0% found this document useful (0 votes)
69 views9 pages

Carrera 2013

This document presents an analysis of free vibration in rotating composite blades using the Carrera Unified Formulation (CUF) method. CUF allows obtaining refined structural theories with variable kinematic descriptions by expanding displacement variables over beam section axes using Taylor expansions of varying order. Both flapwise and lagwise motions of isotropic, composite, and thin-walled structures are considered. Results include natural frequencies obtained by solving governing equations derived via Hamilton's principle using the finite element method.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
69 views9 pages

Carrera 2013

This document presents an analysis of free vibration in rotating composite blades using the Carrera Unified Formulation (CUF) method. CUF allows obtaining refined structural theories with variable kinematic descriptions by expanding displacement variables over beam section axes using Taylor expansions of varying order. Both flapwise and lagwise motions of isotropic, composite, and thin-walled structures are considered. Results include natural frequencies obtained by solving governing equations derived via Hamilton's principle using the finite element method.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Composite Structures 106 (2013) 317–325

Contents lists available at SciVerse ScienceDirect

Composite Structures
journal homepage: [Link]/locate/compstruct

Free vibration analysis of rotating composite blades via Carrera Unified


Formulation
E. Carrera a,b, M. Filippi a,⇑, E. Zappino a
a
Department of Mechanical and Aerospace Engineering, Politecnico di Torino, Corso Duca degli Abruzzi 24, 10129 Torino, Italy
b
King Abdulaziz University, Jeddah, Saudi Arabi

a r t i c l e i n f o a b s t r a c t

Article history: Carrera Unified Formulation (CUF) is used to perform free-vibrational analyses of rotating structures. CUF
Available online 22 June 2013 is a hierarchical formulation which offers a procedure to obtain refined structural theories that account
for variable kinematic description. These theories are obtained by expanding the unknown displacement
Keywords: variables over the beam section axes by adopting Taylor’s expansions of N-order, in which N is a free
Rotating beams parameter. Linear case (N = 1) permits us to obtain classical beam theories while higher order expansions
Finite Element Method can lead to three-dimensional description of dynamic response of blades. The Finite Element Method is
Higher-order theories
used to solve the governing equations of rotating blades that are derived in a weak form by means of
Composites structures
Hamilton’s Principle. These equations are written in terms of ‘‘fundamental nuclei’’, which do not vary
with the theory order (N). Both flapwise and lagwise motions of isotropic, composite and thin-walled
structures are traced. The Coriolis force field is included in the equations. Results are presented in terms
of natural frequencies and comparisons with published solutions are provided.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction opment of theories for the rotating structures based on the Timo-
shenko model, in which the Coriolis term has [8] or has not been
A thorough understanding of the dynamic features of rotating considered [9–12]. In all papers mentioned, the generic rotating
blades serves as a starting point for the study of fatigue effects, blade is assumed to be a compact structure constituted by isotropic
forced-response and flutter instability, which occur in airplane en- material or by orthotropic laminae. However, the design of ad-
gines, helicopters and turbomachinery. Rotating structures where vanced rotor blades has been strongly affected by the advent of
a geometrical dimension is predominant over the others are usu- composite materials which combine a high specific strength and
ally modeled as beams. Many researchers have addressed the prob- stiffness with the capability to be easily modeled. These properties
lem of the rotating beam by simplifying both the equations of produce light and efficient blades, whose dynamic characteristics
motion and the displacement formulations. For instance, Banerjee usually involve phenomena that cannot be detected by the use of
[1,2], Ozge and Kaya [3], Mei [4] and Hodges and Rutkowski [5] the classical models. For this reason, a considerable number of re-
limited their studies to the flexural vibrations of both uniform fined theories have been introduced with the purpose of describing
and tapered Euler–Bernoulli rotating beams by using, respectively, the rotating composite blade behavior. For instance, Song and Lib-
the Dynamic Stiffness Method, the Differential Transform Method rescu presented in [13] a structural model encompassing trans-
and a variable-order finite element. The assumption that the beam verse shear, secondary warping deriving from the assumption of
deforms only in bending is restrictive, since the coupling between the non-uniform torsion along the longitudinal axis and the effect
the axial deformation and the lagwise motion can be significant. In of the heterogeneity of the materials. They observed that by dis-
order to take into account this coupling the introduction of the carding the Coriolis term, the equations described separately the
Coriolis force becomes mandatory. For this reason, Hsiao et al. flap-lag deformation and the extension-twist motion, and within
solved the complete motion equations of Euler beams by using this context, they examined the ply orientation effects. Contrary
the power series solution [6] and the Finite Element Method [7]. to this ad hoc formulation, Chandiramani et al. provided a
Further improvements have been introduced by introducing en- geometrically nonlinear theory for analyzing the rotating compos-
hanced displacement fields over the blade cross-section. Indeed, ite thin-walled box beam [14], in which the non-classical effects
in the open literature, there are many papers devoted to the devel- were captured in a general way. The linearized equations of motion
were solved with the Modified Galerkin Method and the Coriolis
⇑ Corresponding author. Tel.: +39 011 090 6870. term was disregarded. Furthermore, the authors modified their
E-mail addresses: [Link]@[Link] (E. Carrera), matteo.filippi@[Link] formulation for extension to pre-twisted composite blades [15].
(M. Filippi), [Link]@[Link] (E. Zappino). See [16,17], for interesting studies on controlling thin-walled

0263-8223/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
[Link]
318 E. Carrera et al. / Composite Structures 106 (2013) 317–325

composite blades via piezoelectric patches. Jung et al. [18] devel- derivative. The kinetic energy of the whole structure can be written
oped a one-dimensional finite element based on a mixed varia- as the following expression:
tional approach in which both displacement and force Z
1
formulations were used. The walls of the considered structures T¼ qðu_ T u_ þ 2uT XT u_ þ uT XT Xu þ 2u_ T Xr
2 V
were modeled as shell and the global deformation was described
by the Timoshenko beam model. This model is suitable for com- þ 2uT XT XrÞdV ð2Þ
posite structures with open and closed contour. The dynamic of
where q is the density of the structure. In order to take into account
the rotating composite blades clearly represents a complex and
that the body can be loaded by static stresses, such as the centrifu-
interesting topic (see [19,20]), but it seems that a reliable and gen-
gal stress r0, in accordance with the linearized theory, the potential
eral method for its complete analysis is not yet available. In order
energy can be written in the following form:
to overcome the limitation of ad-hoc assumptions about the dis- Z Z
placement fields, this paper presents the Carrera Unified Formula- 1    
U¼ Tl C l dV þ r0 T nl dV ð3Þ
tion (CUF). First of all, CUF offers a procedure to obtain refined 2 V V
structural models by considering the order and the types of theo-
where l and nl are the linear and nonlinear components of the
ries as free input parameters, hence nonclassical effects such as
strain field and C is the matrix of material coefficients. In the case
warping, coupling, and torsion, are captured by simply enriching
in point, the tension due to centrifugal effect is:
the displacement fields. Secondly, by using the Finite Element  
Method, it is possible to solve complete equations of motion by 1 1
r0 ¼ X2 q rh L þ L2  rh y  y2 ð4Þ
including the Coriolis term and both Stiffness and Softening cen- 2 2
trifugal contributions. CUF was first developed for plate and shell
where L and rh are the length of the beam and the dimension of the
models [21–25] and was later extended to the beam model [26].
hub.
The refined displacement fields are written adopting N-order Tay-
lor type expansions (TE) [27] of the section coordinates. The re-
fined finite elements have been evaluated in static and dynamic 3. The unified formulation
analyses carried out on unconventional cross-sections (hollow
cylindrical reinforced structures, airfoil, bridge-like cross-sections, The CUF states that the displacement field, u(x, y, z, t), is an
etc.) in [28–35]. As far as composite beams are concerned, interest- expansion of generic functions, Fs(x, z) for the vector displacement,
ing analyses are presented in [36–39]. Many other papers devoted us(y):
to CUF are available in the literature but, for a thorough and clear uðx; y; z; tÞ ¼ F s ðx; zÞus ðy; tÞ s ¼ 1; 2; . . . ; T ð5Þ
description, see [26].
where T is the number of terms of the expansion and, in according
to the generalized Einstein’s notation, s indicates summation. In
2. The kinetic and potential energies
this work, functions Fs in Eq. (5) are Taylor-like expansions. For
example, the second-order displacement field is:
When a structure is rotating, inertial forces and moments are
observed. In this study, a beam that is free to rotate about one of ux ¼ ux1 þ x ux2 þ z ux3 þ x2 ux4 þ xz ux5 þ z2 ux6
its transverse axes with a constant spin-speed X is considered.
uy ¼ uy1 þ x uy2 þ z uy3 þ x2 uy4 þ xz uy5 þ z2 uy6 ð6Þ
Fig. 1 shows the reference frame of the beam. The absolute velocity
2 2
of the point P is the sum of the relative velocity and the transfer uz ¼ uz1 þ x uz2 þ z uz3 þ x uz4 þ xz uz5 þ z uz6
velocity. while the third-order displacement field becomes:
v abs ¼ v rel þ v tr ¼ u_ þ X  rtot ð1Þ ux ¼ ux1 þ x ux2 þ z ux3 þ x2 ux4 þ xz ux5 þ z2 ux6 þ x3 ux7 þ x2 z ux8 þ xz2 ux9 þ z3 ux10
uy ¼ uy1 þ x uy2 þ z uy3 þ x2 uy4 þ xz uy5 þ z2 uy6 þ x3 uy7 þ x2 z uy8 þ xz2 uy9 þ z3 uy10
2 3 uz ¼ uz1 þ x uz2 þ z uz3 þ x2 uz4 þ xz uz5 þ z2 uz6 þ x3 uz7 þ x2 z uz8 þ xz2 uz9 þ z3 uz10
0 X 0
6 7 ð7Þ
X ¼ 4X 0 05
0 0 0 A remarkable feature is that classical beam theories are obtainable
as particular cases of Taylor expansions. It should be noted that
where u={ux uy uz}T is the displacement vector, rtot = r + u is the classical theories require reduced material stiffness coefficients to
distance of P from the neutral axis and the dot indicates the time contrast Poisson’s locking. Unless otherwise specified, for classical

y
rh

Fig. 1. Reference system.


E. Carrera et al. / Composite Structures 106 (2013) 317–325 319

and first-order models Poisson’s locking is corrected according to € þ GX q_ þ ðK  K X þ K r0 Þq ¼ 0


Mq ð15Þ
Carrera et al. [26].
Now, it is possible to assume a periodic function as solution:
If a classical Finite Element technique is adopted with the pur-
pose of easily dealing with arbitrary lamination schemes and gen-  eixt
q¼q ð16Þ
eric boundary conditions, the generalized displacement vector
where q is the amplitude vector of the motion, ‘i’ is the imaginary
becomes:
unit and x the circular or angular frequency. Substituting Eq. (16)
us ðyÞ ¼ N i ðyÞqsi ðtÞ ð8Þ
and its derivative in Eq. (15), it becomes:
where Ni(y) are the shape functions and qsi(t) is the nodal displace-  eixt ½ðK  K X þ K r0 Þ þ ðGX Þix  ðMÞx2  ¼ 0
q ð17Þ
ment vector:
n oT
qsi ðtÞ ¼ q ux quy q uz ð9Þ
si si si 5. Numerical results

In order to assess the theory proposed, several illustrative


4. Equations of motion in CUF form
examples are presented. The boundary conditions and the beam
dimensions are assumed to be problem parameters. With the pur-
In order to obtain the motion equations, Hamilton’s principle is
pose of enabling a general application of results, if not otherwise
used, and in the usual form it states:
declared, they are presented in non-dimensional form by adopting
Z t1 the following expressions:
d ðT  UÞ dt ¼ 0 ð10Þ
t0
Table 1
where d is the virtual variation of the functional. Dependency of the first three dimensionless natural frequency parameters on the
variations of the dimensionless angular speed and hub dimension of a Euler–Bernoulli
The assumptions made in the above section allow us to rewrite
beam.
Eq. (10) as follow:
Z t1     B.C. x⁄ Theory X⁄ = 1 X⁄ = 5
€ sj þ dqTsi Gijss q_ sj þ dqTsi K ijss  K ijXss þ K ijrss qsj þ dqTsi F iXs r dt
dqTsi M ijss q dt = 0 dt = 1 dt = 2 dt = 3
0
t0
CF 1 [1] 3.6816 3.8888 10.862 12.483
¼0 Present 3.6816 3.8895 10.866 12.488
ð11Þ 2 [1] 22.181 22.375 32.764 35.827
Present 22.178 22.375 32.773 35.840
The matrices are written in form of the fundamental nuclei and 3 [1] 61.842 62.043 73.984 77.935
their explicit forms are: Present 61.836 62.032 73.986 77.939
CP 1 [1] 15.513 15.650 22.663 24.729
M ijss ¼ Iijl / ðF s qIF s Þ. Present 15.512 15.649 22.670 24.738
Gijss ¼ Iijl / ðF s qIF s Þ . 2X 2 [1] 50.093 50.277 60.906 64.382
Present 50.092 50.275 60.919 64.399
ijss
KX ¼ Iijl / ðF s IF s Þ . XT X
q 3 [1] 104.39 104.59 116.99 121.30
ijss Present 104.42 104.62 117.04 121.36
K r0 ¼ Ii;yj;y q
lr0 / ðF s IF s Þ . X X
T
PP 1 [1] 10.022 10.264 19.684 22.078
ijss
K ¼ Iijl / DTnp ðF s IÞ½C np Dp ðF s IÞ þ C nn Dnp ðF s IÞ Present 10.021 10.264 19.690 22.086
ð12Þ 2 [1] 39.642 39.889 53.132 57.235
þ DTp ðF s IÞ½C pp Dp ðF s IÞ þ C pn Dnp ðF s IÞ. Present 39.638 39.886 53.141 57.248
h i 3 [1] 88.991 89.241 103.92 108.93
þ Iij;y
l / DTnp ðF s IÞ þ DTp ðF s IÞC pn F s . þI Xy Present 89.003 89.253 103.94 108.96

þ Ii;yj T
l I Xy / F s ½C np Dp ðF s IÞ þ C nn Dnp ðF s IÞ.

þ Ii;yj;y
l I TXy I Xy / F s C nn F s . Table 2
is Non-dimensional fundamental frequency x⁄ of a cantilever Timoshenko beam as a
FX ¼ Iil / F s qr . X X T
function of the rotating speed.

where: X⁄ [9] Present Diff. (%)


2 3
0 0 1 Z 0 3.4798 3.4831 0.09
6 7 1 3.6452 3.6494 0.11
IX y ¼ 41 0 05 / . . . .X ¼ . . . dX ð13Þ 2 4.0994 4.1064 0.17
X
0 1 0 3 4.7558 4.7667 0.22
4 5.5375 5.5530 0.27
  Z 5 6.3934 6.4144 0.32
ij; i; j i; j; i; j;
Iil ; Iijl ; Il y ; Il y ; Il y y ; Ilry0 y ¼ ðN i ;N i Nj ; N i Nj;y ; Ni;y N j ; N i;y N j;y ; r0 Ni;y Nj;y Þ dy 6 7.2929 7.3205 0.37
l 7 8.2184 8.2538 0.43
ð14Þ 8 9.1596 9.2039 0.48
9 10.109 10.164 0.54
ijss ijss
In addition to the Mass Matrix M and the Stiffness Matrix K the
expression of the kinetic energy introduces other three terms:

 the Coriolis Matrix Gijss


 the Centrifugal Softening Matrix K ijXss
 the Centrifugal Stiffening Matrix K ijrs0s br bt
 the Load vector F iXs

To obtain the natural frequencies and the normal modes of the


rotor, we have to solve the homogeneous equations: Fig. 2. Tapered Beam.
320 E. Carrera et al. / Composite Structures 106 (2013) 317–325

Table 3 Table 4
Dependency of the first three dimensionless natural frequency parameters on the Dependency of the first dimensionless fundamental frequency parameter on the
variations of the dimensionless angular speed for uniform (dt = 1) and tapered (dt = 0) variations of the dimensionless angular speed and hub dimension for the chordwise
beams. motion.

X⁄ x⁄ Uniform Tapered dt X⁄ Present [40] Diff. (%)


[5] Present Diff. (%) [5] Present Diff. (%) 0 2 3.6196 3.6173 0.06
10 4.9700 4.9619 0.16
0 1 3.5160 3.5164 0.01 3.8237 3.8569 0.87
50 7.3337 7.4537 1.63
2 22.034 22.034 0.00 18.317 18.375 0.31
3 61.697 61.685 0.02 47.264 47.442 0.38 1 2 4.3978 4.3960 0.04
10 13.048 13.047 0.01
1 1 3.8888 3.8895 0.18 3.9866 4.0505 1.60
50 41.227 41.346 0.28
2 22.375 22.375 0.00 18.474 18.559 0.46
3 62.043 62.032 0.02 47.417 47.625 0.32 5 2 6.6430 6.6421 0.01
10 27.266 27.276 0.03
5 1 8.9403 8.9443 0.04 6.7344 7.2335 7.41
50 74.003 74.178 0.23
2 29.352 29.360 0.03 21.905 22.538 2.89
3 69.760 69.772 0.17 50.933 51.821 1.74
10 1 16.606 16.615 0.05 11.501 12.638 9.88
2 44.368 44.384 0.04 30.182 31.857 5.55 where Jxx is the moment of inertia about x-axis, E is the Young Mod-
3 89.156 89.169 0.01 60.564 63.043 4.09 ulus and A the area of the cross-section. In the first test case, a thin
sffiffiffiffiffiffiffiffiffiffiffi rotating beam is considered and the non-dimensional results
x X qAL4 rh obtained in CUF form are compared with those presented in [1] in
x ¼ ; X ¼ ; x0 ¼ dt ¼ which the Dynamic Stiffness Method were employed. As can
x0 x0 EJ xx L
sffiffiffiffiffiffiffiffi be seen in Table 1, the results are in strong agreement with the
AL2 reference for all kinds of boundary conditions (B.C.) and for
S¼ the considered hub off-set ratios. The boundary conditions
J xx

(a)

(b) 400
S2
350

300
Frequency parameter (ω*)

250

200
S1
150

100

50

0
0 20 40 60 80 100 120
Rotating Speed Parameter (Ω *)

Fig. 3. Variation of the non-dimension natural frequencies for the speed parameter when d = 0.1 and S = 70 – ‘‘’’: flapwise motion and ‘‘’’: chordwise motion.
E. Carrera et al. / Composite Structures 106 (2013) 317–325 321

Table 5
Material and dimensions of graphite-epoxy beams.

Case Thickness (mm) Orientation (°)


1 2.97 0
2 3.22 15
Property Dimension (GPa)
E11 141.9
E22 = E33 9.780
G21 = G31 6.140
G32 5.520
Fig. 5. Sketch of the box beam.
m12 = m13 0.42
m23 0.54

between the two sets of results is evident, in fact, the relative error
remains below 1%.
are clamped-free (C  F), clamped-supported (C  P) and sup- In the next example, a uniform and a tapered cantilever beam
ported-supported (P  P). It must be highlighted that the compari- are considered, for which hub ratios are assumed to be 1 and 0,
son becomes possible if both Coriolis and the Softening matrices respectively. For the tapered beam, the variation of the depth of
are disregarded. In the second analysis, the structure is no longer the rectangular cross-section is linear and the taper ratio, bt/br is
so thin (S = 30) as to consider the approximations of Euler–Bernoulli assumed to be 0.5 (see Fig. 2). For reference purposes, the results
model valid and, for this reason, the first order shear deformation presented in [5] have been reported. As far as the uniform section
theory is needed. The variation of the fundamental non-dimen- is concerned, the results in Table 3 show a strong agreement be-
sional frequency of a cantilever Timoshenko beam with the rotating tween the present theory and the reference one, whereas there
speed is shown in Table 2 and it is compared with [9]. It is impor- are some discrepancies between the two models for the tapered
tant to note that no shear correction factor is used. The agreement case especially in the computation of the fundamental frequency,

(a)

(b)

Fig. 4. Frequencies versus angular speed – ‘‘’’: present (TE2), ’’}’’: present (TE4), ‘‘h’’: theory [41], ’’4’’: theory [42] and ‘‘’’: experimental data [42]. (a) 0°, and (b) 15°.
322 E. Carrera et al. / Composite Structures 106 (2013) 317–325

Table 6 conceived for tapered structures, while in contrast, in the present


Angular frequencies (rad/s) of the thin-walled box. paper the beam is described with 40 constant cross-section beam
Ply angle x TE2 TE7 2D elements where the width is uniformly decreased along the axis.
0° x1 190.92 164.46 155.99F Possible remedies could be to increase the number of uniform
x2 625.38 599.49 597.50L beams or to use a suitable formulation for tapered beams.
45° x1 40.356 36.788 36.083F/T Since CUF allows us to describe the behavior of a body com-
x2 194.18 152.53 145.07L/T pletely, it is possible to detect all kinds of normal modes of a rotat-
60° x1 32.358 31.899 31.597F/T ing beam, including the chordwise motion. Consequently, in the
x2 133.77 129.98 128.47L/T following analysis these modes have also been studied and their
90° x1 31.372 30.680 30.322F frequencies are compared with those shown in [40]. As can be seen
x2 127.54 125.41 123.44L from Table 4, the results match those of the reference for all values
of non-dimensional speed and hub off-set ratio. Fig. 3 shows how
F: bending mode in z – direction.
L: bending mode in x – direction. the frequency parameter varies with the rotating speed for both
/T: bending/torsional coupled mode. chordwise (empty dots) and flapwise (black dots) motions. In
Fig. 3a, in accordance with [40], it is possible to note that the fre-
for which the error increases significantly with the rotating speed. quencies related to the stretching modes (S) vary with the speed
The reason for this difference is due to the discretization along the in the chordwise motion while, in the flapwise case, these frequen-
longitudinal axis. In fact, in [5], the structure has been approxi- cies remain almost constant. However some remarks are needed
mated by adopting a variable-order Finite Element Method about the occurrence of the divergence stability. In fact, in the first

(a)

(b)

Fig. 6. First angular frequency versus rotational speed for different ply angles – ‘‘solid lines’’: TE2; ‘‘dotted lines’’:TE4. (a) x1 (b), and x2.
E. Carrera et al. / Composite Structures 106 (2013) 317–325 323

Fig. 7. Sketch of the section beam and lamination scheme.

figure, the frequency parameter related to the first bending mode experimental data presented in [42] have been reported in graph-
in chordwise motion decreases to 0 since the non-linear term, ical form. As is shown in Fig. 4, the frequencies obtained with the
uy,y, has been disregarded, but if this term is included the dynamic present theory are in good agreement with the other theoretical
behavior undergoes an observable change. In fact, in Fig. 3b, it is approaches and with the experimental data. It is interesting to note
evident that the dimensionless frequencies, S1 and S2, start to in- that, the second-degree polynomial is enough to detect the bend-
crease with the rotating velocity and, furthermore, the divergence ing modes and their related frequency values, whereas for vibra-
instability no longer occurs. Therefore, the term uy,y is crucial in the tion modes which are dominated by torsion, further refinements
analysis of rotating beams especially when chordwise motion is of the displacement fields are needed in order to improve the solu-
studied. tions. In fact by adopting the fourth-degree polynomial, the fre-
Inasmuch as the Unified Formulation allows us to vary the kine- quencies related to the normal modes shown in Fig. 4 decrease,
matic theories, in these last analyses, higher-order models have getting close to reference values. The remaining frequencies la-
been tested. The structures considered are two graphite-epoxy beled with BC are related to chordwise modes.
rectangular beams. For both beams, the width, the length and the In order to investigate the effects of the ply angle and the rota-
hub off-set are assumed to be 25.4 mm and 800.01 mm and tional speed on the frequencies of more complex and realistic
63.5 mm, respectively. The thickness and the fiber orientations structures, in the following example a thin-walled box is studied.
are different for the two cases and they are listed in Table 5 with The geometry features (see Fig. 5) and the material properties
the material properties. The reference solutions related to the being given by:
flapwise normal modes are taken from [41], in which some
EL ¼ 206:8 GPa ET ¼ 5:17 GPa GLT ¼ 2:55 GPa
GTT ¼ 3:10 GPa
Table 7 mLT ¼ mTT ¼ 0:25; q ¼ 1528:15 Kg=m3
Material properties and dimensions of composite I beam
in Fig. 7. b ¼ 0:0508 m c ¼ 0:2540 m t ¼ 0:01016 m

Property Dimension where L and T denote parallel and transverse directions to the fi-
E11 78.5 (Gpa) bers. The length of beam and the hub-ratio are assumed to be
E22 5.51 (Gpa) 2.032 m, respectively. In [13], the dependency of the first angular
G23 2.07 (Gpa) frequency by the rotational speed and by the ply angle orientations
G31 2.07 (Gpa)
has been evaluated using a refined theory which takes into account
G12 2.34 (Gpa)
m12 0.34 some non-classical effects such as the torsion of the cross-section.
q 1241 (kg/m3) Nevertheless the Coriolis terms were discarded so as to split the
tp 0.2286 (mm) equations describing the flap-lag bending deformation from those
h 12.7 (mm) related to the extension-twist motion. In the present analyses, the
b 25.4 (mm)
Length 844.55 (mm)
Coriolis matrix is included as well as the term uy,y. In Table 6, the
first two angular frequencies computed with TE2 and TE7 are com-
pared with those obtained with shell FEM solutions, in which
QUAD4 of MSC NASTRANÓ have been used. The increasing number
Table 8
The first two frequencies (Hz) of the composite I beam at standstill.
of terms in the expansions leads to a general improvement of the
solution, whose accuracy strongly depends on the ply angle. Fig. 6
TE2 TE4 TE6 Theory [18] Exp. [18] aims to show how the first two angular frequencies change with
F 23.62 23.43 23.35 27.60 23.0 the rotational speed for different fiber orientations. The solid lines
L 28.00 27.09 27.04 29.60 – are obtained with the second-order expansion TE2, whereas the
F: bending mode in z – direction. dotted curves with the TE4 expansion. With these two displace-
L: bending mode in x – direction. ment models, it is possible to detect the flexural-twist coupling
–: value not provided. when the ply angles are different from 0° and 90°. By considering
324 E. Carrera et al. / Composite Structures 106 (2013) 317–325

32 evolution of the natural frequencies and relating modes with


increasing accuracy with respect to the classical theories, also
yielding remarkable results for shell-like deformations. In the light
30 of these considerations, the proposed one-dimensional theory ap-
pears very appropriate for the study of rotating and spinning struc-
Frequency (Hz)

tures, since it combines a simple formulation with the possibility of


28
obtaining refined models of arbitrary accuracy. The research re-
ported in this paper is expected to stimulate further research on
26
dynamic behavior of more complex rotating structural systems.

References
24
[1] Banerjee JR. Free vibration of centrifugally stiffened uniform and tapered
beams using the dynamic stiffness method. J Sound Vib 2000;233:857–75.
[2] Banerjee JR, Su H, Jackson DR. Free vibration of rotating tapered beams using
22
0 200 400 600 800 1000 the dynamic stiffness method. J Sound Vib 2006;298:1034–54.
[3] Ozge Ozdemir Ozgumus, Kaya Metin O. Flapwise bending vibration analysi of
Rotating Speed (rpm) doube tapered rotating Euler–Bernoulli beam by using the differential
transform method. Meccanica 2006;41:661–70.
Fig. 8. Frequencies versus angular speed – ‘‘solid lines’’: present (TE2), ‘‘dotted [4] Mei C. Application of differential transformation technique to free vibration
lines’’: present (TE6), ‘‘’’: theory [18] and ‘‘’’: experimental data [18]. analysis of a centrifugally stiffened beam. Comput Struct 2008;86:1280–4.
[5] Hodges DH, Rutkowski MJ. Free-vibration analysis of rotating beams by a
variable-order finite-element method. AIAA J 1981;19:1459–66.
the graph 6-a it should be noted that, under a certain value of rota- [6] Huang Chih Ling, Lin Wen Yi, Hsiao Kuo Mo. Free vibration analysis of rotating
tional speed depending on the fiber orientation, the dominant mod- euler beams at high angular velocity. Comput Struct 2010;88:991–1001.
al shape that exhibits the lowest frequency is the flapwise mode. [7] Tsai Ming Hsu, Lin Wen Yi, Hsiao Kuo Mo. Investigation on steady state
deformation and free vibration of a rotating inclined euer beam. Int J Mech Sci
Due to the centrifugal effect, this frequency rapidly increases until
2011;53:1050–68.
the frequency lag mode becomes the lower of the two. The same [8] Stoykov S, Ribeiro P. Vibration analysis of rotating 3d beams by p-version finite
phenomenon is observed in graph 6-b for the second angular fre- element method. Finite Elem Anal Des 2013;65:76–88.
quency but, in this case, the threshold speeds determine the inter- [9] Lee SY, Kuo YH. Bending frequency of a rotating timoshenko beam with
generally elastically restrained root. J Sound Vib 1993;162:243–50.
change between the lag/torsional modal shapes and those that [10] Rao SS, Gupta RS. Finite element vibration analysis of rotating timoshenko
involve flap/extensional deformations. Moreover, for both cases, beams. J Sound Vib 2001;242:103–24.
the curves computed with TE4 undergo a translation toward lower [11] R Banerjee J. Dynamic stiffness formulation and free vibration analysis of
centrifugally stiffened timoshenko beams. J Sound Vib 2001;247:97–115.
values, whereas the trends remain almost similar. [12] Yoo Hong Hee, Lee Seung Hyun, Shin Sang Ha. Flapwise bending vibration
In the last analysis, a rotating composite I beam is considered. analysis of rotating multi-layered composite beams. J Sound Vib 2005;286:
The cross-section and the lamination scheme are shown in Fig. 7, 745–61.
[13] Song O, Librescu L. Structural modeling and free vibration analysis of rotating
whereas the beam dimensions and the material properties are composite thin-walled beams. J Am Helicopter Soc 1997;42:358–69.
listed in Table 7. The reference solutions are taken from [18] in [14] Chandiramani NK, Librescu L, Shete CD. On the free-vibration of rotating
which a refined beam element has been developed. This analysis composite beams using a higher-order shear formulation. Aerospace Sci
Technol 2002;6:545–61.
aims to compare the first two frequencies corresponding to the [15] Chandiramani NK, Librescu L, Shete CD. Vibration of higher-order-shearable
flapwise and lagging modal shapes. To this end, Table 8 shows pretwisted rotating composite blades. Int J Mech Sci 2003;45:2017–41.
these frequencies obtained with a number of models at standstill. [16] Na S, Librescu L, Shim JK. Modeling and bending vibration control of
nonuniform thin-walled rotating beams incorporating adaptive capabilities.
As far as the flapwise mode is concerned, the TE expansions are in
Int J Mech Sci 2003;45:1347–67.
good agreement with the experimental data, while in contrast, [17] Song O, Oh SY, Librescu L. Dynamic behavior of elastically taiored rotating
they present considerable discrepancies with the solution pro- blades modeled as pretwisted thin-walled beams and incorporating adaptive
posed in [18]. These differences are lower for the lagging fre- capabilities. Int J Rotating Mach 2002;8:13–25.
[18] Jung Sung Nam, Nagaraj VT, Chopra Inderjit. Refined structural dynamics
quency. As usual, in Fig. 8, the evolutions of the frequencies with model for composite rotor blades. AIAA J 2001;39:339–48.
the rotational speeds are shown and, also in this case, the TE6 [19] Hodges DH. Review of composite rotor blade modeling. AIAA J 1990;28:561–5.
expansion leads to a result near the experimental reference. [20] Jung Sung Nam, Nagaraj VT, Chopra Inderjit. Assessment of composite rotor
blade modeling techniques. J Am Helicopter Soc 1999;44:188–205.
[21] Carrera E. Theories and finite elements for multilayered, anisotropic,
6. Conclusion composite plates and shells. Arch Comput Methods Eng 2002;9(2):87–140.
[22] Carrera E. Theories and finite elements for multilayered plates and shells: a
unified compact formulation with numerical assessment and benchmarking.
In the present paper, Carrera’s Unified Formulation has been Arch Comput Methods Eng 2003;10(3):216–96.
developed and extended to the free vibration of rotating beams. [23] Carrera E, Cinefra M, Nali P. MITC technique extended to variable kinematic
multilayered plate elements. Compos Struct 2010;93:1888–95.
The equations of motion were derived through Hamilton’s Princi-
[24] Carrera E, Miglioretti F, Petrolo M. Accuracy of refined finite elements for
ple and they have been solved with the Finite Element Method. laminated plate analysis. Compos Struct 2011;93:1311–27.
Many rotating beams were considered and, in order to assess the [25] Fazzolari FA, Carrera E. Advanced variable kinematics ritz and galerkin
new theory, the results were compared with those published in formulations for accurate buckling and vibration analysis of anisotropic
laminated composite plates. Compos Struct 2011;94:50–67.
the literature, where possible. The analyses have concerned isotro- [26] Carrera E, Giunta G, Petrolo M. Beam structures. Classical and advanced
pic, composite and non-uniform beams as well as thin-walled theories. Wiley; 2011.
structures. The hierarchical property of CUF has enabled refined [27] Carrera E, Giunta G. Refined beam theories based on a unified formulation. Int J
Appl Mech 2010;2:117–43.
displacement fields with Taylor-type expansions to be obtained. [28] Carrera E, Giunta G, Nali P, Petrolo M. Refined beam elements with arbitrary
The equations are written in a fully three dimensional form. The cross-section geometries. Comput Struct 2010;88(5–6):283–93. http://
stretching and the chordwise motions are coupled by Coriolis [Link]/10.1016/[Link].2009.11.002.
[29] Carrera E, Petrolo M, Zappino E. Performance of CUF approach to analyze the
matrix and, for this reason, the non-linear term uy,y in the centrif- structural behavior of slender bodies. J Struct Eng 2012;138:285–98.
ugal Stiffening matrix appears needed in order to avoid wrong pre- [30] Carrera E, Petrolo M, Nali P. Unified formulation applied to free vibrations
dictions of instability. Furthermore, in the case of laminated beams finite element analysis of beams with arbitrary section. Shock Vib
2011;18(3):485–502. [Link]
and thin-walled boxes, the refined elements have described the
E. Carrera et al. / Composite Structures 106 (2013) 317–325 325

[31] Petrolo M, Zappino E, Carrera E. Refined free vibration analysis of one- [37] Giunta G, Biscani F, Belouettar S, Ferreira AJM, Carrera E. Free vibration
dimensional structures with compact and bridge-like cross-sections. Thin– analysis of composite beams via refined theories. Composites: Part B 2013;44:
Walled Struct 2012;56:49–61. [Link] 540–52.
[32] Carrera E, Petrolo M, Varello A. Advanced beam formulations for free vibration [38] Giunta G, Metla N, Koutsawa Y, Belouettar S. Free vibration and stability
analysis of conventional and joined wings. J Aerospace Eng 2012;25:282–93. analysis of three-dimensional sandwich beams via hierarchical models.
[Link] Composites: Part B 2013;47:5326–38.
[33] Giunta G, Koutsawa Y, Belouettar S, Hu S. Static, free vibration and stability [39] Carrera E, Filippi M, Zappino E. Laminated beam analysis by polynomial,
analysis of three-dimensional nano-beams by atomistic refined models trigonometric, exponential and zig-zag theories. Eur J Mech – A/Solids 2013;
accounting for surface free energy effect. Int J Solids Struct 2013;50:1460–72. 41:58–69.
[34] Biscani F, Giunta G, Belouettar S, Carrera E, Hu S. Variable kinematic beam [40] Chung J, Yoo HH. Dynamic analysis of a rotating cantilever beam by using the
elements coupled via Arlequin method. Compos Struct 2011;93:697–708. finite element method. J Sound Vib 2002;249:147–64.
[Link] [41] Hodges DH, Shang X, Cesnik CES. Finite element solution of nonlinear intrinsic
[35] Carrera E, Zappino E, Filippi M. Free vibration analysis of thin–walled cylinders equations for curved composite beams. J Am Helicopter Soc 1996;41:313–21.
reinforced with longitudinal and transversal stiffeners. J Vib Acoust 2013;135. [42] J Epps J, Chandra R. The natural frequencies of rotating composite beams with
[Link] 011019-1–011019-11. tip sweep. J Am Helicopter Soc 1996;41:29–36.
[36] Giunta G, Crisafulli D, Belouettar S, Carrera E. Hierarchical theories for the free
vibration analysis of functionally graded beams. Compos Struct 2011;94:
68–74.

You might also like