You are on page 1of 12

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/230925280

Volume-Constrained Optimization of Magnetorheological Valves

Article  in  Smart Materials and Structures · September 2004


DOI: 10.1088/0964-1726/13/6/004

CITATIONS READS

88 587

2 authors, including:

Norman M. Wereley
University of Maryland, College Park
529 PUBLICATIONS   11,664 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Norman M. Wereley on 19 June 2014.

The user has requested enhancement of the downloaded file.


INSTITUTE OF PHYSICS PUBLISHING SMART MATERIALS AND STRUCTURES
Smart Mater. Struct. 13 (2004) 1303–1313 PII: S0964-1726(04)86098-1

Volume-constrained optimization of
magnetorheological and
electrorheological valves and dampers
Nicholas C Rosenfeld and Norman M Wereley
Smart Structures Laboratory, Alfred Gessow Rotorcraft Center, Department of Aerospace
Engineering, University of Maryland, College Park, MD 20742, USA

Received 14 July 2004, in final form 14 July 2004


Published 27 September 2004
Online at stacks.iop.org/SMS/13/1303
doi:10.1088/0964-1726/13/6/004

Abstract
This paper presents a case study of magnetorheological (MR) and
electrorheological (ER) valve design within a constrained cylindrical
volume. The primary purpose of this study is to establish general design
guidelines for volume-constrained MR valves. Additionally, this study
compares the performance of volume-constrained MR valves against
similarly constrained ER valves. Starting from basic design guidelines for
an MR valve, a method for constructing candidate volume-constrained valve
geometries is presented. A magnetic FEM program is then used to evaluate
the magnetic properties of the candidate valves. An optimized MR valve is
chosen by evaluating non-dimensional parameters describing the candidate
valves’ damping performance. A derivation of the non-dimensional
damping coefficient for valves with both active and passive volumes is
presented to allow comparison of valves with differing proportions of active
and passive volumes. The performance of the optimized MR valve is then
compared to that of a geometrically similar ER valve using both analytical
and numerical techniques. An analytical equation relating the damping
performances of geometrically similar MR and ER valves in as a function of
fluid yield stresses and relative active fluid volume, and numerical
calculations are provided to calculate each valve’s damping performance
and to validate the analytical calculations.
(Some figures in this article are in colour only in the electronic version)

Nomenclature Fon On-state damping force


hc Coil height
A1 Bobbin core cross-sectional area Iap Applied current
A2 Flux return cross-sectional area L Total valve length
A3 Flange interior area LA Active length
Ad Fluid annulus cross-sectional area (=bd) LP Passive length
Ap Valve face area (=π R 2 ) Q Volume flux through valve
Bi Bingham number Q1, Q2, Q3 Volume flux through volumes 1, 2, 3
b Mean annular circumference R Valve radius
C Newtonian damping Rw Coil resistance
Ceq Equivalent damping ta Bobbin core radius
c Yield stress parameter tb Flange height
d Annular gap width V Total annular fluid volume
F Damping force v0 Mean fluid velocity though annulus
Foff Off-state (Newtonian) damping force vp Piston velocity

0964-1726/04/061303+11$30.00 © 2004 IOP Publishing Ltd Printed in the UK 1303


N C Rosenfeld and N M Wereley

wc Coil width
γ̇ Shear rate
p Total pressure drop across valve
p1 , p2 , p3 Pressure drop across volumes 1, 2, 3
pµ Pressure drop due to viscosity
across valve
pτ Pressure drop due to yield stress
across valve
δ̄ Non-dimensional plug thickness
µ Viscosity
τ Shear stress
τy Yield shear stress
(•)ER Electrorheological parameter
(•)MR Magnetorheological parameter
(•)max Maximum value

1. Introduction

Over the past decade, the use of magnetorheological (MR)


and electrorheological (ER) fluids as the fluid component in
smart damping devices has been given much consideration.
The effectiveness of smart MR and ER dampers has been
demonstrated for a variety of applications including wheeled Figure 1. MR and ER valve comparison.
and tracked land vehicles [1, 2], helicopter rotor systems [3–
5], seismic hazard mitigation [6, 7], and vibration isolation
of the valve, the addition of the magnetic coils will cause an
systems [8]. MR and ER fluids are similar in that they develop
MR valve to have less active fluid volume than a similarly
a controllable yield stress in the presence of magnetic and
constrained ER valve. This paper will present a comparison
electric fields, respectively. However, differences in fluid
of MR and ER valves placed under such a constraint.
properties and actuation requirements of ER and MR fluids
In order to present an unbiased comparison of MR and ER
warrant comparison between the two. Physically, MR fluids
valves, two valves with the same geometric configuration were
can produce yield stresses on the order of ten times greater
constructed. A volume-constrained MR valve optimization
magnitude than ER fluids. A general comparison of MR and
routine was used to set the geometry of the MR valve, while
ER fluids by Carlson [9] demonstrated that this difference
the corresponding ER valve was defined by replacing the
in yield stress will allow an MR device to provide the same
MR valve’s wire coils and flux return with tubular metallic
pressure response as a significantly larger ER device.
electrodes (figure 1). The MR valve was chosen as the initial
Recent comparative studies have dealt, wholly or in part,
design because MR valves’ applied field characteristics were
with the actuation properties of smart dampers, though this
sensitive to bobbin and flux return geometry and material
field is still largely unexplored. Gavin et al have demonstrated
selection [15], whereas ER valves’ applied field actuation
that the power requirement for a switching MR damper is
characteristics were dependent only on the valve gap. The
lower than that for an equivalent switching ER damper [10].
MR and ER valves were used to perform both analytical and
This is particularly true when the ER fluid is subjected to high
temperature or operating current, which will cause dielectric numerical comparisons between the two systems.
breakdown of the fluid. Choi and Wereley have shown
that ER fluids have faster non-dimensional time response 2. MR valve geometric design
characteristics, though MR fluids’ response characteristics are
favorably comparable [11]. Choi et al’s comparison of similar MR and ER fluid valves operate on the same basic principle as
PID-controlled ER and MR clutches showed that the MR clutch a mechanical valve: a high-pressure fluid is forced through a
produced a significantly higher controllable torque, while the small duct, resulting in a drop in pressure of the outgoing fluid.
ER clutch had more favorable response characteristics for The pressure drop across a valve is caused by energy loss in
torque regulation and tracking tasks [12]. the fluid due to plasticity and viscosity. When MR or ER
However, the nature of the actuation methods utilizing ER fluid is activated within the duct, the development of the yield
and MR fluids necessitates other comparisons from a design stress binds a volume of fluid into a solid plug. The remaining
standpoint. A controllable MR valve requires a controllable unbound fluid then experiences greater shear forces, which in
magnetic flux across the active fluid, which is often produced turn results in a greater pressure drop across the valve. MR and
with coils of magnet wire. Controllable ER valves, meanwhile, ER fluids act not only as the fluid component of the system, but
need a controllable electric field across the active fluid, also as a semi-active control element. MR and ER valves are
achieved by electrifying an electrode on one side of the valve attractive because they allow controllable performance with
and grounding the opposite electrode. In many cases, an MR no moving parts. In the context of this study, the MR and ER
valve’s magnetic coils are housed within the valve, adjacent to valves are used as dashpots to provide a usable, controllable
the annulus [13, 14]. If constraints are placed on the volume damping force.

1304
Volume-constrained optimization of magnetorheological and electrorheological valves and dampers

Figure 3. MR valve geometry nomenclature.


Figure 2. Magnetic flux through an MR valve.

volume that MR effects occur. In order to make the valve most


The framework of the optimization procedure was based effective, it was desirable to have a high magnetic flux density
on physical design needs. A damper was considered with a passing through a large active volume. However, producing
fixed cylindrical volume allotted for a valve. It was assumed large magnetic fields required large numbers of magnetic coils.
that the off-state damping requirement had been fixed (e.g., For a valve with a constrained volume, more volume devoted
by safety requirements for minimum damping), and that the to magnetic coils translated directly to a smaller active volume.
damper should achieve the greatest possible on-state damping. Moreover, more volume devoted to the coils left less volume
The optimization procedure was therefore designed to find a for the magnetically permeable carrier materials, which in turn
geometry which maximized the damping coefficient Ceq /C, meant that the valve could saturate at high fields, sapping
the ratio of the magnitude of on-state damping to off-state effectiveness [15]. An optimized circuit would maintain a
damping. In all analyses, quasi-steady operation is assumed. balance between the field produced and power required by the
Typically, ER and MR valves have only one annular gap magnetic coils, and a valve design that would make best use
through which fluid is forced (as drawn in figure 1), but this of the field to activate the MR fluid yield stress.
is not a strict constraint. Analyses of the damping properties For a simplified valve configuration, candidate geometries
of flow through single rectangular [16] and annular ducts [17] were defined as a function of the size of the magnetic coils.
are well known for ER and MR fluids. Configurations with A viable candidate geometry was one in which the various
multiple, concentric annular gaps have been suggested for critical areas though which the magnetic field passes were
both ER [18] and MR [19] valves. A disk-type valve with the same size. This was necessary to keep the magnetic flux
active flow through a radial volume has been suggested for density constant throughout the circuit, which ensured that one
MR valves [13, 20], and such a configuration could be easily region of the magnetic circuit did not saturate prematurely and
extended for use in ER valves. In general, for any particular ER cause a bottleneck effect. There are three critical areas in the
valve configuration, an analogous MR valve configuration can magnetic circuit: the circular cross-section of the bobbin core
be constructed, and vice versa. Although this paper focuses A1 , the annular cross-sectional area of the flux return A2 , and
on a rectangular duct configuration (as an approximation to the cylindrical area at the interior of the bobbin flanges A3 . The
an annular duct), the conclusions reached from this simple geometry of the magnetic circuit was defined by the following
example will be generalizable to any configuration. This will dimensions: radius R, length L, bobbin core radius ta , bobbin
be strictly true for comparisons between similar configurations flange height tb , gap width d, coil width wc , and coil height h c
of ER and MR dampers. Comparisons between differing ER (figure 3). Thus, the critical areas were defined by
and MR configurations would need to take the differing total
fluid volumes of the ER valve and MR valve into account. A1 = πta2 (1)
The MR valve was constrained within a cylindrical volume  
A2 = π R 2 − (ta + wc + d)2 (2)
of fixed height and diameter. Within this volume, candidate
geometries were constructed on the basis of the number of A3 = 2πta tb . (3)
wire coils desired in the valve. These candidate geometries The volume constraint on the circuit was specified by
were imported into an ANSYS magnetic finite element analysis prescribing maximum values for R and L. Small changes
(FEA) routine and the magnetic flux densities through the MR in the valve gap, d, would drastically alter the performance of
valve were calculated. The flux values in the active regions different valves, so a fixed gap was also prescribed to ensure
were then compared with the specific MR properties of Lord an unbiased evaluation. For these constraints, the optimized
MRF-132AD fluid, from which damping coefficient and non- geometry of the valve could be calculated algebraically.
dimensional plug thickness were calculated for each valve. Setting (1) equal to (2) and rearranging into the quadratic form
These performance indicators were used to evaluate which for positive ta yielded
candidate design gave the best overall performance.   
The MR valve was shaped to guide the magnetic flux ta = 12 −(wc + d) + 2R 2 − (wc + d)2 . (4)
axially through the bobbin, across the bobbin flange and fluid
gap at one end, through the flux return, and across the fluid Since R and d were prescribed, (4) allowed ta to be solved for
gap and bobbin flange again at the opposite end (figure 2). a given coil width wc . Setting (1) equal to (3) gave
The volume of fluid though which the magnetic field passes
was defined as the active volume; it is only within this active tb = 12 ta . (5)

1305
N C Rosenfeld and N M Wereley

Table 1. Geometries of candidate valves.


Critical Active
Wraps L A (mm) Total coils area (mm2 ) volume (mm3 )
8 11.3 128 401 1131
10 10.7 170 360 1100
12 10.1 228 320 1066
14 9.42 280 279 1014
16 8.73 336 239 959
18 8.03 414 202 899
20 7.3 480 167 831
22 6.55 572 135 757

Since the magnetic field crossed the MR fluid along a length tb


at either end of the valve, the sum of these lengths was defined
as the active length, L A :

L A = 2tb . (6)
Figure 4. Comparison of magnetic permeability of Hiperco and
Finally, the coil height was calculated from silicon steel.

h c = L − 2tb . (7) model itself consisted of three volume-constrained circuits


stacked axially, with MR fluid filling the fluid gap, and free
The procedure for constructing a candidate valve geometry
space (air) at either end of the circuit. The bobbin and
was as follows: for a valve with fixed values of R, L, and d,
flux return materials were both modeled as high-magnetic-
a coil width wc was chosen. From the preceding equations,
permeability Hiperco alloy and moderately permeable silicon
ta and tb were calculated. Lastly, the coil height h c was
steel. The permeability curves of both there materials were
calculated. It was noted that wc was the only variable necessary
plotted in figure 4. The MR fluid was modeled using the fluid
to characterize the geometry of the valve. As a rule, as wc
properties of Lord MRF-132AD fluid [21]. The wire coils
increased, the size of the magnetic coils increased, the size of
were assumed to have free-space permeability. Following
the critical areas decreased, and the size of the active volume of
a passive actuation method, the model circuit was actuated
MR fluid decreased. That is, as the width of the coils increased,
using a specified current input. A per-wire current density was
the size of the coils, and the magnitude of the magnetic flux that
distributed over the corresponding coil area in the model; the
they created, increased. At the same time, the cross-sectional
middle coil was given a negative current density to simulate
area of the bobbin decreased, causing the magnetic flux density
a reverse-wound coil. The coils were actuated with input
of the circuit to increase. The volume of active MR fluid
amperages in the range of 0.5–2.5 A; the upper value was
decreased, meaning that less fluid in the valve was subject
chosen as a reasonable maximum current that can be safely
to the MR effect. A valve was sought that would balance the
carried through 24-gauge copper wire.
magnitude of the flux density with the volume of active MR
A circuit’s MR performance was calculated as follows.
fluid in order to achieve optimal performance.
The FEM analysis of the circuit was used to calculate the
In practice, the width and height of the wire coils could
distributed magnetic flux density over the active volume. The
not be varied continuously; these dimensions were integer
distributed flux density was averaged across the thickness and
multiples of the diameters of the individual wire strands from
length of the annulus to calculate an average flux density,
which they are made. In this analysis, all wires were sized as
which is used in all subsequent analyses. Though recent
24-gauge (diameter = 0.516 mm). For convenience, the width
findings have shown that the distribution of flux density
of the coil was expressed as the number of circumferential wire
across the gap thickness can have significant effects on
layers, or wraps; for example, a ‘12-wrap’ coil had a coil width
damper performance [22], these non-uniform effects were
equal to 12 wire diameters. When calculating the height of the
neglected in order to simplify analysis. The average flux
coils, the largest multiple of the wire diameter which did not
density in the active volume was compared with an empirical
violate the prescribed valve length constraint was chosen.
characterization of Lord MRF-132AD fluid to determine an
The geometry optimization was conducted with volume
average yield shear stress τy for the fluid. The yield stresses
constraints of L = 20 mm and R = 20 mm, and a prescribed
for the test cases ranged from 8.54 kPa for an 8-wrap silicon
gap width d = 1 mm. Valve geometries were calculated for
steel bobbin powered at 0.5 A to 41.7 kPa for a 14-wrap
8–22 wraps. Table 1 gives a comparison of relevant geometries
Hiperco bobbin powered at 2.5 A. The yield stress for each
for each valve.
case was used to create a Bingham plastic model of the fluid,
characterized by the equation
3. MR valve magnetic optimization
τ = τy sgn(γ̇ ) + µγ̇ (8)
The candidate geometries were imported into an FEM routine
to evaluate their static magnetic performance. A two- where τ is the shear stress of the fluid, γ̇ is shear rate, and
dimensional model of the magnetic valve was created for µ is fluid viscosity. For simplicity, viscosity was chosen to
axisymmetric analysis in the ANSYS FEM program. The be constant, µ = 0.33 Pa s. This assumption allowed the

1306
Volume-constrained optimization of magnetorheological and electrorheological valves and dampers

the central passive volume was defined as volume 2 (figure 5).


The lengths of volumes 1 and 3 were each half the active
length L A , and the length of volume 2 was defined as the
passive length L P . All volumes were assumed to have the same
annular cross-section area Ad . The volume flux for region 2
was approximated from Newtonian analysis for a rectangular
duct as
Ad d 2 p2
Figure 5. Volumes for damping coefficient calculation. Q2 = . (12)
12µL P
calculation of a characteristic Bingham number, Bi , for the The volume fluxes through volumes 1 and 3 were approximated
valve [16]: as that for a Bingham plastic flux through a rectangular
τy duct [16]:
Bi = v0 (9)
µd  
Ad d 2 p1  2 1
where Q1 =  1 1 − δ̄ 1 + δ̄ (13)
Ap 12µ 2 L A 2
v0 = vp . (10)
Ad  
Ad d 2 p3  2 1
Here Ap is the area of the piston face (i.e., π R ), Ad is the
2 Q3 =  1 1 − δ̄ 1 + δ̄ . (14)
12µ 2 L A 2
cross-sectional area of the fluid annulus, v0 is the mean fluid
velocity though the fluid annulus, and vp is the velocity of the In all cases p is the pressure drop along the length of each
piston. Ap and Ad were set by the valves’ geometries and the respective volume. As an aside, note that Ad = bd for the
piston velocity was chosen to be constant vp = 0.05 m s−1 . rectangular duct approximation, so Q i ∝ d 3 —this strong
The Bingham number was then used to determine two non- relationship between volume flux and gap was the impetus for
dimensional parameters, non-dimensional plug thickness and prescribing equal gaps for all candidate geometries. Continuity
damping coefficient, to allow comparison of the valves. Non- dictated that the volume flux through each volume was the
dimensional plug thickness expressed the percentage of the gap same. Furthermore, the volume flux was the product of the
width which behaved as a solid plug due to the MR or ER effect. cross-sectional area of the piston and the velocity at which the
This parameter was related to the activation of the MR or ER piston displaced fluid. Therefore,
fluid itself; larger values indicate stronger induced MR or ER
effects. Non-dimensional plug thickness, δ̄, was calculated by Q 1 = Q 2 = Q 3 = Q = Ap vp . (15)
solving for the root of the following equation [16]:
Substituting (15) into (12)–(14) and solving for the individual
  pressure drops yielded
1 3 3 6
δ̄ − + δ̄ + 1 = 0 0  δ̄  1. (11)
2 2 Bi 12µL A 1
p1 = Ap vp  2  (16)
2Ad d 2 1 − δ̄ 1 + 12 δ̄
4. Damping coefficient derivation
12µL P
Damping coefficient expresses the gain in the equivalent p2 = Ap vp (17)
Ad d 2
damping of an active MR valve compared to the zero-field
(Newtonian) condition. Most analyses of damping coefficient 12µL A 1
p3 = Ap vp  2  . (18)
consider the gain in the active region of the valve only. This is 2Ad d 2 1 − δ̄ 1 + 12 δ̄
valid for ER valves, where ER effects can be activated in the
entire fluid volume. Often, this characterization is extended The total pressure drop for the valve was the sum of the
to MR valves under the assumption that the passive volume of individual pressure drops; furthermore, the total pressure drop
the value has a much wider gap than the active volume, so that was equal to the damping force F over the piston area:
the passive volume’s contribution may be neglected for both

3
F
‘off’ and ‘on’ states. For the simplified geometries considered p = pi = . (19)
in this study, such an assumption does not hold. Moreover, the i=1
Ap
emphasis of this study on optimizing valves with respect to a
constrained total volume suggested that the performance of the Solving (19) for damping force gave

valves be non-dimensionalized with respect to their common 12µL P A2P 12µL A A2P 1
total length as opposed to their varying active lengths. These F= +  2  vp
Ad d 2 Ad d 2 1 − δ̄ 1 + 12 δ̄
issues were addressed by deriving an expression for damping
coefficients which included the effects of the passive volume. = Ceq vp . (20)
This expression is characterized by the ratio of active length
Noting that the sum of L A and L P is the total valve length L,
L A to total length L. For ER valves this ratio is equal to unity,
the equivalent damping Ceq was simplified as
while MR valves have ratios less than unity.
In order to derive the damping coefficient, three volumes 
12µL A2p LA 1
of the valve are identified: the active volumes at either end of Ceq = 1+  2  − 1 . (21)
the valve were defined as volumes 1 and 3, respectively, and Ad d 2 L 1 − δ̄ 1 + 1 δ̄ 2

1307
N C Rosenfeld and N M Wereley

(a) (b)

Figure 6. Non-dimensional plug thickness versus power density as a function of the number of wraps in the MR valve. (a) Hiperco and (b)
silicon steel.

The term outside of the brackets in (21) is the Newtonian (off- where Iap is applied current, and Rw is the calculated wire
state) damping C for the valve. Dividing by this term gave the resistance. Note that as wrap number was increased, Rw
damping coefficient increased and both Ad and L A decreased. Therefore, in cases
where applied current was the same for different valves, valves
Ceq LA 1 with higher wrap numbers had much greater power densities.
=1+  2  −1 . (22) The results of the analysis were plotted in figures 6 and 7.
C L 1 − δ̄ 1 + 1 δ̄ 2 The non-dimensional plug thickness for the Hiperco valves
Examining the limiting cases, as the active length approached was shown in figure 6(a), where each curve represented the
zero the damping coefficient approached unity: performance of an individual valve as applied power was
increased. Taken individually, each valve showed an increase
Ceq in plug thickness as more power was applied. Looking at
lim =1 (23) the maximum achievable plug thickness for all valves, it
→0 C
LA
L
was seen that the maximum plug thickness increased with
which was to say that the valve provided only Newtonian wraps for 8 to 12 wraps, peaked at 14 to 16 wraps, and
damping. Similarly, as active length approached total length slowly decreased with wraps larger than 18. The low-wrap
the equivalent damping coefficient approached trend indicated poor performance due to a deficit in actuation
power—fewer magnetic coils produced a smaller magnetic
Ceq 1 field with which to activate MR effects. The high-wrap
lim = 2  (24) trend, conversely, indicated poor performance due to too much
LA
L
→1 C 1 − δ̄ 1 + 21 δ̄
actuation power; specifically, the larger coils and smaller
which was the result for Bingham plastic flux though a critical areas led to a magnetic flux density that saturated the
fully active rectangular duct [16]. These results were both magnetic permeability of the carrier material, which in turn
physically expected, further validating the analysis. led to decreased performance. Figure 6(b) shows the same
values for the silicon steel valves, which had a lower magnetic
permeability compared to the Hiperco valves. The same trends
5. Optimization results and analysis as were seen for the Hiperco valve were also evident here,
but the high-wrap decrease in maximum plug thickness was
Non-dimensional plug thickness and damping coefficient were more pronounced. This difference provided more evidence
compared as a function of power density for each valve. Power that magnetic flux saturation was the limiting factor for large
density was defined as the electric power applied to the coils wrap numbers.
per unit active volume of fluid. The electric power applied to The damping coefficient for the Hiperco valves was
the coils was calculated as follows. The length of wire in the plotted in figure 7(a). The damping coefficient trends were
coils was approximated by assuming each radial and axial coil qualitatively similar to the plug thickness trends: individual
was a circular loop of wire and summing the circumferences valves showed increased performance with increased applied
of all the circular loops. The resistance of the wire was then power, and the maximum performance of the valves first
calculated using the empirical resistance-per-unit-length value increased, then decreased, with increasing number of wraps.
of 17.2 /100 m for 24-gauge copper wire. Power density However, valuable insight into the importance of valve
was then calculated from geometry was gained by comparing the damping coefficients
with their respective plug thicknesses. The high-wrap drop-off
2
Iap Rw of the damping coefficient was more pronounced and began at
Power density = (25)
Ad L A a lower wrap number compared to the similar trend in plug

1308
Volume-constrained optimization of magnetorheological and electrorheological valves and dampers

(a) (b)

Figure 7. Damping coefficient versus power density as a function of the number of wraps in the MR valve. (a) Hiperco and (b) silicon steel.

damping coefficient for the Hiperco and silicon steel optimized


valves were plotted in figure 9.
This MR design analysis showed that striking the correct
balance between available actuation power and available
actuation volume was key to optimizing the performance
of a volume-constrained MR valve. In general, designing
a volume-constrained valve to obtain only maximum active
fluid volume (small wrap number valves) or only maximum
actuation power (large wrap number valves) will not result in
an optimal design. Furthermore, characterization parameters
must be chosen which illustrate these optimization strategies.
The damping coefficient contained a term which reflected the
effect of the active length on the overall valve performance.
Had this contribution been neglected by considering the
damping ratio of the active regions only (i.e., normalizing the
Figure 8. Maximum damping coefficients and plug thicknesses results over the active volume), the design procedure would
versus candidate geometry. have resulted in a non-optimal valve configuration.

thickness. This indicated that the decrease in the active length 6. ER/MR valve comparison
associated with high wrap numbers was quite significant to the
overall performance of the valve. Specific examples provided Once the 12-wrap MR valve was chosen as the optimal valve
more evidence of this effect. The 14-, 16-, and 18-wrap valves geometry, it was constructive to compare the MR valve with
had essentially the same maximum plug thickness, but their a similar ER valve. ER and MR fluids, while qualitatively
damping coefficient decreased significantly as the number of similar in their development of a field-controllable yield stress,
wraps increased and the active length decreased. Furthermore, differ in actuation methods and physical fluid properties.
the 12-wrap valve, which had a smaller maximum plug While MR fluids are actuated by guiding a magnetic field
thickness but a larger active length than the 14–18-wrap valves, perpendicular to the fluid flow direction, ER fluids are actuated
had the largest damping coefficient of all test cases. Also, the by applying an electric field perpendicular to the fluid flow
8-wrap valve, which had the lowest maximum plug thickness direction. In practice, this is achieved by applying a voltage to
but the longest active length, outperformed the 18-, 20-, and the inner tubular electrode of the valve and grounding the outer
22-wrap valves in terms of maximum damping coefficient. The tubular electrode of the valve. Unlike MR valves, which have
damping coefficients for the silicon steel valves were plotted a passive interior region, ER fluid is activated along the entire
in figure 7(b), and again, similar trends were seen with a more length of the valve. However, this larger active volume is offset
pronounced high-wrap drop-off compared to Hiperco. by ER fluid having a lower maximum yield stress compared to
To provide clarity, the maximum damping coefficients and MR fluid.
plug thicknesses for all candidate valves were plotted with In order to compare ER and MR fluids, separate analytical
respect to wrap number in figure 8. The optimized valve and numerical analyses were performed. The analytical
was chosen as the candidate geometry that gave the greatest analysis sought a simple expression for comparing general ER
damping coefficient. Coincidentally, for both the Hiperco and and MR damping coefficients for geometrically similar valves
silicon steel valves, the 12-wrap valve is the optimized valve in terms of the yield stresses and active volumes of the fluids.
geometry. Plots of the non-dimensional plug thickness and The numerical analysis was used to produce results specific

1309
N C Rosenfeld and N M Wereley

(a) (b)

Figure 9. The influence of magnetic materials on optimal MR design. (a) Non-dimensional plug thickness, (b) damping coefficient.

Table 2. Properties of generic ER/MR fluids. pτ /pη > ∼100) [9]. For an active length that is not equal
ER MR to the total valve length, (27) can be rewritten as
 
τy,max (kPa) 5 50 L A cτy L
pτ = . (28)
µ (Pa s) 0.33 0.33 L d
L A /L 1 0.505
An expression for total fluid volume V of the valve was
found by manipulation of equations (26) and (28). Using the
to the optimal valve and also to validate the analytical model. rectangular duct approximation,
Generic ER and MR fluids were used for these calculations;
V = Lbd 
table 2 shows the relevant valve geometries and fluid properties  2  
for each case. The yield stresses chosen for the ER and MR 12 µ L pτ
= 2 Qpτ . (29)
fluids are the maximum and minimum values, respectively, c τy2 LA pµ
from the analysis performed by Carlson et al [9] (i.e., this
The damping force of the valve in the off-state (Newtonian)
analysis presents the most conservative performance mismatch
condition equals the viscous pressure drop distributed across
between ER and MR fluids; any other values chosen would
the piston face area, which also equals damping times piston
increase the disparity between the valves). In order to simplify
velocity:
analysis, the fluid viscosities µ of the two fluids were assumed
Foff = pµ Ap = Cvp . (30)
to have the same value as that used in the MR optimization.
All geometric properties of the MR and ER valves were the Solving (30) for viscous pressure drop gave
same with the exception of the ratio of active length L A to
vp
valve length L. pµ = C . (31)
Previously, Carlson et al showed that ER dampers would Ap
require an active fluid volume two to three orders of magnitude
Likewise, for the on-state condition, the damping force equals
larger than that of an MR damper to achieve a desired control
the viscous plus controlled pressure drops distributed across the
ratio pτ /pµ at a given flow rate Q and controlled pressure
piston face, which is also equal to equivalent damping times
drop pτ [9]. Starting from the same initial equations,
piston velocity:
a similar analysis was performed to compare the damping
coefficients of MR and ER dampers with equal valve fluid 
Fon = pµ + pτ Ap = Ceq vp . (32)
volumes. From Bingham plastic analysis, the pressure drops
across each valve were divided into two components, a field- Substituting in (31) allows (32) to be solved for controlled
independent viscous component pµ and a field-induced yield pressure drop as
stress component pτ . Approximating the annular valve as a
rectangular duct,  vp
pτ = Ceq − C . (33)
12µQ L Ap
pµ = (26)
bd 3
Substituting (31), (33), and the definition of flow rate from (15)
cτy L into (29) yields
pτ = (27)
d     2
where c is a parameter with a minimum value of 2 12 µ L 2 Ceq
V = 2 − 1 Cvp2 . (34)
(for pτ /pη < ∼1) and a maximum value of 3 (for c τy2 LA C

1310
Volume-constrained optimization of magnetorheological and electrorheological valves and dampers

This is the total fluid volume necessary to achieve a desired


damping coefficient Ceq /C for a set off-state damping C at
piston velocity vp . For an MR device, the total fluid volume is
  2   2
12 µMR L Ceq
VMR = 2 −1 CMR vp2 .
c τy,MR
2 LA MR C MR
(35)
For an ER device which activates fluid across the entire length
of the valve, the total fluid volume is
    2
12 µER Ceq
VER = 2 − 1 CER vp2 . (36)
c τy,ER
2 C ER

Dividing (35) by (36) and setting the total volumes of fluid


equal yields
  2   2
µMR Ceq
Figure 10. MR/ER valve comparison at vp = 0.05 m s−1 .
τy,MR
2
L
LA C
− 1 CMR
1= 
MR
  
MR
2 . (37)
µER Ceq
τy,ER
2 C
− 1 CER For the chosen generic fluids, MR fluid’s maximum yield stress
ER
is ten times greater than ER fluid’s maximum yield stress. By
ER and MR fluids can be assumed to have equal viscosities [9], equation (41), we see that an MR valve with an active length
and by setting the off-state damping of the devices equal, (37) larger than 1/10th the total valve length can achieve a greater
can be rearranged as damping coefficient than an ER valve of the same geometry
  2 operating at the same flow rate. The optimum active length in
Ceq
−1     this study’s example is approximately half of the active length,
C τy,MR 2 L A 2 so the optimized MR damper will outperform the geometrically
 
MR
2 = . (38)
Ceq τy,ER L MR similar ER damper in all cases.
C
− 1
ER The analytical analysis demonstrated the effects of yield
Taking the square root of both sides while noting that, stress and active length on the relative controllable damping
physically, damping coefficients must be equal to or greater of MR and ER valves, but it did not allow calculation of the
actual damping performance of the individual valves. In order
than unity, (38) is rewritten as
to accomplish this calculation, a numerical analysis of each
  valve was performed to allow direct comparison of the valves’
− 1 τ  
Ceq
C y,MR LA exact performance.
 MR = . (39)
Ceq
−1 τy,ER L MR For numerical analysis, a piston velocity of vp =
C ER 0.05 m s−1 was chosen and the Bingham number for each
fluid was calculated from (9) for a range of yield stresses
It can be shown that the term in the numerator and denominator
up to τy,max . The maximum Bingham numbers for these
of the left-hand side, (Ceq /C)−1, is exactly equal to the control
conditions were Bi ER,max = 25 and Bi MR,max = 254. Recall
ratio pτ /pµ as defined by Carlson. This value expresses
that these values represent the least possible disparity between
the ratio of a valve’s controllable effects to its passive effects.
the general ER and MR fluids. Noting that non-dimensional
Equation (39) shows that geometrically similar MR and ER
plug thickness could be expressed as a function of Bingham
dampers having the same passive damping performance will
number, the damping coefficient calculation (22) for each valve
have a ratio of controllable performance that is directly
was simplified as
proportional to the ratio of the yield stresses and the ratio of
the MR valve’s active length to its total length. Note that Ceq LA
for damping coefficients (Ceq /C)  1—i.e., for conditions =1+ [ f (Bi ) − 1] (42)
C L
where total damping is approximately equal to controllable
damping—(39) approaches where the damping coefficient for a fully active duct, as given
by (24), was approximated by the rational function [23]
   
Ceq    Ceq 1
C τy,MR LA = 2 
 MR = . (40) C L A =1 1 − δ̄ 1 + 12 δ̄
Ceq τy,ER L MR L
C ER = f (Bi )
Examining the relationships in (39) and (40), we see that the ∼ 5.17Bi
= 1.019 + 0.172Bi + . (43)
MR damping coefficient will be greater than the ER damping 129 + Bi
coefficient for the condition The damping coefficient was plotted in figure 10 as a function
    of Bingham number. The active length ratio dictated the slope
τy,MR L of the plot for each valve. The ER valve, having a larger
> . (41)
τy,ER L A MR active length ratio, had a steeper slope; for a given Bingham

1311
N C Rosenfeld and N M Wereley

geometries. An MR valve will outperform a similar ER valve


for an easily attainable case where the ratio of maximum MR
yield stress to ER yield stress is greater than the ratio of total
fluid volume to active MR fluid volume. This fact, along with
MR fluid’s comparatively greater operating temperature range,
lower power requirements, and comparable time response
characteristics as compared to ER fluid, makes MR fluid a
more attractive choice for use in damping applications.
As an end note, the comparison between differing volume-
constrained ER and MR configurations can be revisited in the
light of this study’s findings. It has been concluded that, for the
same fluid volume, an MR valve will allow greater controllable
damping than an ER valve. Expanding this idea, it is clear
that increasing the fluid volume within the given constrained
volume will allow for higher damping performance. A multi-
annular or disk-type valve affords more fluid volume than a
single-annulus valve. A recent study of multi-annular ER
Figure 11. MR/ER valve comparison at τy,MR = τy,MR,max and
τy,ER = τy,ER,max . dampers [24] (i.e., relatively large total fluid volumes) suggests
that such configurations can achieve damping performance
Table 3. MR/ER damping coefficient comparison. comparable to that of a similar-sized, single-annulus MR
Numerical Analytical Numerical
damper [25] (i.e., relatively small total fluid volume). This,
vp (m s−1 ) (Ceq /C)ER (Ceq /C)MR (Ceq /C)MR MR/ER ratio in turn, suggests that a multi-annular or disk-type MR valve
would be able to give still higher performance. Thus, one can
0.05 6.25 24.8 27.5 3.97
tentatively conclude that MR dampers will provide superior
0.1 3.67 13.3 14.5 3.62
0.5 1.56 3.65 3.83 2.34 damping in any cross-configuration comparison.

7. Conclusions
number, the ER valve provided a greater damping coefficient
than the MR valve. However, the larger achievable yield stress This study investigated a means of optimizing MR and ER
of MR fluid resulted in a greater achievable Bingham number, smart valves and dampers in cases where the volume available
which in turn allowed the MR valve to attain a larger maximum for the valve was constrained. A procedure for developing
damping coefficient. Operating at maximum yield stress, the simple candidate valve geometries within the constrained
volume-constrained MR valve achieved a maximum damping volume was presented. Magnetic FEM analysis was used
coefficient (Ceq /C)MR,max = 24.8, nearly four times greater to calculate the performance of each candidate valve and
the ER valve’s (Ceq /C)ER,max = 6.25. two parameters, non-dimensional plug thickness and damping
A second numerical comparison was conducted by setting coefficient, were used to characterize and compare the different
the yield stress to τy,max and varying piston velocity to illustrate geometries. The 12-wrap geometry was chosen as the optimal
the maximum achievable damping coefficients for piston MR valve. This optimal MR valve was then compared with
velocities of vp = 0.05, 0.1, and 0.5 m s−1 . This analysis a geometrically similar ER valve through both analytical and
was plotted in figure 11; note that the range of Bingham numerical calculations. The major conclusions of this study
numbers was the same as for the previous analysis. As piston were:
velocity increased, the Bingham number decreased. For all
cases, the MR valve produced a higher damping coefficient (1) The performance of an MR valve is highly dependent on
than the ER valve at the same operating condition. These both actuation power and active fluid volume, so a volume-
results were then used to validate the analytical analysis. For constrained valve must balance these two factors. In the
each numerically calculated ER damping coefficient, (39) was optimization procedure, the 8-wrap geometry (i.e., large
used to calculate the corresponding MR damping coefficient. active volume) had a maximum damping coefficient of
The analytically calculated MR damping coefficients match ∼19.5 while the 22-wrap geometry (i.e., large actuation
well with the numerically calculated MR coefficients. Table 3 power) had a maximum damping coefficient of ∼14. The
shows a comparison of all calculated values. In the limiting optimum 12-wrap geometry (balanced active volume and
case as piston velocity approached infinity, the MR and ER actuation power) had a damping coefficient of ∼22.5.
valves would both converge to a damping ratio of unity. (2) A given MR damper will have a larger damping coefficient
This analysis demonstrated that for the given constrained than a geometrically similar ER damper in the case where
volume, the optimized MR valve showed significant the ratio of the achievable maximum MR yield stress to
performance benefits over the geometrically similar ER valve maximum ER yield stress is greater than the ratio of the
in all conditions. While the ER valve had a greater active total length of the valves to the active length of the MR
volume of fluid, the greater yield stress of the MR fluid valve. This study’s conservative maximum yield stress
translated directly into a larger achievable damping coefficient. ratio of 10 is larger than the optimized total-to-active
Moreover, the analysis provided a criterion for comparing length ratio of ∼2, indicating that the MR damper will
the performances of general MR and ER valves with similar have the greater damping coefficient.

1312
Volume-constrained optimization of magnetorheological and electrorheological valves and dampers

(3) The superior performance of the MR damper over a Improved Seismic Performance in Urban Regions (Seattle,
geometrically similar ER damper was validated through WA, Aug. 2001) pp 225–36
[11] Choi Y T and Wereley N M 2002 Comparative analysis of the
numerical analysis. Calculations of damping coefficient
time response of electrorheological and magnetorheological
at vp = 0.5 m s−1 gave an MR maximum damping dampers using nondimensional parameters J. Intell. Mater.
coefficient of 3.65, over twice the ER one’s maximum Syst. Struct. 13 443–52
damping coefficient of 1.56. A slower speed of vp = [12] Choi S B, Hong S R, Cheong C C and Park Y K 1999
0.05 m s−1 gave an MR maximum damping coefficient Comparison of field-controlled characteristics between
of 24.8, nearly four times greater than the ER one’s 6.25 ER and MR clutches J. Intell. Mater. Syst. Struct.
10 615–9
maximum damping coefficient.
[13] Carlson J D and Chrzan M J 1992 Magnetorheological fluid
This optimized MR valve provided a greater range of dampers US Patent Specification 5,277,281
[14] Gordaninejad F and Breese D G 1997 Magneto-rheological
controllable damping than a geometrically similar ER valve.
fluid damper US Patent Specification 6,019,201
This result, along with comparable or superior range of [15] Bölter R and Janocha H 1997 Design rules of MR fluid
operating conditions, power requirements, and response actuators in different working modes Proc. SPIE Smart
characteristics as compared to the ER valve, makes MR Structures and Materials 1997: Passive Damping and
technology more attractive for volume-constrained conditions. Isolation vol 3045 (Bellingham, WA: SPIE Optical
Engineering Press) pp 148–59
[16] Wereley N M and Pang L 1997 Nondimensional analysis of
References semi-active electrorheological and magnetorheological
dampers using approximate parallel plate models Smart
[1] Gordaninejad F and Kelso S P 2000 Fail-safe Mater. Struct. 7 732–43
magneto-rheological fluid dampers for off-highway, [17] Kamath G M, Hurt M K and Wereley N M 1995 Analysis and
high-payload vehicles J. Intell. Mater. Syst. Struct. 11 testing of Bingham plastic behaviour in semi-active
395–406 electrorheological fluid dampers Smart Mater. Struct. 5
[2] Park D W, Choi S B, Suh M S and Shin M J 2001 ER 576–90
suspension units for vibration control of a tracked vehicle [18] Stanway R, Sproston J L and El-Wahed A K 1996
Proc. SPIE 4327 159–64 Applications of electro-rheological fluids in vibration
[3] Kamath G M, Wereley N M and Jolly M R 1999 control: a survey Smart Mater. Struct. 5 464–82
Characterization of magnetorheological helicopter lag [19] Namuduri C S, Alexandridis A A, Madak J and Rule D S 2001
dampers J. Am. Helicopter Soc. 44 234–48 Magnetorheological fluid damper with multiple annular
[4] Gandhi F, Wang K W and Xia L 2001 Magnetorheological flow gaps US Patent Specification 6,279,701
fluid damper feedback linearization control for helicopter [20] Dogruer U, Gordaninejad F and Evrensel C A 2003 A new
rotor application Smart Mater. Struct. 10 96–103 magnetorheological fluid dampers for high-mobility
[5] Zhao C, Choi Y T and Wereley N M 2004 Semi-active multi-purpose wheeled vehicle (MNNWV) Proc. SPIE
damping of ground resonance in helicopters using Smart Structures and Materials 2002: Damping and
magnetorheological dampers J. Am. Helicopter Soc. 49 Isolation vol 5052 (Bellingham, WA: SPIE Optical
at press Engineering Press) pp 198–206
[6] Dyke S J, Spencer B F Jr, Sain M K and Carlson J D 1997 An [21] Lord Corporation 2003 MR Fluid Product Bulletins Online
experimental study of MR dampers for seismic protection http://www.rheonetic.com/fluid begin.htm
Smart Mater. Struct. 7 693–703 [22] Gavin H 2001 Annular Poiseuille flow of electrorheological
[7] Hiemenz G, Choi Y T and Wereley N M 2003 Seismic control and magnetorheological fluids J. Rheol. 45 983–94
of civil structures utilizing semi-active MR braces J. [23] Hu W and Wereley N 2003 Nondimensional equivalent
Comput. Aided Civil Infrastruct. Eng. 18 31–44 damping analysis of flow-mode magnetorheological and
[8] Choi Y T, Wereley N M and Jeon Y-S 2002 Semi-active electrorheological dampers Proc. ASME Int. Mechanical
vibration isolation using magnetorheological isolators Proc. Engineering Congress and Exposition (Washington, DC,
SPIE Smart Structures and Materials 2002: Damping and Nov. 2003) ASME paper no IMECE2003-43135
Isolation vol 4697 pp 284–91; AIAA J. Aircr. at press [24] Gavin H P 1998 Design method for high force
[9] Carlson J D, Catanzarite D M and St Clair K A 1996 electrorheological dampers Smart Mater. Struct. 7 664–73
Commercial magnetorheological fluid devices Int. J. Mod. [25] Carlson J D and Spencer B F Jr 1996 Magneto-rheological
Phys. B 10 2857–65 fluid dampers for semi-active seismic control Proc. 3rd Int.
[10] Gavin H, Hoagg J and Dobossy M 2001 Optimal design of MR Conf. on Motion and Vibration Control (Chiba, Japan,
dampers Proc. US Japan Workshop on Smart Structures for 1996) vol 3, pp 35–40

1313

View publication stats

You might also like