You are on page 1of 8

Smart Mater. Struct. 5 (1996) 607–614.

Printed in the UK

A model of the behaviour of


magnetorheological materials
Mark R Jolly, J David Carlson and Beth C Muñoz
Lord Corporation, Thomas Lord Research Center, 405 Gregson Drive, Cary, North
Carolina 27511, USA

Received 28 May 1996, accepted for publication 19 July 1996

Abstract. Magnetorheological materials are a class of smart materials whose


rheological properties may be rapidly varied by application of a magnetic field.
These materials typically consist of micron-sized ferrous particles dispersed in a
fluid or an elastomer. A quasi-static, one-dimensional model is developed that
examines the mechanical and magnetic properties of magnetorheological materials.
This model attempts to account for magnetic non-linearities and saturation by
establishing a mechanism by which magnetic flux density is distributed within the
composite material. Experimental evidence of the viscoelastic behaviour and
magnetic properties of magnetorheological fluids and elastomers suggests that the
assumptions made in the model development are reasonable. It is shown that the
model is semi-empirical in that it must be fit to the experimental data by adjusting a
parameter that accounts for unmodelled magnetic interactions.

1. Introduction et al 1994, Carlson et al 1995) and brakes and clutches


for exercise equipment (Carlson et al 1995, Gentry et
Magnetorheological (MR) materials consist of micron al 1995, Anon 1995, Chase 1996). MR elastomers
size (typically 3 to 5 microns) magnetically permeable will ultimately find use in applications where stiffness or
particles suspended in a non-magnetic medium. Upon resonance tunability is sought (Jolly et al 1996, Kasahara
application of a magnetic field, the rheological properties et al 1994). Other applications may also make use of
of these materials are rapidly and reversibly altered. The the unique mechanical, magnetic and thermal anisotropic
mechanism responsible for this bulk effect is the induced properties of these elastomer composites (Bowen et al
magnetic interaction of particles within the matrix. In 1995).
the case of MR fluids, this mutual interaction amongst Most models of MR material behaviour are based
the particles causes the formation of chain-like structures on magnetic dipole interactions between adjacent particles
aligned roughly parallel to the applied magnetic field. The within a particle chain (Harpavat 1974, Spasojevic et
configuration and rigidity of this structure will depend upon al 1974). These interparticle interactions are then
several factors including the strength and distribution of averaged over the entire sample to yield a model of the
the applied magnetic field. A finite stress must develop to bulk magnetorheological effect. Many such models are
yield these chain structures. Hence, MR fluids are often borrowed from models developed to understand similar
modelled as Bingham plastics with a field dependent yield interparticle interactions within electrorheological (ER)
stress (Spasojevic et al 1974). Yield stresses of nearly fluids (e.g., Block and Kelly 1988 and references therein,
100 kPa have been reported in commercially available Gast and Zukoski 1989). Still a weakness unique to MR
fluids (Lord 1994). MR solids have also been developed material modelling is an understanding of how magnetic
by incorporating magnetically permeable particles into flux density distributes itself within the particle network.
elastomer media (Rigbi and Jilken 1983, Muñoz and Jolly Such an understanding needs to take into account magnetic
1995, Jolly et al 1996). Typically, magnetic fields are nonlinearities and the effects of particle saturation. There
applied to the elastomer during crosslinking such that has been recent interest in this problem using finite
particle chain structures form and become locked in place element approaches (Ginder and Davis 1994) and analytical
upon final cure (see figure 1). At low strains (< 2%), approximations (Carlson 1991, Ginder et al 1995, Jolly et
these elastomers exhibit a field dependent modulus. A field- al 1996).
induced change in modulus of up to 40% has been reported In this paper, the simple dipole model of particle energy
(Jolly et al 1995). interaction is reviewed. From this we obtain a model of
MR fluids are beginning to find commercial success the MR effect as a function of particle magnetization. In
in applications where controlled energy dissipation is particular, it is seen that MR material stress is quadratically
required. Such applications include controllable dampers related to particle magnetization. This theory is then
for primary and secondary vehicle suspensions (Carlson augmented with an approximate model of the relationship

0964-1726/96/050607+08$19.50
c 1996 IOP Publishing Ltd 607
M R Jolly et al

then considered within a viscoelastic material, then the


viscoelastic properties of the composite are the sum of
the viscoelastic properties of the composite with no field
applied (which are not modelled here) and the elastic/plastic
properties induced by interparticle magnetic force.
The model is quasi-static, therefore the dynamic effects
of the magnetic field as well as inertial and rate-dependent
effects of the particles are ignored. The particles are
assumed to be uniform, homogenous spheres that can
be magnetically modelled as identical induced dipole
moments. It is also assumed that particles are aligned
in perfect chains (albeit with gaps between particles), and
that quasi-static shear strains and associated stresses are
uniformly distributed over the length of each particle chain.

2.1. The relationship between rheology and particle


polarization
Figure 2 shows two adjacent particles (dipoles) in a chain
along with the direction of applied magnetic field. The
interaction energy of the two dipoles of equal strength |m|
and direction is (Rosensweig 1985)
|m|2 (1 − 3 cos2 θ )
E12 =
4π µ1 µ0 |r|3
r02  (1)
|m|2 1 − 3 r 2 +x 2
= 0

4π µ1 µ0 (r02 + x 2 )3/2
where µ1 is the relative permeability of the medium and
where some basic trigonometric identities have been used
in the second equality. By defining the scalar shear strain
of the particle chain as  = x/r0 , the interaction energy
can be written as
( 2 − 2)|m|2
E12 = (2)
4π µ1 µ0 r03 ( 2 + 1)5/2
Now it is assumed that the particles are aligned in
long chains and that there is only magnetic interaction
between adjacent particles within the chain, i.e., there
are no multipole interactions. Then the total energy
Figure 1. SEM images of magnetic particles in a polymeric density (energy per unit volume) associated with the one-
elastomer matrix; (upper) randomly dispersed particles, dimensional shear strain can be calculated by multiplying
(lower) particles aligned by applied magnetic field. The the particle-to-particle energy, given by equation (2), by the
width of the images is 70 microns. total number of particles and dividing by the total volume:
3φ( 2 − 2)|m|2
between particle magnetization and average composite flux U= (3)
2π 2 µ1 µ0 d 3 r03 (1 +  2 )5/2
density—a quantitiy that can be measured. This model
takes into account the gradual saturation of the permeable where φ is the volume fraction of particles in the composite
particles as applied field increases. This theory is then and d is the particle diameter. The stress induced by the
compared to experimental data on MR fluids and MR application of a magnetic field can be computed by taking
elastomers. the derivative of inter-particle energy density with respect
to the scalar shear strain:

2. Theory ∂ 9φ(4 −  2 )|m|2


σ = U= . (4)
∂ 2π µ1 µ0 d 3 r03 (1 +  2 )7/2
2

The present model examines the magnetomechanical Next, it is noted that the average particle polarization Jp is
properties of parallel chains of magnetically permeable the dipole moment magnitude per unit particle volume, i.e.
spherical particles. In particular, a magnetic field is |m| = Jp Vi , where Jp has SI units of Tesla:
applied parallel to the chains and the shear stress of
the chains caused by interparticle magnetic force is φ(4 −  2 )Jp2
σ = (5)
modelled. It is assumed that if these particle chains are 8µ1 µ0 h3 (1 +  2 )7/2
608
A magnetorheological material behaviour model

Experimental evidence has indicated that the actual particle


chain yield strain is 0.01–0.02 in MR elastomers (Jolly et al
1995) and substantially less than 0.01 in fluids (Weiss et al
1994). Theories that a majority of strain and the ultimate
yielding of a given chain occurs between a single (or a few)
pairs of particles (Tang et al 1995) are typical justification
for this large discrepancy between theory and experimental
evidence.
A small strain approximation of stress given by
equation (5) is

φJp2
σ '  < 0.1 (8)
2µ1 µ0 h3
for which a 3.6% error results when strain  = 0.1. The
preyield modulus G of the particle network is simply stress
divided by strain, or, from equation (8)

φJp2
G'  < 0.1. (9)
2µ1 µ0 h3
Furthermore, the maximum possible field induced change
in stress (and modulus) occurs when the aligned particles
become magnetically saturated, i.e. when Jp = Js .
It can be seen that the maximum field-induced stress
rises quadratically with saturation magnetization Js of the
Figure 2. (a) Geometry of two particles of diameter d
within a particle chain. Adjacent chains are a distance y
particle material. This observation advocates the use of a
apart and h = r0 /d . (b) Magnetic interaction of the two particulate material with a high saturation magnetization.
particles modelled as dipole moments m and sheared with Pure iron has the highest saturation magnetization of
respect to one another. known elements with Js = 2.1 T. However, alloys of
iron and cobalt (e.g. Permendur) have higher saturation
where we have defined h = r0 /d which is an indication of magnetizations with Js ≈ 2.4 T.
the gap between particles in a chain. This is illustrated in
figure 2. 2.2. The relationship between particle polarization and
Note that the stress-strain relationship given in average composite flux density
equation (5) is independent of particle size. The size of
The average composite flux density is relatively straightfor-
the gap between the particles in a chain can be seen to be a
ward to measure experimentally. For example, flux meters
very important parameter—magnetic field induced stress is
employing field probes or sense coils can typically measure
inversely proportional to this gap cubed. Say, for example,
the flux density at the composite sample directly. Alterna-
that the gap between particles is equal to one quarter of a
tively, if assumptions are made regarding the reluctance of
particle diameter (h = 1.25). Gaps of this size will cause a
the magnetic circuit, the flux density at the composite sam-
factor of two reduction in the magnetically induced stress.
ple can be calculated as a function of the applied magneto-
From figure 2, simple geometry indicates that adjacent
motive force. On the other hand, the particle polarization
particle chains will touch (y = d) when the particle gap is
is impossible to measure directly and is very difficult to
such that h = π/6φ. And particles will be evenly spaced
compute. In this section, theory is developed on the basis
within the matrix (y = dh) when h3 = π/6φ.
that saturation begins in the particle’s polar regions at a
The yield stress of the particle chains occurs at the strain
very low applied field and increases towards total particle
for which stress is maximum. This is found by setting the
saturation as applied field increases. Average particle polar-
first derivative of stress with respect to quasi-static shear
ization is then computed as a function of average composite
strain to zero:
flux density.
∂σ (4 4 − 27 2 + 4)φJp2 Consider the schematic in figure 3. We, again, isolate a
= =0 (6) single chain of particles from an array of parallel chains. It
∂ 8µ1 µ0 h3 (1 +  2 )9/2
is assumed that uniform saturation occurs in each particle
from which yield strain and corresponding yield stress can in region (s) shown in figure 3. Furthermore, distance
be found: s decreases from r to zero as average composite flux
p √ density increases from zero to some large level. Letters
27 − 665 in parentheses in figure 3 correspond to subscripts used in
y = = 0.389 (7a)
23/2 the following development. The magnetic model is one-
0.1143φJp2 dimensional, therefore vector quantities are expressed as
σy = . (7b) scalar magnitudes.
µ1 µ0 h3
609
M R Jolly et al

more complex than simple linear chains, i.e., columnar


structures with more than two contact points per particle
that may locally be face center cubic or body center
cubic arrangements. Parameter k may also be used to
accommodate for situations where µm 6= 1. Such a situation
may arise when there are particles within the matrix that
are not within a chain. These particles may contribute
little to the desired field dependency of the composite yet
may act to increase the effective permeability of the matrix
Figure 3. Particle saturation model. Quantities are (analogous to the concept of effective dielectric constant
consistent with those defined in figure 2. Letters in
parentheses indicate different regions: (s) region where
in electrorheological fluids discussed by Adriani and Gast
particle is saturated (shaded), (u) region where particle is 1988). The following area ratio can be calculated
unsaturated, (m) matrix region with assumed relative
As π(r 2 − s 2 ) 3 
permeability µm = 1. Letters in parentheses also = 2
= φ 1 − α2 . (14)
correspond to subscripts used in theory development. A y 2
Now, by substituting equations (13) and (14) into (10), the
It is assumed that the flux through the unsaturated quantity α as a function of B can be solved for in terms of
and saturated particle regions is uniformly distributed and known parameters.
remains constant, i.e., 8p = Bp Ap = Bs As = Bu Au . We now seek to develop a relationship between B and
Given that the reluctance of the matrix Rm and the H , the applied field. It is known that
reluctance of the particle Rp are in parallel, the flux

µ0 H = B − Jf (15)
distribution between the two regions can then be expressed
as where hJf i is the average polarization density of the
1 composite. Since it has been assumed that µm = 1, all
8p = Bs As = BA (10)
1 + Rp /Rm polarization is due to the polarization of the particles such
that

where As is the particle area at x = s, and A = y 2 is the
µ0 H = B − φ Jp (16)
total area. It is further assumed that the reluctance of the
particle is the sum of the reluctance of the unsaturated and where hJp i is the average polarization density of the
saturated regions of the particle: particles and may be calculated by
Z s Z r 

1  
Rp = Ru + Rs Jp = J π r 2
−x 2
dx+ J π r 2
−x 2
dx
Z s 2/3π r 3 0
x s
2 1 2 r −s s
= dx + (17)
µp µ0 0 π(r − x )
2 2 µ0 π(r 2 − s 2 ) where Js is the saturation polarization of the particles and
2 r −s Jx is the polarization density at a distance x within the
' (11)
µ0 π(r 2 − s 2 ) unsaturated region of the particle and may be written as

and the reluctance of the matrix is (r 2 − s 2 )


Jx = Bs . (18)
(r 2 − x 2 )
2r
Rm = (12) Since, in equation (11), it has been assumed that the
µ0 (y 2 − πr 2 )
reluctance of the unsaturated portion of the particle is
where the following additional assumptions have been negligible compared to that of the saturated region (i.e.,
made: µp  1 implying that reluctance of the unsaturated all of the field appears across the saturated region), it can
region is negligible; flux lines are perpendicular to the be stated that Bs ≈ Js + µ0 H , which when combined with
secants that define the saturated region; the relative (16) and (17) yields
permeability of the matrix and the saturated portion of the  Z
3/2φ  s 2 
particle are unity, i.e., µm = µs = 1; and for the sake µ0 H = B − 3 Js + µ0 H r − s 2 dx
r 0
of brevity the gap between particles within a chain are Z r 
assumed negligible, i.e., h = 1. Inclusion of h > 1 is not 
+ Js r 2 − x 2 dx . (19)
difficult but would detract from the clarity of exposition s
sought here. Integration of equation (19) with some algebra yields
The ratio of reluctances required by equation (10) can
now be calculated as B − φ(1 − α 3 )Js
µ0 H = (20)
Rp 2/3φ − 1 1 + 3/2φα 3
=k (13)
Rm (1 + α) and the average polarization density (or intrinsic induction)
of the composite is hJf i = B − µ0 H = φhJp i, where, from
where α = s/r, and particle volume fraction φ = 2π r 2 /3y 2 equation (16), the average particle polarization density is:
has been substituted, and k ≥ 1 is a factor that has been

3/2α 3 B + (1 − α 3 )Js
added to accommodate for magnetic interaction between Jp = Jp = . (21)
adjacent particle chains and packing arrangements that are 1 + 3/2φα 3

610
A magnetorheological material behaviour model

Figure 4. Magnetic induction curves for three MR fluids of Figure 5. Experimental yield stress as a function of applied
different iron volume percents. Experimental (thin lines) magnetic field for two MR fluids containing approximately
and model (thick lines) results are shown. 30% iron by volume. Theoretical model results are also
shown for k = 8 (solid) and k = 5 (broken).

3. Magnetorheological fluids
3.2. Magnetorheological effect
3.1. Magnetic properties of MR fluids

The magnetic properties of three MR fluid samples were Experimental evidence has confirmed that MR fluids in
assessed using a magnetic hysteresigraph (KJS Assoc. Inc., the presence of a magnetic field exhibit both a pre-yield
Model HG-500, per ASTM A773-91). The fluid samples regime, characterized by an elastic response, and a post-
contained three different iron loadings: 10%, 20% and 30% yield regime, characterized by a viscous response. The
by volume. The magnetic induction curves (also called the transition between these two regimes has been reported to
B–H curves) of the three samples are shown in figure 4. occur at a yield strain on the order of 0.5% (Weiss et al
Also shown are theoretical model results computed from 1994). Since MR fluids typically operate in a continuous
equation (20). As the theory predicts, magnetic induction yield regime, they are usually characterized by their field
increases with field strength until saturation of the fluid dependent yield stress.
occurs. This can be seen to occur at an intrinsic induction Figure 5 shows published results of yield stress for
of φJs . For example, the MR fluid sample loaded with 30% two MR fluids formulated with approximately 30% (V/V)
iron (which in bulk form saturates at 2.1 T) can be seen carbonyl iron (Lord 1994, Weiss et al 1993). Also
to saturate at about (0.3)(2.1 T) = 0.63 T. A very gradual shown are theoretical model results for two different values
increase in fluid induction is observed after saturation of the of parameter k (from equations (7b) and (21)). The
fluid. This effect—which has been observed during testing model results are seen to be in relatively good agreement
of other MR fluids (Carlson 1996)—is attributed to particle with the experimental data. The figure suggests that the
restructuring over time (Felici et al 1994) and increasing experimentally applied fields were not sufficient to saturate
field. The permeabilities of these fluids were found to be the MR fluid. This is suggested by the trend demonstrated
between approximately 2 µ0 and 6 µ0 when measured at by the model and also by the observation that increases
very low fields. These permeabilities are consistent with in field at higher experimental fields are still resulting in
results of similar experiments conducted by Göktürk et al corresponding substantial increases in yield stress.
(1993). It is anticipated that experimental hysteresigraph results
Parameter k, which was introduced in equation (13), can be used to calibrate the theoretical model. As was
was adjusted to fit the model to the experimental data. done in figure 4, such a calibration entails choosing a
Recall that parameter k is used to account for magnetic k that allows for a good fit between the model and the
interactions other than those between adjacent particles experimental induction curves. The values of k, which
within a chain. The values of k used to fit the model were are weakly dependent upon iron volume fraction, can then
5.0, 4.6, and 4.4 for the 30%, 20% and 10% iron by volume be used with the model for characterization and design
MR fluid samples, respectively. Since these values are purposes. Recall that k = 5 was used in conjunction with
accurate within about 10%, it can be seen that the model, the 30% (V/V) iron MR fluid sample in figure 4. The
when applied to MR fluids, is relatively insensitive to the experimental data in figure 5 can be seen to fall to the right
parameter k. In other words, a value of, say, k = 4.7 would of the theoretical model result for k = 5. It is speculated
provide reasonable model accuracy over a wide range of that this discrepancy is a result of losses in the magnetic
iron volume fractions. circuit used to generate the experimental results.

611
M R Jolly et al

Figure 6. Double lap shear test specimen. The elastomer


segments (shaded) are 20 mm by 7.5 mm by 1.0 mm thick
and are sandwiched between an inner iron plate and two
outer iron plates. The vertical arrow indicates the direction
of applied field.

4. Magnetorheological elastomers

Double lap shear specimens of MR elastomer were prepared


in a mould using silicone oil loaded with a specified volume Figure 7. The response of elastic modulus to applied field
percent of carbonyl iron particles of a 3–4 µm mean for 10% (h = 1.29, k = 5), 20% (h = 1.08, k = 12) and 30%
diameter (see figure 6). This formulation was then cured (h = 1.0, k = 15)) iron by volume elastomer specimens.
in the presence of a magnetic field of 0.8 × 106 A m−1 ◦
Experimental data (×, , M) and theoretical model data
(——) are shown.
applied in the direction shown in figure 6. This field
causes parallel particle chains to form while the silicone
is still liquid. Upon crosslinking (during which polymer
molecules interconnect), this particle structure is locked in Because the maximum change in modulus is independent
place. Three double lap shear specimens of MR elastomer of parameter k, these two parameters (k and h) can be
were prepared containing 10, 20 and 30% iron by volume. adjusted independently to best match the model with the
Testing of the double lap shear specimens was data. It can be seen that an increase in particle volume
conducted using a Dynastatr material testing system. This fraction results in a pronounced increase in k, implying that
system applies a fixed oscillatory strain to the specimen magnetic interaction between neighboring particle chains
and measures the amplitude and phase of the output is increasing. This is consistent with intuition because
force, from which stress and modulus can be calculated. as particle volume fraction increases, the distance between
Testing involved recording the complex modulus of various neighbouring chains decreases. It is also noted that there is
specimens at various frequencies, strains and applied much greater sensitivity to k when the model is applied to
magnetic fields. Magnetic fields were applied to the test MR elastomers as compared to MR fluids. This is attributed
specimens using a horseshoe-shaped electromagnet with to variations in particle structure for the two cases, resulting
poles that efficiently focused magnetic flux across the from the corresponding variations in the level of particle
elastomer segments in the direction shown in figure 6. The mobility in the presence of a field. It was also observed
induction field (in Tesla) across the elastomer segments was that as volume fraction of iron increases, the parameter
experimentally measured using an integrating flux meter h decreases suggesting that gaps between particles within
(Walker Scientific, Model MF-3D) and a sense coil wrapped a chain decreases. It is speculated that at lower volume
directly around the elastomer segments. fractions, particle forces that induce tight, gap-free particle
Figure 7 shows the effect of induced composite flux
chains are lower within the precured elastomer as a result
density on elastic modulus for the three test specimens. As
of an increase in the mean distance between particles.
can be seen, the maximum change in modulus increases
Onset of particle chain yielding in elastomer composites
monotonically with increasing volume percentage of iron.
has been observed in the vicinity of 1–2% strain (Jolly et
While the change in modulus increases to nearly 0.6 MPa
al 1996). As shown in figure 8, this yielding is manifested
as iron loading increases to 30% (V/V), we see from
table 1 that the zero field modulus also increases with by a pronounced drop off in the magnetorheological effect
particle loading. As a result, the percentage of maximum and an increase in field dependent energy dissipation
increase in modulus for the three samples remains relatively (tan δ). The theoretical yield strain has been observed
constant between 30% and 40%. The model results provide to be much higher than experimental evidence indicates.
reasonable predictions for the maximum change in modulus As with MR fluids, an explanation for this discrepancy
obtained at large saturating flux densities for the 20% (V/V) is that the majority of strain and the ultimate yielding of
and 30% (V/V) iron samples. Once again, in using the a given chain occurs between a single (or a few) pairs
present model at intermediate fields, the parameter k, which of particles. In other words, at a local level, it is likely
accounts for magnetic interaction between neighbouring that the theoretical yield strain is achieved, but the strain
particle chains, must be adjusted in order to ‘fit’ the averaged over an entire chain is considerably smaller. This
model to the experimental data. In addition, parameter phenomenon is attributed to imperfect chain structures and
h, which accounts for gaps between particles in a chain, non-uniformity in particle geometry and chain structure.
can be adjusted to match the maximum change in modulus Still, the solids exhibit a yield strain that is several times
predicted by the model to that observed experimentally. higher in magnitude than that of analogous MR fluids.

612
A magnetorheological material behaviour model

Table 1. Enumeration of the nominal absolute modulus (no field applied), the experimental maximum change in modulus for
specimens tested at 1.0% strain and 2.0 Hz, and the maximum change in modulus predicted by the model assuming no gaps
between particles (h = 1).

Nominal Max. change in


modulus, Max. change in modulus,
no field modulus, exper. model, h = 1
Test specimen (MPa) (MPa) (% change) (MPa)
30% (V/V) iron 1.80 0.56 (31%) 0.55
20% (V/V) iron 0.74 0.29 (39%) 0.36
10% (V/V) iron 0.26 0.08 (30%) 0.17


Figure 8. The effect of peak strain on field induced change in modulus ( ) and change in tan δ (M). This data corresponds to
the 30%(V/V) iron specimen measured at 10 Hz and 0.85 T.

5. Conclusions Experimental pre-yield data taken on MR elastomers


was also compared with model results. Once again it was
A model that examines the behaviour of MR materials observed that the model could be calibrated to provide a
has been presented. The model is based on the dipolar good fit with experimental data. The maximum change in
interaction of particles within an assumed structure. This modulus was observed to increase monotonically with iron
dipolar model was then augmented with a proposed loading—nearly 0.6 MPa was observed with an iron loading
mechanism of how magnetic flux density distributes itself of 30% (V/V). However, it was also observed that the
within the particle network. This mechanism takes into zero field modulus increases with particle loading such that
account magnetic non-linearities associated with the gradual the percentage of maximum increase in modulus remained
saturation of the permeable particles as applied field relatively constant between 30% and 40%.
increases. The theory is capable of examining both the Ongoing work on deterministic modelling of MR
magnetic and mechanical properties of MR materials. material behaviour will examine the magnetic properties
The developed theory was compared with experimental of more complex particle structures.
data taken on MR fluids and elastomers. Magnetic and post-
yield properties of MR fluids were examined. In particular,
the model was fit to experimental magnetic induction curves References
by adjusting a parameter that accounts for unmodelled
magnetic interactions between neighboring particle chains. Adriani P M and Gast A P 1988 A microscopic model of
electrorheology Phys. Fluids 31 2757–68
Indications that these magnetic interactions decreased with Anon 1995 Brake cuts exercise-equipment cost Design News
decreasing particle volume fraction were observed. It was Dec. 4
proposed that this fit of the model with magnetic induction Block H and Kelly J P 1988 Electro-rheology J. Phys. D: Appl.
curves be a basis for calibrating the model for various Phys. 21 1661–77
Bowen C P, Shrout T R, Newnham R E and Randall C A 1995
particle loadings. The calibrated model was then observed Tunable electric field processing of composite materials J.
to be in relatively good agreement with published yield Intelligent Mater. Syst. Struct. 6 159–68
stress data for MR fluids. Carlson J D 1991 Private communication

613
M R Jolly et al

Carlson J D 1994 The promise of controllable fluids Proc. Jolly M R, Carlson J D, Munoz B C and Bullions T A 1996 The
Actuator 94 (Bremen, Germany 1994) pp 266–70 magnetoviscoelastic effect of elastomer composites
Carlson J D 1996 Private communication consisting of ferrous particles embedded in a polymer
Carlson J D, Chrzan M J and James F O 1994 matrix J. Intell. Mater. Syst. Struct. submitted
Magnetorheological fluid devices US Patent No. 5,284,330 Kasahara A, Yamada A, Yoshizawa T, Wada K and Yamasaki H
Carlson J D, Catanzarite D M and St Clair K A 1995 1994 Viscoelastic substance and objective lens driving
Commercial magneto-rheological fluid devices 5th Int. apparatus with the same US Patent No. 5,337,865
Conf. ER, MR Suspensions Assoc. Technol. (Sheffield, Lord 1994 Lord Product Information: MR Fluid MRX-100 and
England, 1995) MRX-135CD, Lord Corp., Cary, NC, USA
Chase V D 1996 Cutting edge Appl. Manuf. May p 6 Muñoz B C and Jolly M R 1995 Mechanical response of
Felici N, Foulc J N and Atten P 1994 A conduction model of elastomers to magnetic fields 4th Ann. Workshop: Advances
electrorheological effect Electrorheological Fluids: in Smart Mater. for Aerospace Appl. (NASA Langley
Mechanisms, Properties, Technology and Applications ed R Research Center, 1995), 210th ACS National Meeting
Tao and G D Roy (Singapore: World Scientific) pp 139–52 (Chicago, IL, 1995)
Gast A P and Zukoski C F 1989 Electrorheological fluids as Rigbi Z and Jilken L 1983 The response of an elastomer filled
colliodal suspensions Adv. Colloid Interface Sci. 30 153–202 with soft ferrite to mechanical and magnetic influences J.
Gentry S B, Mazur J F and Blackburn B K 1995 Muscle training Magnetism Magn. Mater. 37 267–76
and physical rehabilitation machine using Rosensweig R E 1985 Ferrohydrodynamics (Cambridge:
electro-rheological magnetic fluid US Patent No. 5,460,585 Cambridge University Press)
Ginder J M and Davis L C 1994 Shear stresses in Spasojevic D, Irvine T F and Afgan N 1974 The effect of a
magnetorheological fluids: role of magnetic saturation Appl. magnetic field on the rheodynamic behaviour of
Phys. Lett. 65 3410–2 ferromagnetic suspensions Int. J. Multiphase Flow 1 607–22
Ginder J M, Davis L C and Elie L D 1995 Rheology of Tang X, Chen Y and Conrad H 1995 Structure and interaction
magnetorheological fluids: models and measurements 5th force in a model magnetorheological system 5th Int. Conf.
Int. Conf. ER, MR Suspensions Assoc. Technol. (Sheffield, ER, MR Suspensions Assoc. Technol. (Sheffield, England,
England, 1995) 1995)
Göktürk H S, Fiske T J and Kalyon D M 1993 Electric and Weiss K D et al 1993 High strength magneto- and
magnetic properties of a thermoplastic elastomer electro-rheological fluids Proc. SAE SAE paper no 932451
incorporated with ferromagnetic powders IEEE Trans. Weiss K D, Carlson J D and Nixon D A 1994 Viscoelastic
Magn. 29 4170–6 properties of magneto- and electro-rheological fluids J.
Harpavat G 1974 Magnetostatic forces on a chain of spherical Intell. Mater. Syst. Struct. 5 772–5
beads in a uniform magnetic field IEEE Trans. Magn. 10
919–22

614