+ All Categories
Home > Documents > Enhanced structural behavior of flexible laminated composite beams

Enhanced structural behavior of flexible laminated composite beams

Date post: 28-Nov-2023
Category:
Upload: techio
View: 0 times
Download: 0 times
Share this document with a friend
15
Enhanced structural behavior of flexible laminated composite beams Yulia Fridman, Haim Abramovich * Faculty of Aerospace Engineering, Technion IIT, 32000 Haifa, Israel Available online 3 June 2007 Abstract The aim of the present study is to investigate analytically and numerically the structural behavior of laminated composite beams under axial compression using piezoelectric layers. A mathematical model was developed based on a first order shear deformation theory (FSDT) which includes shear deformations, usually neglected in the classical lamination theory of composite structures. Closed form solutions for the bending angle and the axial and lateral displacements along the beam are presented for various boundary conditions. Natural frequencies and their associated mode shapes, as well as, buckling loads were computed for beams with and without piezo- electric layers influence, having various boundary conditions and lay-ups. Next, the influence of the piezoelectric layers on the axial compression load and the natural frequencies is investigated to yield an enhancement of the structural behavior of the beam. This is done using a proportional control load, in which the sensed voltage on the beam is fed back (after being amplified using a constant gain G), onto the PZT actuators which prevent the premature buckling of the flexible beam by actively increasing its stiffness. A parametric investigation was performed for beams with various lay-ups, and it was shown that for a given feedback gain value, the natural frequencies and the buckling loads can be increased by a factor of two, when using the enhancement procedure developed within the present study. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: Laminated composite beam; PZT; Buckling load; Gain; Enhanced buckling load; Piezoelectric patch 1. Introduction The issue of increasing the buckling loads and natural frequencies of beams and plates using intelligent materials such as piezoelectric or shape memory alloy (SMA) mate- rials had recently become very popular. However most of the studies concentrate on basic mod- els, like Euler–Bernoulli beam theory, while most of the emphasis is made on the control part. Also most of the works had only focused on the control of structural vibra- tions. Typical references are next highlighted. Abramovich and Livshits [1] investigated the dynamical behavior of piezo-laminated composite beams with general non-symmetric lay-up. The natural frequencies of the beam with various boundary conditions were calculated and the influence of continuous piezoelectric layers bonded on the top and bottom of the beams, acting as actuators in the open loop, was investigated. Song et al. [2] studied the active vibration control of composite beams using piezoelectric ceramic patches as sensors and actua- tors. Two control algorithms were developed to achieve the active vibration damping. The results were calculated based on the theoretical Euler–Bernoulli model, and numerical simulations were obtained using the ANSYS finite element code. A good correlation was shown between the numerical predictions and the experimental results. Huang and Sun [3] developed a beam model based on the Reissner–Mindlin plate theory to demonstrate the dynamic analysis of composite beams with bonded or embedded composite sensors and actuators. Waisman and Abramovich [4] studied the stiffening effects of smart composite beams with piezo-ceramics layers or patches bonded on the surface of the beam. The analysis considers the linear piezoelectric constitutive relations and the first 0263-8223/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.compstruct.2007.05.007 * Corresponding author. Tel.: +972 4 8293199; fax: +972 4 8292303. E-mail addresses: [email protected] (Y. Fridman), haim@ aerodyne.technion.ac.il (H. Abramovich). www.elsevier.com/locate/compstruct Composite Structures 82 (2008) 140–154
Transcript

www.elsevier.com/locate/compstruct

Composite Structures 82 (2008) 140–154

Enhanced structural behavior of flexible laminated composite beams

Yulia Fridman, Haim Abramovich *

Faculty of Aerospace Engineering, Technion IIT, 32000 Haifa, Israel

Available online 3 June 2007

Abstract

The aim of the present study is to investigate analytically and numerically the structural behavior of laminated composite beamsunder axial compression using piezoelectric layers. A mathematical model was developed based on a first order shear deformation theory(FSDT) which includes shear deformations, usually neglected in the classical lamination theory of composite structures. Closed formsolutions for the bending angle and the axial and lateral displacements along the beam are presented for various boundary conditions.

Natural frequencies and their associated mode shapes, as well as, buckling loads were computed for beams with and without piezo-electric layers influence, having various boundary conditions and lay-ups.

Next, the influence of the piezoelectric layers on the axial compression load and the natural frequencies is investigated to yield anenhancement of the structural behavior of the beam. This is done using a proportional control load, in which the sensed voltage onthe beam is fed back (after being amplified using a constant gain G), onto the PZT actuators which prevent the premature bucklingof the flexible beam by actively increasing its stiffness.

A parametric investigation was performed for beams with various lay-ups, and it was shown that for a given feedback gain value, thenatural frequencies and the buckling loads can be increased by a factor of two, when using the enhancement procedure developed withinthe present study.� 2007 Elsevier Ltd. All rights reserved.

Keywords: Laminated composite beam; PZT; Buckling load; Gain; Enhanced buckling load; Piezoelectric patch

1. Introduction

The issue of increasing the buckling loads and naturalfrequencies of beams and plates using intelligent materialssuch as piezoelectric or shape memory alloy (SMA) mate-rials had recently become very popular.

However most of the studies concentrate on basic mod-els, like Euler–Bernoulli beam theory, while most of theemphasis is made on the control part. Also most of theworks had only focused on the control of structural vibra-tions. Typical references are next highlighted.

Abramovich and Livshits [1] investigated the dynamicalbehavior of piezo-laminated composite beams with generalnon-symmetric lay-up. The natural frequencies of thebeam with various boundary conditions were calculated

0263-8223/$ - see front matter � 2007 Elsevier Ltd. All rights reserved.

doi:10.1016/j.compstruct.2007.05.007

* Corresponding author. Tel.: +972 4 8293199; fax: +972 4 8292303.E-mail addresses: [email protected] (Y. Fridman), haim@

aerodyne.technion.ac.il (H. Abramovich).

and the influence of continuous piezoelectric layersbonded on the top and bottom of the beams, acting asactuators in the open loop, was investigated. Song et al.[2] studied the active vibration control of composite beamsusing piezoelectric ceramic patches as sensors and actua-tors. Two control algorithms were developed to achievethe active vibration damping. The results were calculatedbased on the theoretical Euler–Bernoulli model, andnumerical simulations were obtained using the ANSYSfinite element code. A good correlation was shownbetween the numerical predictions and the experimentalresults. Huang and Sun [3] developed a beam model basedon the Reissner–Mindlin plate theory to demonstrate thedynamic analysis of composite beams with bonded orembedded composite sensors and actuators. Waismanand Abramovich [4] studied the stiffening effects of smartcomposite beams with piezo-ceramics layers or patchesbonded on the surface of the beam. The analysis considersthe linear piezoelectric constitutive relations and the first

Y. Fridman, H. Abramovich / Composite Structures 82 (2008) 140–154 141

order shear deformation theory. The influence of the actu-ators is evaluated by means of the pin-force model andtheir size and location along the beam are taken intoaccount. The numerical solution of the equations ofmotion are compared with the FE results and it was foundthat piezoelectric bonded actuators are yielding significantchanges in the natural frequencies and mode shapes of thebeam. Raja et al. [5] derived a finite element formulationcapable of modeling extension–bending and shear inducedactuation in an adaptive composite sandwich beam basedon Timoshenko’s beam theory. The efficiency of bothactuators in controlling the bending vibration was studiedusing modal control analysis. The shear actuator performsbetter in controlling the first three bending modes than theextension-bending actuator. Sloss et al. [6] investigated theeffect of axial load, the piezo-sensor and the actuator feed-back control on the vibration frequencies and the modeshapes of the beam. The first three vibration frequenciesof controlled and uncontrolled beam were presented forvarious values of the feedback gain, axial load and patchsizes. Karagulle et al. [7] showed that the Ansys/Multi-physics finite element code can be used for the simulationof the active vibration control of smart structures.

Mukherjee and Chaudhuri [8] presented an exact solu-tion for the feedback vibration control of piezo-laminatedcolumns, in which the signals from the sensors attachedto the structure are fed back to the actuators with a gainmultiplier. The analytical solution has been validated withexperimental studies for tip loaded piezo-laminated cantile-ver beam with collocated PVDF layers using as sensors andas actuators.

In their previous work, Mukherjee and Chaudhuri [9]developed an imperfection approach for exact solutionsof the instability of piezo-laminated symmetrical columnsunder static and dynamics axial loads. Their solution isbased on Euler–Bernoulli beam theory. A constant gainfeedback control algorithm is derived using modified stiff-ness yielding an increased Euler’s buckling load while usingpiezoelectric sensing and actuation. A limiting actuationfeedback gain is derived. Meressi and Paden [10] observedthat the buckling of a flexible Euler–Bernoulli beam can bepostponed beyond the first and under the second criticalload by stabilization of the first bending mode by meansof a feedback control using piezoelectric actuators andstrain sensors. This is followed by the state-space modelof the reduced order system and designing of a controllerby using standard linear quadratic regulator (LQR) withconstant feedback gain. Chase et al. presented optimal sta-bilizations of column buckling [11] and plate buckling [12]using MEMS-based strain sensors and embedded piezo-electric ceramic patches. The column is fixed with pinnedends and the axial load is applied dynamically. The stabil-ity of the resulting controllers is based on multi-input,multi-output (MIMO) methodology using Lyapunov’smethods. The finite element column model is presented instate-space equations. The optimal buckling controllerswere tested on a column made of G-10 fiberglass yielding

an increase in the critical buckling load by a factor of2.9. In Ref. [12] the derivation of Ref. [11] is expanded tothe case of axially loaded composite plates clamped onall four sides. The controller is designed for Pdesired =1.5Pcr. For this case, the calculations became complicatewith a high requirement for computer time and controlpower.

Thompson and Loughlan [13] performed experimentson the active buckling control of pin-ended composite col-umn strips made of graphite-epoxy using lateral deflectiondisplacement sensor and surface bonded piezo-ceramicactuators. The test procedure is outlined and load–deflec-tion plots, obtained with and without active control, arepresented. For the lay-up configurations considered, theincrease in the load carrying capability is of the order of19.8–37.1%.

Chandrashekhara and Bhatia [14] presented a finite ele-ment analysis for active buckling control of laminatedcomposite plates using piezoelectric sensors and actuators.The finite element model is based on the first order sheardeformation plate theory in conjunction with linear piezo-electric theory. The sensor output is used to determine theinput to the actuator using proportional control algorithm.The presented finite element solutions show effectiveness ofpiezoelectric materials in enhancing the buckling loads.Berlin [15] presented results of an analysis based onEuler–Bernoulli beam theory together with an experimen-tal work and showed that buckling can be preventedtrough computer-controlled adjustment of dynamicalbehavior. He used piezo-ceramic actuators bonded on thesurface of a steel column to counteract buckling. Activecontrol of the buckling allows this column to support 5.6times more axial load then the original buckling load. Thiswas done with a complicated control law mechanism.Wang and Quek [16] showed the increase of the fluttervelocity and buckling capacity of a fixed-free column, sub-jected to a follower force using a pair of piezoelectriclayers.

In the present study, a mathematical model of a piezo-laminated composite beam was developed, formulatedand applied. It is based on a first order shear deformationtheory (FSDT) which includes shear deformations, usuallyneglected in classical lamination theory of composite struc-tures, and linear piezoelectric constitutive relations. Thethree coupled partial differential equations of motion of ageneral non-symmetric piezo-laminated composite beamsubjected to axial and lateral tractions are presented andsolved for the dynamic case – to find natural frequenciesand mode shapes for an axially compressed beam, withand without axially loads, and for the static case – to findthe buckling load, the bending angle, the axial and lateraldisplacements.

Moreover, the buckling load of a laminated compositebeam with various boundary conditions is enhanced usingpiezoelectric sensors and actuators. Various lay-ups areconsidered, including symmetric and non-symmetric ones.The enhanced buckling loads are shown to be twice the

y

z

hb

hp

hp

y

z

c

h

b c

PZT 0 Structural layer 90 Structural layer

L

zq

P xP

a

Fig. 1. A laminated composite beam with continuous piezoelectric layers.(a) The axial and lateral loads applied on the beam, (b) a symmetriclaminated beam [PZT/0/90/90/0/PZT] and (c) n non-symmetric laminatedbeam [PZT/0/90/0/90/PZT].

142 Y. Fridman, H. Abramovich / Composite Structures 82 (2008) 140–154

normal ones (without using the piezoelectric layers actua-tion (G = 0)).

2. Formulation of the problem

Fig. 1 shows a piezo-laminated type beam, referred to asystem of Cartesian coordinates with the origin of the mid-plane of the beam and the x-axis being coincident with thebeam axis. It is assumed that the piezoelectric layers havethe same length as the carrying structure (the case of con-tinuously piezoelectric layers1), are perfectly glued to it andthe thickness of the adhesive can be neglected (thus neglect-ing the effect of shear lag).

The axial displacement of the beam uðx; zÞ and the lat-eral wðx; zÞ are assumed to have the following form (implic-itly assuming incompressibility in the z direction, see Ref.[17]):

Uðx; z; tÞ ¼ U 0ðx; tÞ þ zUðx; tÞ ð1ÞW ðx; z; tÞ ¼ W 0ðx; tÞ ð2Þ

where U 0ðx; tÞ and W 0ðx; tÞ are the axial and lateral dis-placements of a point on the mid-plane and Uðx; tÞ is thebending rotation of the normal to the mid-plane.

The total strain vector is the sum of the mechanicalstrain vector and the actuator induced strain vector

feg ¼ femg þ feag ð3Þwhere the mechanical normal and transverse shear strains,em

x ; emz and cm

xz, as well as the actuator induced strains eax ; e

az

and caxz, are defined as

1 It can be shown that one can solve the issue of piezoelectric patchesbonded on a flexible laminated composite beam using the present modelwith continuous piezoelectric layers, by dividing the beam into threeregions, two without the patch and one (in the middle) with it andenforcing continuity conditions at the part with the patch. The results of alaminated composite beam with bonded piezoelectric patches will bereported in a separate paper to be issued in due time.

emx ¼

oU 0

oxþ z

oUox

ð4Þ

emz ¼ 0 ð5Þ

cmxz ¼ Uþ oW 0

oxð6Þ

eax ¼

XNa

k¼1

V kðxÞdk31

zak � za

k�1

ð7Þ

where Na is the number of actuators, V kðx; tÞ is the appliedvoltage to the kth actuator having a thickness of ðza

k � zak�1Þ

and dk31 is the piezoelectric constant. The other induced

strains vanish:

eaz ¼ ca

xz ¼ 0 ð8Þ

The beam constitutive equations can be written as

Nx

Mx

Qxz

8><>:

9>=>; ¼

A11 B11 0

B11 D11 0

0 0 A55

264

375

oU0

oxoUox

/þ oW 0

ox

8><>:

9>=>;þ

E11

F 11

0

8><>:

9>=>; ð9Þ

where

Nx ¼Z h=2

�h=2

crx dz

Mx ¼Z h=2

�h=2

crxzdz

Qxz ¼Z h=2

�h=2

csxy dz

ð10Þ

rx and sxz being the normal and shear stresses respectivelyand c is the width of the beam and h is the beams totalthickness. A11, B11, D11 and A55 are the usual extensional,bending-extensional, bending and transverse shear stiffnesscoefficients defined according to the lamination theory (seeAppendix A).

E11 and F11 are the actuator induced axial force andbending moment, respectively, defined as

E11 ¼ cXNa

k¼1

ðQ11ÞakV kðx; tÞðdk

31Þ ð11Þ

F 11 ¼c2

XNa

k¼1

ðQ11ÞakV kðx; tÞðdk

31Þðzak � za

k�1Þ ð12Þ

where zk is the distance of the kth layer from the x-axis, Na

is the number of piezoelectric layers, ðQ11Þak is calculatedaccording to Eq. (13) using the material properties of pie-zoelectric material (see Table 6).

Q11 ¼ Q11 cos4 hþ Q22 sin4 hþ 2ðQ12 þ 2Q66Þ sin2 h cos2 h

ð13ÞQ55 ¼ G13 cos2 hþ G23 sin2 h ð14Þ

The angle h is the angle between the fiber direction andthe longitudinal axis of the beam. The constants Q11,Q12, Q22 and Q66 are the usually used material constants(see Table 6).

Y. Fridman, H. Abramovich / Composite Structures 82 (2008) 140–154 143

3. Equations of motion

Using a first order shear deformation theory, the gov-erning equations of motion of a general shaped piezo-lam-inated beam (see Ref. [17])

o

oxA11

oU 0

oxþ B11

oUoxþ E11

� �¼ o

ot½ðI1

_U 0Þ þ ðI2_UÞ� ð15Þ

o

oxA55 Uþ oW

ox

� �� P

oWox

� �¼ o

ot½ðI1

_W Þ� þ q ð16Þ

o

oxB11

oU 0

oxþD11

oUoxþ F 11

� �� A55 Uþ oW

ox

� �¼ o

ot½ðI3

_UÞ þ ðI2_UÞ�

ð17Þ

where

ðI1; I2; I3Þ ¼ cZ h=2

�h=2

qð1; z; z2Þdz ð18Þ

q is the mass density of each layer, the flux denotes the par-tial derivative with respect to time, t. q denotes the trans-verse distributed loading, P is the axial compressive forceapplied at both ends of the beam (see Fig. 1). Together withthe boundary conditions for the general case:

Nx ¼ A11

oU 0

oxþ B11

oUoxþ E11 ¼ P or U 0 ¼ 0 ð19Þ

Qxz ¼ A55 Uþ oWox

� �� P

oWox¼ 0 or W ¼ 0 ð20Þ

Mx ¼ B11

oU 0

oxþ D11

oUoxþ F 11 ¼ 0 or U ¼ 0 ð21Þ

Table 1 presents the boundary conditions used in the pres-ent study.

4. Sensing effect

Due to mechanical stresses applied on the beam an elec-tric charge is generated in the piezo layer. Using the Gausslaw the charge accumulated on the piezoelectric electrodesis given by

QðtÞ ¼Z

SD31dS ð22Þ

where S is the area of the PZT layer and D31 is the electricaldisplacement. For a surface mounted flexural type PZTcontinuous layer, the electric displacement, D31 is given by

D31 ¼ Q11d31emx ¼ e31e

mx ð23Þ

Table 1Boundary conditions at beam’s ends

Boundaryconditions

x = 0 x = L

Simple–Simple W = 0 Mx = 0 U0 = 0 W = 0 Mx = 0 Nx = P

Clamped–Clamped

W = 0 U = 0 U0 = 0 W = 0 U = 0 Nx = P

Clamped–Simple

W = 0 U = 0 U0 = 0 W = 0 Mx = 0 Nx = P

Clamped–Free W = 0 U = 0 U0 = 0 Qxz = 0 Mx = 0 Nx = P

Substituting Eq. (23) into Eq. (22) and assuming that theelectrodes width is equal to the beam width, we get the totalinduced charge on the electrode, by integrating over thebeam length:

QðtÞ ¼Z

Se31e

mx dS ¼ c

ZL

e31emx dx ð24Þ

where c is the beam width.The voltage sensed per unit length of the beam is defined

as

V s ¼QðtÞ

C¼ �q

C¼ e31em

x hp

�eð25Þ

where C is the PZT capacitance given as

C ¼ �eS=hp ð26Þwhere hp is PZT layer thickness and �e is a dielectric con-stant. C and �q are the capacitance and electric charge perunit of beam’s length defined as

C ¼ CL¼ �ec=hp ð27Þ

�q ¼ ce31emx ð28Þ

5. The control law (actuation mechanism)

To enhance the axial loading capability of a laminatedcomposite beam, one has to control the deflections due toapplication of mechanical loads, by using sensing and actu-ation mechanisms. When a beam is subjected to axial com-pressive loads, it deflects laterally due to the bendingmoment. A large amount of lateral deflection is reachedwhen the axial load approaches the critical buckling load.To reduce this deflection and to make the beam nearly per-fect again we need to sense the change in the deformationof the beam and to apply a ‘‘counteractive’’ bendingmoment by using the accumulated voltage in the sensorsand fed it back into the actuators. This is given by thefollowing basic proportional control law:

V a ¼ GV s ð29Þwhere G is a constant feedback gain.

Substitution of Eqs. (4) and (25) into Eq. (29) yields thevoltage applied to the actuator as function of the displace-ments field:

V a ¼ Ge31exhp

�e¼ G

e31hp

�eoU 0

oxþ z

oUox

� �¼ G

e31hp

�eoU 0

oxþ h0

oUox

� �

ð30Þ

h0 ¼hp þ hb

2

� �ð31Þ

where hp; hb-beam dimensions (see Fig. 1).The substitution of Eq. (30) into Eq. (12) yields the

expression for the induced moment F11 as function of themechanical strain, the gain G, and the piezoelectric proper-ties of both the sensing and the actuation continuous lay-ers. In our case, we consider two sensors and two

144 Y. Fridman, H. Abramovich / Composite Structures 82 (2008) 140–154

actuators because each piezo-ceramic layer can operatesimultaneously both as sensor and as actuator:

E11 ¼ 2e2

31chp

eG

oU 0

oxþ h0

oUox

� �¼ Gf

h20

oU 0

oxþ G�f

h0

oUox

ð32Þ

F 11 ¼e2

31chp

�eGh0

oU 0

oxþ h0

oUox

� �¼ G�f

h0

oU 0

oxþ G�f

oUox

ð33Þ

where

�f ¼ e231chph2

0

�eð34Þ

The limiting condition of the system stability is that thecounteractive bending moment due to the induced bendingmoment F11 (for the case of symmetrical lay-ups) must beequal and in an opposite direction to the bending momentproduced by the externally applied axial load P.

F 11 6 Mx ð35Þwhere Mx-bending moment due to mechanical load only(see Eq. (9)).

For an axially loaded beam, the lateral displacementW 0ðx; tÞ is larger than the axial displacement U 0ðx; tÞ andtherefore we can assume that:

U 00ðx; tÞ << U0ðx; tÞ ð36ÞLeading to the following expressions:

F 11 ¼ G�f U0ðx; tÞ ð37ÞMx ¼ D11U

0ðx; tÞ ð38ÞThis yields the value of the limiting gain, Glim,

Glim ¼D11

�fð39Þ

5.1. The dynamical case – calculating the beam’s natural

frequencies and mode shapes

In the present study we apply only induced moments bytwo forces with equal magnitude and opposite direction, sothe total induced axial force E11 is zero. For the dynamicsolution we assume the lateral load to vanish, q = 0. Thegeneral solution form has the following general form:

U 0ðx; tÞ ¼ u0ðxÞ � eixt

W ðx; tÞ ¼ wðxÞ � eixt

Uðx; tÞ ¼ /ðxÞ � eixt

8><>: ð40Þ

where u0ðxÞ;wðxÞ and /(x) are the amplitudes of the axialand lateral displacements and bending rotation, respec-tively, time harmonically varying with a frequency x.

Substitution of Eq. (39) into Eqs. (15)–(17) yields:

A11u000 þ B11/00 ¼ �x2I1u0 � x2I2/ ð41Þ

A55/0 þ ðA55 � P Þw00 ¼ �x2I1w ð42Þ

B11 þG�fh0

� �u000 þ ðD11 þ G�f Þ/00 � A55ð/þ w0Þ

¼ �x2I3/� x2I2u0 ð43Þ

Together with the boundary conditions:

A11u00 þ B11/0 � P ¼ 0 or u0 ¼ 0 ð44Þ

A55ð/þ w0Þ � Pw0 ¼ 0 or w ¼ 0 ð45Þ

B11 þG�fh0

� �u00 þ ðD11 þ G�f Þ/0 ¼ 0 or / ¼ 0 ð46Þ

where the prime denotes differentiation with respect to x.To calculate the natural frequencies and mode shapes of

a beam, the solution process is divided into two cases:

(1) A symmetric case – a composite beam with a symmetriclay-up (see fig. 1b).

(2) A non-symmetric case – a composite beam with anon-symmetric lay-up (see fig. 1c).

For each case, two piezoelectric layers are bonded on thetop and bottom surfaces of the beam.

5.2. The symmetric case solution

Since for the symmetric case B11 = 0,I2 = 0 and underthe assumption from Eq. (36), the equations of motion,Eqs. (41)–(43) become:

A11u000 þ x2I1u0 ¼ 0 ð47ÞA55/

0 þ ðA55 � P Þw00 ¼ �x2I1w ð48ÞðD11 þ G�f Þ/00 � A55ð/þ w0Þ ¼ �x2I3/ ð49Þ

Together with the boundary conditions:

A11u00 � P ¼ 0 or u0 ¼ 0 ð50ÞA55ð/þ w0Þ � Pw0 ¼ 0 or w ¼ 0 ð51ÞðD11 þ G�f Þ/0 ¼ 0 or / ¼ 0 ð52Þ

Eq. (46) can be solved separately assuming the followingsolution form:

u0ðxÞ ¼ B1 sinðkxÞ þ B2 cosðkxÞ; k ¼ x

ffiffiffiffiffiffiffiI1

A11

rð53Þ

where B1 and B2 are constants to be found according to thebeam’s in-plane boundary conditions. Eqs. (48) and (49)can be decoupled yielding the partial differential equationfor w only:

a2wIV þ a1w00 þ a0w ¼ 0 ð54Þ

a0 ¼ x2I1 � x4I3I1

A55

a1 ¼ � ðD11þG�f Þx2I1

A55� P þ x2I3 �1þ P

A55

� �

a2 ¼ ðD11 þ G�f Þ �1þ PA55

� �

8>>>><>>>>:

ð55Þ

The characteristic equation of Eq. (53) has the followingform:

a2s4 þ a1s2 þ a0 ¼ 0 ð56ÞThe solution for the lateral displacement w(x) and thebending rotation /(x) is

Y. Fridman, H. Abramovich / Composite Structures 82 (2008) 140–154 145

wðxÞ ¼ C1 sinhðk1xÞ þ C2 coshðk1xÞ þ C3 sinðk2xÞ þ C4 cosðk2xÞð57Þ

/ðxÞ ¼ E1 sinhðk1xÞ þ E2 coshðk1xÞ þ E3 sinðk2xÞ þ E4 cosðk2xÞð58Þ

where

k1 ¼ffiffiffiffis1

p; k2 ¼

ffiffiffiffiffiffiffiffi�s2

p ð59Þ

where s1 and s2 are roots of the characteristic Eq. (55).C1–C4 are constants, which will be found from the

boundary conditions of the beam. The other four con-stants, E1–E4 are depended on the first four constantsC1–C4, and can be found from the coupled equation,Eq. (48). See Appendix A for these relations.

5.3. The non-symmetric case solution

For the non-symmetric case, the three coupled equationsof motion Eqs. (41)–(43) should be first decoupled. Afterdecoupling, it is straight forward to obtain the followingform for the solutions (details see in Appendix A)

wðxÞ ¼ C1 sinhðm1xÞ þ C2 coshðm1xÞ þ C3 sinðm2xÞ ð60Þ

þ C4 cosðm2xÞ þ C5 sinðm3xÞ þ C6 cosðm3xÞ

u0ðxÞ ¼ A1 sinhðm1xÞ þ A2 coshðm1xÞ þ A3 sinðm2xÞ ð61Þ

þ A4 cosðm2xÞ þ A5 sinðm3xÞ þ A6 cosðm3xÞ

/ðxÞ ¼ E1 sinhðm1xÞ þ E2 coshðm1xÞ þ E3 sinðm2xÞ ð62Þ

þ E4 cosðm2xÞ þ E5 sinðm3xÞ þ E6 cosðm3xÞ

where

m1 ¼ffiffiffiffis1

p; m2 ¼

ffiffiffiffiffiffiffiffi�s2

p; m3 ¼

ffiffiffiffiffiffiffiffi�s3

p ð63Þand s1, s2 and s3 are the roots of the characteristic uncou-pled equation for w(x).

C1–C6 are constants which are found from the beam’sboundary conditions, while A1–A6, E1–E6, are constantsdependent on C1–C6, which can be found from the cou-pled equations, Eqs. (41)–(43) (for more details seeAppendix A).

5.4. The static case – calculation of buckling loads and the

beam’s displacement field

We shall now restrict ourselves to the static case omit-ting in Eqs. (15)–(17) the inertial terms on the right-handside. The equations of motion for the static case and con-stant properties along the beam and q = const are:

A11u000 þ B11/00 ¼ 0 ð64Þ

A55/0 þ ðA55 � PÞw00 ¼ q ð65Þ

B11 þG�fh0

� �u000 þ ðD11 þ G�f Þ/00 � A55ð/þ w0Þ ¼ 0 ð66Þ

Together with the boundary conditions of the general case,Eqs. (44)–(46).

5.5. Symmetric beams

For a symmetric laminated beam B11 = 0 and it isassumed that ou0

ox << z o/ox.

The coupled equations of equilibrium have the followingform:

A11u000 ¼ 0 ð67ÞA55/

0 þ ðA55 � P Þw00 ¼ q ð68ÞðD11 þ G�f Þ/00 � A55ð/þ w0Þ ¼ 0 ð69Þ

In this case, the longitudinal beam motion can be solvedseparately from the lateral motion yielding

u0ðxÞ ¼ g1xþ g0 ð70Þ

where g1 and g2 are constants to be found according to thebeam’s in-plane boundary conditions.

Eqs. (68) and (69) can be decoupled to yield:

KwIV þ Pw00 ¼ �q ð71ÞK/000 þ P/0 ¼ q ð72Þ

where

r ¼ 1� PA55

� �; K ¼ ðD11 þ G�f Þr ð73Þ

For a given distribution of the transverse loading, q, Eqs.(52) and (53) can be solved by looking for the homoge-neous and particular solutions. For the case of q = constthe solutions have following general form:

wðxÞ ¼ A1 cosðkxÞ þ A2 sinðkxÞ þ A3xþ A4 �qx2

2Pð74Þ

/ðxÞ ¼ B1 cosðkxÞ þ B2 sinðkxÞ þ B3 þqxP

ð75Þ

where

k ¼ffiffiffiffiPK

rð76Þ

There are seven unknowns to be determined (A1–A4, B1–B3) and only four boundary conditions. The substitutionof (74) and (75) into the uncoupled equations of motions(67) and (68) yields the additional three relations:

B1 ¼ �rkA2; B2 ¼ rkA1; B3 ¼ �A3 ð77ÞThe relations in Eq. (77) with the specific boundary condi-tions for each case give us the exact solution for w(x) and /(x). Table 2 presents the constants A1–A4 for differentboundary conditions.

For a Clamped–Simply supported beam see Appendix A(as the expression is too long).

Demanding the lateral displacement to tend to infinity(the normal definition of the buckling load) yields thebuckling load of the beam. Table 3 summarizes the closedform solutions for the buckling loads based on a FSDTapproach for different boundary conditions.

The buckling load is the minimal critical load. For Sim-ply–Simply supported and Clamped–Free symmetrical

Table 2The expressions for A1–A4-the lateral displacement solution of symmetric lay-up beams

B.C A1 A2 A3 A4

S–S � q

rk2PqðcosðkLÞ � 1Þ

Pk2r sinðkLÞqL2P

q

rk2P

C–C � qLðcosðkLÞ þ 1Þ2rkP sinðkLÞ

� qL2rkP

qL2P

qLðcosðkLÞ þ 1Þ2rkP sinðkLÞ

C–F q

P 2rk

ðLP sinðkLÞ � KkÞcosðkLÞ � qL

rPkqLP

� q

P 2rk

ðLP sinðkLÞ � KkÞcosðkLÞ

Table 3Closed form expressions for critical loads of symmetric lay-up beams

S–SPcrðnÞ ¼ p2n2ðD11 þ G�f ÞA55

p2n2ðD11 þ G�f Þ þ A55L2; n ¼ 1; 2; . . .

C–FPcrðnÞ ¼ p2n2ðD11 þ Gf ÞA55

p2n2ðD11 þ Gf Þ þ A55L2; n ¼ 1; 2; . . .

C–S F(k) � (sin(kL) � Lkrcos(kL)) = 0

C–CF ðkÞ � sinðkL

2Þ 2 sin

kL2

� �� krL cos

kL2

� �� �¼ 0

146 Y. Fridman, H. Abramovich / Composite Structures 82 (2008) 140–154

beams the analytical expression yields the minimal load forn = 1. For Clamped–Simply supported and Clamped–Clamped beams the minimal root of the characteristicequations F(k) would give the buckling load. For G = 0we get the critical loads for a symmetric laminated beamwithout the influence of the piezoelectric actuationeffect.

Table 4The expressions a1–a4; c3–c4 for lateral and axial displacements solutions of n

a1 a2

S–S �B11

A11

� q

r�k2Pa1ðcosð�kLÞ � 1Þ

sinð�kLÞ

C–C� qLðcosð�kLÞ þ 1Þ

2rkP sinðkLÞ� qL

2r�kP

C–F tanð�kLÞqL

r�kP� qK

rP 2 cosð�kLÞ� qL

Pr�k� B11

A11 cosð�kLÞ

5.6. Non-symmetric beams

For the non-symmetric beam case, the three coupledequations (64)–(66) with boundary conditions (44)–(46)have to be solved.

Eqs. (44)–(46) can be decoupled to yield:

KwIV þ Pw00 ¼ �q ð78ÞK/000 þ P/0 ¼ q ð79ÞKuIV

0 þ Pu000 ¼ 0 ð80Þ

where

B11 ¼ B11 þG�fh0

; D11 ¼ D11 þ G�f ;

r ¼ 1� PA55

; K ¼ ðD11 �B11

A11

B11Þr ð81Þ

For a given distribution of the transverse distributed load-ing q, Eqs. (78)–(80) can be solved by looking for homoge-neous and particular solutions. For the case of q = constthe solutions have the following general form:

on-symmetric lay-up beams

a3 a4 c3 c4

qL2P

�a1 � B11qA11P

þ PA11

�ar2�k

B11

A11

qL2P

�a1 � qB11 � P 2

A11P

qL2P

B11

A11

qLP

�a1 � B11qA11P

þ PA11

qLP

B11

A11

Table 5Closed form expressions for critical loads of non-symmetric lay-up beams

S–S

PcrðnÞ ¼p2n2 D11 � B11

A11B11

� �A55

p2n2 D11 � B11

A11B11

� �þ A55L2

n ¼ 1; 2; . . .

C–F

PcrðnÞ ¼p2 D11 � B11

A11B11

� �A55ð2n� 1Þ2

p2 D11 � B11

A11B11

� �ð2n� 1Þ2 þ 4A55L2

n ¼ 1:2; . . .

C–S F ð�kÞ � ðsinð�kLÞ � �kr cosð�kLÞÞ ¼ 0C–C

F ð�kÞ � sin�kL2

� �2 sin

�kL2

� �� �krL cos

�kL2

� �� �¼ 0

Y. Fridman, H. Abramovich / Composite Structures 82 (2008) 140–154 147

wðxÞ ¼ a1 cosð�kxÞ þ a2 sinð�kxÞ þ a3xþ a4 �qx2

2Pð82Þ

/ðxÞ ¼ b1 cosð�kxÞ þ b2 sinð�kxÞ þ b3 þqxP

ð83Þ

uðxÞ ¼ c1 cosð�kxÞ þ c2 sinð�kxÞ þ c3xþ c4 ð84Þ

where

�k ¼ffiffiffiffiP

K

rð85Þ

There are 11 unknowns to be determined ða1–a4; b1–b3;c1–c4Þ and only six boundary conditions. The substitutionof Eqs. (82)–(84) into the uncoupled equations of motions(64)–(66) yields the additional five relations:

b1 ¼ �r�ka2; b2 ¼ r�ka1; b3 ¼ �a3

c1 ¼ �b1

B11

A11

; c2 ¼ �b2

B11

A11

ð86Þ

The relations Eq. (86) with the specific boundary condi-tions for each case will give us the exact solution forwðxÞ;/ðxÞ and u(x). Table 4 presents the coefficients a1–a4

and c3; c4 for different boundary conditions.For a Clamped–Simply supported beam see Appendix A –

the expression is too long to be included in the table.Table 5 summarizes the closed form expressions for crit-

ical loads of a beam with a non-symmetric lay-up.The buckling load is the minimal critical load and was

found in the same way as for the symmetric lay-up case.In Table 5, the expressions for D11;B11 depend on the

gain G (see Eq. (81)). For G = 0 we get the critical load

Table 6Material properties and constants

Graphite-Epoxy PZT-5H

E1 (N/m2) 144.8 * 109 6.3 * 109

E2 (N/m2) 9.65 * 109 6.3 * 109

G12 (N/m2) 7.1 * 109 24.8 * 109

G13 (N/m2) 7.1 * 109 –G23 (N/m2) 5.92 * 109 –m12 0.3 0.28�e (f/m) – 1.593 * 10�8

d31 (m/V) – �166 * 10�12

for a non-symmetric laminated beam with structural influ-ence of piezoelectric layers but without the influence of thepiezoelectric actuation effect.

6. Results and discussion

The main aim of the present research is the investigationon the enhancement of the stability of axially loaded com-posite beams with continuous PZT layers.

Buckling loads, natural frequencies, and lateral displace-ments for beams with piezoelectric layers, having variousboundary conditions like Simply–Simply, Clamped–Clamped, Clamped–Free, Clamped–Simply, and symmetricand non-symmetric lay-ups, were computed based on aMatlab [18] written code.

Fig. 1b–c shows the cross-section of a cross-ply symmetriclaminate [PZT/0�/90�/90�/0�/PZT] and an anti-symmetric[PZT/0�/90�/0�/90�/PZT] piezo-laminated beams, respec-tively. Both structures consists of six layers where the topand bottom are PZT layers and four internal layers madeof graphite-epoxy.

The geometrical parameters of the beam and themechanical properties of the two materials (graphite-epoxyand PZT-5H) are listed in Table 6. The lateral distributedload is taken as, q = 0.05 [N/m] for all cases. The axial loadP varies between the first and second buckling load and itdepends on the feedback gain G.

The first natural frequency as function of the axial loadfor various gains for a Simply–Simply supported symmetri-cal beam is presented in Fig. 2a. For P = 0 we get the free

Graphite-Epoxy PZT-5H

h (m) 1.27 * 10�4 2 * 10�4

L (m) 0.254 0.254c (m) 0.0254 0.0254Q11 N/m2 1.457 * 1011 6.836 * 1010

Q22 N/m2 9.708 * 109 6.836 * 1010

Q12 N/m2 2.878 * 109 1.626 * 1010

Q66 (N/m2) 7.1e * 109 0q(kg/m3) 1560 7600

Fig. 2. A Simply–Simply supported symmetrical beam. (a) First natural frequency vs. axial load. (b) First squared natural frequency normalized by thefirst free frequency vs. axial load normalized by the first buckling load.

148 Y. Fridman, H. Abramovich / Composite Structures 82 (2008) 140–154

vibration frequency of a beam with a gain G and for f1 = 0the buckling load of a beam is achieved. Fig. 2b presents therelative improvement of the first natural frequency vs. therelative improvement in critical load. Fig. 3a and b showthe same results for a Clamped–Free non-symmetricalbeam. The first three natural frequencies of symmetricand anti-symmetric beams for various gains without axialload influence were calculated and presented in Tables 7and 8, respectively.

Figs. 4 and 5 show the differences of behavior between asymmetric and an anti-symmetric Clamped–Simply sup-ported beam.

As expected, one can see that the natural frequencies ofa symmetric beam are higher than the anti symmetric one,for the same boundary conditions. Increasing the gain G

Fig. 3. A Clamped–Free non-symmetrical beam. (a) First natural frequency vsfrequency vs. axial load normalized by the first buckling load.

Table 7Natural frequencies of a symmetric beam

P = 0 Simply–Simply Clamped–S

f1 (Hz) f2 (Hz) f3 (Hz) f1 (Hz)

G = 0 27.45 109.5 246.29 42.73G = 5 30.01 120.16 270.48 47.03G = 15 34.77 139.26 313.14 54.35Glim = 24.3 38.75 154.86 348.15 60.56

yields an increase in the natural frequency. The relativeimprovement is in the same order for the various boundaryconditions and for a constant gain value it is higher for abeam with non-symmetric lay-up.

The first and the second critical loads and their ratio fordifferent boundary conditions for the present symmetricand anti-symmetric beams were calculating by substitutingG = 0 into the equations presented in Tables 3 and 5 andare listed in Table 9.

As expected the buckling loads of the symmetrical beamare higher then the non-symmetrical one.

The limiting gain value was found to be Glim = 24.303for a symmetric beam and Glim = 21.5531 for a non-symmetric one. The values of the buckling loads fordifferent G, and the ratios between these loads and

. axial load. (b) First squared natural frequency normalized by the first free

imply Clamped–Free

f2 (Hz) f3 (Hz) f1 (Hz) f2 (Hz) f3 (Hz)

138.62 288.95 9.78 61.11 171.17152.15 317.28 10.74 67.16 187.88176.18 367.41 12.33 77.66 217.49195.92 408.47 13.76 86.42 241.84

Table 8Natural frequencies of an anti-symmetric beam

P = 0 Simply–Simply Clamped–Simply Clamped–Free

f1 (Hz) f2 (Hz) f3 (Hz) f1 (Hz) f2 (Hz) f3 (Hz) f1 (Hz) f2 (Hz) f3 (Hz)

G = 0 25.57 102.27 230.05 39.99 129.47 269.99 9.1 57.08 159.8G = 5 28.68 114.69 257.98 44.79 145.13 302.7 10.22 64.02 179.2G = 15 34.05 136.18 306.28 53.09 172.22 359.26 12.13 76 212.75Glim = 21.55 37.15 148.57 334.14 57.88 187.86 391.89 13.24 82.93 232.11

Fig. 4. First natural frequency vs. axial load for a Clamped–Simply supported beam. (a) Symmetric lay-up and (b) non-symmetric beam.

Fig. 5. First squared natural frequency normalized by the first free frequency vs. axial load normalized by the first buckling load A Clamped–Simplysupported beam. (a) Symmetric lay-up and (b) non-symmetric beam.

Table 9Critical loads without PZT actuation (G = 0)

Boundary conditions 1st Critical load 2nd Critical load Critical loads ratio

Pcr1 (N) Pcr2 (N) Pcr2/Pcr1

C–C symmetric beam 76.50 156.46 2.045C–C non-symmetric beam 66.73 136.48 2.04S–S symmetric beam 19.13 76.51 3.999S–S non-symmetric beam 16.69 66.74 3.999C–S symmetric beam 39.13 115.64 2.955C–S non-symmetric beam 34.13 100.87 2.955C–F symmetric beam 4.78 43.04 9.004C–F non-symmetric beam 4.17 37.54 9.002

Y. Fridman, H. Abramovich / Composite Structures 82 (2008) 140–154 149

the first buckling load for each case, are presented inTable 10.

The value of the feedback gain, G, depends on the flex-ural rigidity and the geometric configuration of the piezo-

electric material. Different Glim values were found forsymmetric and anti-symmetric laminates. Increasing thevalue of G, would yield a higher buckling load. The ratiobetween the enhanced buckling load and the first critical

Table 10Buckling loads with PZT actuation

Boundary condition G = 5 G = 15 G = Glima

Pcr (N) Pcr/Pcr1 Pcr (N) Pcr/Pcr1 Pcr (N) Pcr/Pcr1

C–C symmetric 92.2430 1.2057 123.7084 1.6169 152.9745 1.9995C–C non-symmetric 83.9300 1.2577 118.3148 1.7729 140.8431 2.1105S–S symmetric 23.0666 1.2057 30.9375 1.6171 38.2594 1.9999S–S non-symmetric 20.9873 1.2577 29.5882 1.7732 35.2242 2.1109C–S symmetric 47.1836 1.2057 63.2816 1.6171 78.2557 1.9997C–S non-symmetric 42.9303 1.2580 60.5210 1.7731 72.0483 2.1108C–F symmetric 5.7670 1.2057 7.7350 1.6172 9.5658 2.0000C–F non-symmetric 5.2471 1.2577 7.3976 1.7732 8.8069 2.1110

a Note the different values of Glim for symmetric and non-symmetric laminate.

Fig. 7. A Simply–Simply supported [PZT/0/90/90/0/PZT] beam withvarious ply thickness. (a) Sensed voltage vs. normalized beam thickness.(b) Applied voltage with Glim vs. normalized beam thickness.

150 Y. Fridman, H. Abramovich / Composite Structures 82 (2008) 140–154

load depends only on the laminate structure and is higherfor non-symmetric laminates. No influence of the bound-ary condition was found. For G = Glim the enhanced buck-ling load is 2 * Pcr1 for a symmetric laminate and is2.11 * Pcr1 for a non-symmetric one.

In Fig. 6, a comparison between the enhanced bucklingload vs. feedback gain and its relation to the first bucklingload for symmetrical and non-symmetrical laminateclamped–free beam is presented. From this figure one cansee that the increase in the enhanced buckling load is linearwith the gain. The slope of anti-symmetric beam is higherthan the symmetric one as a result of in-plane out-of-planecoupling between bending moment and extension of themiddle surface (see Eq. (41)).The smallest values of Pcr

were obtained for G = 0 (the first buckling loads for bothtypes of beams).

Fig. 7 presents the sensed voltage for beams with variousgraphite-epoxy ply thicknesses where the lowest ply thick-ness is h = 1.27 * 10�4 [m] , and the other three ply thick-nesses are 2 h, 4 h and 8 h. The sensed and the appliedvoltage were calculated for different laterals loads whichwould lead to the same lateral deflection, Wmax = � 1.3 *10�4 [m] or Wmax/L = 5.23 * 10�4 for all beams loadedby axial loads of P = 0.95 * Pcr1. The applied voltage cal-culated for the limiting gain G = Glim values (see Eq.(39)) is shown in the fig. 7b. From these results one can

Fig. 6. Clamped–Free symmetrical and non-symmetrical beam. (a) Enhancedbuckling load vs. gain.

see that the real actuation power is limited by the appliedvoltage capability of the piezoelectric layer which is nor-mally for monolithic PZT under 200 V.

Fig. 8 presents the lateral displacement of a beam undervarious axial loads without piezoelectric actuation. From

buckling load vs. gain. (b) Enhanced buckling load normalized by first

Y. Fridman, H. Abramovich / Composite Structures 82 (2008) 140–154 151

these figures one can see the usual displacement form fordifferent boundary conditions.

Figs. 9 and 10 show the lateral displacement along thebeam length with PZT actuation. The PZT actuation effectis visible in these figures.

Fig. 11 presents the mid-span lateral displacement of thebeam with a symmetric lay-up, (PZT/0�/90�/90�/0�/PZT)

Fig. 8. Lateral displacement vs. beam length for various axial load for G = 0,Clamped–Free non-symmetrical beam.

Fig. 9. Lateral displacement vs. beam length for constant axial load and variousymmetrical beam.

Fig. 10. Clamped–Clamped non-symmetrical beam. (a) Lateral displacementdisplacement vs. beam length for various axial loads and gains.

vs. the axial load P, for various directions of the lateralload, q. The different lines present the displacement ofthe beam for different values of gain G. Fig. 12 presentsthe same results for a non-symmetric Clamped–Simplysupported beam (PZT/0�/90�/0�/90�/PZT). As expected, alarge lateral deflection is reached when the axial loadapproaches critical buckling load.

q = 0.5 [N/m]. (a) Clamped–Simply supported non-symmetrical beam. (b)

s gains. (a) Simply–simply supported symmetrical beam. (b) Clamped–Free

vs. beam length for constant axial load and various gains. (b) Lateral

Fig. 11. Axial load vs. beam mid-span deflection for various gains – ASimply–Simply supported symmetrical beam.

Fig. 12. Axial load vs. lateral mid-span displacement for various gains- AClamped–Simply supported beam.

152 Y. Fridman, H. Abramovich / Composite Structures 82 (2008) 140–154

Fig. 13 presents the mid-span lateral displacement of aClamped–Clamped beam vs. the axial load P normalizedby the first critical load for symmetric and non-symmetriclay-ups.

Fig. 13. Axial load normalized by first buckling load vs. beam mid-span deflectand (b) non-symmetrical lay-up.

Fig. 14. Lateral mid-span displacement vs. lateral load for a Simply–Simply sload P = 0.95 * Pcr1. (b) For various axial loads P = 0.95 * Pcr.

From Fig. 14, one can see that for a beam with an anti-symmetric lay-up, the direction of the lateral displacementdepends on the values of the lateral load, q, and itsdirection.

ion for various gains – a Clamped–Clamped beam. (a) Symmetrical lay-up

upported anti-symmetrical beam for various gains. (a) For constant axial

Y. Fridman, H. Abramovich / Composite Structures 82 (2008) 140–154 153

7. Conclusions

A mathematical model is presented for active control ofthe instability of symmetric and non-symmetric piezo-lam-inated composite beams subjected to axial and lateralloads.

The active control is obtained by using piezo-ceramiclayers as an actuators and sensors acting in closed-loop.The voltage transferred from the sensors to the actuatorshas to be amplified through a constant feedbackgain G.

An exact mathematical model, based on a first ordershear deformation theory (FSDT), was developed anddescribed. Closed form solutions for the bending angleand the axial and lateral displacements along the beamare presented for various boundary conditions.

Using the present model, the critical buckling loads andthe natural frequencies of beams and their respective modeshapes were computed for cases with and without the feed-back gain.

A parametric investigation was performed for beamswith different lay-ups, and it was found that for a givenfeedback gain value one can increase the critical bucklingload and or the natural frequency of a beam with a non-symmetric lay-up in a higher ratio then one can achievefor a symmetric lay-up. The limiting feedback gain for astable system was derived. A linear relationship was foundbetween the enhanced buckling load and the feedbackgain.

The present results are consistent with similar onespresented in the literature (see for example Refs. [8,10,13,15]), namely the buckling load of a beam or its naturalfrequencies can be actively increased above their criticalvalues, using piezoelectric sensors and actuators actingin a closed loop. The range of this increase varies froma few percents up to more than five times the criticalbuckling load, depending on the properties of the beamsand the relevant control types. The present methodologyyielded an enhancement factor of 2 for a symmetric lay-up and 2.11 for a non-symmetrical one, comparable tothe other existing results, yielding a sound and reliablegeneric model.

The sensed and the applied voltage were calculated forvarious laterals loads which would lead to the same lateraldeflection, in the order of Wmax = � 1.3 * 10�4 [m] forbeams loaded at P = 0.95 * Pcr1. It was shown that thesensed voltage is proportionally increasing, while theapplied voltage, at G = Glim, is increasing in a squaredmanner with the beam’s thickness.

One should note that the required voltage to preventbuckling is dependent on the beam’s moment of inertia.Therefore, to control laminated composite beams of prac-tical dimensions (let us say, length in meters and momentsof inertia in the range of 10+4 cm4), the applied voltageswould be a few hundred of volts, provided the piezoelectriclayers can carry this electrical load.

Appendix A

A11 ¼ cXN

k¼1

ðQ11Þkðzk � zk�1Þ ðA:1Þ

B11 ¼c2

XN

k¼1

ðQ11Þkðz2k � z2

k�1Þ ðA:2Þ

D11 ¼c3

XN

k¼1

ðQ11Þkðz3k � z3

k�1Þ ðA:3Þ

A55 ¼ cK0

XN

k¼1

ðQ55Þkðzk � zk�1Þ ðA:4Þ

where zk is the distance of the kth layer from the x-axis, N

is the number of layers (including the piezoelectric ones),K0 is a shear correction factor (chosen to be 5/6).

Relations for the dynamic solution for a symmetric case:

E1 ¼ n1C2; E2 ¼ n1C1; E3 ¼ n2C4; E4 ¼ �n2C3

ðA:5Þ

n1 ¼ �k2

1ðA55 � P Þ þ x2I1

A55k1

; n2 ¼k2

2ðA55 � P Þ � x2I1

A55k2

ðA:6ÞRelations for dynamic solution for a non-symmetric case:

a3wVI þ a2wIV þ a1w00 þ a0w ¼ 0 ðA:7Þa0 ¼ �b1b4

a1 ¼ b0b4 þ b6 � b1b3

a2 ¼ b3b0 þ b5 � b1b2

a3 ¼ b0b2

8>>>><>>>>:

ðA:8Þ

b0 ¼ PA55� 1

b1 ¼ I1x2

A55

8<: ðA:9Þ

b2 ¼ B11 þ G�fh0

� �a1

a4

b3 ¼ B11 þ G�fh0

� �ða2�a3Þ

a4þ x2I2

a1

a4þ ðD11 þ G�f Þ

b4 ¼ �A55 þ x2I3 þ x2I2ða2�a3Þ

a4

b5 ¼ B11 þ G�fh0

� �a2

a4

b6 ¼ �A55 þ x2I2a2

a4

8>>>>>>>>>>><>>>>>>>>>>>:

ðA:10Þ

a1 ¼ B11 þ G�fh0

� �B11 � A11ðD11 þ G�f Þ

a2 ¼ A11A55

a3 ¼ A11x2I3 � B11 þ G�fh0

� �x2I2

a4 ¼ A11x2I2 � B11 þ G�fh0

� �x2I1

8>>>>>>><>>>>>>>:

ðA:11Þ

The characteristic equation:

a3s6 þ a2s4 þ a1s2 þ a0 ¼ 0 ðA:12ÞE1 ¼ �g1C2; E2 ¼ g1C1; E3 ¼ g2C4

E4 ¼ �g2C3; E5 ¼ g3C6; E6 ¼ �g3C5 ðA:13Þ

154 Y. Fridman, H. Abramovich / Composite Structures 82 (2008) 140–154

g1 ¼b0m2

1 � b1

m1

; g2 ¼�b0m2

2 � b1

m2

; g3 ¼�b0m2

3 � b1

m3

ðA:14ÞA1 ¼ c1C2; A2 ¼ c1C1; A3 ¼ c2C4;

A4 ¼ �c2C3; A5 ¼ c3C6; A6 ¼ �c3C5 ðA:15Þ

c1 ¼a1g1m2

1

a4

þ ða2 � a3Þg1

a4

þ a2m1

a4

c2 ¼ �a1g2m2

2

a4

þ ða2 � a3Þg2

a4

� a2m2

a4

ðA:16Þ

c3 ¼ �a1g3m2

3

a4

þ ða2 � a3Þg3

a4

� a2m3

a4

Clamped–Simply supported symmetrical beam:

wsymc�s ðxÞ ¼ Ac�s

1 cosðkxÞ þ Ac�s2 sinðkxÞ þ Ac�s

3 xþ Ac�s4 � qx2

2PðA:17Þ

AC�S1 ¼ � q

2rP 2

ðsinðkLÞð2K þ L2PrÞ � 2LkrKÞðsinðkLÞ � Lkr cosðkLÞÞ

AC�S2 ¼ q

2rP 2

ðcosðkLÞð2K þ L2PrÞ � 2KÞðsinðkLÞ � Lkr cosðkLÞÞ ðA:18Þ

AC�S3 ¼ � qk

2P 2

ðcosðkLÞð2K þ L2PrÞ � 2KÞðsinðkLÞ � Lkr cosðkLÞÞ

AC�S4 ¼ �AC�S

1

Clamped–Simply supported non-symmetrical beam:

wn�symc�s ðxÞ ¼ ac�s

1 cosð�kxÞ þ ac�s2 sinð�kxÞ þ ac�s

3 xþ ac�s4 � qx2

2PðA:19Þ

ac�s1 ¼ ðB11ðB11qþ P 2Þ �D11A11qÞðr�kL� sinð�kLÞÞ þ 0:5qL2PA11 sinð�kLÞ

P 2A11ðsinð�kLÞ � r�k cosð�kLÞÞ

ac�s2 ¼ ðB11ðB11qþ P 2Þ �D11A11qÞðcosð�kLÞ � 1Þ � 0:5qL2PA11 cosð�kLÞ

P 2A11ðsinð�kLÞ � r�k cosð�kLÞÞac�s

3 ¼ �r�kac�s2 ; ac�s

4 ¼ �ac�s1

cc�s4 ¼ �r�k

B11

A11

ac�s2 ; cc�s

3 ¼ � B11qA11P

þ PA11

ðA:20Þ

References

[1] Abramovich H, Livshits A. Dynamic behaviour of cross-ply lami-nated beams with piezoelectric layers. Compos Struct 1993;25:371–9.

[2] Song G, Qiao PZ, Binienda WK, Zou GP. Active vibration dampingof composite beams using smart sensors and actuators. J AerospaceEng 2002;15(3):97–103.

[3] Huang D, Sun BH. Approximate solution on smart compositesbeams by using MATLAB. Compos Struct 2001;54:197–205.

[4] Waissman H, Abramovich H. Active stiffening of laminated com-posite beams using piezoelectric actuators. Compos Struct2002;58:109–20.

[5] Raja S, Prathap G, Sinha PK. Active vibration control of compositesandwich beams with piezoelectric extension-bending and shearactuators. Smart Mater Struct 2002;11:63–71.

[6] Sloss JM, Bruch JC, Sadek IS, Adali S. Piezo patch sensor/actuatorcontrol of the vibrations of a cantilever under axial load. ComposStruct 2003;62:423–8.

[7] Karagulle H, Malgaca L, Oktem HF. Analysis of active vibrationcontrol in smart structures by ANSYS. Smart Mater Struct2004;13:661–7.

[8] Mukherjee A, Saha Chaudhuri A. Active control of piezolaminatedcolumns – exact solutions and experimental validation. Smart MaterStruct 2005;14:475–82.

[9] Mukherjee A, Saha Chaudhuri AJ. Active control of dynamicinstability of piezolaminated imperfect columns. Smart Mater Struct2002;11(2002):874–9.

[10] Meressi T, Paden B. Buckling control of a flexible beam usingpiezoelectric actuators. AIAA J Guid Control Dyn 1992;16:977–80.

[11] Chase JG, Yim J. Optimal stabilization of column buckling. ASCE JEng Mech 1999;125:987–93.

[12] Chase JG, Bhashyam S. Optimal stabilization of plate buckling.Smart Mater Struct 1999;8:204–11.

[13] Thompson SP, Loughlan J. The active buckling control of somecomposite column strips using piezoceramic actuators. ComposStruct 1995;32:59–67.

[14] Chandrashekhara K, Bhatia K. Active buckling control of smartcomposite plates-finite element analysis. Smart Mater Struct1993;2:31–9.

[15] Berlin AA. 1994 Towards intelligent structures: active control ofbuckling. PhD thesis, Department of Electrical Engineering andComputer Science, Massachusetts Institute of Technology.

[16] Wang Q, Quek ST. Enhancing flutter and buckling capacity ofcolumn by piezoelectric layers. Int J Solids Struct 2002;39:4167–80.

[17] Abramovich H. Deflection control of laminated composite beamswith piezoceramic layers – closed form solution. Comp Struct1998;43:217–31.

[18] Matlab User’s Manual version 6.5.01.


Recommended