Anda di halaman 1dari 45

The Pennsylvania State University

The Graduate School


College of Engineering

MODELING MAGNETO-ACTIVE ELASTOMERS USING


THE FINITE ELEMENT METHOD

A Thesis in
Mechanical Engineering
by
Robert M. Sheridan

2014 Robert M. Sheridan

Submitted in Partial Fulfillment


of the Requirements
for the Degree of

Master of Science

August 2014

The thesis of Robert M. Sheridan was reviewed and approved* by the following:

Paris von Lockette


Professor of Mechanical Engineering
Thesis Adviser

Mary Frecker
Professor of Mechanical Engineering

Karen Thole
Department Head of Mechanical and Nuclear Engineering

*Signatures are on file in the Graduate School.

ii

ABSTRACT
Magneto-active elastomers (MAE) are a new type of smart materials that consist of hardmagnetic particles such as barium ferrite in an elastomer matrix. Under the application of a
uniform magnetic field, the MAE material undergoes large deformation as the material bends
due to magnetic torques acting on the distribution of hard-magnetic particles. This behavior
demonstrates the potential of MAEs to act as remotely-powered actuators. MAEs are different
from magnetorheological elastomers (MRE) which use soft-magnetic iron particles in place of
the hard-magnetic particles. Traditional MREs, formed with carbonyl iron, are primarily driven
by magnetic interactions between particles that cause magnetostriction and associated
phenomena.
In this work, MAEs were fabricated using 30% v/v 325 mesh M-type barium ferrite (BaM)
particles in Dow Corning HS II silicone elastomers. Prior to curing, the samples were placed in
a uniform ( 2 Tesla) magnetic field to align the magnetic particles and produce a magnetization
oriented in the direction of the applied magnetic field. For this work, the magnetic particles were
physically oriented either in-plane or through-plane. The geometries studied in this work
consisted of a cantilever beam MAE specimen (where experimental data is collected in [1]), a
symmetric two-segment and an asymmetric two-segment accordion structure, and a foursegment accordion structure. In the four-segment and both two-segment geometries, MAE
patches were bonded to a magnetically inert polydimethylsiloxane (PDMS) substrate. The ability
of these structures to exert load over a range of prescribed displacements highlights the ability of
the MAE actuators to perform useful work. The structures studied in this work deformed in
either a bending or folding mode. The data collected from the experiments included tip force vs.
displacement vs. magnetic field for the cantilever geometry, axial force and bend angle vs.
magnetic field for the two-segment geometries, and average bend angle vs. magnetic field for the
four-segment geometry. Results show increases in the measured tip force, axial force, and bend
angle versus the applied field for the aforementioned geometries.
The experimental results were then compared to results from a finite element analysis (FEA).
The FEA methodology, performed in the commercial FEA software package Comsol, employs
the Maxwell stress tensor applied as a traction boundary condition at the interface between
magnetic and non-magnetic domains. Results show fair to excellent agreement for all structures
studied. Discrepancies between the experimental results and the simulations may be attributed to
the inability to capture the actual distribution of the magnetizations of the particles, which are
assumed perfectly aligned, in the finite element model; the assumption that the Maxwell surface
stress accurately models the particle-field and particle-particle interactions of the MAE; or the
failutre to capture any non-uniformities in the applied magnetic field. Future work will focus on
using the platform developed in this work to optimize structure design with regard to the
magnitude and direction of the MAE's magnetization, size of the active material, and placement
of the active material.

iii

TABLE OF CONTENTS
List of Tables
List of Figures
Acknowledgements

v
vi
vii

1. INTRODUCTION AND BACKGROUND


1.1 Problem Statement
1.2 Origami Engineering
1.3 Active Materials
1.4 Magneto-Active Elastomers
1.5 Project Goals

1
1
1
2
3
6

2. METHODOLOGY
2.1 Material Fabrication
2.2 Experiments
2.3 Finite Element Modeling
2.4 The Finite Element Method
2.5 Mesh Convergence Study

7
7
9
12
14
16

3. RESULTS AND DISCUSSION


3.1 Cantilever Beam
3.2 Two-Segment Accordion
3.3 Asymmetric Two-Segment Accordion
3.4 Four-Segment Accordion

20
20
21
22
24

4. CONCLUSIONS

26

Bibliography

28

Appendix

31

iv

LIST OF FIGURES
Figure 1.1: The decomposition of the origami crane into its two-dimensional crease
Figure 1.2: Schematic detailing the stages of deformation for a shape memory polymer
Figure 1.3: (a) Depicts a soft-magnetic particle subjected to a uniform magnetic field
Figure 1.4: The two orientations of the magnetic particles within the elastomer matrix
Figure 1.5: Schematic which details the various particle alignments for MRE and MAE
Figure 1.5: Schematic detailing the perfect alignment of the magnetic particles (left)
Figure 2.1: Representative image which depicts a typical acrylic mold used
Figure 2.2: Illustrates the dimension variables, orientation of magnetization
Figure 2.3: (A) schematic of undeformed shape (B) schematic of measurements
Figure 2.4: (A) schematic of undeformed shape (B) schematic of measurements
Figure 2.5 (A): Two-segment accordion structure consisting of MAE patches
Figure 2.6: (A) schematic of undeformed shape (B) schematic of measurements
Figure 2.7: An illustration of the boundary condition satisfied at the boundary
Figure 2.8: Finite element meshes corresponding to the (A) four-segment accordion
Figure 2.9: A mesh convergence study for the cantilever beam structure where
Figure 2.10: A mesh convergence study for the two-segment structure
Figure 2.11: A mesh convergence study for the four-segment structure where
Figure 3.1: Results of simulations of the (A) remanent magnetization in red arrows
Figure 3.2: Total force data for the cantilever beam specimen. The experimental
Figure 3.3: Simulation results of the (A) remanent magnetizations of the MAE
Figure 3.4: Measured bend angle (left) and resulting axial force (right) as a function
Figure 3.5: Experimental (symbols) vs. FEA simulation data (lines) for the total
Figure 3.6: Experimental (symbols) vs. FEA simulation data (lines) for the elastic
Figure 3.7: Experimental (symbols) vs. FEA simulation data (lines) for the magnetic
Figure 3.8: Experimental (symbols) vs. FEA simulation data (lines) for the magnetic
Figure 3.9: Depiction of the (A) remanent magnetizations of the MAE segments
Figure 3.10: Experimental (symbols) vs. FEA (solid-line symbols) and MES
Figure 4.1: Water bomb origami structure which uses MAE patches integrated
Figure A.1: Details of the parametric study conducted on the cantilever beam MAE
Figure A.2: Details of the parametric study conducted on the two-segment accordion
Figure A.3: Stress-strain curve for a PDMS specimen that is subjected
Figure A.4: Details of the parametric study conducted on the symmetric
Figure A.5: Details of the parametric study conducted on the four-segment
Figure A.6: Details of the parametric study conducted on the four-segment
Figure A.7: A macroscopic illustration of the cantilever beam experiment
Figure A.8: A macroscopic illustration of the two-segment accordion
Figure A.9: A macroscopic illustration of the four-segment accordion

1
3
4
4
4
5
7
8
9
9
11
11
13
17
18
18
19
20
20
21
22
22
23
23
24
24
25
27
31
34
34
35
35
36
36
37
37

LIST OF TABLES
Table 2.1: Dimensions for the cantilever beam, two-segment accordion, and four
Table 2.2: Effective material properties for the cantilever beam, two-segment
Table A.1: Details of the parametric study conducted on the cantilever beam MAE

8
14
31

vi

NOMENCLATURE
Ahi
BH
DE
DOF
F
FEA
FEM
H
M
MAE
MES
MRE
Mrem
MSS
n
SMA
SMP
T
u,v,w

1
2
H1
H2

Maximum magnetic vector potential


Magnetic Induction
Dielectric elastomer
Degree of Freedom
Measured Force in Newtons.
Finite element analysis
Finite element model/method
Applied magnetic field vector
Magnetization
Magneto-active elastomer
Minimum Energy Structure
Magneto-rheological elastomer
Remanent Magnetization
Maxwell Surface Stress
Normal component
Shape memory alloy
Shape memory polymer
Magnetic torque
x, y, and z displacements respectively
Elastic stress variable
Internal stress component of MSS
External stress component of MSS
Internal magnetic stress component of MSS
External magnetic stress component of MSS

vii

ACKNOWLEDGEMENTS
I would like to gratefully acknowledge the support of the National Science Foundation EFRI
grant number 1240459 and the Air Force Office of Scientific Research. Any opinions, findings,
and conclusions or recommendations expressed in this material are those of the authors and do
not necessarily reflect the views of the National Science Foundation.
I would also like to thank the support of my adviser Dr. Paris von Lockette for his help and
expertise given during the research process which is summarized in this work. He has helped in
both my academic and professional development for which I am grateful. I would also like to
thank Dr. Mary Frecker for her guidance, not only in writing this work, but also during the
research process. Dr. Zoubeida Ounaies was also a great mentor and helped in my research
process in many ways.
Carrie Tedesco was a student working on the experimental side of the project and I would
like to thank her for her perseverance and willingness to help out with the project in any way she
could. Additionally, I would like to thank Adrienne Crivaro for her help with COMSOL and
getting acclimated to the Penn State environment. I would also like to thank Landen Bowen and
Joe Calagero for their help in the areas of mathematics, kinematics, and the finite element
method. Finally, I would like to thank all of the remaining faculty and students on the Origami
EFRI project for their help, guidance, and companionship.

viii

1. INTRODUCTION AND BACKGROUND


1.1 Problem Statement
In the field of origami engineering, it is of special interest to develop materials that are able
to self-fold given an applied field. These smart materials can be actuated fields such as
magnetic, thermal, or electric. In this work, magneto-active elastomers (MAE) are studied as the
primary active material; this material is actuated using a magnetic field. Magneto-active
elastomers are composites that consist of silicone rubber and physically-aligned barium ferrite
magnetic particles. In a magnetic field, the MAE is able to actuate through a distributed torque
that acts on each individual particle; the summation of these torques creates the macroscopic
response of the MAE. In order to develop origami structures that can actuate using MAE as an
active material, it is ideal to be able to model the behavior of the MAE given arbitrary boundary
conditions, geometries, loading conditions (i.e. magnetic field), and effective material properties.
This work seeks to develop and validate a finite element model (FEM) that is able to predict the
behavior of MAE composites given arbitrary conditions. The knowledge gained from this
research will then be used to optimize parameters such as the required magnetization, structure
geometry, and the magnetic orientation of the particles.
1.2 Origami and Origami Engineering
Origami is the Japanese art of creating various shaped objects by folding paper [2]. These
folds can be either mountain (pointing up) or valley (pointing down) folds. The complexity of
origami structures has increased due to mathematical techniques which allow the structures to be
modeled and the design parameters can be
characterized [3,4]. For example, a single sheet
of paper can be partitioned with a series of
mountain and valley fold-lines to create a
predetermined three-dimensional structure such
Figure 1.1: The decomposition of the origami crane into
as the crane. The flapping wing crane is an its two-dimensional crease pattern [13].
example of an action origami structure that is
able to actuate given external stimuli (in traditional origami, this refers to a force exerted using
your fingers). Action origami varies from traditional origami as it uses external stimuli to
produce dynamic movements within the structure whereas traditional origami typically generates
static structures [6]. An example of a crane deconstructed from its three-dimensional form into
its planar crease pattern can be seen in Figure 1.1. It is crucial to recognize the importance of not
only the crease pattern to develop these three-dimensional objects, but also the order by which
the folds are applied. It is of interest to use active materials, i.e. materials that are able to actuate
given external stimuli, as they can be used to induce the desired folding process. While active
materials may be able to actuate a structure to its folded position, other techniques are required to
actuate the folds in a pre-determined sequential manner; it is also evident that the placement of
the active materials with respect to the creases will also be relevant to this problem. Thus,
origami engineering can be employed to combine these various fields.
Origami engineering emerged when origami was influenced by areas such as engineering and
kinematics. This new field combines the knowledge of origami structures and their folding
behavior with the work-potential of active materials such as dielectrics [5], shape memory alloys
1

[6], and magneto-active elastomers (MAEs) [1,7]. Combining these two fields of research
allows researchers to actuate origami structures (such as the crane) through external stimuli such
as thermal, electric, or magnetic fields. Engineers and designers alike seek to understand the
concepts behind origami engineering as it can be used in deployable devices making it suitable
for space applications [9]. Furthermore, it is of interest to utilize engineered materials such as
silicone and polydimethylsiloxane as opposed to paper. While paper is traditionally used in
origami, engineering materials give researchers the ability to program the material's properties
and incorporate active materials directly into the substrate.
When using active-materials, origami engineering is used to develop and optimize material
systems which can fold and, in some cases, unfold from the aforementioned external stimuli [1012]. The systems are optimized to reduce the amount of required active materials in relation to
the structure they are actuating. The systems also optimize the structure by determining the
position of the active material on the structure which provides the highest efficiency for the
structure; this efficiency can be quantified as the relationship between the input, which in this
case is the external stimuli such as a magnetic field, and the output, or the actuation of the active
material. In order to develop and optimize an active origami structure with an active material
component, it is important to be capable of modeling the behavior of the active material. The
ability to model the material in simple systems that contain the basic mountain-valley building
blocks will allow researchers to form a basis for which they can develop and optimize more
complicated systems. In this work, magneto-active elastomers are studied and the prediction of
their performance requires the numerical coupling of elastic behavior and electromagnetic
phenomena.
1.3 Active Materials
In the field of origami engineering, active materials are used to actuate the structure in such a
manner that it is able to self-fold. These materials consist of, but are not limited to, shape
memory alloys (SMA), shape memory polymers, dielectric elastomers, terpolymers, photothermal polymers, and photo-chemical polymers and respond to either electric, magnetic, or
thermal fields [13-19].
Shape memory alloys are a type of smart material with the ability to undergo large
deformation. The alloys can include metals such as nickel, titanium, and iron. After unloading
the material, or upon heating the material, it is able to return to its original shape [13]. The SMA
is able to return to its initial configuration due to a reversible phase transformation that the
material undergoes. This martensitic transformation, as it is known, can be produced as a result
of applied stresses or changes in the temperature [13]. Similar to shape memory alloys, shape
memory polymers are also able to recover from a large deformed state due to an environmental
stimulus such as temperature, light, or humidity [14]. While the mode of deformation varies
between the different stimuli used to influence the polymers, the process between all modes is
similar. As an example, a shape memory polymer system studied in [14] provides an illustrative
example. The material begins in its original shape and is heated above its transition temperature
(Tg) and subjected to an applied force to enter a temporary deformed state. Next the material is
cooled below its Tg to enter an intermediate shape between the deformed and undeformed shape.
Finally, the material is heated above the Tg without an applied force to return to its original shape
[14]. This process is shown in Figure 1.2.
2

Figure 1.2: Schematic detailing the stages of deformation for a shape memory polymer.

Dielectric elastomers (DE) are another type of smart material that actuate by converting
electrical energy into mechanical energy. They achieve this actuation through a Maxwell stress.
This Maxwell stress is a result of two charged parallel compliant electrodes clamped around a
soft elastomer film. When a voltage is applied to the electrodes, opposing charges form at both
electrodes which results in electrostatic forces; these forces compress the elastomer through its
thickness and allow the DE to expand in the planar direction [15]. When used in reverse, the DE
can also act as a generator by applying mechanical work against an electric field to generate
electrical energy [15].
Terpolymers are another branch of smart materials that display electromechanical behavior.
While dielectric elastomers use electrostatic forces to compress the elastomer film, terpolymers
use electrostrictive forces to deform [16]. These polymers are semi-crystalline and have a higher
level of conversion from electrical to mechanical energy when compared to dielectric elastomers
and other electromechanical materials. However, these materials require high electric fields to
undergo deformation [16].
The final material that will be discussed in the context of active materials is a magneto-active
elastomer. Magneto-active elastomers are a form of active material that couple the elastic
properties of an elastomer compound and magnetic properties of a distribution of magnetic
particles to produce a composite that is able to actuate given an applied magnetic field.
Magneto-active elastomers are studied over the other mentioned active materials as they exhibit
large deformation, produce a sizeable distributed torque, may be actuated remotely, and are bidirectional. The aforementioned active materials do not exhibit all of these behaviors; a
comparison of these materials is detailed in [21]. The discussion of the MAE material is
described in detail in Section 1.4.
1.4 Magneto-Active Elastomers
Before discussing MAEs, it is relevant to briefly discuss magnetorheological elastomers
(MREs). Magnetorheological elastomers are a class of smart materials that preceded MAEs.
Magnetorheological elastomers combine soft magnetic particles with a nonmagnetic elastic
medium [18]. The soft magnetic particles are typically iron or nickel and typically have particle
sizes around 1-2 m [18, 19]. These particles can either be randomly distributed or aligned in
the elastomer matrix. The combination of these materials allows the composite to deform under a
uniform magnetic field through a magnetostrictive effect [20, 22, 22, 25]. The phenomenon
(magnetostrictive effect) occurs due to dipole-dipole interactions between the soft magnetic
particles. When chains of magnetic particles are sheared within an applied magnetic field this
generates a local demagnetizing field that in turn generates a restoring couple that seeks to
minimize the energy of the system. As a result, MRE structures typically seeks to returns to their
unperturbed state which results in an increase in the shear modulus of the MRE [1] [28].

Figure 1.3: (a) Depicts a soft-magnetic particle subjected to a uniform magnetic field, H (blue arrows). The magnetization,
M (black arrow), has no preferred orientation prior but aligns with an applied magnetic field; this produces no net torque on
the particle. (b) Depicts a hard-magnetic particle subjected to a uniform magnetic field. The particle has a remanent
magnetization Mrem (black arrow) which is not aligned with the applied magnetic field; however, the magnetization, M,
remains nominally aligned with Mrem which results in a magnetic torque, T [1].

Magneto-active elastomers, the focus of this work, are a relatively new smart material in
which ferromagnetic particles, which are hard magnetic particles, are embedded in an elastomer
matrix prior to the cross-linking process. With the
addition of the magnetic particles (barium hexaferrite),
the otherwise inactive rubber is able to exhibit magnetosensitive behaviors caused by inter-particle and particle- Figure 1.4: The two orientations of the
field interactions that can produce actuation such as magnetic particles within the elastomer matrix.
The left MAE patch shows the magnetic
magnetostriction [20], bending[1, 11], and shearing particles oriented in-plane while the left MAE
[24]. In the experiments detailed in Chapter 2, and patch shows the magnetic particles oriented
found in the literature [1], the magnetic particles are through-plane.
aligned during the fabrication of the MAE to physically
orient to the particles magnetizations in the direction of the applied field (~ 2 Tesla). The result
is an anisotropic material where the particle-field behavior is dependent on the direction and
magnitude of the applied magnetic field in
relation to the orientation of the magnetic
particles.
When a magnetic field is applied, a
magnetic torque is generated at the particle
level as T = m x H where H is the applied
magnetic field, m is the magnetization of
the MAE, and T is the magnetic torque
generated as a result of the applied field
(detailed in Figure 1.3). Each particle
experiences a torque and the summation of
the torques, across the distribution of
particles, in the MAE produces a
macroscopic response in the material (i.e. Figure 1.5: Schematic which details the various particle
the structure actuates) [7].
The alignments for MRE and MAE materials [1]. In [1], specimens
magnetization of the MAEs in this work can of each particle alignment are characterized and compared.
4

be oriented either in-plane or through-plane as shown in Figure 1.4. Various schematics of the
classes of MRE and MAE alignments can be seen in Figure 1.5. The alignment of the magnetic
particle's dipoles is created by applying a magnetic field during the curing of the silicone
elastomer. This process allows a majority of the
magnetic particles to align their dipoles with the applied
magnetic field. Though the alignments are assumed to
be perfectly aligned in this work, in most cases they
may have a conical distribution about the primary
magnetization axis as shown in Figure 1.6. This may be
due to the inherent nature of the MAE system to return Figure 1.5: Schematic detailing the perfect
to an equilibrium state after the magnetic field is alignment of the magnetic particles (left) and a
representative conical distribution of the
removed from the curing specimen.
magnetic particles (right).
In recent works, finite element models were used to
study the behavior of both MREs and MAEs. In [25], MAEs with periodic and random
distributions of circular and elliptical fibers were modeled in a FEM. The fibers were hard
magnetic in nature. The work sought to determine effective properties of the material which is
similar to the goal of this work. The work also was similar as it used the Maxwell stress to
couple the elastic and magnetic behaviors of the MAE in the model. Their work studied the roles
of deformation, particle concentration, particle shape, and particle distribution on the
magnetostrictive phenomenon, actuation stress, and magnetically-induced change in stiffness.
Key outcomes of the work include the ability to model the magneto-elastic behavior on a microscale through inclusion of the magnetically induced strains detailed in [26]. Numerical data was
not compared to experimental data; however, variations in the normalized mechanical stress
concentration, normalized magneto-elastic tangent Young's modulus, normalized effective
susceptibility, and normalized magneto-elastic coupling coefficient were shown for rectangular
and hexagonal finite elements. Data showed large variations in data collected between the two
elements used for the numerical analysis.
In [27] and [28], a MRE finite element model was developed for a thin flat plate fixed about
its rim. The MRE was magnetized by a uniform magnetic field Ho normal to the membranes
surface. The model used a boundary condition that included a traction formed from the Cauchy
stress tensor equated to an energy product of the magnetization vector to couple elastic and
magnetic behavior. In equilibrium, the magnetic and elastic pressures, described in the
aforementioned boundary condition, were balanced. In the presence of a uniform magnetic field,
the MRE achieved equilibrium through magnetostriction; magnetostriction is a result of dipoledipole interactions which induce strains in the soft magnetic elastomer composite. In these
works, the thin flat plate, positioned vertically, bulged in the center in the presence of a
horizontal uniform magnetic field. The finite element model and experimental data measured the
displacement of the MRE at the center of the plate as a function of the applied magnetic field.
The results of their modeling efforts showed excellent agreement with the experimental data.
The data showed that for increases in the applied magnetic field, the displacement of the MRE at
the center of the thin plate also increases quickly but began to asymptote around field strengths
of 100 kA/m; however, this mode of deformation only began to occur at a finite field strength
(when H 50 kA/m).

1.5 Project Goals


This work seeks to accurately model MAEs using the finite-element method. The
information attained from this work will then be used to optimize the designs of origami
structures. This overall goal can be accomplished by addressing the following matters:
Determine effective material properties that model the experimental behavior of the MAE
accurately.
Develop a finite element model that uses the magnetic and elastic material properties and
couples the underlying physics of the elastic and magnetic behavior.
Determine sufficient mesh density, which inherently relates to the Degrees of Freedom
within the FEM, along with a sufficient bounding air-box size (used to model the external
magnetic fields).
Model and experimentally validate arbitrary geometries to further test the validity of the
modeling methods.
With the methods developed in this work, computational modeling of MAEs can be used in the
future to optimize MAE material placement, the magnitude and orientation of the MAE
specimens magnetization, and the specimens size on folding origami structures across a wide
design space more efficiently than trial and error fabrication.

2. METHODOLOGY
2.1 Material Fabrication
The MAE material used in this work consisted of DOW HS II RTV silicone rubber mixed at
20:1 base to catalyst ratio. Prior to the curing of the silicone rubber, barium hexaferrite (325
mesh) was mixed into the silicone rubber at a 30% v/v. Acrylic molds were used to cast the
specimens into sheets. A representative mold is detailed in Figure 2.1. The closed molds were
rectangular with prescribed dimensions to create specimens detailed in Figure 2.2 with their
respective dimensions described in Table 2.1. MAEs were aligned by subjecting the specimens
to a ~2 Tesla magnetic field, H, in the plane of the sheet which ideally orients the particle
magnetizations.

Figure 2.1: Representative image which depicts a typical acrylic mold used to cast MAE and polydimethylsiloxane specimens.

In some cases, an unfilled polydimethylsiloxane (PDMS) compound was used as passive


substrate for the MAE specimens. A similar casting procedure was used with the PDMS
materials that were used in this work. The PDMS substrate was created using Gelest OE 41 a
two-part compound mixed 1:1 by weight. The compound was also cast in rectangular molds of
desired thickness and allowed to cure producing sheets from which desired cross sections were
either used as cast or cut to the desired shape. In cases where a PDMS substrate was used, the
MAEs were bonded to the PDMS substrate, with the magnetizations oriented in opposing
directions, using Loctites plastic bonding system two-part adhesive. This adhesive is designed
for elastomer on elastomer bonding. The final composite structures analyzed in this work, along
with dimensions, can be seen in Figure 2.2. This figure includes schematics along with images
of the specimens. They include a cantilever beam, a two-segment accordion structure, an
asymmetric accordion structure, and a four-segment accordion structure.
The cantilever specimen is so-named due to its deformed shape; the beam is subjected to a
fixed boundary condition on one boundary and is free on all other boundaries. When a magnetic
field is applied perpendicular to the primary axis of the structure, the structure bends (much like
a cantilever beam with a point load on the free-end of the beam). The aforementioned accordion
structures are so-named due to their mode of deformation; upon applying a magnetic field
perpendicular to the primary axis of each accordion (and also of the magnetization of the MAE
specimens), the PDMS deforms into a mountain or valley fold (two-segment accordion) or a
series of mountain and valley folds (four-segment accordion). The direction of the magnetic
field and orientation of the magnetic particles net magnetization determine if the fold will be a
mountain or valley fold. This is illustrated in Figure 2.5 where the mode of deformation is
dependent upon the torques generated by the MAE patches.

Table 2.1: Dimensions for the cantilever beam, two-segment accordion, and four-segment accordion (shown in Figure 2.2).
Two-Segment
Asymmetrical
Four-Segment
Cantilever
Accordion,
Two-Segment
Accordion,
Beam
Prime
Accordion, Prime
Double Prime
Parameter
Value (mm)
Value (mm)
Value (mm)
Value (mm)
MAE Thickness (A)
20.0
4.85
4.85
15.0
MAE Height (B)
5.0
3.1
4.0
3.18
MAE Length (C)
50.0
10.0
10.0
20.3
MAE Spacing (D)
25.0
10.0
10.0
9.53
PDMS Thickness (E)
4.85
4.85
15.0
PDMS Height (F)
2.8
1.0
3.18
PDMS Free Length (H)
52.0
25.0
119.0
PDMS Length (G)
10.0
80.0
-

Figure 2.2: Illustrates the dimension variables, orientation of magnetization (green arrows), and physical boundary conditions
for the (A) cantilever geometry, (B) two-segment accordion structure (includes the asymmetric accordion), and (C) the foursegment accordion structure. The hatched area depict a fixed-boundary condition where the x, y, and z displacements (u,v,w)
are set to zero. (a) is an image of an cantilever specimen, (b) is an image of the symmetric two-segment accordion specimen,
(b) is an image of the asymmetric two-segment accordion specimen, and (c) is an image of the four-segment accordion
specimen. Note that image (c) is in an applied magnetic field (no zero-field image is available).

2.2 Experiments
2.2.1 Cantilever Beam
In [1], cantilever bending
behavior of hard-magnetic MAE
specimens was observed.
The
specimen, shown in Figure2.2A, was
fabricated using methods described
in the Section 2.1.
The MAE
specimens had a free length of 50
mm with a cross-sectional area of 20
A
B
C
x 5 mm2. The MAE was fixed at the
base and subjected to a forcedFigure 2.3: (A) schematic of undeformed shape (B) schematic of
measurements on deformed shape, (C) image of representative
displacement ranging from 0 to 8
deformed shape. Note: Figures not drawn to scale.
mm at the free-end of the cantilever
beam in the horizontal direction (at the top left corner shown in Figure 2.3B). The horizontal
displacement was applied using a Shimpo force gauge (Model FGV-0.5XY with an accuracy of
+/- 0.1%) attached to non-ferrous extension arm. Simultaneously, a magnetic field was applied
perpendicular to direction of the magnetization in the materials original configuration. The
strength of the applied magnetic fields ranged from 0 to 0.09T. The resulting transverse force at
the point where the displacement was prescribed was observed using the Shimpo force gauge.
The experimental set-up can be seen in Figure 2.3. Further details of the experiment can be
found in [1]. Results of the experiment are discussed in Chapter 3 (Figure 3.2). These data from
the literature, shown in [1], will be used to help validate the finite element modeling techniques.
A schematic of the experimental set-up is shown in Figure A.7 in the Appendix.
2.2.2 Asymmetric Two-Segment Accordion Structure
The asymmetric two-segment accordion structure, shown in Figure 2.2B was fabricated using
methods discussed in Section 2.1. The experimental data collected in this experiment was
attained from [33] and is summarized in this work. The structure was placed between the polefaces of a C-shape electromagnet such that the magnetic field and magnetization of the MAEs
were perpendicular to each other. This is depicted in Figure 2.4A-B where the direction of the

Figure 2.4: (A) schematic of undeformed shape (B) schematic of measurements (C) image of representative deformed
shape for the asymmetric two-segment accordion structure. Note: Figures not drawn to scale.

magnetic field and magnetization are shown in blue and green arrows, respectively. The
asymmetric two-segment accordion structure was fixed on the upper portion of the PDMS
substrate and attached to a Shimpo force gauge (Model FGV-0.5XY with an accuracy of +/0.1%) on the bottom portion of the PDMS. A view of the experimental set-up can be seen in
Figure 2.4C. The electromagnet consisted of a C-shape ferrous core with wound solenoid to
produce magnetic fields. The current-field relationship of the magnet was recorded using a 475
DSP Gaussmeter.
The experiment recorded the axial force applied by the MAE accordion structure to the
Shimpo force gauge as the magnetic field was increased in increments of ~10 mT. A high
resolution Canon EOS 7D camera (5184x3456 pixels) was used to photograph the deformed
specimen for the application of each magnetic field. This also allowed bend angles to be
measured using AutoCAD software. One sample was tested three times. Results are discussed
in Chapter 3. A macroscopic view of the experimental set-up is shown in Figure A.8 in the
Appendix.
2.2.3 Symmetric Two-Segment Accordion Structure
In a similar experiment, a symmetric two-segment accordion, shown in Figure 2.2B, was
subjected to conditions similar to the conditions applied to the asymmetric two-segment
accordion experiment detailed in Figure 2.4. The experimental data collected in this experiment
was attained from [33]. The aforementioned structure is referred to as the symmetric twosegment accordion for the remainder of this work. However, a longitudinal displacement,
ranging from 0.0 to 6.5 mm, was applied at the bottom boundary of the PDMS substrate to
deform the structure into a buckled configuration. The structure was then subjected to magnetic
fields varying in strength from 0.0 - 0.1T. The resulting force and prescribed displacement at the
lower boundary of the PDMS substrate were observed. The experimental details can be seen in
Figure 2.5.
To characterize the elastic behavior of the PDMS substrate, test specimens were placed in an
Instron Model 4202 #2 tensile loader and conditioned at strain rates of 10%/min and 0.1%/min
up to strains of 90 percent. One specimen was tested with dimensions 60.4 x 6.2 x 1.1 mm3. The
resulting stress-strain curves were then compared to stress-strain curves developed from a
Comsol model that used the Mooney-Rivlin strain-energy formulation. The stress-strain plots
and relevant data can be found in the Appendix. The results are discussed in Chapter 3. A
macroscopic view of the experimental set-up is shown in Figure A.8 in the Appendix.

10

Figure 2.5 (A): Symmetric two-segment accordion structure consisting of MAE patches with remanent magnetization shown
in green arrows and the PDMS substrate with fixed upper and lower boundaries (with the ability to prescribe a displacement
along the longitudinal axis of the PDMS) (B) Symmetric two-segment accordion structure subjected to a magnetic field, H,
in blue arrows that results in a torque on each MAE specimen, shown in red, and an axial force, F, on the lower fixedsegment that creates (C) the representative deformed configuration of the symmetric two-segment accordion when subjected
to a magnetic field and prescribed displacement . Note: Figures not drawn to scale.

2.2.4 Four-Segment Accordion Structure


The four-segment accordion structure was fabricated using methods discussed in the Section
2.1. The geometry of the structure is detailed in Figure 2.2C. The structure was placed between
the poles of the C-shaped core such that the
magnetic field and magnetization of the MAE
specimens in the undeformed configuration were
perpendicular. This is depicted in Figure 2.6 where
the direction of the magnetic field and
magnetization are shown in blue and green arrows,
respectively. The PDMS substrate was fixed at the
top and the structure was subjected to magnetic
fields ranging from 0.0 - 0.2T. The deformed
configurations were photographed using a high
A
B
C
resolution Canon EOS 7D camera (5184x3456
pixels). These images were then analyzed in
AutoCAD to determine the bend angles as a Figure 2.6: (A) schematic of undeformed shape (B)
schematic of measurements (C) image of representative
function of the applied magnetic field. Results are deformed shape for the four-segment accordion structure.
Note: Figures not drawn to scale.
discussed in Chapter 3.

11

2.3 Finite Element Modeling


To model the behavior of the MAE composite structures, a finite element analysis was done
using Comsol Multiphysics software as it has the capability to couple magnetic and elastic
behavior [19]. The finite element approach taken in this work uses an Arbitrary LagrangianEulerian (ALE) formulation with finite deformation kinematics to account for the large
deformation. For the remainder of this work, discussion of the model will be done using a
Lagrangian description. The domains, and their corresponding meshes, for the three finite
element models developed in Comsol can be seen in Figure 2.8.
Consider the quasi-static deformation of a body from its initial configuration (material
configuration) to its deformed configuration. The material configurations material points can be
defined through a position vector X. When a magnetic field is applied to an MAE-actuated
structure, the body deforms; the new position of the previously said material points can now be
described by the position vector x. The local deformation of the material can be characterized by
the deformation gradient F = Grad x such that J = det(F) > 0 where J is the Jacobian. The
Jacobian is a measure of the materials volume change that occurs during deformation. In this
work, the material is assumed to be incompressible which states that J = 1 since the volume
remains unchanged. This is modeled internally in Comsol by modifying the second PiolaKirchoff stress tensor to remove the effect of volumetric strain from the original stress tensor but
adds additional constraints [34]. From conservation of mass, we can say that the mass density in
the undeformed solid changes from po to p in the deformed state where p = po J [20].
Modeling the magnetic and elastic behavior of the various geometries uses a method that
seeks to solve the static conservation of momentum in parallel with Maxwells equations for
magneto-statics. The three primary equations include
(1)
(2)
(3)
where is the Cauchy stress, H is the magnetic field vector, and BH is the magnetic induction.
The magnetic behavior of the domains in the models can be defined through
BH = oH + M
(4)
where M is the remanent magnetization of the domain and o is the permeability of free space.
The elastic behavior of the PDMS substrate, and MAE in the cantilever model, is assumed to
follow the Mooney-Rivlin strain energy density function, U, given as
U = C10( 1 + 3) + C01 ( 2 + 3)+ (J-1)2
(5)
where C01, C10, and K1 are material properties, J is the Jacobian matrix, and 1 and 2 are the first
and second invariants of the deviatoric component of the Cauchy-Green deformation tensor [16].
The C01 and C10 parameters and K1, or bulk modulus, are material properties. The
incompressibility of the material implies that there is no net volume change during deformation
(i.e. J = 1). Comsol models this by modifying the second Piola-Kirchoff stress tensor to remove
the effect of volumetric strain from the original stress tensor; however no additional constraints
are added to the model [25]. The Mooney-Rivlin model was chosen for its ability to model
elastomers by fitting model constants to experimental data [16-17] and due to the only slight
non-linearity seen in tensile testing of the PDMS (not shown). As previously stated, the MAE
material was assumed to behave linearly.

12

The elastic and magnetic behavior of the structures were coupled through use of the Maxwell
Surface Stress (MSS); we seek to solve the following series of equations
n (2 - 1) = 0
(6)
1 = + H1
(7)
2 = H2
(8)
where n is the outwards normal to the surface, 1 and 2 are the stress tensors at the interfaces
corresponding to the MAE and air domains, respectively, the H subscript denotes the full
Maxwell stress tensor. The component of the magnetic stress tensor, H, is given as
H =

(BHiBHj - 0.5 ij BH2)

(9)

where H is the magnetic component of the stress calculated on the boundary of the MAE due to
an applied magnetic field. This is detailed in Figure 2.7.

Figure 2.7: An illustration of the boundary condition satisfied at the boundary between the MAE and air domain. This boundary
condition states that the normal internal traction (elastic and magnetic) must equal the normal component of the external traction
(magnetic only). These boundary conditions are detailed in Equations (6) - (9).

Material properties for the MAE and PDMS substrate were determined through parametric
curve-fitting or by using constants determined from experimental testing. The magnetic
behavior in the simulation was modeled by defining an effective magnetization in the desired
orientation for the MAE material in Comsol and by applying a magnetic field, uniform in the farfield approaching the outer boundaries of the air domain, across all domains. The strength of the
magnetic field used in the model was set to be the strength of the magnetic field used in each
experimental configuration.
The effective magnetization of the MAE specimens in the Comsol model was determined by
conducting a parametric sweep on the magnetization until an observed quantity converged on the
experimentally measured quantity. For the cantilever geometry, the total force was observed as a
function of the magnetic field for various magnetizations. For the two-segment geometries, the
axial force was observed as a function of the magnetic field for various magnetization. Lastly,
the four-segment geometry varied the magnetization and observed the average bend angle versus
the applied field. Note that, in Table 2.2, all values for the magnetization are not identical; the
magnitude of the magnetizations determined through the parametric curve fitting did not
converge to similar values. Though all samples were prepared in an identical fabrication
process, experimental error is a possibility. It is also important to note that all samples were not
13

experimentally quantified; hence, results of the parametric curve fitting have no means for
comparison.
The elastic behavior of the model was characterized using a linear elastic model for the twosegment and four-segment accordion structures MAE patches or a hyperelastic two-parameter
Mooney-Rivlin model (where the parameters are defined as C01 and C10) for the PDMS substrate
and MAE cantilever beam. For the asymmetric two-segment accordion structure, the
hyperelastic constants were found through a tensile test which is described in detail in Section
2.2.3. For all other geometries, these parameters were determined through parametric curvefitting by observing a specific quantity and conducting a parametric sweep on the variable in
question. This variable was the resulting tip force for the cantilever structure, the resulting axial
force for the asymmetric and symmetric two-segment accordion structure, and the bend angle for
the four-segment accordion structure. When the observed quantity converged on the
experimentally measured quantity, the variable was set to the corresponding value. The
parameters that were used to model the various structures studied in this work can be found
below in Table 2.2. The data from the parametric studies can be found in the Appendix. In the
same context as the magnetization, the PDMS and MAE specimens were fabricated using
identical methods; variations in the constants determined through parametric curve fitting and the
tensile test may be attributed to experimental error. It is also important to note that the two
materials were not experimentally quantified; hence, results of the parametric curve fitting have
no means for comparison.
Table 2.2: Effective material properties for the cantilever beam, two-segment, symmetric two-segment, and four-segment
accordion structures along with range of field strengths the structures were subjected to.

Parameter
H (mT)
m (Amps/m)
C10
C01
EMAE (Pa)

Cantilever
Beam

Two-Segment
Accordion

Value
0 to 40 mT
55,000
300,000
275,000
-------------------

Value
0 to 45
35,000
63,000
31,000
1,450,000

Symmetric
Two-Segment
Accordion
Value
0 to 100
15,000
380,000
500,000
1,450,000

Four-Segment
Accordion
Value
0 to 200
35,000
63,000
31,000
1,450,000

2.4 The Finite Element Method


This work seeks to employ the finite element method to model the behavior of MAEs. While
the governing equations describe the physics involved in the model, they do not describe how a
solution is approximated based on the governing equations and select boundary conditions. The
finite element method begins by developing a geometric model and evolving it into a
mathematical model through applied differential equations and boundary conditions [22].
Next, the mathematical model is discretized by partitioning the geometric domain into a
mesh of finite elements. This process converts the fully continuous field (i.e. domain) into a
piecewise continuous field (i.e. a finite number of nodes and elements). This discretization
introduces error as the solutions between nodes are approximated via interpolation functions.
Even if the process of discretizing a domain could reduce error to zero, other sources of error lie
14

within the process. For instance, a computer uses a finite precision, or number of digits, to
represent the data. Thus numerical error is introduced.
After a model is discretized, the finite element analysis follows the following computational
procedure [22]:
1.) Convert the governing equations into matrices that describe the element behavior.
2.) Connect all elements together and assemble the element matrices into a single structure
matrix.
3.) Provide loads and boundary conditions for nodes which capture the experimental
behavior. (boundary loading is used to capture the magneto-elastic coupling through the
Maxwell surface stress)
4.) Solve the system of algebraic equations for each node and interpolate along the elements
using shape functions (also referred to as basis functions).
5.) Compute gradients which include strain and magnetic field gradients for this work. This
is an iterative process.
In this method, the quadratic-elements used to connect the nodes can form various geometric
mesh elements. These elements include, but are not limited to, linear triangles, quadratic
triangles, and quadratic rectangles in two-dimensional analysis and tetrahedron and hexahedral
elements in three-dimensional analysis.
In the FEA, the solutions are obtained using numerical methods such as the Galerkin method.
The universal step that all numerical methods use is converting the strong form of the governing
equations into their weak form. The strong form of the governing equations implies that all
conditions must be met at every material point whereas the weak for (or integral form) states that
conditions must be met on an average basis [22]. Next, the finite element solver seeks to solve
the weak form of the equations and generates the solutions (dependent variables). In Comsol,
the solvers used consist of the Direct solver and the PARDISO solver. The choice in solver is
determined by characteristics of the problem such as matrix symmetry and the presence of nonlinearity. The solutions are then compared to initial values to generate a residual. Typically, this
value is very small (approximately 10-6). If the residual is too large, Comsol uses an iterative
process to reduce this residual. The next step in the solution is generally determined by using the
gradient of the solution to project a new approximation for the solution. This solution is then
compared to initial values and a residual is calculated. The process is then repeated until a finite
residual is achieved; this is typically set by the user or the program. These methods are
described in detail in [22].
In this work, the magneto-elastic coupling is approximated by evaluating the MSS over the
boundary of the MAE. Since the MSS drives the deformation of the structures, high mesh
quality along the boundary is imperative as inaccuracies in the MSS is detrimental to the validity
of the models.

15

2.5 Mesh Convergence


The finite element analysis for the cantilever, two-segment (both geometries), and foursegment geometries employed both a mesh convergence study to determine a sufficient value of
degree of freedom density in each model as well as a convergence study on the size of the
bounding air box needed to model the application of the magnetic field. The studies showed
that as the degrees of freedom increased, the solution of the model converged on a finite axial
force and magnetic field (axis plotted on a log scale). Converged meshes for all four cases are
shown in Figures 2.8A-D. In each case a sufficient number of elements and size of the bounding
box was used to ensure both uniformity in the magnetic field at the boundaries (the far field) and
convergence of the reaction forces (cantilever and two-segment accordion) or average bend angle
(four-segment accordion) models. Note that meshes were refined at locations where the solution
was expected to vary sharply relative to other areas.
The cantilever and two-segment accordion structures were modeled with triangular elements
of quadratic order and periodically remeshed to retain mesh quality. Two separate mesh
geometries were analyzed for the four-segment accordion geometry. First attempts used
quadratic, triangular elements and remeshing as in the cantilever and four-segment geometries. A
second method used quadratic, quadrilateral elements (without remeshing). Both methods
yielded similar results. The mesh convergence studies for the four structures can be seen in
Figures 2.9 - 2.11. The converged meshes had degree of freedom densities of 11.1, 14.2, 14.1,
and 11.7 DoF/mm2 for the cantilever, asymmetric two-segment, symmetric two-segment, and
four-segment geometries respectively. The convergence study observed both a reaction force at a
fixed node boundary (which is at the boundary of the PDMS for the accordion structures and the
boundary of the MAE for the cantilever beam structure) along with the average magnetic field in
the air domain. Convergence was established when the observed values differed by less than one
percent in subsequent simulations. For the four-segment accordion model, the converged solution
used a mesh with DOFs that were an order of magnitude lower than the other three geometries as
seen in Figure 2.11; this may be attributed to the uniqueness in the geometry. It may also be true
that an increase in the DOFs could result in a deviation from the asymptote. However, the model
fails to solve at larger mesh densities than those used in the convergence study.

16

Figure 2.8: Finite element meshes corresponding to the (A) four-segment accordion structure, (B) cantilever beam, (C) symmetric
two-segment accordion, and (D) two-segment accordion constructed using COMSOL Multiphysics. Note: Images are not scaled
relative to each other.

17

100

0.152
0.150
0.148

Reaction
Force
Magnetic
Field

0.146
0.144

10

0.142
0.140
0.138
0

100000

200000
300000
Degrees of Freedom

400000

Magnetic Flux Density, B (mT)

Reaction Force, X-Component (N)

0.154

1
500000

Figure 2.9: A mesh convergence study for the cantilever beam structure where the observed reaction force was located at the topleft node of the MAE structure and the observed magnetic flux density was the average over the entire air domain. Note: The
right axis is plotted on a log-scale.

100

Reaction
Force
Magnetic
Field

Reaction Force, X-Component (N)

0.020

0.015

10
0.010

0.005

0.000
9000

Magnetic Flux Density, B (mT)

0.025

1
209000

409000
609000
Degrees of Freedom

809000

1009000

Figure 2.10: A mesh convergence study for the two-segment structure (along with the symmetric two-segment accordion
structure as the results were nearly identical) where the observed reaction force was located at the right boundary of the PDMS
substrate the observed magnetic flux density was the average over the entire air domain. Note: The right axis is plotted on a logscale.

18

3.685

100

3.675
3.670

Reaction Force

3.665

Magnetic Field
10

3.660
3.655
3.650

Magnetic Flux Density, B (mT)

Reaction Force, X-Component (N)

3.680

3.645
3.640
14000

19000

24000
29000
Degrees of Freedom

34000

1
39000

Figure 2.11: A mesh convergence study for the four-segment structure where the observed reaction force was located at the right
boundary of the PDMS substrate the observed magnetic flux density was the average over the entire air domain. Note: The right
axis is plotted on a log-scale.

19

3. RESULTS AND DISCUSSION


3.1 Cantilever Beam
In [1], experimental testing of the
cantilever beam specimen (detailed in
Figure 2.2) is described. The cantilevered
specimen was subjected to a forceddisplacement on the fixed-end of the beam
and the resulting force at the location of
the applied displacement was measured
and compared to FEA data. The typical
results of a simulation can be seen in
Figure 3.1. Figure 3.1 details the three
stages of actuating the MAE structure.
First, the structure is given a
magnetization and elastic properties.
(A)
(B)
(D)
(C)
Next, the specimen is subjected to a
Figure 3.1: Results of simulations of the (A) remanent
uniform magnetic field which is applied
magnetization in red arrows, (B) magnetic field in red
streamlines, (C) remanent magnetization in red arrows for the
perpendicular to the orientation of the
cantilever deformed geometry and (D) Maxwell surface
MAEs magnetization in the reference
tractions in blue arrows for the cantilever deformed
configuration. Finally, the structure is
geometry.
actuated through the Maxwell surface
stress into its deformed configuration. This stress is illustrated in Figure 3.1D.
The resulting total force was then calculated at the top left corner of the cantilever specimen
for each forced-displacement and applied magnetic field. The results can be seen in Figure 3.2.
Experimental and simulation results show fair agreement. Results show that as the tip deflection
increases for a given applied field, the force exerted by the cantilever beam increases linearly
(both experimentally and computationally). For a given displacement, increases in the magnetic
field also result in increases in the observed force.
As stated in the Finite
Element Modeling section,
the elastic and magnetic
values were determined
through parametric curvefitting.
Elastic constants
(C01
and
C10)
were
determined
by
fitting
simulation data to the zerofield experimental data
presented in Figure 3.2.
Magnetization of the MAE
specimen was determined
by fitting simulation data to
the
zero-displacement
Figure 3.2: Total force data for the cantilever beam specimen. The experimental
data (points) is compared to Comsol simulations (lines).

20

experimental data presented in Figure 3.2


Inconsistencies between the experimental and simulation data may be attributed to many
factors such as the conical distribution of magnetic particles in the experimental specimens, the
accuracy of the Maxwell surface stress in the FEM, along with the non-uniformity of the
magnetic field between the pole faces of the C-shape electromagnet. These are discussed in
detail in Chapter 4.
3.2 Asymmetric Two-Segment Accordion
Results for a simulation for the
asymmetric two-segment accordion
(including the symmetric geometry)
can be seen in Figure 3.3. The figure
details the process by which the MAE
structure actuates and is described in
Section 3.1. However, in this case,
the MAEs are bonded to a nonmagnetic substrate. The MAEs are
assumed to be stiffer than the PDMS
while the PDMS substrate is able to
deform. When the MAE is subjected
to a uniform magnetic field, as shown
in Figure 3.3B, a Maxwell stress is
applied to the surface which generates
a torque. Note that the magnetic field
in Figure 3.3B may not appear to be
uniform as the magnetization of the
MAE patches interacts with the
applied magnetic field to generate
local non-uniform fields. Since the
MAE are stiffer than the PDMS, the
torque is transmitted to the PDMS
(A)
(B)
(C)
substrate which causes the structure to
Figure 3.3: Simulation results of the
(A) remanent
actuate
into
its
deformed
magnetizations of the MAE segments in red arrows, (B) the
configuration as shown in Figure
magnetic flux, in green streamlines, and (C) of Maxwell surface
3.3C.
stress tensor for the asymmetric two segment accordion structure
with the relative size of the tractions depicted in blue arrows. Note
Results of the experimental testing
the traction would produce opposing torques on the MAE patches
(points) and simulations (lines) can be
due to their opposing remanent magnetizations. Also, please note
that the asymmetric two-segment accordion behaves in a similar
seen in Figure 3.4. Figure 3.4 shows
manner.
the measured bend angle and resulting
axial force as a function of the applied magnetic field. As shown, the model has excellent
agreement with the experimental results. As the applied magnetic field increases, the bend angle
between the two MAE patches also increases. However, since the bend angle is measured from
its reference configuration, which is assumed to be a 180 bend angle, the measured angle
appears to decrease in a numerical sense. The measured bend angle is detailed in Figure 2.4B.
However, the degree of deformation does increase. Also shown in Figure 3.4 is the measured
21

axial force as a function of the magnetic field. As the magnetic field increases, the resulting
axial force also increases. The model accurately predicts the axial force data presented in Figure
3.4 (right data plot) and the bending behavior presented in Figure 3.4 (left data plot).
Effective material values for the asymmetric two-segment accordion structure were also
determined through parametric studies. The character of the resulting axial force and bend angle
versus magnetic field was established through parameterizing and sweeping the C01 and C10
values. The simulation data was then fit to the experimental data by varying the magnetization
until convergence was achieved.

Figure 3.4: Measured bend angle (left) and resulting axial force (right) as a function of the applied magnetic field for the twosegment accordion structure. Experimental results (open shape) are compared with Comsol simulations (lined filled shapes).

3.3 Symmetric Two-Segment Accordion


Results of the experimental testing and the comparison to the finite element analysis for the
symmetric two-segment accordion can be seen in Figure 3.5; these results are described as the
total force data sets since the observed force is a combination of the magnetic and elastic
forces. Figure 3.5 shows that increasing the magnetic field, H, for a given displacement results
in an increase in the axial force observed in the PDMS specimen; also, increases in the
prescribed displacement applied to the PDMS for a given magnetic field result in a decrease in
the axial force. The latter decrease in force denotes a compressive reaction force against the
force gauge while the former increase in force stems from the tensile nature of the accordion
structures actuation.

Figure 3.5: Experimental (symbols) vs. FEA simulation data (lines) for the total axial forces measured on the PDMS substrate.

22

Using the total force observed at a 0 mT field, the elastic data could be extracted from the
total force data and is shown in Figure 3.6. The data shows increases in the axial force as the
prescribed displacement increases from 0.0 to 2.0 mm. However, from 2.0 to 6.0 mm, the axial
force becomes asymptotic about 0.10 N. Note that elastic constants were determined from
tensile testing.

Figure 3.6: Experimental (symbols) vs. FEA simulation data (lines) for the elastic axial forces measured on the PDMS substrate.

Using the total force and elastic force plots, the magnetic force can be calculated by
subtracting the elastic force from the total force. The resulting magnetic force data can be seen
in Figure 3.7 where data is presented; as the magnetic field increases for a given displacement,
the magnetic force also increases. For a given applied field, as the magnitude of the prescribed
displacement increases, the magnitude of the magnetic force decreases. As the magnetic field
increases, the behavior of the magnetic force versus the prescribed displacement shifts from
approximately a constant function (illustrated by the data plots on the bottom of Figure 3.7) to a
polynomial function (illustrated by the data plots on the top of Figure 3.7). This may be
attributed to non-uniformity in the magnetic field created by the C-shape electromagnet along
with the characterization of the magnetization. The magnetization given to the specimen is a
perfectly aligned approximation of the magnetization; however, the magnetization in
experimental testing is predominantly aligned along the primary axis of the MAE specimens.
The distribution of particles about this primary axis is not captured in the finite element model.

Figure 3.7: Experimental (symbols) vs. FEA simulation data (lines) for the magnetic axial forces that are generated from the
magnetic torques detailed in Figure 2.5B and 2.5C.

23

Next, if the magnetic force versus displacement is integrated over the displacement, one is
able to obtain the magnetic work potential for the two-segment accordion structure. The
magnetic work potential as a function of the applied displacement can be seen in Figure 3.8. As
shown in Figure 3.8, increases in the magnetic field for a given displacement (with magnitude
greater than zero) result in an increased magnetic work potential; increases in the magnitude of
the prescribed displacement for a given magnetic field also result in an increased magnitude of
the magnetic work potential.

Figure 3.8: Experimental (symbols) vs. FEA simulation data (lines) for the magnetic work potential of the two-segment
accordion structure.

3.4 Four-Segment Accordion


Results for a typical simulation of the foursegment accordion model can be seen in Figure
3.9. The structure actuates in a manner similar to
that of the asymmetric two-segment accordion
structure and is detailed in the section Section
3.2; however, this structure has two additional
MAE patches. The deformed structure displays
both mountain and valley folds regardless of the
direction of the applied magnetic field. Note that
the magnetic field does not appear be uniform in
Figure 3.9B; this is due to the interaction between
the magnetic field and the MAE patches which is
detailed in Section 3.2.
In both the experiment and simulations, the
average of the bend angles, shown in Figure
2.6B, was calculated as a function of the applied
magnetic field. The data is presented in Figure
3.10 with the experimental data (symbols)
compared to the simulation data (lined-symbols).
As the applied magnetic field increases, the
average bend angle also increases in magnitude.
As shown in Figure 3.10, the experimental data
and Comsol simulation data show qualitative

(A)

(B)

(C)

Figure 3.9: Depiction of the (A) remanent


magnetizations of the MAE segments in red arrows,
(B) magnetic flux, in green streamlines, and the (C)
MSS vectors for the four segment accordion
structure with the relative size of the tractions
depicted in blue arrows. Note the traction would
produce opposing torques on the MAE patches.

24

agreement. However, the magnitude of the measured bend angle is not captured. This could be
due to reasons that were previously mentioned such as the inability to capture the conical
distribution of the aligned magnetic powder or the assumption that the applied magnetic field
was perfectly uniform.
The elastic and magnetic material properties for the MAE and PDMS substrate were
determined through parametric curve fitting. The methodology parallels that presented in
Section 3.2. Note that the experimental data and simulation data for the bend angle shown in
Figure 3.10 show disagreement as the applied magnetic field increases in magnitude. While
modeling efforts sought to determine effective material properties (C01, C10, and M), variations
in the magnetization and hyperelastic parameters resulted in negligible changes in the observed
bend angle. As seen in Figure A.5 and A.6, the magnitude of the observed bend angle saturated
at a maximum value of 110 degrees. This indicates that there is a numerical issue with the model
as further changes to these parameters resulted in similar behavior of the structure.

Figure 3.10: Experimental (symbols) vs. FEA (solid-line symbols) simulations for the four-segment accordion structure.

25

4. CONCLUSIONS
This work highlights the ability of the finite element method to model magneto-active
elastomers. Furthermore, it shows the ability of the FEM to model complex structures that
consist of MAE materials and inactive elastomer substrates consisting of multiple fold lines. The
method uses the Maxwell surface stress to couple the elastic and magnetic behaviors of the
MAE. In this work, four structures were analyzed and consisted of a cantilever beam MAE
specimen, a two-segment and symmetric two-segment accordion structure, and a four-segment
accordion structure. The method was able to predict, to an extent, the magnetic and elastic
behavior of the four geometries. The model was unable to capture the magnitude of the bending
behavior in the four-segment accordion model; however, the mode of the structure's deformation
in the simulation was similar to that of the deformation observed in the experiment.
This may attributed to the fact that the models failed to capture the conical distribution of the
magnetic powder in the elastomer matrix. Instead, the model assumed a magnetization in a
prescribed orientation; however, this magnetization was permitted to rotate with the MAE
specimens as they rotated and/or deformed. The model also has shortcomings in predicting the
true magnetic and elastic behavior of the MAE and PDMS materials when using parametric
curve fitting. Since this model uses the FEM, the solution is approximated on a finite point by
point basis. The method does not satisfy the governing equations over the entire domain;
instead, the solution uses the weak form of the governing equations and satisfies them in their
integral form (i.e. on an average basis). Additionally, this model fails to capture the microscopic
phenomena, such as the particle-field and particle-particle interactions. The model simply tries
to capture the deformation these interactions have through the MSS. For instance, particles may
be rotating within the elastomer matrix as opposed to generating a torque through their restriction
within the matrix in an applied magnetic field. This phenomena is not accounted for in the
Comsol model.
This work still also requires a way to predict effective constants for the various material
parameters. Currently, the method uses parametric studies (on material properties) to fit the
simulation data to experimental data. Future methods will use experimentally measured material
properties to develop the finite element models; currently, only the symmetric two-segment
accordion employed methods that experimentally quantified the elastic constants. However, the
values may need to be adjusted in the model to account for assumptions that are made within the
model that are not true in situ such as non-uniformites in the magnetic field and the distribution
of magnetic particles. These adjusted material properties are considered to be effective material
properties as they are not the experimentally measured quantities.
The findings from this work does show promise in modeling magneto-active elastomers
using the finite element method. While previous works of literature have studied the material
properties and modeled the small deformation of MAE and MRE in the presence of magnetic
fields, the results presented in this work highlight the ability of the FEM to model MAE
undergoing large deformation in the presence of a magnetic field. Furthermore, this work allows
researchers to understand the modes of deformation based on the geometry given. Substrates
with MAE patches placed with their magnetization in opposing directions (separated by a
prescribed distance) generate mountain or valley folds. This is expected as the patches create
opposing torques to generate the desired fold. The characteristic of the fold (mountain or valley)
26

is determined through the direction of the applied magnetic field along with the orientations of
the MAE patches relative to the field. The degree of bending is also affected by the compliance
of the substrate used (or the elastomer used to house the barium ferrite for the cantilever beam
geometry). Using more compliant (i.e. less stiff) substrates results in larger degrees of bending;
however, when the compliance is too low, the material may deform under its own weight. Thus,
an optimal value exists between compliance of the elastomer and the desired actuation you seek
to achieve.
Additionally, this work allows researchers to understand the effects of the magnetizations
orientation and magnitude of the MAE patches. Patches with opposing orientation bonded to a
substrate a finite distance apart will generate opposing torques that generate mountain and valley
folds, assuming the substrate is more compliant that the patches, if oriented correctly (i.e. in
opposing directions) and with a sufficient magnetization. The magnetization required to actuate
the structure will vary depending on the materials used and the magnitude and direction of the
applied magnetic field. The placement of the MAE patches on a substrate is also important as
they affect the degree of bending. Patches placed closer together, relative to an initial
configuration, form a stiff joint (where the joint is the substrate material between the patches)
whereas patches place further apart, relative to the same initial configuration, form a more
compliant joint. These structures with the two patch placements will require larger and smaller
magnetic torques, respectively.
While optimization is not addressed in this work, the FEM developed in this work is capable
of studying variations in the current structures to optimize the material placement, the magnitude
and orientation of the magnetization, and the specimen's size.
Future work in this field of study will incorporate other active materials such as terpolymers
into the action origami structures; this active material uses electric fields to produce a form of
actuation. Action is achieved when the terpolymer is subjected to an electric field. The electric
dipoles within the terpolymer material seek to reach a minimum energy state by aligning with the
applied electric field. This creates electrostriction which seeks to compress the material through
its thickness and expand in the planar directions. When the planar surface on one side of the
terpolymer is fixed, the structure is restricted on one boundary and free on all other boundaries.
The result is a structure capable of bending.
Preliminary modeling of the material seeks to modify the equations of equilibrium within the
Comsol software. This is done by introducing a new set of terms to the equilibrium equations
which includes an electromechanical coupling coefficient and the strain
induced by the applied electric field. These two terms act through the plane
of the terpolymer to produce a compression through the thickness of the
material and an expansion in the planar direction of the material. While
current efforts seek to add these terms to the linear model, future efforts will
require incorporating these terms into the hyperelastic model of the
Figure 4.1: Water
terpolymer.
origami
Future work will also develop action origami structures with MAE bomb
structure which uses
patches integrated into the substrate layer. This is detailed in Figure 4.1. MAE
patches
The figure depicts a waterbomb structure that is actuated with MAE integrated into a
substrate to
patches. The substrate is a PDMS material while the patches are placed PDMS
actuate in an applied
within the PDMS prior to curing the specimen.
magnetic field.
27

BIBLIOGRAPHY
[1] P. von Lockette, S. Lofland, J. Biggs, J. Roche, J. Mineroff and M. Babcock, Investigating
new symmetry classes in magnetorheological elastomers: cantilever, Smart Materials and
Structures (20) p.105022., 2011.
[2] T. Kanade, A Theory of Origami World, Artificial Intillegence (13) pp. 279-311., 1980.
[3] L. Bowen, W. Baxter, S. Magleby and L. Howell, A position analysis of coupled spherical
mechanisms found in action origami, Mechanism and Machine Theory (77) pp. 13-24.,
2014.
[4] T. Tachi, Simulation of rigid origami, R. Lang (Ed.), Origami 4: Fourth International
Meeting of Origami Science, Mathematics, and Education, A K Peters, Ltd. pp. 175188
(vol.), 2009.
[5] R. Lang, A computational algorithm for origami design, Proceedings of the Twelfth Annual
Symposium on Computational Geometry, ACM, pp. 98105., 1996.
[6] R. Lang, Origami in Action, New York, NY: St. Martin's Griffin, 1997.
[7] R. D. Kornbluh, R. Pelrine, J. Joseph, R. Heydt, Q. Pei and S. Chiba, High-field
electrostriction of elastomeric polymer dielectrics for actuation, Symposium on Smart
Structures and Materials, International Society for Optics and Photonics, 149-161., 1999.
[8] M. Leary, F. Schiavone and A. Subic, Lagging for control of shape memory alloy actuator
response time, Materials & Design, 31 (4), 2124-2128., 2010.
[9] P. von Lockette and R. Sheridan, Folding Actuation and Locomotion of Novel MagnetoActive Elastomer (MAE), SMASIS 2012-3222, 2013.
[10] A. Heller, A giant leap for space telescopes, Sci. Technol. Rev., pp. 1218., 2003.
[11] M. Barham, D. Steigmann and D. White, Magnetoelasticity of highly deformable thin films:
Theory and simulation, International Journal of Non-Linear Mechanics (47) pp. 185-196.,
2012.
[12] A. Benine and M. Farshad, Magnetoactive elastomer composites, Polymer Testing (23) pp.
347-353., 2004.
[13] A. Dorfmann and I. Brigadnov, Constitutive modeling of magneto-sensitive Cauch-elastic
solids, Computational Materials Science (29) pp. 270-282., 2003.
[14] A. Cladera, B. Weber, C. Leinenbach, C. Czaderski, M. Shahverdi and M. Motavalli, Ironbased shape memory alloys for civil engineering structures: An overview, Construction and
Building Materials (63) pp. 281-293, 2014.
[15] P. Ghosh and A. Srinivasa, Development of a finite strain two-network model for shape
memory polymers using QR decomposition, International Journal of Engineering Science
(81) pp. 171-191., 2014.
[16] R. Pelrine and H. Prahlad, Dielectric Elastomers as Electromechanical Transducers,
Fundamentals, Materials, Devices, Models and Applications of an Emerging Electroactive
Polymer Technology, pp. 145-155., 2008.
[17] J. Capsal, J. Galineau, M. Lallart, P. Cottinet and D. Guyomar, Plasticized relaxor
ferroelectric terpolymer: Toward giant electrostriction, high mechanical energy and low
28

electric field actuators, Sensors and Actuators (A 207) pp. 25-31., 2014.
[18] H. Okuzaki, T. Saido, H. Suzuki, Y. Hara and H. Yan, "A Biomorphic Origami Actuator
Fabricated by Folding a Conducting Paper," Journal of Physics: Conference Series, vol.
127, p. 012001, 2008.
[19] L. Bowen, M. Frecker, T. Simpson and P. von Lockette, "A dynamic model of magnetoactive elastomer actuation of the waterbomb base," in Proceedings of the ASME 2014
Internation Design Engineering Technical Conferences & Computers and Information in
Engineering Conference, Buffalo, New York, 2014.
[20] R. Martinez, C. Fish, X. Chen and G. Whitesides, "Elastomeric Origami: Programmable
Paper-Elastomer Composites as Pneumatic Actuators," Advanced Functional Materials, vol.
22, pp. 1376-1384, 2012.
[21] L. Bowen, M. Frecker, T. Simpson and P. von Lockette, "A Dynamic Model of MagnetoActive Elastomer Actuation of the Waterbomb Base," in ASME International Design and
Engineering Technical Conference, Buffalo, NY, 2014.
[22] G. Du and X. Chen, "MEMS magnetometer based on magnetorheological elastomer,"
Measurement, vol. 45, pp. 54-58, 2012.
[23] A. Boczkowska and e. al., "Mechanical properties of magnetorheological elastomers under
shear deformation," Composites, vol. 43, no. B, pp. 636-640, 2012.
[24] G. Du and X. Chen, MEMS magnetometer based on magnetorheological elastomer,
Measurement (45) pp. 5458., 2012.
[25] J. Ginder, S. Clark, F. Schlotter and M. Nichols, "Magnetostrictive phenomena in
magnetorheological elastomers," International Journal of Modern Physics, vol. 16, no. B,
pp. 472-478, 2002.
[26] J. Martin, R. Anderson and D. Dead, "Magnetostriction of field-structured
magnetoelastomers," Phys. Rev., vol. 74, no. E, pp. 051507-051524, 2006.
[27] X. Wang, F. Gordaninejad, M. Calgar, L. Yanming, J. Sutrisno and A. Fuchs, "Sensing
behavior of magnetorheological elatomers," Journal of Mechanical Design, vol. 131, pp. 16, 2009.
[28] G. Zhou, Shear properties of a magnetorheological elastomer, Smart Matererials and
Structures (12) pp. 139146., 2003.
[29] E. Galipeau, S. Rudykh and e. al., "Magnetoactive elastomers with periodic and random
microstructures," Int. Journal of Solids Struct., no. Accepted manuscript, 2014.
[30] E. Galipeau and P. Ponte Castaneda, "The effect of particle shape and distribution on the
macroscopic behavior of magnetoelastic composites," Int. J. Solids Struct., vol. 49, pp. 1-17,
2012.
[31] Y. Raikher and O. Stolbov, "Deformation of an ellipsoidal ferrogel sample in a uniform
magnetic field," Journal of Applied Mechanics and Technical Physics, vol. 46, no. 3, pp.
434-443, 2005.
[32] Y. Raikher, O. Stolbov and G. Stephanov, "Shape instability of magnetic elastomer
membrane," J. Phys. D: Appl. Phys., vol. 41, pp. 1-4, 2008.
[33] C. Tedesco, INVESTIGATING THE EFFECTS OF A MAGNETO-ACTIVE ELASTOMER
29

EXPOSED TO AN EXTERNAL MAGNETIC FIELD AND AT FIXED DISPLACEMENTS,


State College, PA: Pennsylvania State University, 2014.
[34] Comsol, Comsol Multiphysic's User Guide, AB: Comsol, 2010.
[35] A. Bower, Applied Mechanics of Solids, Boca Raton, FL: CRC Press, 2012.
[36] A. Bower, Applied Mechanics of Solids, CRC Press 1st Edition, 2012.
[37] R. Cook, D. Malkus, M. Plesha and R. Witt, Concepts and Applications of Finite Element
Analysis, Danvers, MA: John Wiley & Sons. Inc., 2002.
[38] B. Margherita, "Wolfram Math World," MathWorld, [Online]. Available:
http://mathworld.wolfram.com/KawasakisTheorem.html. [Accessed 5 June 2014].
[39] B. Kim, S. Beom Lee, J. Lee, S. Cho, P. H., S. Yeom and S. Han Park, A comparison
among Neo-Hookean model, Mooney-Rivlin model, and Ogden model for chloroprene
rubber, International Journal of Precision Engineering and Manufacturing Vol 13, Issue 5,
pp. 759-764., 2012.

30

APPENDIX
Cantilever Beam Specimen: Parametric Studies
0.200
0.180
Experimental

0.160

M=25k

Total Force (N)

0.140

M=30k

0.120

M=35k

0.100

M=40k

0.080

M=45k

0.060

M=50k

0.040

M=55k

0.020

M=60k

0.000

M=65k
0

0.01

0.02

0.03

Magnetic Flux Density (T)

Figure A.1: Details of the parametric study conducted on the cantilever beam MAE specimen to
determine the effective magnetization.
Table A.1: Details of the parametric study conducted on the cantilever beam MAE specimen to
determine the effective C01 and C10 values.
Elastic Force Simulations:
E=1450000, c01=55800, c10=200000

E=145000,c01=155000,c10=300000

x_disp

Reaction force, x component (N)

x_disp

Reaction force, x component (N)

1.00E-05

1.02E-07

1.00E-05

1.82E-07

1.00001

0.01022

1.00001

0.01818

2.00001

0.02039

2.00001

0.03627

3.00001

0.03054

3.00001

0.05432

4.00001

0.04069

4.00001

0.07238

5.00001

0.05087

5.00001

0.09049

0.0611

0.10868

0.0714

0.12701

0.08181

0.14551

0.09234

0.16424

10

0.10302

10

0.18324

11

0.11389

11

0.20257

E=1450000,c01=100000,c10=200000
x_disp

Reaction force, x component (N)

E=125000,c01=155000,c10=300000

8
x_disp

Reaction force, x component (N)

31

1.00E-05

1.20E-07

1.00E-05

1.82E-07

1.00001

0.01198

1.00001

0.01818

2.00001

0.02391

2.00001

0.03627

3.00001

0.03582

3.00001

0.05432

4.00001

0.04772

4.00001

0.07238

5.00001

0.05966

5.00001

0.09049

0.07166

0.10868

0.08374

0.12701

0.09594

0.14551

0.10829

0.16424

10

0.12082

10

0.18324

11

0.13357

11

E=1450000,c01=100000,c10=175000

x_disp

Reaction force, x component (N)

x_disp

Reaction force, x component (N)

1.00E-05

1.10E-07

1.00E-05

1.82E-07

1.00001

0.01099

1.00001

0.01818

2.00001

0.02192

2.00001

0.03627

3.00001

0.03283

3.00001

0.05432

4.00001

0.04375

4.00001

0.07238

5.00001

0.05469

5.00001

0.09049

0.06569

0.10868

0.07676

0.12701

0.08795

0.14551

0.09927

0.16424

10

0.11075

10

0.18324

11

0.12244

11

0.20257
E=100000,c01=250000,c10=300000

E=145000,c01=145000,c10=200000

x_disp

Reaction force, x component (N)

x_disp

Reaction force, x component (N)

1.00E-05

1.38E-07

1.00E-05

2.20E-07

1.00001

0.01378

1.00001

0.02197

2.00001

0.0275

2.00001

0.04384

3.00001

0.04119

3.00001

0.06567

4.00001

0.05488

4.00001

0.08749

5.00001

0.06861

5.00001

0.10938

0.0824

0.13137

0.0963

0.15352

0.11033

0.17589

0.12454

0.19853

10

0.13894

10

0.2215

11

0.1536

11

10

x_disp

Reaction force, x component (N)

1.00E-05

1.58E-07

0.24486
E=100000,c01=275000,c10=300000

E=145000,c01=145000,c10=250000
5

0.20257
E=100000,c01=155000,c10=300000

11

x_disp

Reaction force, x component (N)

0.0

0.000

32

1.00001

0.01578

1.0

0.023

2.00001

0.03149

2.0

0.046

3.00001

0.04716

3.0

0.069

4.00001

0.06284

4.0

0.091

5.00001

0.07855

5.0

0.114

0.09435

6.0

0.137

0.11026

7.0

0.161

0.12632

8.0

0.184

0.14258

9.0

0.208

10

0.15908

10.0

0.232

11

0.17586

11.0

0.256

E=145000,c01=155000,c10=275000

x_disp

Reaction force, x component (N)

1.00E-05

1.72E-07

1.00001

0.01718

2.00001

0.03428

3.00001

0.05134

4.00001

0.0684

5.00001

0.08551

0.10271

0.12003

0.13751

9
10

0.15522
0.17317

11

0.19144

33

Two-Segment Accordion Structure: Parametric Studies

Resulting Axial Force (N)

0.06
0.05
0.04

Experimental
Simulation M=35000

0.03

Simulation M = 45000

0.02

Simulation M = 55000

0.01

Simulation M = 65000

0
0

10

20

30

40

50

Mag. Field, B (mT)

Figure A.2: Details of the parametric study conducted on the two-segment accordion structure to
determine the effective magnetization.
Symmetric Two-Segment Accordion Structure: Parametric Studies
2.9
2.4

Tensile Stress (Mpa)

Conditioned 0.1%/min
Conditioned 10%/min

1.9

Mooney-Rivlin Data Fit

1.4

Comsol Using C10/C01

0.9
0.4

-0.1

0.2

0.4

0.6

0.8

1.2

Strain (mm/mm)
Figure A.3: Stress-strain curve for a PDMS specimen that is subjected to a tensile test at a
0.1%/min and 10%/min strain rate up to strains of approximately 100%. The specimen tested
was 60 mm long with a cross-section of 6.0 x 1.0 mm2. The conditioning curves are shown in
red and orange. The curves were matched with a Mooney-Rivlin model, both analytically and
numerically (in Comsol), up to strains of approximately 20%. These curves are shown in gray
and yellow, respectively.

34

0.15

Resulting Axial Force (N)

0.1

Experimental 2.0 mm
Displacement

0.05

Simulation M=15000
0
0

20

40

60

80

100

120

Simulation M = 25000

-0.05

Simulation M = 35000
-0.1

Simulation M = 50000
-0.15

-0.2

Mag. Field, B (mT)

Figure A.4: Details of the parametric study conducted on the symmetric two-segment accordion
structure to determine the effective magnetization.
Four-Segment Accordion Structure: Parametric Studies
180

Bend Angle (degrees)

160
140
120
100
80
60
40
20
0
0.0000

Simulation Data C01=31,


C10=63
Simulation Data C01=61,
C10=63
Simulation Data C01=31,
C10=31
Simulation Data C01=63,
C10=63

20.0000

40.0000

60.0000

80.0000 100.0000 120.0000 140.0000 160.0000 180.0000

Field Strength (mT)

Figure A.5: Details of the parametric study conducted on the four-segment accordion structure to
determine the sufficient values for the C01 and C10 parameters.

35

180

Bend Angle (degrees)

160
140
120
100
80
60
40
20
0
0.0000

Simulation M =
35000 A/m
Simulation M =
45000 A/m
Simulation M =
55000 A/m
Simulation M =
65000 A/m

20.0000

40.0000

60.0000

80.0000 100.0000 120.0000 140.0000 160.0000 180.0000

Field Strength (mT)

Figure A.6: Details of the parametric study conducted on the four-segment accordion structure to
determine the sufficient values for the magnetization. Note that large changes were not seen
between elastic and magnetic curve fitting.
Experimental Set-Up: Cantilever, Two-Segment Accordions, and Four-Segment Accordion
Geometries

Figure A.7: A macroscopic illustration of the cantilever beam experiment detailed in Chapter 2.
The equipment used consists of a Shimpo force gauge, a non-ferrous extension to prescribe the
displacement, an MAE specimen, and a magnetic coil with a magnetic flux directed as shown in
the figure. Details of this experiment are described in [1].
.

36

Figure A.8: A macroscopic illustration of the two-segment accordion experiments detailed in


Chapter 2. The equipment used an Instron Model 4202 #2 tensile loader, a two-segment
accordion specimen, a C-shape electromagnet, and a power supply.

Figure A.9: A macroscopic illustration of the four-segment accordion experiments detailed in


Chapter 2. The equipment used a high resolution Canon EOS 7D camera (5184x3456 pixels) to
collect images, a gauss meter to measure the magnetic field, a four-segment accordion specimen,
and an electromagnet (specifications unknown).

37