chandra and Chopra, 1991~).
Induced
Twist
Induced
0.03 Response .
rad.
0.02 : Bending
. Slope
0.01 i
L 1
I-Beam 5
30
I-Beam 4
15
Fig. 13. Bending slope and induced twist at tip for bending-torsion
I-beams subjected to unit tip bending load.
coupled graphite-epoxy
0.05,
1.04 1.03 Bending
Slope
rad.
Box beam
1.02
c
ox beam 1
to1901
15
30
45
Fig. 14. Tip bending slope due to unit tip bending loads for graphite-epoxy box beams.
367
Induced
Tip Twist
rad.
0.010
0.005
0.000
45
Box beam 2
Box beam 3
Box beam 4
Fig. 15. Induced tip twist due to unit tip bending loads for graphite-epoxy box beams.
very much depends on the lay-up of the composite beams. Note that the induced twist is
about five times the bending slope for these beams under bending load.
Figures 14 and 15 present the influence of ply orientation on the tip bending slope
and induced twist of graphite-epoxy box beams under unit tip bending loads. Box beam
1 consists of cross-ply laminates and has no bending-torsion
coupling, whereas Box
beams 2-4 consist of angle-ply laminates and have bending-torsion
coupling. There is a
good correlation between theory and experiment. Note that the bending slope and induced
twist increase with an increase in ply orientation.
Figure 16 shows the tip twist of [15], and [O/30], graphite-epoxy box beams under
unit tip torsional load. These beams have antisymmetry with respect to their mid-axes and
result in extension-torsion
and bending-shear couplings. The results of the present
analysis correlate well with experimental data. Figure 17 shows the induced tip twist of
these beams under unit axial load. Again, the extension-torsion
coupling depends upon
the composite lay-up.
Figure 18 shows the tip bending slope of several extension-torsion
coupled blades
under a unit tip bending load. Results corresponding to the one-cell theory are obtained
by neglecting the web and treating the blade section as a single cell. As expected, the onecell approximation
overestimates the bending slope as compared to the two-cell theory,
and the experimental results are generally closer to the two-cell analysis. The correlation
between the two-cell analysis and experiment is within 7%. This figure also indicates the
0.004
Twist
rad.
0.002
Box Beam 5
D516
Fig. 16. Twist of extension-torsion
Box Beam 6
W3013
R.
368
CHANDRA
and I.
CHOPRA
0.0002
Induced
Twist
rad. 0.0001
Box Beam 5
1154
0.0000
Box Beam 6
[0/3013
0.030
0.020
Bending
Slope
rad.
0.010
0.000
Fig. 18. Influence of lay-up on bending slope of extension-torsion coupled blades of slenderness
ratio 72 under unit tip bending load (Chandra and Chopra, 1991~).
0.0030
111
Experiment
Tip Twist
rad.
0.0015
0.0000
Blade 5
Fig. 19. Influence of lay-up on twist of extension-torsion coupled blades of slenderness ratio 72
under unit tip torsional load (Chandra and Chopra, 1991c).
369
0.0005
n Theory (one-cell)
0 Theory (two-cell)
0.0003
Tip Twist
-rad.
0.0002
0.0001
0.0000
Blade 8
Blade 9
Fig. 20. Difference between single-cell and two-cell predictions of twist of extension-torsion
coupled blades of slenderness ratio 36 under unit tip torsional load (Chandra and Chopra, 1991~).
Blade 7
370
R. CHANDRA
and I.
CHOPRA
0.010
Theory
n Experiment I
0.008
0.006
Response
rad.
0.004
I
0.002
0.000 i
Tip Bending slope
Fig. 22. Response of bending-torsion coupled rotor blade (Blade 10) under unit tip bending load
(Chandra and Chopra (1991d).
0.003 /
0 Theory
n Experiment I
0.002
Response
rad.
0.001
Tip Twist
Fig. 23. Response of bending-torsion coupled rotor blade (Blade 10) under unit tip torsional load
(Chandra and Chopra, 1991d).
Fensile 40.0
strain
microstrain 30.0
20.0
10.0
0.0
Blade 10
Blade 11
Fig. 24. Structural response of bending-torsion coupled blades under extensional loads (Chandra
and Chopra, 1991d)
Bending
slope
371
Induced
twist
Fig. 25. Structural response of a 15 solid beam at the tip due to a unit tip bending load.
extensional load, the strain levels in its different branches were computed. The maximum
strains were of the order of 800 microstrains at an axial load of 1000 lbs.; these values are
well within the material allowables of the blade. This implies that the blade may be able
to resist a 1000 lb. axial load. Note that the induced twist rate of this blade at 1000 lbs.
would be of the order of 0.25 deg. in:. This value may be sufficient in satisfying the
requirement for the design of extension-twist coupled tilt rotor blades JVX and XV-15
rotors (Nixon, 1987).
Figure 22 shows the tip bending slope and induced twist of the bending-torsion
coupled blade, Blade 10, under unit tip bending load. Good correlation between theory
and experiment for bending slope and induced twist is achieved. Figure 23 illustrates the
tip twist and induced slope for this blade under unit tip torsional load. Again, good
correlation
between theory and experiment is noted. Note that the amount of
bending-torsion
coupling in this blade is rather low as compared to the box beams due to
the difference in lay-up and geometry of the section.
Figure 24 shows the uniform tensile strain in bending-torsion coupled blades (Blades
10 and 11) under tensile load. Good correlation between analysis and experiment is seen
in this figure. The analytical results include extension-transverse shear couplings. The
influence of extension-transverse shear coupling in the bending-torsion coupled blade is
to reduce its extensional stiffness. This reduction for the blades examined in the present
study is small (about 5% for Blade 10).
The structural response of the [15],, bending-torsion
coupled graphite-epoxy solid
beam is presented in Figs 25 and 26. The dimensions of this beam are: width = 1.75 in.,
Structural
response
rad.
0.02
0.01
Twist
Induced
bending
slope
Fig. 26. Structural response of a 15 solid beam at the tip due to a unit tip torsional load.
372
thickness = 0.11 in. and effective length = 30.25 in. The bending load and torque are
applied separately at the tip and tip bending slope and twist are determined. Good correlation between theory and experiment is achieved. For this beam, the induced twist and
bending slope due to unit tip bending load are of the same order.
CONCLUSION
The Vlasov theory is expanded to perform a linear analysis of open- and closedsection beams made out of general composite laminates. Transverse shear deformation of
the beams is accounted for in the theory. In order to provide experimental correlation to
the theory, graphite-epoxy beams of various cross-sections like solid rectangular, I-beam
and box beam, and two-cell airfoil covering parameters, such as ply orientation and
slenderness ratio, were fabricated and tested for their structural response under tip bending,
torsional and extensional loads. The structural response in terms of bending slope and
twist was measured using a mirror and laser system. Good correlation between theory and
experiment was achieved. Based on this study, the following conclusions are drawn:
(3)
(4)
(5)
(6)
Acknowledgements-This
research work was supported by the Army Research Office under contract number
DAAL-03-88-C-022. The Technical Monitors are Dr Robert Singleton and Dr Tom Doligalski.
REFERENCES
Bauchau, 0. A. (1987). Large displacement analysis of naturally curved and twisted composite beams. AIAA
JI 25, 1469-1415.
Borri, M. (1986). A large displacement formulation for anisotropic beam analysis. Meccanica 21, 30-37.
Chandra, R. and Chopra, I. (1991a). Experimental and theoretical analysis of composite I-beams with elastic
couplings. AZAA Jt 29, 2197-2206.
Chandra, R. and Chopra, I. (1991b). Vibration characteristics of composite I-beams with elastic couplings under
rotation. Proc. 47th Annual Forum of American Helicopter Society, Phoenix, Arizona.
Chandra, R. and Chopra, I. (1991~). Structural behavior of two-cell composite rotor blades with elastic
couplings. Proc. 17th European Rotorcraft Forum, Berlin, Germany.
Chandra, R. and Chopra, I. (1991d). Coupled composite rotor blades under bending and torsional loads. Proc.
AHS Specialists Meeting on Rotorcraft Structures, Williamsburg, VA.
Chandra, R. and Chopra, I. (1992). Experimental-theoretical
investigation of the vibration characteristics of
rotating composite box beams. Accepted for publication in the J. Aircraft.
Chandra, R., Stemple, A. D. and Chopra, I. (1990). Thin-walled composite beams under bending, torsional and
extensional loads. J. Aircraft 21, 619-627.
Ciiavotto, V. et al. (1983). Anisotropic beam theory and application. Comput. Struct. 16, 403-413.
313
Gjelsvik, A. (1981). The Theory of Thin-walled Bars. John Wiley, New York.
Hodges, D. H., Atilgan, A. R., Fulton, M. V. and Rehfield, L. W. (1991). Free-vibration analysis of composite
beams. J. Am. Helicopter
Sot. 36, 36-41.
Hodges, D. H. and Dowell, E. H. (1974). Nonlinear equations of motion for the elastic bending and torsion of
twisted nonuniform rotor blades. NASA TN D-7818.
Hong, C. H. and Chopra, I. (1985). Aeroelastic stability of a composite blade. J. Am. Helicopter
Sot. 30,
51-61.
Kosmatka, J. B. (1991). Extension-bend-twist coupling behavior of thin-walled advanced composite beams with
initial twist. Proc. 32nd AZAA/ASME/ASCE/AHS/ASC
Structures,
Structural
Dyanamics
and Materials
Conf. AIAA, Washington, DC.
Megson, T. H. G. (1974). Linear Analysis of Thin-walled
Elastic Structures.
John Wiley, New York.
Minguet, P. and Dugundji, J. (1990a). Experiments and analysis for composite blades under large deflections,
Part 1: Static behavior. AZAA JI 28, 1573-1579.
Minguet, P. and Dugundji, J. (1990b). Experiments and analysis for composite blades under large deflections,
Part 2: Dynamic behavior. AZAA JI 28, 1580-1588.
Nixon, M. W. (1987). Extension-twist coupling of composite circular tubes with application to tilt rotor blade
design. Proc. 28th AZAA/ASME/ASCE/AHS/ASC
Structures,
Structural
Dynamics
and Materials
Conf.
AIAA, Washington, DC.
Rehfield, L. W. and Atilgan, A. R. (1989). On the buckling behavior of thin walled laminated composite open
section beams. Proc. 30th AZAA/ASME/ASCE/AHS/ASC
Structures,
Structural
Dynamics
and Materials
Conf. AIAA, Washington, DC.
Rehfield, L. W., Atilgan, A. R. and Hodges, D. H. (1988). Structural modeling for multicell composite rotor
blades. Proc. 28th AZAA/ASME/ASCE/AHS/ASC
Structures,
Structural
Dynamics
and Materials
Conf.
AIAA, Washington, DC.
Rehfield, L. W., Atilgan, A. R. and Hodges, D. H. (1990). Nonclassical behavior of thin-walled composite
beams with closed cross-sections. J. Am. Helicopter
Sot. 35, 42-51.
Smith, E. C. and Chopra, I. (1991). Formulation and evaluation of an analytical model for composite box
beams. J. Am. Helicopter
Sot. 36, 23-35.
Vlasov, V. Z. (1961). Thin-walled
Elastic Beams (translated from Russian). National Science Foundation and
Department of Commerce, U.S.A.
APPENDIX: THE RELEVANT COEFFICIENTS OF STIFFNESS MATRIX [K] OF
BENDING-TORSION
AND EXTENSION-TORSION
COUPLED BEAMS
A, =
#oflayers
C Qr(h,+,
k=l
Bji = +
D, = +
- h,J
c.
Q$(h;+,
- hi)
Q;(h;+,
- h:)
(AlI
(43)
k=,
where Q$ refers to stiffness matrix of the kth layer or web in the sz plane. hr+, and h, are the coordinates of
the kth layer in the n direction from the mid-plane of the laminates as the reference surface.
(44)
K,,=S[
-2B,,
+ $%$
66
K,,
ds
(A51
(46)
A,,cosBds
15
K,,
[A,,y+
2B,,ycosB+
(48)
D,,cos21?]ds
sI
Kz, =
S[
Kz6 =
[A,ey
5
L =
-2B,,y
+ A,,Gsy
-
A66
24, COSe +
B,,G,
COSe ds
66
(Al3
(Al 1)
374
R. CHANDRA
6,
4D,,+A-4-G,2
66
Ke6 =
KT7 =
and I.
CHOPRA
&Gs
466
ds
C4W
A,, cos2 0 ds
(A131
(A141
ss