Home Search Collections Journals About Contact us My IOPscience

Evaluation of the characteristics of a shape memory alloy spring actuator

This content has been downloaded from IOPscience. Please scroll down to see the full text.

2000 Smart Mater. Struct. 9 817

(http://iopscience.iop.org/0964-1726/9/6/311)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 134.151.40.2
This content was downloaded on 06/09/2014 at 17:26

Please note that terms and conditions apply.

Smart Mater. Struct. 9 (2000) 817–823. Printed in the UK PII: S0964-1726(00)17095-8

Evaluation of the characteristics of a

shape memory alloy spring actuator
Hyo Jik Lee and Jung Ju Lee†
Department of Mechanical Engineering, ME3033, Korea Advanced Institute of Science and
Technology, 373-1, Kusong-dong, Yusong-gu, Taejon 305-701, Korea (South)
E-mail: [email protected]

Received 25 February 2000, in final form 21 August 2000

Abstract. A shape memory alloy (SMA) spring actuator for an active catheter was fabricated
by the heat treatment of wound SMA wire on a mandrel. In order to evaluate the
characteristics of the fabricated SMA spring actuator, four types of tests were performed, as
follows: isothermal loading and unloading, measurement of shape recovery force,
temperature follow-up, and load follow-up tests. In this work, it was first developed that the
prediction method of the shape recovery force was based on the neural network theory. The
prediction results were better than the results from the conventional constitutive equation of
SMA. From the experimental results, it was also proved that the following characteristic for
loads was better than that for temperature and the following characteristics could be enhanced
according to cooling and control methods.
(Some figures in this article are in colour only in the electronic version; see www.iop.org)

1. Introduction

In many fields of application of intelligent systems there

is a growing need for active structures that are able to
modulate their mechanical characteristics and to transfer
their action to surrounding or neighboring passive structures.
This need is widely present in the area of advanced robotics
and bioengineering. This becomes critical in miniaturized
devices [1].
Among many potential applications of intelligent
systems, we focus our attention on the field of medical
engineering. Medical operation is a field that demands
specialty, accurate skill, and promptness. Recently, to
enhance the quality of medical service, medical robots,
new operation methods, and surgical tools are receiving
much attention. Medical robots should be able to perform
sophisticated work and surgery on the planned path, or can Figure 1. Concept of an active catheter with a large bending angle
be manipulated in real time by surgeons [2]. in blood vessels.
In recent years, as minimal invasive surgeries are be-
coming more popular, intravascular catheter therapy instead
skill of maneuvering the catheter as well as a short handling
of conventional open surgery has been increasingly in de-
time is quite important to reduce the patients’ pain. There-
mand [1]. However, conventional catheters cannot move ac-
fore, new types of catheter capable of active movement with
tively because steering and operation with catheters requires
intelligent functions are in demand [3–9]. Active catheters
certain human skill by an expert surgeon. The aforemen-
generally require a large bending angle (over 90◦ ) to facili-
tioned catheter is a hollow flexible tube that can be inserted
tate placement in the tortuous vascular anatomy and, also, an
into a vessel or cavity of the body to withdraw or to instill
ultra-slim outer diameter to allow easy insertion into narrow
fluids, monitor various types of information, and visualize a
lumen, as shown in figure 1.
vessel or cavity. The limited activity and controllability of
conventional catheters results in great pain and some damage In the catheterization, microactuators are increasingly
to patients as well as fatigue to surgeons. Thus, handling expected to replace the operator’s hands. The actuator
technologies under development include shape memory
† Author to whom correspondence should be addressed. alloys (SMAs), piezoelectric actuators, magnetostrictive

0964-1726/00/060817+07$30.00 © 2000 IOP Publishing Ltd 817

Hyo Jik Lee and Jung Ju Lee

Table 1. Selective properties of actuator materials.

Density Stress Strain Life
Class (kg m−3 ) (MN m−2 ) (%) cycles
Piezoelectric 7500 35 0.09 >108
SMA 6450 >200 >5 >105
Polymer 1500 180 >2 >105
Magnetostrictive 9250 70 0.2 >105
Electrostatic 1061 0.04 >10 ?

actuators, contractile polymers, and electrostatic actuators.

No technology has dominance over any other, but the choice
of actuator depends on the mechanism and where they are
used. Table 1 shows a comparison of different actuators [10].
Among many actuators, SMA actuators exhibiting the
shape memory effect (SME) have been receiving attention for
the design of actuators [11–13]. The SME does not simply
mean that the previous shape will be restored freely when
Figure 2. Experimental set-up and control block diagram.
the SMA is heated. Restraining a deformed SMA spring
while heating the spring above the transition temperature
will generate a large recovery stress. This unique behavior the SMAs. The martensitic transformation may be simply
is one of the fundamental characteristics exploited in many illustrated by the change in the martensitic volume fraction
applications of SMAs as actuators. The major advantages of with respect to temperature and stress, as in [12]. These
SMA actuators over others are the high power to volume ratio thermomechanical behaviors have been described by a
and the large maximum recovery strain. Therefore, SMA is a variety of constitutive models [14], and the differences
very useful material as an actuator for an active microcatheter. between the actual behavior and the model have been
Some actuating mechanisms for manipulating active presented previously. Therefore, this paper suggests the
catheters have been proposed [4–9]. Actuating mechanisms neural network model based on the experimental results. In
can be largely classified into catheters of serial type or parallel this work, the suggested neural network model predicts the
type. The representative catheter of the serial type is the shape recovery forces, which is the important factor to be well
active endoscope developed by Maeda et al [7]. Since predicted in terms of control problems of SMA actuators.
the structure of the active endoscope is very simple, the
fabrication process is comparatively easy. However, this 3. Experimental details
simple structure keeps the active endoscope from bending
in various directions. On the other hand, the representative 3.1. Fabrication of SMA spring actuator
catheter of the parallel type is the multi-link active catheter The most popular and useful form of an NiTi SMA for
by Lim et al [8]. This catheter can be bent in any direction actuators is the helical coil spring. Such springs can
with three degrees of freedom (3-DOF) per link, but the provide an impressively large stroke and be designed to exert
fabrication process is so complicated. In consideration of the significant forces. Thus, this type of actuator is suitable for
size of the SMA spring fabricated in this work, the actuators the actuator of the active catheter.
are proposed for the actuators of the catheter of the serial A heat treatment is needed in order to make the SMA
type. Basically, this paper covers only the evaluation of spring. The detailed process for the fabrication of SMA
the characteristics of actuator part among the entire active springs is well described in [15]. An SMA spring was
catheter systems. fabricated from 0.38 mm SMA wire of Shape Memory
In this paper, to evaluate the potential of an SMA spring Applications, Inc. The spring used in this experiment has
actuator for application to an active catheter, four types 2 mm inside diameter, 11.5 turns, and a linear end shape.
of tests of SMA spring actuator were performed, such as
isothermal loading and unloading, measurement of the shape 3.2. Testing system
recovery force, temperature follow-up, and load follow-up
tests. Based on these test results, simulation analysis by a A PC-based test set-up was built to investigate the
conventional constitutive equation was performed to predict performance of the SMA spring. As shown in figure 2, this
the shape recovery force of SMA spring actuators, and then experimental set-up has the following characteristics.
a newly developed neural network method was suggested for (1) The testing system was designed to evaluate the
the prediction of the shape recovery force of the SMA spring isothermal loading and unloading, shape recovery force,
actuator. temperature follow-up, and load follow-up tests by
using a linear variable differential transformer (LVDT),
2. Thermomechanical behavior of SMA loadcell, and thermocouple.
(2) Electrical resistance heating was used to increase the
The martensitic transformation is the basic characteristic temperature of the SMA. Therefore accurate and high-
of an SMA involved in all the unique characteristics of speed response voltage control was required. The

818
 Characteristics of a shape memory alloy spring actuator

Figure 3. Isothermal force–displacement curves of an SMA Figure 4. Recovery stress against temperature at various initial
spring actuator. tensile forces.

voltage source used was the current amplified analog and the corresponding stress range for the reverse
output signal from an analog to digital/digital to transformation is given by
analog (AD/DA) acquisition board.
(3) During a measurement of the SMA spring temperature, CM (T − Ms )
σeq
CM (T − Mf ) (2)
the thermocouple was electrically insulated with the √
SMA spring to avoid significant noise on the signal of where σeq is the equivalent stress (σeq = 3τ ) and CA
the temperature sensor because of the current flowing and CM are material constants related to the influence of the
through the SMA wire. We used a K-type thermocouple stress on the phase transformation. The shear stress, τ , from
with a very fine diameter (0.012 mm) and wound it onto Hooke’s law for a helical coil spring, is given by
the SMA spring wire to avoid point contact and to obtain 2F R
a stable signal. We used only one thermocouple at the τ= (3)
πr 3
left end part of the spring for measuring the temperature
of the SMA. The placement of the thermocouple should where F is the tensile force, R is the radius of the spring
be at the linear part since the thermocouple is apt to be coil, and r is the radius of spring wire. In figure 3, an
detached from the coil part due to the movement. We increase of temperature increases the transformation stresses
assumed that the cross sectional area of the SMA wire as in equations (1) and (2). Since no residual strain remains
was constant throughout the entire spring actuator. This above 30 ◦ C in figure 3, we can determine that Ms will be
means that the resistance of the SMA wire is constant near 30 ◦ C. A residual strain remains after the isothermal
throughout the entire spring actuator and thus the heat loading and unloading process decreases below about 30 ◦ C
dissipation is constant. Additionally, the condition of in figure 3 and then heating above Af can restore the SMA
convection was assumed to be same throughout the entire spring with residual strain to the original shape. We can
part since the spring actuator is small. Thus, the use obtain CA , CM , Af , and Ms of equations (1)–(3) from the
of one thermocouple would not give an error in the results of an isothermal force–displacement experiment as
measurement of the average temperature. shown in figure 3, but the other two constants cannot be
(4) The SMA spring displayed a very slow cooling rate obtained because the stroke of the SMA spring is limited.
under the natural convection. Therefore, to enhance the Thus, when the shape recovery force is predicted in the
response speed, a CPU (central processing unit) cooling following section, two constants will be properly assumed.
fan was used for ventilation.

4.2. Shape recovery force

4. Evaluation of SMA characteristics Figure 4 shows the result of the restrained recovery of an
SMA spring at various initial spring deflections. Restraining
4.1. Isothermal force–displacement relation
a deformed SMA spring while heating the spring above its
Figure 3 shows the complete loading and unloading process transition temperature will generate a large recovery force,
of the SMA spring under various isothermal conditions. as shown in figure 4. This unique behavior can offer active
From Liang and Rogers’ model [12], the stress range in which movement to conventional catheters, which require motorless
the martensite to austenite transformation may occur, can be motion. The important feature, as an actuator, is that the
expressed as follows: greater the initial deformation of the SMA spring the larger
the recovery force will be. These characteristics are typically
CA (T − Af )
σeq
CA (T − As ) (1) shown in figure 4.

819
Hyo Jik Lee and Jung Ju Lee

Figure 5. Comparison of experimental data and calculated results

of the SMA spring by conventional constitutive modeling.

Based on the multidimensional constitutive relation, the Figure 6. The architecture of an MLP for the prediction of the
shape recovery force.
one-dimensional shear stress and strain relation of an SMA
can be expressed as [12]
 
dτ = G dγ + √ dξ + √ dT (4)
3 3
where G is the elastic shear modulus,  is the phase
transformation tensor (which can be calculated from −DεL
where D is the elastic modulus), εL is the maximum
recoverable strain of an SMA, and  is the thermoelastic
tensor, which is related to the thermal expansion of SMA.
T is the temperature and ξ is the internal variable, which
describes the degree of the martensitic transformation. This
parameter, referred to as the martensitic fraction, is defined
as the ratio of martensite to the total volume of the SMA. In
order to predict the experimental results using equation (4),
the 12 following material constants are required:

DA DM CA CM Figure 7. Comparison of the experimental data and the predicted

results of the SMA spring by the neural network.
Mf Ms As Af
Unlike the conventional modeling methods, which use
ν εL  ξ0
mathematical expressions to approximate the experimentally
where DA and DM are Young’s moduli in austenite phase observed behavior of materials, neural networks offer a
and martensite phase, respectively; ν is Poisson’s ratio, εL fundamentally different approach in the modeling of the
is the maximum recoverable strain, and ξ0 is the initial constitutive behavior of materials. They have unique
martensitic fraction. Calculated results using equation (4), learning capabilities, which can be used in learning complex
with the assumptions of some material properties, do not nonlinear relationships. The learning capabilities allow
show good agreement with the experimental data, as shown neural networks to be directly trained with the results of
in figure 5. This conventional constitutive modeling has experiments. They learn the constitutive material behavior
the following disadvantages. It is not practical because the presented in the experimental data and then store this
constitutive modeling requires many experimental data to knowledge in their connection weights. If the training data
determine the 12 material constants. To obtain the shape contains the relevant information on material behavior, then
recovery force, equation (4) should be solved by iteration the trained neural network can generalize novel cases as well
because of nonlinearity of equation (4). Thus, it is a time as its training data. We used the multilayer perceptron (MLP)
consuming method. with the error back-propagation algorithm [21]. The neural
Such disadvantages can be improved by a new method network methodology was first applied to the SMA modeling
of constitutive modeling using a neural network, which in this work in order to predict the shape recovery force.
was originally proposed by Ghaboussi et al [16]. This The architecture of MLP for the prediction of the shape
methodology was then applied to constitutive modeling of recovery force is presented in figure 6. The input layer was
concrete [17], sand [18, 19], and composite materials [20]. made up of two sensory units, which are the initial deflection

820
 Characteristics of a shape memory alloy spring actuator

and current temperature of the SMA spring. One output

neuron, which produces the shape recovery force, constituted
the output layer of the neural network. Two hidden layers
were made up of four neurons each. Among the nine
experimental data sets with different initial deflections from
each other, five experimental data sets were used in training
the neural network and the remainder, four experimental data
sets, were compared with the predicted curves that resulted
from the trained neural network. The training data are
51 points, which are randomly selected, in five experimental
data sets. Figure 7 shows that the trained neural network
predicts the shape recovery force very well.
The developed neural network has two advantages
compared with the conventional constitutive equations.
Firstly, the prediction of the shape recovery force is
possible immediately through the real-time monitoring of the
(a)
temperature of the SMA because the neural network method
does not require a history of materials. However, if the
conventional constitutive equations are used, it takes time
to obtain the result because nonlinear materials such as SMA
are path dependent. Secondly, the developed neural network
method is so simple, since it requires only two input data.
However, the conventional constitutive equation based on
Liang and Rogers’ theory requires several input data, which
may be impossible to obtain by conventional experiments.
The application of the neural network is able to predict
the almost exact shape recovery force with only two input
variables. These two input parameters are the initial
deflection and temperature of the SMA spring actuator. The
initial deflection is the most important design factor in the
design of an active catheter and the temperature may be
a feedback variable in the control of the SMA actuator.
Thus, the developed neural network method can be applied to
(b)
designing an active catheter and explaining the mathematical
model of the SMA actuator in a controller design. Figure 8. Temperature follow-up test with respect to (a) a 0.1 Hz
square wave and (b) a 0.1 Hz sine wave in the use of proportional
control.
4.3. Temperature follow-up test
The temperature follow-up test is necessary in the case of cooling are very important issues. In the almost robotic
temperature feedback control. As described in the previous SMA actuator, electric heating in which electric currents
section, if the temperature and the initial deflection are known flow through an SMA wire was used. The most important
then the shape recovery force can be obtained by the neural problem in electrical heating, however, is the cooling method,
network technology. The relation between the recovery force because the response of the actuator mainly depends on the
and the bending angle of the active catheter can also be cooling rate. In figure 8, the forced convection by the CPU
obtained if the actuating mechanism is determined. Thus cooling fan significantly improves the temperature follow-up
the temperature follow-up test is valuable when we select the characteristics. In the application of a micro SMA coil spring
temperature, from several feedback variables (for example, in the actuator of an active catheter, new cooling methods
temperature, force, displacement and resistance), for the should be required to minimize the size of catheter.
control of the active catheter. In the temperature follow-up
tests and load follow-up tests, two control methods are used. 4.4. Load follow-up test
One is the on–off control method in which a constant current
flows through the SMA by on–off switching. The other is When an active catheter moves in the human cavity, the
the proportional control method in which the amplitude of force control or the position control of the active catheter
current is controlled by the difference between a reference is required. The SME of the SMA spring actuator may
input and a sensor’s output. disappear because an unpredictable excessive load can be
Figure 8 presents the temperature follow-up character- imposed during the insertion of the catheter by position
istics of an SMA spring actuator about the input of a square control. Thus, load control is preferred to activate the SMA
wave of 0.1 Hz and a sine wave of 0.1 Hz when proportional in the working range of the shape recovery force.
current control is used. The load follow-up test is necessary when the force
Since the heating and cooling of the SMA must be feedback control is performed in active catheterization. This
repeated to operate the actuator, the methods of heating and control method is possible under the condition that micro

821
Hyo Jik Lee and Jung Ju Lee

(a) (a)

(b) (b)
Figure 9. Load follow-up test with respect to (a) a 0.1 Hz square Figure 10. Load follow-up test with respect to (a) a 0.1 Hz square
wave and (b) a 0.1 Hz sine wave in the use of on–off control. wave and (b) a 0.1 Hz sine wave in the use of proportional control.

force sensors in the active catheter are able to measure the the combination of SMA modeling and neural network
force of SMA actuators. As in the temperature follow-up technology was first proposed. The predicted results by
test, since the relevance between the recovery force and the the trained neural network based on the experiments showed
bending angle of the active catheter can be determined by the good agreement with the experimental results. In terms of
actuating mechanisms to be selected, the load follow-up test the shape recovery force, this technique requires only two
is valuable. input parameters (initial deflection and temperature), which
Figures 9 and 10 show the experimental results of the can be important design parameters of the active catheter.
load control system shown in figure 2. This experiment The load- and temperature-following experiments were
was also carried out under the input of a square wave of conducted in order to evaluate the feedback control
0.1 Hz and a sine wave of 0.1 Hz. Proportional current characteristic when the feedback variables were assumed
control shows a better load follow-up response and stability to be temperature and force in the active catheter. The
than on–off current control. In terms of the input signal experimental results showed that cooling and control methods
follow-up response, the following characteristic for loads were able to enhance the following characteristic of the SMA
is better than for temperatures. The important aspect here spring actuator. It also informed us that the force feedback
is that the temperature sensor is difficult to attach to the control is better than the temperature.
SMA wire with a small diameter and cannot obtain as stable
a signal as the force sensor. Therefore, force feedback
Acknowledgment
control is recommended over temperature feedback control
for controlling an active catheter.
This study was supported by HWRS-ERC project of KOSEF
(Korea Science and Engineering Foundation). We would like
5. Conclusions to thank KOSEF for this support.

The evaluation of the characteristics of an SMA spring

actuator was performed in this work. The fabricated SMA References
spring actuator shows a difference between the experimental [1] Santa A D, Mazzoldi A and Rossi D D 1996 Steerable
results and the predicted results from the constitutive model microcatheters actuated by embedded conducting polymer
by Liang and Rogers [12]. Therefore, in this work, structures J. Intell. Mater. Syst. Struct. 7 292–300

822
 Characteristics of a shape memory alloy spring actuator

[2] Bien Z 2000 Development of state-of-the-art medical robot [11] Reynaerts D and Brussel H V 1998 Design aspects of shape
systems HWRS-ERC Newsletter 1 20–1 memory actuators Mechatronics 8 635–56
[3] Haga Y, Tanahashi Y and Esashi M 1998 Small diameter [12] Liang C and Rogers C A 1997 Design of shape memory
active catheter using shape memory alloy Proc. 11th Ann. alloy springs with applications in vibration control
Int. Workshop on MEMS pp 419–24 J. Intell. Mater. Syst. Struct. 8 314–22
[4] Arai F, Ito M, Fukuda T, Negoro M and Naito T 1994 [13] Kuribayashi K 1991 Improvement of the response of an
Intelligent assistance in operation of active catheter for SMA actuator using a temperature sensor Int. J. Robotics
minimum invasive surgery IEEE Int. Workshop on Robot Res. 10 13–20
and Human Communication pp 192–7 [14] Brinson L C and Huang M S 1996 Simplifications and
[5] Fukuda T, Guo S, Kosuge K, Arai F, Negoro M and comparisons of shape memory alloy consitutive models
Nakabayashi K 1994 Micro active catheter system with J. Intell. Mater. Syst. Struct. 7 108–18
multi degrees of freedom IEEE Int. Conf. on Robotics and [15] Shape Memory Applications, Inc. 1999 Making shape
Automation pp 2290–5 memory springs in NiTi smart sheets, webpage
[6] Guo S, Fukuda T, Nakamura T, Arai F, Oguro K and http://www.sma-inc.com
Negoro M 1996 Micro active guide wire catheter [16] Ghaboussi J, Garrett J H Jr and Wu X 1991 Knowledge-based
system—characteristic evaluation, electrical model and modeling of material behavior with neural networks
operability evaluation of micro active catheter Proc. 1996 J. Eng. Mechan. Division, ASCE 117 132–53
IEEE Int. Conf. on Robotics and Automation [17] Wu X and Ghaboussi J 1993 Modeling unloading
(Minneapolis, MN, April 1996) pp 2226–31 mechanism and cyclic behavior of concrete with adaptive
[7] Maeda S, Abe K, Yamamoto K, Tohyama O and Ita H 1996 neural networks Proc. 2nd Asian–Pacific Conf. on
Active endoscope with SMA (shape memory alloy) coil Computational Mechanics (Sydney, Australia, 1993)
springs Proc. 9th Ann. Int. Workshop on MEMS [18] Ghaboussi J and Sidarta D E 1998 New nested neural
pp 290–5 networks (NANN) for constitutive modeling Comput.
[8] Lim G, Park K, Sugihara M, Minami K and Esashi M 1996 Geotechnics 22 29–52
Future of active catheters Sensors Actuators A 56 [19] Sidarta D E and Ghaboussi J 1998 Constitutive modeling of
113–21 geomaterials from non-uniform material tests Comput.
[9] Peirs J, Reynaerts D and Brussel H V 1998 Design of a shape Geotechnics 22 53–71
memory actuated endoscopic tip Sensors Actuators A 70 [20] Ghaboussi J, Pecknold D A, Zhang M and Haj-ali R M 1998
135–40 Autoprogressive training of neural network constitutive
[10] Hunter I W and Lafontaine S 1992 A comparison of muscle models Int. J. Numer. Meth. Eng. 42 105–26
with artificial actuators Techn. Digest IEEE 5th [21] Haykin S 1999 Neural Networks 2nd edn (Englewood Cliffs,
Solid-State Sensor and Actuator Workshop pp 178–85 NJ: Prentice-Hall)

823