Review of Current Status of Smart Structures and

Integrated Systems
Inderjit Chopra
Professor and Director
Center for Rotorcraft Education and Research
University of Maryland, College Park, MD 20742

1. Introduction
A smart structure involves distributed actuators and sensors, and one or more microprocessors that
analyze the responses from the sensors and use distributed-parameter control theory to command the actuators
to apply localized strains to minimize system response. A smart structure has the capability to respond to a
changing external environment (such as loads or shape change) as well as to a changing internal envfronment
(such as damage or failure). It incorporates smart actuators that allow the alteration of system characteristics
(such as stiffness or damping) as well as of system response (such as strain or shape) in a controlled mariner.
Many types of actuators and sensors are being considered, such as piezoelectric materials, shape memory alloys,
electrostrictive materials, magnetostrictive materials, electro-rheological fluids and fiber optics. These can be
integrated with main load-carrying structures by surface bonding or embedding without causing any significant
changes in the mass or structural stiffness of the system.

Numerous applications of smart structures technology to various physical systems are evolving to
actively control vibration, noise, aeroelastic stability, damping, shape and stress distribution. Applications
range from space systems, fixed-wing and rotary-wing aircraft, automotive, civil structures and machine tools.
Much of the early development of smart structures methodology was driven by space applications such as
vibration and shape control of large flexible space structures, but now wider applications are envisaged for
aeronautical and other systems. Embedded or surface-bonded smart actuators cn an airplane wing or helicopter
blade will induce alteration of twist/camber of airfoil (shape change), that in turn will cause variation of lift
distribution and may help to control static arid dynamic aeroelastic problems.

Applications of smart structures technology to aerospace and other systems are expanding rapidly.
Major barriers are: actuator stroke, reliable data base of smart material characterteristics, non-availability of
robust distributed parameter control stratgies, and non-existent mathematical modeling of smart systems. The
objective of this paper is to review the state-of-the-art of smart actuators and sensors and integrated systems
and point out the needs for future research.

2. Charcteristics of Smart Actuators and Sensors

Piezoelectrics are the most popular smart materials. They undergo surface elongation (strain) when an
electric field is applied across them and produce voltage when surface strain is applied, and thus can be used
both as actuators and sensors. Under an applied field, these materials unfortunately generate a very low strain
but cover a wide range of actuation frequency. Piezoelectric materials are relatively linear and bipolar, but
exhibit hysteresis. The most widely used piezoceramics (such as lead zirconate titariite) are in the form of thin
sheets that can be readily attached or embedded in composite structures. Among other smart materials, shape
memory alloys (SMA) appear attractive as actuators because of the possibility of achieving large excitation

20 ISPIE Vol. 2717 0-8194-2092-1/96/$6.OQ

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 forces and displacements. These materials undergo phase transformation at a specific temperature. When
plastically deformed at a low temperature, these alloys recover to their original undeformed condition if its
temperature is raised above the transformation temperature. This process can be repeated again. A remarkable
characteristic of SMA is its large change of modulus of elasticity with heating (typically three to five times of
room temperature value). The most common SMA material is Nitinol (nickel titanium alloy) and is available
in the form wires of different diameters. Though heating is carried out internally (electrically), response is
very slow (less than 1 Hz). Electrostrictive materials are quite identical to piezoelectric materials, with
slightly better strain capability, but very sensitive to temperature. Also, electrostrictive materials have a
monopolar nonlinear relation between an applied field and induced strain, but exhibit negligible hysteresis.
Magnetostrictive materials such as Terfenol-D elongate when exposed to a magnetic field. These materials are
monopolar and nonlinear and exhibit hysteresis. These materials generate low strains and moderate forces over
a wide frequency range. Because of coil and magnetic return path, these actuators are often bulky. Electro-
rheological (ER) fluids consist of suspensions of fine dielectric particles in an insulating fluid that exhibit
controlled rheological behavior in the presence of large applied electric fields (up to 1-4 Ky/mm). Application
of an electric field results in a significant change of shear loss factor that helps to alter damping of the system.
Fiber optics are becoming popular as sensors because they can be easily embedded in composite structures with
little effect on structural integrity and also have the potential of multiplexing.

Piezoelectric and electrostrictive materials are also available in the form of 'stacks' where many
layers of materials and electrodes are assembled together. These stacks generate large forces but small
displacements in the direction normal to the top and bottom surfaces. Bimorphs or bending actuators are also
available commercially where two layers of these materials (piezoceramic) are stacked with a thin shim
(typically of brass) between them. If an opposite polarity is applied to two plates, a bending action is created.

Smart structures are becoming feasible because of the (1) availability of smart materials commercially,
(2) ease of embedding devices in laminated structure, (3) exploitation of material couplings such as between
mechanical and electrical properties, (4) potential of a substantial jump in performance improvement at a small
price, and (5) advances in microelectronics, information processing and sensor technology. Key elements in the
application of smart structures technology to a system are: actuators, sensors, control methodology, and
hardware (computer and power electronics).
Actuators
Typical actuators consist of piezoceramics, magnetostrictives, electrostrictives, and shape memory
alloys. These directly convert electric signals into actuation strain/displacement. Piezoelectrics and
electrostrictors are available as ceramics, where as magnetostrictors and shape memory alloys are available as
metal alloys. Piezoelectrics are also available in polymer form. Most important performance parameters of
actuators indude maximum stroke or strain, maximum block force, stiffness and bandwidth. Other important
parameters include linearity, sensitivity to temperature, brittleness, weight density, compactness, and
efficiency. The induced strain is treated like thermal strain. The total strain in the actuator is assumed to be
the sum of the mechanical strain caused by the stress plus the induced strain caused by the electric field. The
strain in the host structure is obtained by establishing the displacement compatibility between the host
material and the actuator. In a piezoelectric material, when an electric field is applied, the monopoles of the
material are pulled causing strain in the material. This relation of strain versus voltage is linear in the first
order. In an electrostrictive material, there is an interaction between the electric field and electric dipoles that
is inherently nonlinear. The magnetostrictive response is based cn the coupling of magnetic field and magnetic
monopoles in the material, a nonlinear effect. Shape memory is a result of phase transformation due to
temperature of the material (caused by an electric field). Phase transformation is vey much a nonlinear
phenomenon.

A common piezoceraniic material is lead zirconate titanate (PZT) and its maximum actuation strain is
about 1000 microstrain. Polyvinylidenefluoride (PVDF) is a polymer piezoelectric film and its maximum
actuation strain is about 700 microstrain. A common ceramic electrostrictive material is lead magnesium niobate
(PMN) and its maximum actuation strain is about 1000 microstrain. PZT and PMN are available in the form of

SPIE Vol. 2717121

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 thin sheets which can be either bonded or embedded in a structure. Terfenol, a rare earth magnet-like material,
can create a maximum actuation strain of about 2000 microstrain. It needs a large magnetic field to cause this
actuation strain. Nitinol (nickel titanium alloy) available in the form of wires or sheets, can create free strain
up to 20000 microstrain (2% strain).

Actuators PZT G-1195 PZT-5H PVDF PMN Terfemol Nitinol

DZ
Shape
Actuation Piezo- Piezo- Piezofilm Electro- Magneto Memory
Mechanism ceramic ceramic strictive strictive Alloy
Free Strain
Amax , JL strain 1000 1000 700 1000 2000 20000
4 for
Modulus E 9 10 .3 17 7 (martensite)
lO6psi 13 for
(austenite)
8max for
400 350 10 500 580 8500
tc
aluminumbeam
Band Width High High High High Moderate Low
Strain-Voltage First order First First Non Non Non
Linearity Linear Order Order Linear Linear Linear
Linear Linear

Sensors
Typical sensors consist of strain gauges, accelerometers, fiber optics, piezoelectric films and
piezoceramics. Sensors convert strain or displacement (or their time derivatives) into electric field.
Piezoelectric sensors are generally made of polymers such as polyvinylidene fluoride, PVDF. It can be easily
formed into very thin sheets (films) and adhered to any surface. Key factors for sensors are their sensitivity to
strain or displacement, bandwidth, and size. Other less important factors include temperature sensitivity,
linearity, hysteresis, electromagnetic compatibility, embeddibility, and needed associated electronics (size
and power requirement). The sensitivity for resistor gauge is 30 volts per strain, for semiconductor gauge is i0
volts per strain, and for piezoelectric and piezoceramic gauges is i0 volts per strain. The sensitivity for fiber
optics sensors are defined differently, and is about 106 degree per strain.

Table 2: Comparison of Sensors

Sensor Resistance Semiconductor Fiber Optics Piezofilm Piezoceramics

gauge gauge 0.04' .001" .001'
1OV 1OV interferometer thickness Thickness
excitation excitation gauge length
Sensitivity 30 VIE 1000 VIE 106 deg/E 10 VIE 2x10 VIE
Localization 0.008 0.03 0.04 <.04 <.04
(inches)
Bandwidth 0 Hz - 0 Hz - 0 Hz - .1 Hz .1 Hz
acoustic acoustic acoustic - GHZ - GHZ

22/SPIEVoI. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 Piezoelectric Constitutive Relations
Piezoceramic materials are assumed to be linear and the actuation strain is modeled like thermal strain.
Piezoceramics can be idealized as an orthotropic material like composite unidirectional laminate. The
constitutive relations are based ai the assumption that the total strain in the actuator is the sum of the
mechanical strain induced by the stress, the thermal strain due to temperature and the controllable actuation
strain due to the electric voltage. The axes are identified by numerals: 1 corresponds to x axis, 2 corresponds to y
axis, and 3 corresponds to z axis. The 3 axis is assigned to the direction of the initial polarization of the
piezoceramic, and 1 and 2 axes lie in plane perpendicular to 3 axis.

3
IE

Coupled electromechanical constitutive relations are:

eTE+C+
=4 ++_1i:
where
= stress vector,N/mm2 or lb/in2
= strain vector
= vector of applied field, V/mm or V/in
= vector of piezoelectric coefficients, N/V-mm or lb/V-in
4 = vector of piezoelectric stain coefficients, mm/V or in/V
= vector of stiffness coefficients, N/mm2 or lb/in2
= vector of compliance coefficients, ium2/N or in2/lb
= vector of thermoelastic coefficients, N/mm2K or lb/in2K
= vector of thermal coefficients of expansion, 1/K
= temperature change, K

Rewriting above stress-strain equation

SPIE Vol. 2717123

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 E1 S12 S13 0 0 0 00
S11

02 00 d31
c:y1 cx1

E2 S12 S11 S13 0 0 d31 a2

3 S13 S13 S33 0 0 0 00 d33
E
E2 a3
0 0 0 S44 0 0 0 d15 0
'Y23

Y31 0000S0 t23

d15 0 0 E3

Y12 0 0 0 0 0 S66 12 0 00 0

Tn a plane perpendicular to the piezo-polarization, it has isotropic properties, i.e., transversely

isotropic material in the plane 1-2. For orthotropic material, there is no temperature shear strain. However,
there is a shear strain induced due to electrical field Ei and E2.
For piezoceramic materials:
0 0 d31
0 0 d31
Actuation Strain A =
0 0d E1

E3
d15
d5
where d33, d31, and d15 are called piezoelectric strain coefficients of a mechanical free piezo element.
. d31 characterizes strain in the 1 and 2 directions to an electric field E3 in the 3 direction
. d33 relates strain in the 3 direction due to field in the 3 direction
. d15 characterizes 2-3 and 3-1 shear strains due to field E2 and Ei respectively.
Thus, if an electric field E3 is applied to a free piezo element, it causes longitudinal strains Ei, E2 and 3. This is
very similar to thermal strain. If an electric field El or E2 is applied, the material reacts with shear strain y
and 123 respectively. For orthotropic materials, there is no corresponding thermal strain. To overcome this
problem, it is better to assume, piezoelectric materials as anisotropic.
If a compressive force is applied in the polarization direction (axis 3), or tensile force is applied in the
plane perpendicular to polarization direction (axis 2 or 3), it will result in a voltage that has the same
polarity as the original poling direction.
Note: Unless the level of mechanical force is high, there is no effect of mechanical bias strain a-i piezoceramic
properties.
Piezoelectric Characteristics
During the manufacture of a piezoceramic, a large (greater than 1KV/mm) field is applied across the
ceramic to create polarization. This is called a coercive field. During subsequent testing, if a field greater than
the coercive field, Ec, is applied opposite to the polarization direction, the ceramic will lose its piezoelectric
properties - called depoling. However, it is possible to repole the material.
If an applied field is aligned with the initial polarization direction, there is no depoling. A
sufficiently high voltage can cause arcing or a brittle fracture. Depoling is also possible if the temperature
exceeds the Curie temperature or if a large stress is applied.

24/SPIE Vol. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 A1 = d31 E3

3
IE
field
volts 1

Strain (microstrain)
A linear model fits well only for a small field. For large strain, a nonlinear variation occurs and a cubic
variation fits better.

d1 =

1*
31

small field

Induced Strain
There is a significant hysteresis in stress-strain plot. Aspect ratio of the hysteresis loops increases with higher
strains and the hysteresis causes an unmodeled phase lag. It can be modeled using complex quantities to
represent the mechanical, dielectric and piezoelectric coefficients.

SPIE Vol. 2717/25

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 Field

Strain

At high frequencies, energy from mechanical losses can generate enough heat to severely affect the operation of
the actuator. For static problems, the presence of residual strains can be quite important. The piezoelectric
induced strain fails to return to zero when the applied field is removed. Creep and the strain rate dependence of
d* je small but measurable which become more significant for larger strains and lower frequencies. For low
frequencies or static applications, the tendency of piezoceramics to creep under prolonged application of electric
fields. For high frequencies, creep can be ignored. The degree of creep is reduced when piezoceramics are
elastically constrained. Also, the characteristics of piezoelectric ceramics may change gradually with time,
generally a logarithmic change with time. This means that they becomes more stable with a longer time after
poling.
Manufacturing Issues
There are several issues concerning building of smart structures. These are:
1) Electrical contact on both sides of piezo is required. One way to overcome this problem is to drill a hole in the
substructure and use nonconducting epoxy.
2) Piezo has to be insulated from structure. By anodizing the beam, this problem can be solved.
3) For proper transfer of induced strain to main structure, bond layer thickness need to be thin and uniform. For
this, pressure is applied during curing.
Embedding vs. Surface-Mounting: With surface-mounted actuators, there is a ease of manufacturing, access for
inspection, and less maintenance cost. Because of exposure, actuators are more susceptible to damage. Also, the
functioning of actuators is dependent cai structural surface. Embedded actuators become inaccessible for
inspection. The devices, however, are better protected and interconnections with other devices become easy.
Also with embedding, the piezoelectric must have an elastic modulus comparable to the host structure in order
to avoid structural discontinuity, and the Curie temperature should be higher than the curing temperature of
composites. Further, piezo devices must be electrically insulated from the host structure. This means that the
piezo elements can be directly embedded in glass/epoxy laminae, but need an insulating layer with
graphite/epoxy laminae. However, an insulating layer can reduce the effectiveness of the actuator. It appears
appropriate to wrap the piezo device in a .05 mm thick Kapton film (Dupont) and use acrylic epoxy to reduce
slippage between piezo and insulating layer.
Embedding Electronics: For embedding integrated circuits, it is essential to have their electrical insulation and
mechanical isolation. For a minimal degradation of structure, it is important to have a minimum ply
interruption.
At this time, a reliable data base of characteristics for smart actuators and sensors is not available.
Thus, there is a need to undertake extensive testing of engineering specimens (macro level) to represent most of
the operating conditions such as stress, strain, temperature, and voltage. Building smart structures is the state-
of-the-art and requires considerable experience and expertise in fabricating complex systems with embedded or
surface-mounted smart actuators and sensors.

26 ISPIE Vol. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 3. Modeling: Beam with Induced Strain Actuation
To model one-dimensional structures such as slender beams, three different models are adopted. These are
block force model, uniform strain model and Euler-Bernoulli model.
Simple Block Force Model

Piezoelectric
elment

Basic structure
isotropic
V

Let us say, a piezoelectric element of length £, width b, and thickness t is attached to an isotropic beam. If
an electric field V(Voltage) is applied, then maximum actuator strain will be:

Emax = A = d31[Y

The A is called free strain. Maximum block force (zero strain condition) is
Fb = d31EbV
where E is Young modulus of elasticity of piezo and d31 is piezo constant. A piezo attached to the beam
structure results in an axial force F in the beam due to potential V. The reactive force in the piezo element will
be -F. Then the strain in piezo becomes:
V F
——____
31
btE
Plotting this force-strain relation for constant field V,

max

F Fm
Figure: Force-strain characteristic of a piezo-element

SPIE Vol. 2717/27

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 This plot can also be used to determine the properties of piezo materials experimentally.
(t
d31

E =—-- C
bt
Two identical actuators mounted on the surface of a beam, one on either surface, can produce pure bending as well
as pure extension.

F
I 1 Compression
-
// _____________________
-
Extension
F
// I— —I
F
Pure Extension (Same Voltage to Top & Bottom Piezos)

F
// _________________________________________
I —- ---- I compression
//
//
F ____
- ____ tension
F compression
//
_________
I— —-I tension
F
Pure Bending (Equal but Opposite Voltage to Top & Bottom Piezos)

For a pure extension case, same potential is applied to top and bottom actuators. The induced force is
F = d31V EAbEAC — Fb -
EAb
2tC EAb + EAC EAb + EA
where Fb is a block force of each piezo element. The extensional stiffnesses are defined as
EAC = 2ECbCtC
EAb = EbbCtb

increasing V
E
2F
Ebbbtb

28/SPIEVo!. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 If piezo stiffness EA >> EAb (beam stiffness), the actuation force becomes zero though the actuation
strain equals free strain. On the other hand, if piezo stiffness EA <xEAb, the actuation strain becomes zero
though actuation force equals block force.
For a pure bending case, an equal and opposite potential is applied to top and bottom actuators and the
induced bending moment is

____ _______
EIbEIC EIb
______
tctb )1b1CJ[E1b1CJ
M—[2d31
where Mb is block moment and it is equal to Fbtb. The bending stiffnesses are defined as

(EI) = 2(bt)(
2) E
(EI)b EbIb
Again, if EI >> EIb, actuation moment becomes zero and if EI << EIb, actuation strain becomes zero.

Uniform Strain Beam Model

This approach is also called Pin Force Method. Let us consider two identical strain-induced actuators
are bonded to an isotropic beam, one to the top surface and the second to the bottom surface of the beam. Between
the actuator and the beam surface, there is a finite thickness elastic bond. Each actuator is assumed to induce a
uniform strain across its thickness. There is linear distribution of strain in the host structure.

z Piezoactuator
tc
I ÷ - Adhesive
-÷ ÷ .4- 1— 1— - 1-
1— 4— -4- 4-. 4— 4— -+ -÷ -÷ -+ -* -* Surface Shear

Beam mid-axis
tb
(a) Definition of thickness and coordinate system (b) transmission of shear across adhesive layer
Diagram of
beam mounted with symmetric actuators

I da c dx
+
___j— o.c
dx
'V

M d

a ______ I
da
dx
CdX
Diffrential Element for Bending Case

SPIEVo!. 2717/29

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 ts = adhesive thickness
tc = piezo element thickness
tb beam thickness
Assume that the normal stress on actuator (Y does not vary across its thickness.
Defining strains as
du
piezo: E —s-
dx

beam: 8b
S du
dx
1 —
= —[ut
bond:
u]
where is the u-deflection of the outer fiber of the beam. The solution of equilibrium equations becomes
a [1cosh(]T)
Aa+'P[ coshf
a [11+—'Pcosh(F)
A a+'P[ a coshF
where a=2 for pure extension and a=6 for pure bending. Other constants are defined as

=—---, O=-,
t 0S

=-- y=E
= + -&f= = + att S
t5E Ebtb) 4 tL 'Ytctb

= (EA)b = Ebtb
EA Et
The G is shear-lag parameter. The shear stress in adhesive layer becomes
- G5 sinh(F)
—
A FE5cosh(F)
For pure extension:

u(x) =
aA [
)[X
t — sinh(F2x / £)1 + aA (t [ tahri(F)
)[1
—
(a + 2Fcosh(F) j (a + ) F

For pure bending:

30/SPIE Vol. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 (t\
I-\2 I 2x
—- I coshi F—
I

w(x)=
2aA —— x2 2)
I
£ + aA [ C ______
xtanh(F)
tb(a+'P) 2 F2cosh(F) tb(CC+'F)[ F

+ C[ __
aAt2 1 tanh(F)+—
1
2F 4
tb(a+'P)[2F
A very thin bond ( F>30) represents a perfectly bond condition and strain distributions reduce to
E=E= a
A A a+'F
This means that the induced strain on the surface of a host structure is equal to the actuator strain and it is
proportional to the product of the actuation strain, A ,(that can be commanded to the actuation material) and
the reciprocal of one plus the stiffness ratio (structural stiffness/actuator stiffness). The second term is as a
result of the impedance matching. Higher the stiffness of the actuator, more effective the strain transfer.

Bernoulli-Euler Beam Model

The Bernoulli-Euler model is a consistent strain model and generally gives more accurate results than the
uniform strain model. It considers the beam, adhesive and actuator as a continuous structure and follows the
Bernoulli's assumption. This means that a plane section normal to the beam axis remains plane and normal to
the beam axis after bending. There is a linear distribution of strain in the actuator and host structure.
z

Figure 1. Coordinates of Differential Beam Element for Bernoulli-Euler Model.

The strain distribution is
E(Z)=Eo—ZK
where k is bending curvature = -w,,
Enet E(z)_A(z)
(Y(z) = E(Z)Enet
This gives the normal stress for the complete beam in the x-direction. Since the piezo is assumed as an integral
part of the beam, E varies in the z-direction.
Total axial force in beam:

F=$''
—h/2
b(z)z(z)dz
where b(z) is beam width.
Total moment in beam:

SPIE Vol. 2717/31

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 rh/2
M= I b(z)zc(z)dz
J—h/2
The force and moment expressions can be wriften as
F + FA 1 _
S [(EA)0 (ES)0 iS
1M + MA I L (ES)0 (EI)0 fiw"
where
(EA)tot = $h/2
—h/2
b(z)E(z)dz

(ES)tot
= fh/2
b(z)E(z)zdz
J—h/2

(El)tot = J—h/2
1h2 b(z)E(z)z2dz

FA
$b(z)E(z)A(z)dz
MA $b(z)E(z)A(z)zdz
where FA and MA are respectively axial force and bending moment due to induced stress. The (ES)0 is
equivalent to a coupling term. If the placement of actuators is symmetric, this term will be zero. If an actuator is
attached only on one side, this term will be non-zero (extension-bending coupling).
For an isotropic beam with pure bending actuation:

MA ECbAtC(tb + t)

÷t: ____
;:::::c!t :
MA
6I1+A
T)t
E=z
(EI)0 (W+6)+—+----
T T2
where

:T Extensional stiffness ratio = _____

Et EA
Beam Thickness
T— =
Actuator Thickness t
The thickness ratio, T, determines whether the strain variation across the piezo element affects the analysis.
Following figure shows the variation of the normalized curvature with the thickness ratio. For a case with
piezo stretched across the beam width, and with a thickness ratio of 12.5, there is a 2.7% difference between
the predictions due to the uniform strain model and the Bernouffi-Euler model. For thin beams, the uniform
strain model overpredicts strain (curvature). For large thickness ratios, the predicted induced bending
carvatures are identical using both methods.

32 ISPIE Vol. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 1.0

0.8

0.6

0.4

0.2

0.0
0 5 10 15 20
T
C

Single actuator normalized curvature comparison for a perfect bond condition

Embedded Actuators:
Assume two identical actuators are embedded in an isotropic beam at an equal distances from the mid
plane of the beam resulting in a symmetric configuration. A very thin bond layer exists between the actuators
and the beam such that a perfect bond assumption is valid. An equal voltage applied to both actuators results
in pure extension whereas an equal but opposite voltage applied to both actuators causes pure bending of the
beam.

lv:

tb
—d _-_ _
I

I
l_y__t

-1
-ff
Combining bending-extension relations into matrix
I F + F4 1 rEAtota, EStotai r
=
M + M4 [EStotai EItotai

Energy Formulation:
Using the same basic assumptions as made in the force equilibrium formulation, the principle of virtual
work can provide the governing equations and boundary conditions which can be easily adapted to dynamic
equations of motion. Assuming that the only allowable modes of deformation are actuator extension, adhesive
shear, and beam bending and extension, the strain energy relations may be directly written.

SPIEVo!. 2717/33

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 IC (—\ o\2
Beam
Extension
•
o_l1i laub
2 -f
Beam • K_1f (a2w
Bending 2 '-f ax2

Substrate.• u S— G SA S
q' 7S
Shear )2dx
2
Actuator . — 1 j- (au
Extension C
2 '--- cI\ ax
Defining
'b
EbAb
UK
b 1$2
2--f (s _ s)2dx
Us 11GA (u -u)2dx
The Principle of Virtual Work for static behavior is mathematically stated
WE 6U

Extension-Bending-Torsion Model (Skewed Piezo)

This section presents a force equilibrium formulation of a uniform strain model of an isotropic rectangular
beam with an induced strain actuator bonded to one surface. Figure below shows an induced strain actuator
mounted to the top surface of a beam with a finite thickness adhesive layer. The actuator axis is offset from the
beam axis by an angle J3 . It induces extension bending and torsion.

Crystal Axis offset from the Beam Axis

The mechanism which induces torsion in the system is a two-dimensional strain state, however global beam
torsion is adequately represented by a one-dimensional model. Assuming that the beam may only extend, bend
and twist and that the bending inertia is much greater than I, and chordwise extensional stiffness, EA.
is much greater then longitudinal extensional stiffness, EAT, transverse deflections, vb, may be neglected. This
assumption has the effect of aligning the principle strain axes with the beam axes and setting the transverse
principle strain identically zero
The piezo element is assumed to have a high aspect ratio, thereby behaving as a line element and only
inducing strain in its longitudinal direction. As with the previously presented uniform strain model, the strain
distribution is assumed to be uniform through the piezo thickness and linear through the beam thickness. The
adhesive is considered a linear shear layer which only transfers load in the piezo longitudinal axis direction.
A differential element of the beam and actuator is shown below.

34 ISPIE Vol. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 Components of the Differential Element

The governing differential equations with respect to the actuator axes are

Ns —
cos2()) =0
alV cos(13) -
äX Nb t2S (8_cos2(1))0
2Eo ____ -
cos(13)) =0
—
Nib

— tbbC G5 sin(f3) —
cos(f3)) =0
The normalized bending curvature is obtained as

2A
= cos()(a -1) - cosh(T )
Nib +acos( cosh(F) J
The twist rate for a rectangular isotropic beam is given by the expression
—
3T
Gbbbt
Integration with respect to x provides the twist angle for a cantilevered beam as
= 3 1 Eb cosQ)sin(f3) (sinh(F ) + sinh(r' )
A bGb Nib + acos( Fcosh(F)
— +1)]

SPIE Vol. 2717/35

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 0.15 0.15

0.12 0.12

0.09 0.09

-t

0.06 0.06

0.03 0.03

0.00 0
0 15 30 45 60 75 90
J3 [deg]
Analytical strains and deflections (yb=38)

At b =90, the strains all approach zero values due to the one dimensional assumptions. If two
dimensional strains were considered, the actuator strain would approach the limit compatible with the
stiffness in the transverse direction. The longitudinal beam strains, however, would stifi approach zero. Near
b =45, the analysis predicts maximum twist for a fixed actuator length.

Validation of Beam Theories

Experimentation was performed on rectangular aluminum beams with surface bonded G-1195 piezoceramic
actuators to validate the beam theories [Park93]. Crawley and de Luis [Craw87] previously correlated the
analysis for a pair of piezo elements surface-bonded to an aluminum beam. In ref. [Park93], static tests were
performed to validate beam bending results for a pair of piezos with different width ratios, a single piezo
aligned with the beam axis, and a single piezo not aligned with the beam axis.

A necessary ingredient for the analysis is the free crystal strain as a function of the applied electric field.
Strain data for an unconstrained G-1195 piezoceramic element was obtained versus applied voltage and zeroed
for each data point. This was done in order to minimize the effects of hysteresis and creep. For use with the
analytical predictions, the experimental data are fit with a 6th order polynomial.

The test specimens were 1/16 inch and 1/32 inch thick 2" x 16' aluminum beams with 9.5 mu G-1195
piezoceramic elements. Configurations include actuators which span the full width and halfwidth of the
beam. The piezoceramics are bonded to the specimens using a cynoacrylate (CA) adhesive to minimize the bond
layer effects. Bending slope and twist data were obtained by measuring the relative horizontal travel of a
laser beam reflected off a mirror at the tip of the beam.

36 ISPIE Vol. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 300.0

200.0

100.0

0 0.0
0

-100.0

-200.0

-300.0
-30 -20 -10 0 10 20 30
Volts/mu

G-1195 piezoceramic element free behavior

1.00

0.75

0.50

0.25

0.00

-0.25

-0.50

-0.75

-1.00
-20 -15 -10 -5 0 5 10 15 20
Volts/mu

Bending slope data for 1/32 inch thick aluminum beam

Experimental torsion and bending results for cantilevered beams with actuators bonded to one surface and
oriented with an angle b relative to the beam axis were compared with analytical predictions from the
combined extension-bending-torsion uniform strain modeL The test specimens were for 1/32 inch thick aluminum
beams with three 2" x 1/4" x 7.5 mu G-1195 piezoceramic elements distributed along the 16" beam in 4"
intervals. Since actuators of aspect ratio 8 were used to approximate the theoretical line element assumption,
three piezos were required to obtain measurable deflections using optical measurement system. Superposition of
the analytical results is assumed for comparison with the experiment and independently verified for the piezo
spacing.

SPIEVo!. 2717/37

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 insulated grip

G - 1195 piezoceramic crystal

Typical torsion test specimen

As shown below, the bending slope predictions are within 20% of the experimental values up to b =45.
Beyond this point the theory significantly diverges from the test results. The torsion analytical results follow
the experimental trend over the full range of b but overpredict the experimental magnitude by 35%-100% in the
midrange (15<b<75). Beam torsion and bending are adequately modeled one-dimensionally but the mechanism
which produces torsion in this system is inherently two dimensional. The primary source of discrepancy can be
attributed to the one-dimensional approximation of the strain state. Effects of chordwise and lateral bending
have been neglected and may constitute significant error depending upon the beam cross-section aspect ratio.
Overall, the results indicate that a one-dimensional model is not satisfactory to predict combined torsion,
bending, arid extension of beams with surface bonded induced strain actuators.
0.40

0.35

0.30

I 0.25

0.20
I0 0.15

0.10

0.05

0.00
0 15 30 45 60 75 90
13 [deg]
Effects of actuator orientation on beam bending slope and twist angle
Yb 38; Ys 0.00204
This section examined several one-dimensional structural models which predict the behavior of different
configurations of beams with induced strain actuators. Shear lag effects of the adhesive were identified as a
potential source of analytical error for systems with a finite bond layer. The bending component of the strain
was shown to converge with the Bernoulli-Euler solution at a lower thickness ratio, Tc, than the extensional
strain. The single actuator uniform strain model governing equations can be also formulated using the Principle
of Virtual Work as an alternative method which is easily adapted to dynamic applications. A one-
dimensional treatment of a strain actuated beam in coupled extension, bending, and torsion was examined and
experimentally determined insufficient to predict system structural behavior within acceptable limits.
However, since the torsion trend is predicted, analytical accuracy may be improved by integrating a local two
dimensional model of the actuation mechanism with a global one dimensional system modeL

38/SPIE Vol. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 4. Modeling: Plate with Induced Strain Actuation
Induced strain actuation can be used to control the extension, bending and twisting of a plate. With
tailored anisotropic plates, control of specific static deformation can be augmented using piezo actuators. Plates
with distributed induced strain actuators can be used: to control pointing of precision instruments in space; to
control structural borne noise; and to change aerodynamic shape for vibration reduction, flutter suppression and
gust alleviation. To develop a consistent plate model with induced strain actuation, the following assumptions
are made.
1. Actuators and substrates are integrated as plies of a laminated plate.
2. Consistent deformation assumed in the actuators and substrates.
3. Assumption of thin classical laminated plate theory adopted.
z

For systems actuated in extension, the strains are assumed to be constant across the thickness of
actuators and plate. For systems actuated in pure bending, strains are assumed to vary linearly through the
entire thickness. The strain in the system therefore depend on mid-plane strain €° and the curvature K.
E C0 + ZK

co

E0
x
1
xy 1

'cx
K lC,
=
7c

The constitutive relation for any ply of a laminated plate

SPIE Vol. 2717/39

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 stress vector

Y=

Actuation strain vector

A=

Ny

Nxy
x,u
Nxy

Nx
Nx

y,v

x,u
Mx

My

y,v
/
Matrix Q is the transformed reduced stiffness of the plate and QA represents an equivalent stress due to
actuation. By substituting the assumed deformation into the stress strain equations and integrating through the
thickness t of the plate for net forces and moments.

40/SPIE Vol. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 iiV Zi 9i Zig 9i NVI
N Nv'(
Zi ZZ 9ZV Zig ZZg 9Zg
CXN —
— 9i 9Z 99 : 9Zg 99 —
—
V(X
xJ7V iig Zig 9ig 11ci Zi 9iG "V
vx
Zig ZZg 9Zg Zia ZZU 9ZG "V
v(
9Zg 99g 9Zçj 99a vcxI/v
OJOtJM

uosux :ssjs = zp
N
= O) y) - ([
Z - Z
uqdnoj :ssujjs g = ¼ =
(y
= =

1O4flpSaDioj pu :suouiom N = Zp\Tj VFs = zpzvO1J

)p;uRUAs su!tU,-1
SUtUJ q4M Ajd sdn-Aej ;q are qIM
podsar 04 uejdp!m 4UJflSJ suoinb
DLIOURUAS
uonp jo pamuiAs 'Sa;rruij
-p' sutu
=0
op iou ump is
Ajqdurrs AJqropSuoD ioj DLIWUIAS
ssaDoJd aiauj si ou rqdnoD u;q upuq pu
&iqdnoD SSOUJ;S samozoq •oiaz IosTv
•uo!Sua:xa

UoSua1x :uoi3np
IIcj Zty 91 x

ZIV V 9ZV VA

91V 9zV Ak (
VXN I
AN = I
it-I
VAXN 0
si ou ponprn &r.reoqs r13s
uipua :uotnpv

(; v I
=A
J

IvjcxNl
[
-I
+
U
1&
arqj Si ou ursiw ;o •o4lIr1UT

7°/t3IdS Lfr/LLLZ

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 Case I: Symmetric laminates with multiple isotropic layers
(' 11)K
\ —122)K
\ _ 1_i2
Ek 1
"
\ —1— \
16)K ' 26)K
k

(\_DkEk
k'
12)K 1—o 1\_
" 66)K
Ek
2(1+uk)
A16=A26=O D16=D26=O
A11 = A22 D11 = D22

Case II: Symmetric laminates with multiple specially orthotropic layers

For a special orthotropic ply

[ J
E1
(11)K [ 12 21JK
(12)12
12 21 K
(66)K (G12)K

()K [E2J 12 21 K
(16)K =(26)K °
Also, A16 = A26 D16 D26 0
The behavior of this laminate is very similar to single layer specially orthotropic lamma. To be a symmetric
laminate, it should consist of odd number of plies.

Case III: Symmetric Laminates with multiple generally orthotropic layers

Terms A16, A26, D16 and D26 are non-zero. The behavior of this laminate is very similar to a single layer
generally orthotropic lamina. However, there is a coupling between normal forces and shearing strain and vice
versa, and twisting moment and bending curvature and vice versa.

Case IV: Symmetric laminates with multiple antisymmetric layers

For example, a 5 layers laminate will be -cC/cf/-cf/cf/-cf
The behavior of this laminate is similar to that of a symmetric laminate with general orthotropic layers. For a
many layered angle-ply laminate, the values of A16, A26, D16 and D26 becomes quite small as compared to
other terms and the laminate behaves more like a laminate with specially orthotropic layers.

Antisymmetric Laminates
Laminates with ply lay-ups that are antisynimetric with respect to midplane. Behavior of such
laminates can be quite different from that of symmetric laminates. An antisymmetric laminate should have
even number of plies.
(Q16)+cx = {Q16)_a
(Q26)+cz
=
For this case:

A16 = (Ql6)k(Zk — Zk14 =0

A26 = D16 = D26 = 0
Thus, for an antisymmetric laminate, the extensional-shear couplings and bending-twist couplings are zero.
Also, there is a bending-extension coupling with antisyinmertic laminates.
For symmetric piezo actuation:
A11 A12 Us B11 B12 B16 K N
A12 A22 0 + B12 B B26 K =
0 0 A66 y, B16 B26 B66

42/SP!EVo!. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 B11 B12 B16 D11 D12 0K M,
B12 B B26 D12 D 0 K M
B16 B26 B66 0 0 D66 K, M,
Antisymmetric Cross-Ply Laminates
These are even number of O and 9O alternating plies.
Two ply example is O/9O arid six ply example is
(Q11)0. = (Q)90.
Qil Ql2 0 (r \ _(r \
'4 _
::: — I-' C'22)o k'*:11)9O.
'212 '22 I \_I \
0 0 I' 12)o 12)9g.
Q66
(Q66)0. = (Q66)90.
—1 2 2
ii — k=i
—(ii)k( k k_i)

1'-'ii _— '(ç \ 2I + (r\ \ 2

2'ii)O*

B22 =—.(Q)0.t2 +(Q)90.t2

= —(Q11)90. t2 + (Q11)90. t2
= —B11
=
B66 —-(Q66)0. t2 + .(Q66)90. t2 0
B16 = B26 = 0

A11 A12 0 0s
k N B11 0
A12 A 0 E + 0 —B11 0 k =
0 0 0kN
A66 0 0

B11 0 0
0 k
+ D12 D 0 k =
M D11 D12
0 —B11 0 M
0 0 0 0 0 D66 k,(), M
Antisymmertic Angle-Ply Laminates
Example of a two-layered antisymmertic laminate is a/-a. both plies are of equal thickness. For a more
general lay-up, if a lamina of +a degrees orientation is placed cn one side of mid-plane, then equal thickness
lamina of -a degrees orientation is placed on the other side. For symmetric induced action, the force and moment
relations become:
A11 A12 0 0 0 B16 k N
A12 A22 0 +0 0 B26 k =
0 0 A66 y), B16 B26 0kN
SPIE Vol. 2717/43

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 0 0 B16 D11 D12 0kM
0 0 B26 + D12 D 0 k = M
B16 B26 0 0 0 D66 k, M,
Approximate Solution (Energy Approach)
For many problems, it is not possible to obtain exact solution. Thus, it becomes necessary to obtain
approximate solution using energy approach such as Rayleigh-Ritz method. The strain energy of the system is
J {s° T
u = U2A s° -
BDB] {
KT) [A } dA AJ[NAA
J M ] {K dA

yx

xy
Rayleigh-Ritz approximate solution

u=Øq1 v=Ø.q w=øWq

v =[H}q
w
where

[GUi' U2 - UM 1 [0] [0]

H= [0] [Vi' V2 - VNJ [0]
[0] [0] [Wi/ W2 - Wp} 3 xr
r=M+N+P
=
{}
Substitution in strain energy

U = L qTKq - Qq

44/SP!E Vol. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 K= AB (DH) cM
J • (DH)T BD
A

Q1% = JA{ N }
dA

The matrix K is of size r x r and the actuation forcing matrix QA is of size r x 1. Applying Lagrange's equation,
solve for model amplitudes. Assume a simple solution for extension and bending deflection:
x
U= —q v=O
f \2
w= (x
— xy q
q + ——
L
This results in
L)
I I

Lc
2 2
A11
-B11 -B16
- : B11 -D11
L L
—f-
CL D16 q1 NXAL
q2 -2MA
2 4 4 q3 0
——B —D —D
C 16 CL 16 C2 66

where C is chord (or width) and L is length of uniform plate.

Bending-Twist Coupling:
SettingB=O B16=O
NL
-1
D11 D16
Jq2 — 1 1] C J-2MA
q3J4 CL C2
10
D66 - D16
1 J-2MA
(D11D66 - D6 - 1)16 P11 1 0
CT] C2L2)
CL L2

C2L2 1 21)16
q3 = - CLMA
D11D66
CL 16
l166 - D6 2D16M
CL A
—
2
Twist curvature depends on the ratio of the coupling stiffness D16 to the bending stiffness D66.
=
1)16
bending-twist coupling

Extension-twist coupling:
B16 is nonzero. Setting Bli =0 and D16 =0 results in the following relation.

SPIE Vol. 2717145

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 A11 0 B16
q1 NxAL
0
-D11 0 q2 = —2MA
4 q3 0
2B D66
-C16 C2
This results in

-D11q2 = -2MA

q1 JNXL
1q3 {o
D66
1 C16 JNXL
D11D66 + C2 B2
16 [.B16 A11
I
to
N
q3 =
B16L

D66 - B6)
=CL
2 [A11 D66 -
B16
2
B16 ]
NL
B16 is extension-twist coupling.
Twist depends on B16, extensional stiffness and torsional stiffness.
B16
extension-twist coupling parameter 1VB =

So far there have been limited validation studies of these plate models with experimental data or
with detailed 3-D finite element analyses. Modeling of adhesive layer is non-existent. Thus, there is a need to
further refine plate models with induced strain actuation.

46ISPIEVo!. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 5. Shape Memory Alloys (SMA)
Shape memory alloys have a special characteristic of memorizing a certain stretched or bent shape, and
recovering that shape at higher temperature. Buehler and Wily [Buel6l] discovered a Nickel-Titanium alloy in
1961 called NITINOL that exhibited the best shape memory effect. A 100% recovery of strain up to a maximum of
8% extensional prestrain was achieved in this alloy. Another interesting feature noticed was a 3 to 5 times increase
in Young's modulus at high temperature phase as compared to its low temperature value. Restraining the
recovering strain at a higher temperature results in large recovery stress (several times more than the initial stress
required to prestrain at room temperature). This means that the work required to deform the SMA at low
temperature is much lower than the work produced at high temperatures. Following figure explain the shape
memory effect schematically.
Original Austenite
Shape L Phase
Cooling
Original Martensite
Shape Phase
L
Deforming
Martensite
Phase
L +ziL
Heating
100 % recovery Austenite
Phase
Schematic of Shape Memory Effect in Stretching

The shape that a SMA memorizes can be assigned or reassigned through an annealing process. The SMA
is heated above the annealing temperature (say 500°C) and constrained to the desired shape. At room
temperature, the SMA is in martensite phase and its undeformed crystal structure is twinned (less symmetric). In
this phase, it is easy to move twin boundaries that are mobile, and thus the material modulus of elasticity and
yield stress are quite low. On application of extensional stress to the material above its yield stress causes
detwinning (change orientation of crystal twin angle) and hence a plastic deformation. On application of heat to
this material, the plastic prestrain can be completely recovered (pseudo-elastic effect). The material is in austenite
phase and its crystals result in right angles ordered lattice. In the austeriite phase, it is more difficult to deform the
twin boundaries than martensite phase resulting in a higher yield stress and modulus of elasticity. Figure below
sketches the shape memory process. A plastic strain of more than 8% at low temperature (fully detwinned
martensite phase) can introduce permanent or irrecoverable plastic strain (dislocation generation and movement).
If SMA is constrained with a spring (such as the case with embedding in a host structure), the alloy is prevented
from returning to the original shape on heating that results in the generation of recovery stress. On the other
hand, if SMA is not constrained, there will be no recovery stress (free recovery) and on cooling there wifi be no
change of strain (from austenite to martensite phase, neglecting small thermal strain). For SMA embedded in a
host structure, the recovery stress will decrease on cooling and if this stress is stifi higher than martensite yield
stress, it will again result in a plastic strain (two way motion actuator).

SPIE Vol. 2717/47

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 Original Shape

No
Change
in
Shape

Marten site
(Twinned)
Process of Shape Memory Effect

The therino-mechanical behavior of SMA material depends upon the internal crystalline structure (phase)
that in turn is dependent on temperature, stress and history of the material. In the heating cycle, for temperatures
below A5, the material is in the 100% martensite phase, whereas for temperatures above Af, the material is in the
100% austenite phase. Higher stress increases A5 because more energy is needed to move the lattice structure
under imposed stress. In the cooling cycle, for temperatures above M5, the material is in 100% austenite phase
whereas for temperature below Mf, the material is in 100% martensite phase. Under an imposed stress, the
material will start to martensite at a higher temperature. Hysteresis can be viewed as the friction associated with
the movement of twin boundaries. At any other temperature between M5 and Mf, the material is partly in the
martensite phase and partly in the austenite phase. This state of the material is characterized by the volume
fraction of the martensite phase, x. The ability of the shape memory alloys to recover large strains comes from the
reversible phase transformation characteristics. Figure below schematically ifiustrates the change in martensite
volume fraction with temperature.

Martensite
volume
fraction

Temperature
Schematic of phase transformation

48/SPIE Vol. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 Constitutive Models
Many constitutive models have been developed to describe the thermo-mechanical behavior of SMA
materials. Some models are based primarily on thermo-mechanics and others are based on a combination of
thermo-mechanics and SMA phenomenology, and/or statistical mechanics. One of the most popular models is
Tanaka's model [Tana86] and is based on thermo-mechanics. In this model, the second law of thermodynamics is
written in terms of the Helinholtz free energy and then its rate form is derived. It is assumed that strain,
temperature and the martensite volume fraction are the only state variables. Tanaka developed an exponential
expression to describe the stress and temperature, rather than determining the free energy expression. Liang and
Rogers[Lian9O] presented a model which is based on the rate form of the constitutive equation developed by
Tanaka. In their model, Tanaka's equation is integrated with respect to time and it is assumed that the coefficients
in the equation are constant. Brinson and Lammering [Brin93J conducted a nonlinear finite element analysis to
predict the constitutive behavior of SMA.

The constitutive model assumes that stress, strain, temperature, and the martensite volume fraction are
the complete set of state variables for the shape memory alloys. In the constitutive equation, the stress consists of
three parts: the mechanical stress, the thermo—elastic stress and the stress due to phase transformation.
(o-—o.o)= E(E—E0)+e(T—I)+c(—0)
where and E respectively stress and strain, E is Young's modulus, E is thermo-elastic constant, T is
temperature, is phase transformation constant and is martensite volume fraction. The terms associated
with subscript 'o" refer to the initial conditions of the wire. During heating, the martensite volume fraction is:
= M exp[aA (A — T) + bAa]
where
aA = ln{(O.O1 I M ) /(A1 — Ac )}
hA a4 " CA
The 1S initial martensite volume fraction and CA is stress influence coefficient (Austenite phase) and is given
as:
1
CA
dA
d
During cooling, the martensite volume fraction is:
= (1— A)[1 — —
T)+ bMa}]
exp{aM(MS
where

aM =ln[1_ 1199]/(MM)
bM = aM ICM
where A is initial martensite volume fraction and CM, stress influence coefficient (Martensite phase) and is
given as:
1
CM =
dM
d
The coefficients used in the above constitutive relations, E, Q, W, x, M5, Mf, A, Af, CM , and CA , are
determined through testing of the SMA wires.
For the constrained recovery stress,
E— =0
The constrained recovery stress is obtained as:

SPIE Vol. 2717149

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 (a-a0)=®(T--i)+-0)
Note that the constrained recovery stress in the wire depends upon thermo-elastic constant, temperature, phase
transformation constant and martensite volume fraction.

Testing of SMA wires

Before a constitutive model can be used to predict recovery forces, the coefficients in the model have to be
determined through extensive experimentation. Although there have been many studies towards the
understanding of the thermo-mechanical behavior of Nitinol since its discovery in 1961, there is no reliable data
base available at this time. Thus, it is important to determine the characteristics of the material prior to its use.
Also, the thermo-mechanical properties of the SMA wire depend on many variables: prestrain, wire diameter and
condition of the wire (uncycled or cycled). The following procedure is normally employed prior to using the wire
as an actuator. First, to ensure that the wire is in the 100% martensite phase before testing, it is heated to its
austenite finish temperature and cooled. Second, to ensure its repeatable performance, the SMA wire is subjected
to thermo-mechanical cycling before any further testing. To cycle the wire, the wire is clamped at one end and
loaded by suspending a weight from the other end. Note that different load levels are used to create different pre-
strains. The wire is then unloaded and heated to its austenite finish temperature to recover most of the
deformation. The test is conducted in an enclosed chamber, so that the surrounding conditions do not influence
the temperature of the wire. To obtain the repeatability of recovery strain or stress, it becomes necessary to cycle
material thermo-mechanically many times to stabilize its micro-stmcture. Deforming the wire and then heating it
to its austenite finish temperature constitutes one cycle. Typically this cycle is repeated 30-50 times. Figure below
shows the influence of cycling on recovery strain of 20 mil diameter wire. Note that the recovery strain stabilizes
after about 40 cycles.

0 . 03

O.02
.
Pre-strain Applied force= 3.3 lbs
Stablized strain= 2.8 %
0.01•

0.00- I I I I
0 10 20 30 40 50
Number of cycles

The influence of thermo-mechanical cycling on recovery strain of a 20 mil diameter SMA wire.

Stress-strain Behavior
There is a marked variation of mechanical behavior of shape memory alloys at different temperatures.
The stress-strain curve for the martensite phase can be divided into three regions. The first region is an initial low
plateau pure elastic results due to detwinning type of mechanism. The second region is usually linear but not
purely elastic and takes place at a higher stress. It provides heat recoverable plastic strains of detwinned
martensite. In the third region, there is an onset of irrecoverable plastic strains similar to conventional metals. The
stress-strain curve can be noticeably different for tension, compression and torsion loads. Three identical SMA

SO/SPIE Vol. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 wires of 25 inches length and 20 mu diameter were tested to determine their thermo-mechanical characteristics.
Following figure shows the stress-strain curves at various temperatures for one cycled wire. It is to be noted from
this figure that the modulus and elastic limit of the wire increase with the increase in temperature due to phase
transformation.

Stress
(lb/in2)

Strain (in/in)
Stress-strain curves for cycled 20 mil diameter SMA wire at different temperatures-wire #1

There is a considerable variability between the stress-strain curves for identical wires at room
temperature. Note that at room temperature, the wire is expected to be completely in the martensite phase. The
variation of stress at low values of strain from wire to wire is low whereas at high strains, there is a considerable
variation of stress. The variation between the stress-strain curves between different identical wires is less at higher
temperature when the material is in austenite phase.
Linear segment of stress-strain curve was used to determine the Young's modulus of SMA wire. Figure
shows the variation of this modulus of the SMA wire with temperature. The modulus increases with the
increase in temperature due to a phase transformation. At 110°F, the modulus is about 6 times that of the
modulus at room temperature. The plot shows good repeatability of data at lower and higher ends of
temperatures, however, for in between temperatures the repeatability is not as good. For example, there is a good
repeatability at 75° F and at 110° F. At these temperatures, the wire resides in one phase- all martensite phase at
75°F and all austenite phase at 110° F. It is easier to control the temperature of the wire while it resides in one
phase. Also note that 95°F is the austenite start temperature of the wire and the amount of phase transformation
taking place at a given time varies during the testing of the wire. The difficulty in controUing the temperature of
the wire in this transition state was responsible to the large scatter of data.

SPIEVo!. 2717/51

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 1 1O ________________________________________

.0 Wire#1
Wire#2
0
810.6 • Wire#3
0 •
6 1O.
Modulus
lb/in2
4 1O.

S
21O. 0

0100. I
70 80 90 100 110 120
Temperature, °F
Variation of the modulus with temperature for two 20 mil diameter SMA wires.

Phase transformation temperatures depends upon stress level in the wire. These temperatures are
determined by testing the wire for recovery strain-temperature characteristics at different loads. Figure below
shows the variation of transition temperatures with stress. There is a linear relationship of phase transition
temperature with stress. The stress influence coefficients , CA and CM are obtained as inverse of slopes of these
curves.

140

A (a)
120

M (a)
100

Temperature A (a)
80 —————— -
M(a)
60

Stress, lb/in2

Variation of phase transition temperatures with stress

Following figure exhibits the variation of constrained recovery stress with temperature. This recovery
stress in the wire results from the shape memory effect under constrained displacement conditions (zero
displacements). The recovery stress for a given material depends upon pre-strain, temperature and martensite
volume fraction which in turn depends upon the temperature. The recovery stress in the temperaturerange
where there is no phase transformation is controlled by thermoelastic coefficient. Hence, the thermo-elastic

52 /SPIE Vol. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 coefficient is obtained by considering the range of temperature below As. The thermo-elastic coefficient is the
slope of the linear segment of the curve.

20000

1500(
Stress e = Slope of plot before
A is reached
(1b/in2)1000
c0=l .7%

500(

80 ico
- •i: - •izb 10
Temperature, °F

Recovery stress versus temperature

Table 3. shows the thermo-metharücal properties of SMA wires as obtained from the described tests. Note
a 5.6 times increase in the modulus due to phase transformation from martensite to austenite.

Table 3. Thermo-mechanical Properties of SMA wires

Coefficients! Units Values Coefficients! Units Values

Properties Properties
Q psi/°F 194.66 A5 °F 71.643
w psi -O.132041e+6 Af °F 110.71
CA psi/°F 617.6 M5 °F 100.0
- CM psi/°F 4219.9 Mf °F 70.714
EA psi 8.36e+6
EM psi 1.49e+6

30 ___________________________________
I • Test 1
25 a Test 2 a
20 a
Applied Load=3.3 lbs
Recovery 1 5 Pre-strain= 2.8% 1
Force (lb.)
10 I
5 I
I
o
70
- 80 90 100 110 120 130 140 150 160
p
I
Temperature, °F
Recovery force versus the temperature for a 20 mil diameter SMA wire

SPIE Vol. 2717/53

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 Above figure shows the constrained recovery force developed in 20 mu diameter SMA wire at different
temperatures. Note that the austenite finish temperature for this material is 158 F.

Composite Beams with Embedded SMA Wires

SMA wires have been used to alter the natural frequencies of composite structures in several studies.
Barker and Rogers [Bark9O] showed an increase in the natural frequencies of a composite beam due to the
activation of SMA wires embedded directly in the structure. In this experimental study, the beam and SMA wires
were independently clamped. When the SMA wires were heated, the beam was subjected to an axial force due to
the shape memory effect and the elastic constraints due to the beam on the wire. They demonstrated a 200%
increase in the natural frequency of graphite-epoxy beams by using a 15% volume fraction of SMA wires.
Correlation of experimental results with predictions were predictions were not carried out.
Baz, Imam and McCoy [Baz9O] conducted a study on the active vibration control of flexible beams.
Experiments were conducted on flexible beams with SMA wires mechanically constrained on the exterior of the
structure. The recovery force due to mechanically-constrained, prestrained SMA wires at higher temperatures
was used to demonstrate active vibration control of flexible beams. In such an application, external access to the
substructure becomes essential in order to clamp SMA wires to an independent support. For many aerospace
structures like rotor blades, it may not be possible to follow this scheme. In another study, Baz et al [Baz9lJ
inserted SMA wires into flexible beams with sleeves to control their buckling and vibration behavior. They used
the finite element method to correlate with their experimental results. The experimental results compared well
with the theory. They showed that the buckling load of a flexible fiberglass composite beam could be increased
three times when compared to the buckling load of art uncontrolled beam. Epps and Chandra [Epps95] presented
an experimental-theoretical study on the active tuning of graphite-epoxy rectangular solid section beams with
SMA wires inserted in embedded sleeves and showed a 22% increase in the first natural frequency using one 20
mil diameter wire. The volume fraction of SMA wires in this configuration was 2%. Good correlation between
theory and experiment was achieved in this study.
Composite beams with fused silica tubes filled with 'dummy steel wires were fabricated using an
autoclave molding technique. In order to assure a good bond between the tubes and the graphite-epoxy material,
a ifim adhesive was used. The number of prepreg plies were eight. The inner and outer diameters of the silica
tubes were 27 and 33 mils respectively. After curing the composite beam, the steel wires were replaced with pre-
strained SMA wires. Thus graphite-epoxy beams of solid rectangular cross-sections were built.

SMA wires
Graphite-epoxy beam with SMA wires.

The composite beams were tested for their bending frequencies under clamped-clamped boundary
conditions. The natural frequencies of a composite beam with SMA wires inserted into sleeves embedded in the
beam depend upon the beam parameters and the SMA characteristics. The beam parameters are length, thickness,
width, material and boundary conditions. The SMA wire parameters are recovery force in each wire (which in
turn depends on pre-strain, mechanical properties and temperature) and number of wires. Following figure
shows the first bending frequency of a graphite-epoxy composite beam activated by one 20 mil diameter SMA
wire. The dimensions of this beam are as follows: clamped length = 18.0 inches, width = 0.25 inches, and thickness

54 ISPIE Vol. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 = 68.0 mils. The increase in the fundamental frequency due to 100% SMA activation (temperature =
160°F) is
21.8%. The correlation between theory and experiment is within 5%. Note that the experimentally obtained
recovery force is used in the prediction of the frequency. The SMA wires are treated as an elastic foundation for
the composite beam and stiffness of foundation depends on the constraint force. It is important to note that the
prediction of frequency correlates with the experimental results within acceptable limits only when the beam on
elastic foundation analysis is used.

S Experiment
140 Theory-Beam on Elastic Foundation
--S Theory-Beam with Axial Force

120- a-.
a

Frequency a
Hz
i:o

60 80 100 120 140 160 180

Temperature, deg F

Fundamental frequency of clamped-clamped graphite-epoxy beam activated by one 20 mu dia SMA wire, beam
dimensions: length = 18 in., width = 0.25 in., thickness = 0.068 in.

It is feasible to increase the frequencies of beams using SMA wires. Different mechanisms of using these
wires described are: external connection to the substructure, embedding in composite substructure and inserting
in sleeves embedded in substructure. The mechanism of inserting wire in embedded sleeve appears promising in
changing the frequencies.

6. Applications
Applications of smart structures technology to various physical systems are focused to actively control
vibration, noise, and stability. Applications range from space systems, fixed-wing and rotary-wing aircraft,
automotive, civil structures and machine tools. Scope and potential of smart structures applications to
aeronautical systems are expanding. Embedded or surface-bonded smart actuators on an airplane wing or
helicopter blade will induce alteration of twist/camber of airfoil (shape change), that in turn will cause variation
of lift distribution and may help to control static and dynamic aeroelastic problems. Ehlers and Weisshaar [Ehie]
showed that the piezoelectric actuators can be used to control static aeroelastic behavior of wings that include lift
effectiveness, divergence and roll effectiveness. Lazarus, Crawley and Lin [Laza] showed the potential of flutter
suppression using piezo actuators. Scott and Weisshaar [Scot] studied control of panel flutter with piezo actuators

SPIE Vol. 2717/55

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 and shape memory alloy actuators. Heeg [Heeg92] carried out an experimental investigation for flutter
suppression of a simple wing model (beam) with piezoelectric actuators. There are other studies [Barr94, Barr95,
Ross93} that use shape control to stabilize flutter of missile fins and aircraft wings.

Compared to fixed-wing aircraft, helicopter suffer from severe vibratory loads, more susceptibility to
aeromechanical instability, excessive noise levels and poor flight stability characteristics. To reduce these
problems to an acceptable level, numerous passive devices arid many ad hoc design fixes, are resorted to with
resultant weight penalities and reduced payloads. The primary souce for all these problems is the main rotor
which operate in an unsteady and complex aerodynamic environement. Currently, considerable research has
been directed towards application of smart structures technology to rotor systems to improve their performance.
Numerous light—weight compact actuators and sensors are embedded or surface-mounted at different stations on
the blades and optimal distributed foces applied with the help of modem control theory. At the University of
Maryland, two types of Froude-scale smar rotor models were built: controllable twist models incorporating
embedded piezoceramic elements, arid trailing-edge flap models actuated with smart actuators. For the first
concept, a six-foot diameter four-bladed bearingless rotor model was built, where each blade was embedded with
banks of specially-shaped piezoelectric actuators at +1- 45 degree respectively on the top and bottom surfaces
(Figure below). When the same potential was applied to the top and bottom banks, it cause pure twisting of the
blade. The rotor was tested on a hover stand and tip twisting of the order of .2 degree was achieved at the
operating speed of 900 RPM for a 4/rev excitation [Chen93, Chen94, Chen95]. Though this dynamic twist
amplitude is an order of magnitude lower than needed to suppress vibration (need about 2 deg.), it shows the
potential for further research. Such a scheme shows the advantage of distributed actuation and sensing, allowing
for spanwise and azimuthal control of blade lift to maximize rotor performance.

trailing edge banks

fiberglass cloth skin
[0/90}

Rohacell rigid foam

fiberglass spar —' .25 chord

NACA 0012 Airfoil Section

Piezoceramic Blade Cross Section Details; with embedded Piezo Elements at +/450

For the second concept, a six-foot diameter two-bladed bearingless rotor model was built, where each
blade was mounted with a trailing-edge flap actuated with a piezo bimorph. The bending deflection of a
cantilevered piezo bimorph is amplified using a mechanical leverage system to rotate a trailing-edge flap (Figures
below). Significant oscifiatory flap deflections have been sustained in both fixed-wing testing and in rotation (up
to 2 degrees at 900 RPM) [Walz94, Benz95]. A major drawback of this scheme is its complex hinges and leverage
system where considerable losses take place especially at higher rotational speeds. Even though this flap
amplitude is about one-fourth of the needed amplitude to suppress vibration, this concept appears more
promising.

MAIN ROTOR BLADE WITH FLAP ACTUATOR

56/SPIE Vol. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 Bimorph

Deflection
Fig. 3: Bimorph Flap Actuation System

Another novel concept of trailing-edge smart flap concept has been investigated where a magnetostrictive
actuator is used in conjunction with extension-torsion coupled composite tube to provide rotation of a flap for
full-scale applications (Figure shown below). This scheme does not require complex mechanical hinges and
leverage system. Because of poor structural efficiency of extension-torsion tube, this concept does not appear
viable [Both94}. To obtain improved trailing-edge flap performance in rotation, a new actuation concept is
investigated where a bending-torsion coupled composite beam with surface-bonded piezoceramic is used to
actuate the flap [Bern96J. A novel spanwise variation of ply lay-up and piezoceramic elements phasing is used to
minimize bending response and maximize twist response at the tip of a cantilevered beam. This flap mechanism
did not involve any mechanical amplification system. A six foot diameter Froude-scaled rotor model was tested
on hover stand and trailing-flap defection of 4 degrees was obtained at the operating speed of 900 RPM. It is
important to note that unlike earlier concepts, the flap performance did not degrade with rotational speed.

Trailing edge
flap

Extension-Torsion
oupled composite
tube

Schematic of rotor blade trailing edge flap actuated by extension-torsion coupled composite tube and
magnetostrictive actuators

From early tests on smart rotors showed that high force, high displacement actuators are needed to
achieve any realistic vibration reduction or performance improvement. The commercially available piezo
bimorphs, by virtue of their design, are not capable of generating the required actuation force. To achieve high
force, high displacement actuation, two types of actuators respectively using piezoelectric and electrostrictive
stacks as drivers, in conjunction with mechanical leverage system were developed [Sama94]. The first actuator
with piezo stack was developed to actuate a 'flaperon' which consisted of a small movable surface to trip the
boundary layer, located on the top surface of a wing model with span and chord each of 12 inch and of NACA
0012 airfoil. The second actuator with electrostrictive stack as a driver was designed to move a leading edge
droop flap hinged at 25% chord of a wing model with span of 8 inch, chord of 4 inch and of VR-12 airfoil.
Experiments were performed on both stacks to evaluate their important characteristics such as block force, free

SPIE Vol. 2717/57

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 displacement, stiffness that were essential in the design of the actuators. The results showed that the block force
obtainable from piezo stack was higher and that of electrostrictive stack was lower than that specified by the
respective manufacturers while the free displacements are about the same. A major drawback with these
actuators was their mechanical amplification system. The lever arm supported on two knife edges often slipped
during testing (Figure below). Recently in another study [Spen96], high-force piezo stack was used to actuated a
trailing-edge flap on a 12-inch chord fixed-wing model. Mechanical amplification was used to amplify small
deflections of piezo stack. Dynamic deflection on the order of 4 degrees were obtained in the free-jet wind tunnel
at a free stream velocity of 90 ft/sec. At higher speeds, flap performance degraded considerably. From dynamic
test, it became clear that an amplification system without hinges could be effectively designed to produce high
force and high displacement actuators for application to rotor systems.

Frame

Piezo Stack Spring

Fig. 6: Flaperon Actuator

Stroke Amplification: For most applications, there is a need for compact, high force and large displacement
actuators. Most actuators, in particular piezoceramic actuators, are low force and low stroke devices. To increase
actuation force, multi-layered actuators such as piezo stacks can be used. A key challenge is to amplify the stroke
of these actuators. Large mechanical amplification using a compact leverage system often leads to substantial
losses at hinges and slippage at knife edges. Replacing mechanical hinges with flexure can overcome some of
these problems, but requires enormous effort and experience to perfect such systems. To amplify the stroke of
piezo devices, specially shaped actuators are being built. Typical ones are: rainbow actuators, moonie actuators
and C-block actuators. C-block actuators are multi-layered arched bimorph piezoelectric actuators and a large
axial displacement can be achieved using a series arrangement of C-actuators [Brei95]. So far, most of these
specialized actuators have shown limited success. For practical applications, it is necessary to develop large stroke
actuators and hence requires special attention.

7. Status and Recommendations for Future Work

The summary of the state-of-the-art as presented in the paper together with recommendations for future work
as as follows.

1. Development of Large Stroke Smart Actuators: At this time, most of the commercially available smart
actuators (such as piezoceramics and magnetostrictors) are low stroke and low force devices. For most
applications, there is a need for compact, high force and large displacement actuators. A key challenge is to
amplify the stroke of these actuators. Large mechanical amplification using a compact leverage system often leads

58/SPIE Vol. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 to substantial losses at hinges and slippage at knife edges. The goal should be to increase displacement capability
of smart materials actuators by 300 to 500%.

2. Data Base for Smart Materials Characteristics: For design development of smart structures, it is essential to
have a reliable data base of characteristics of smart materials. At this time, such a data base for smart materials is
non-existent. For example, basic characteristics such as Young's modulus and yield stress for Nitinol at different
temperatures and stresses are not available. Thus, there is need to focus on the development of a reliable data
base of smart material characteristics through extensive testing of engineering specimens (macro level).

3. Beam Modeling: Because the local strain distribution near a piezo actuator is two-dimensional, it is important
to refine 1-D beam models that capture this effect. Simple models for beams with actuators not aligned with the
beam axis are inadequate and should be improved. Most analyses assume linear actuation strain with field, and
hence need to be improved to include nonlinear actuation strains; in particular to predict the response for a higher
voltage and a lower frequency.

4. Plate Modeling: There is a general lack of validation studies with experimental data or with detailed analyses.
Most plate models assume a perfect bond condition that is too restrictive. There is a need to develop detailed 3-D
finite element analyses including nonlinear actuation strain to check the limitations of existing laminated plate
analyses as well as to understand the diffusion of actuation strain.

5. Constitutive Relations for Magnetosrictives and Electrostrictives: At this time, most of the smart structures
studies are focused on piezoelectric and SMA actuators. Simplified constitutive relations for magnetostrictive and
electrostrictive actuators are non-existent. To exploit these actuators for some special applications, it is important
to develop simplified constitutive relations and validate them with experimental data.

6. Shape Memory Alloys: Systematic validation of constitutive models of SMA for different temperatures and
strains are not available and is a major barrier for design development of smart structures. Building of composite
structures with embedded SMAs is still a challenge and needs a focused effort. Analysis of structures with
embedded SMA has so far shown mixed success in terms of correlation of predicted results with measured data.
To exploit the potential of SMA to various applications, it is important to refine and simplify analyses and carry
out systematic validations.

7. Application to Realistic Systems: Currently, this emerging technology is applied to models/simulations that
are often unrealistic. For a proper assessment of this technology, it is necessary to apply it to realistic structures.
Major barriers are: actuator stroke, non-availability of robust distributed parameter control strategies, and non-
existent mathematical modeling of the smart system. The objective should be to build and test dynamically-scaled
models and evaluate performance of actuators and feedback control algorithms under different operating
conditions. Using these test data, validate comprehensive analyses and then carry out multidisciplinary
optimization studies to develop a smart rotor system.

8. Expand Smart Structures Applications to Improve Performance and Minimize Acoustics: Currently, most
applications of smart structures technology to aero systems such as rotors are focused on vibration minimization.
There is an enormous pay-off to exploit this technology to improve system performance and minimize
external/internal noise. It is necessary to build dynamically-scaled models and test them in different operating
conditions. Key barriers are: large bandwidth-large stroke actuators, non-existent active noise control analytics,
robust acoustic sensors, well-defined objectives, and unavailability of distributed parameters control algorithms.
The objective should be to build and test dynamically-scaled models and evaluate performance of smart actuators
and feedback controllers under different flight conditions. Using these test data, validate comprehensive analyses
and then carry out multidisciplinary optimization studies to develop a smart structures system to simultaneously
increase system performance and minimize noise and vibration.

9. Smart Structures Application to Increase Damping: There are systems that are inherently low damped and in
need of damping augmentation. For example, to overcome the aeromecharücal instability of helicopters, rotors are
installed with either complex mechanical dampers or expensive elastomeric dampers. To reduce the maintenance

SP!E Vol. 2717/59

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 and procurement cost substantially, there is a potential application of smart structures technology toaugment
rotor damping. Hybrid active/passive constrained layer damping in conjunction with ER (Electro-Rheological)
and MR (Magneto-Rheological) fluid dampers as well as shape control devices (such as flaps) can be used to
augment lag mode damping. Current barriers are: inadequate analytical models, insufficient basic test data, and
impractical simulations. Road maps should be established to develop and validate different analytical elements
of damping augmentation, as well as to apply these to dynamically-scaled models and evaluate their performance
under different operating conditions.

10. Structural Health Monitoring of Systems: Aerospace systems are highly susceptible to damage because of
severe vibratory and fatigue loads. The problem is addressed by frequent inspection of damage-sensitive parts
and, if required, replacement of such parts. This contributes to higher operating and maintenance costs. The
health monitoring of aerospace systems offer the potential to lower operating costs and enhance flight reliability.
At this time, there have been only limited experimental studies on this topic and analytical tools areprimitive.
Major barriers are unavailability of robust sensors, reliable system identification algorithms, and proper ifiters.
Keeping in view the future goal to reduce maintenance costs of aerospace systems by at least 25 -50%, there is a
need of focused fundamental research activities related to health monitoring of aerospace and other systems
using smart sensors.

Acknowledgements
The author wishes to acknowledge the support of the Army Research Office under the Grant DAAL 03-
92-G-0121 with Dr. Gary Anderson as Technical Monitor.

References
[Barr94} Barrett, R., "Active Plate and Missile Wing Development Using DAP Elements," AIAA Journal, Vol. 32,
No. 3, pp. 601-609, Feb. 1994.

[Barr95 I Barrett, R., Gross, R. S. and Brozosky, F. "Missile Flight Control Using Active Flexspar Actuators," SPIE
Conference on Smart Structures and Materials, Feb. 26-March 3, 1995, San Diego, California.

[Baz9O] Baz, A., Imam, K., McCoy, J., "Active Vibration Control of Flexible Beams Using Shape Memory
Actuators," Journal ofSound and Vibration, Vol. 140 (3), 1990, pp. 437-456.

[Baz9l] Bar, A., Ro, J., Mutua, M., Githeany, J., "Active Buckling Control of Nitinol-Reinforced Composite Beams,"
Conference on Active Material and Adaptive Structures., Alexandria, VA, Nov. 1991, pp. 167-176.

[Benz95] Ben-Zeev, 0., and Chopra, I., "Advances in the Development of an Intelligent Helicopter Rotor
Employing Smart Trailing-Edge Flaps" Proceedings ofthe 1995 North American Conference on Smart Structures and
Materials, February 1995, San Diego, California.

[Bem96] Bernhard, A. P. F. and Chopra, I.,"Development of a Smart Rotor with a Trailing-Edge Flap Activated by
a Bending-Torsion Coupled Composite Beam," Proceedings of the 1995 North American Conference on Smart
Structures and Materials, February 1996, San Diego, California.

[Both94] Bothwell, M. C., Chandra, R. and Chopra, I., "Torsional Actuation with Extension-Torsion Composite
Coupling and a Magnetostrictive Actuator"Proceedings of the 35st AIAA /ASME/ASCE/AHS/ASC Structures,
Structural Dynamics and Materials Conference, Hilton Head, SC, April 1994.

[Brei95] Brei, D. E. "Force-Deflection Behavior for C-Block Actuators Architectures,"Proceedings of the 1995 North
American Conference on Smart Structures and Materials, February 1995, San Diego, California.

60/SPIE Vol. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 [Brin93] Brinson, L. C., and Lammering, R., 'Finite Element Analysis of the Behavior of Shape Memory Alloys and
Their Applications,' International Journal of Solids and Structures, VoL 30, 1993, pp. 3261-3280.

[Chan93} Chandra, R. and Chopra, I., "Structural Modeling of Composite Beams with Induced Strain Actuation,"
AIAA Journal, Vol. 31, No. 9, Sept. 1993, pp. 1692-1701.

[Chen94] Chen, P. C. and Chopra, I., "Induced Strain Actuation of Composite Beams and Rotor Blades with
Embedded Piezoceramic Elements," 1994 North American Conference on Smart Structures and Materials, Orlando,
Florida, Feb. 1994.

[Chen93J Chen, P. C., and Chopra, I., "A Feasibility Study Build a Smart Rotor: Induced-Strain Actuation of Airfoil
Twisting Using Piezoelectric Crystals' Proceedings of the 1993 North American Conference on Smart Structures and
Materials, February 1993, Albuquerque, NM.

[Chen94] Chen, P. C., and Chopra, I., "Induced Strain Actuation of Composite Beams and Rotor Blades with
Embedded Piezoceramic Elements' Proceedings of the 1994 North American Conference on Smart Structures and
Materials, February 1994, Orlando, Florida.

[Craw87J Crawley, E. and de Luis, J., "Use of Piezoceramic Actuators as Elements of Intelligent Structures," ALAA
Journal, Vol. 25 No. 10, October 1987, pp. 1373-1385.

[Craw89] Crawley, E. and Anderson, E., 'Detailed Models of Piezoceramic Actuation of Beams," Proceedings of
the 30th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference, AIAA,
Washington, DC, April 1989.

[Cros69] Cross, W. B., Kariotis, A. H., Stimler, F. J., Nitinol Characterization Study," NASA CR-1433, September
1969.

[Duer9O] Duerig, T. W., et al, "Engineering Aspects of Shape Memory Alloys," Butterworth-Heinemarin Ltd, Great
Britain, 1990.

[Dye9O] Dye, T. E., "An Experimental Investigation of the Behavior of Nitinol," Masters Thesis, Department of
Mechanical Engineering, Virginia Polytechnic Institute and State University, Blacksburg Virginia, August,
1990.

[Ehle9O} Ehiers, S. M. and Weisshaar, T. A., "Static Aeroelastic Behavior of an Adaptive Laminated Piezoelectric
Composite Wing," 31st AIAA 5DM Conference, April 1990.

[Eh1e92J Ehiers, S. M. and Weisshaar, T. A., "Effect of Adaptive Material Properties on Static Aeroelastic Control,"
33rd AIAA SDM Conference, April 1992.

[Epps9sa] Epps, J. and Chandra, R., "Shape Memory Actuation for Active Tuning of Composite Beams", SPIE
Conference on Smart Structures and Materials, Feb. 26-March 3, 1995, San Diego, California.

[Haje9l] Hajela, P. and Glowasky, R., " Application of Piezoelectric Elements in Supersonic Panel Flutter
Suppression," AIAA Aircraft Design Systems and Operations Meeting, Sept. 1991.

[Heeg92] Heeg. J., "An Analytical and Experimental Investigation of Flutter Suppression Via Piezoelectric
Actuators," 33rd AIAA SDM Conference, April 1992.

[Laza9l] Lazarus, K. B., Crawley, E. F. and Lin, C. Y., "Fundamental Mechanism of Aeroelastic Control with
Control Surface and Strain Actuation," 32nd AIAA SDM Conference, April 1991.

SPIE Vol. 2717/61

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms

 [Lin94} Lin, C. Y. and Crawley, E. F., "Towards Optimal Aeroelastic Control Using Elastic and Induced Strain
Anisotropy," 35th AIAA SDM Conference, April 1994.

[Lian9OJ Liang, C., and Rogers, C. A. , "One-Dimensional Thermomechanical Constitutive Relations for Shape
Memory Material,' Proceedings of AIAA/ASME/ASCE/AHS/ASC 31st Structures, Structural Dynamics, and
Materials Conference, Long Beach, CA, April 1990.

[Park93} Park, C., Walz, C. and Chopra, I., "Bending Torsion Models of Beams with Induced Strain Actuation,' 1993
North American Conference on Smart Structures and Materials, Albuquerque, New Mexico, Feb. 1993.

[Park94} Park, C.. and Chopra, I., "Modeling Piezoceramic Actuation of Beams in Torsion," 35th
AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference and Adaptive Structures
Conftrence, Hilton Head, SC, April 1994.

[Roge9O} Rogers, C. A. arid Barker, D. K. , " Experimental Studies of Active Strain Energy Tuning of Adaptive
Composites," Proceedings of AIAA/ASME/ASCE/AHS/ASC 31st Structures, Structural Dynamics, and
Materials Conference, Long Beach, CA, April 1990.

[Ross93] Rossi, M., Austin, F. and VanNostrand, W., 'Active Rib Experiment for Shape Control of an Adaptive
Wing," 34th AIAA 5DM Conference, La Jolla, Calif., April 1993.

[Sama93] Samak, D. K., and Chopra, I., "A Feasibility Study to Build a Smart Rotor: Trailing Edge Flap Actuation"
Proceedings ofthe 1993 North American Confrrence on Smart Structures and Materials, February 1993, Albuquerque,
NM.

[Sama94] Samak, D. K., and Chopra, I., "Design of High Force, High Displacement Actuators for Helicopter
Rotors" Proceedings of the 1994 North American Conference on Smart Structures and Materials, February 1994,
Orlando, Florida.

{Sche79} Schetky, L. M., " Shape Memory Alloys," Scientific American, Nov. 1979,Vol. 241, No. 5, pp.74-82.

[Sche68} Schuerch, H. U., " Certain Physical Properties and Applications of Nitinol," NASA CR-1232, November
1968.

[Scot9l] Scot, R. C. and Weisshaar, T. A., "Controlling Panel Flutter Using Adaptive Materials," 32rd AIAA SDM
Conference, April 1991.

[Spen96] Spencer, B. T. and Chopra, I., "Design and Testing of a Smart Trailing-Edge Flap Using Piezoelectric
Stacks," Proceedings of the 1996 North American Conference on Smart Structures and Materials, February 1996, San
Diego, California.

[Stre92] Strehiow, H. and Rapp, H., "Smart Materials for Helicopter Rotor Active Control," AGARDISMP Specialist's
Meeting on Smart Structures for Aircraft and Spacecraft, Lindau, Germany, October 1992.

[Tana86] Tanaka, K., "A Thermomechartical Sketch of Shape Memory Effect: One-dimensional Tensile Behavior,"
Res. Mechanica, 18, p. 251-263, 1986.

[Wa1z94} Walz, C. and Chopra, I., "Design and Testing of a Helicopter Rotor Model with Smart Trailing-Edge
Flaps" Proceedings of the 35st AIAA /ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials
Conference, Hilton Head, SC, April 1994.

62 ISPIE Vol. 2717

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/12/2015 Terms of Use: http://spiedl.org/terms