HI

1
4

 

ES: \lll‘klllllIWIHWIIWHH4Mll!lH1WlH|Wl

 

THS

    

IIIIIII IIIIIIIIIIIIIII

"9‘8!" uemnss

IIIIII III/III

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the

thesis entitled

FEEDBACK CONTROL OF
ELECTRORHEOLOGICAL FLUIDS

presented by

RUTH MARIE ANDERSLAND

has been accepted towards fulﬁllment
of the requirements for

MASTERS degree in MECHANICAL ENGINEERING

 

Major professor

Date 3/

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

PLACE It RETURN BOX to man this chockout from your record.
To AVOID FINES Mom on or More data duo.

DATE DUE DATE DUE DATE DUE

       

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU to An motiv- ActionlEqull Opportunity Intuition
W

ans-9.!

FEEDBACK CONTROL OF ELECTRORHEOLOGICAL FLUIDS
By
Ruth Marie Andersland

A THESIS

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Mechanical Engineering

1995

ABSTRACT
FEEDBACK CONTROL OF ELECTRORHEOLOGICAL FLUIDS
By
Ruth Marie Andersland

Electrorheological (ER) ﬂuids have electrically controllable stiffness, viscosity, and
heat transfer properties. Since the 19405 researchers have attempted to model the
properties of ER ﬂuids and have proposed applications which attempt to utilize their
special characteristics in the operation of hydraulic valves, soft clutches, and active
suspension systems. Early attempts to make these applications commercially successful
were hampered by the relatively slow, nonlinear response of these ﬂuids to the
application of an electric ﬁeld. These problems in response were attributed to the
inﬂuence of particle volume concentration and moisture content, and to ER ﬂuid
temperature. In contrast, successful applications will require fast, precise control of the
response of ER ﬂuids, independent of those properties.

This study presents a new approach to the control of ER ﬂuids that overcomes the
problems of imprecise, slow, nonlinear response. A feedback-based approach to the
control of ER ﬂuid response was developed and compared to the conventional feed-
forward control approach. A sensor was used to indicate the ER ﬂuid state in a layered
composite window. Feedback control employs the sensor and high initial electric ﬁeld
strength to speed ER state response, then lowers the ﬁeld strength to the level required to
achieve the desired ER ﬂuid state. Predicted responses were compared to experimentally
measured responses and showed excellent agreement. In fact, it was demonstrated for the
ﬁrst time that the proportional feedback control system responded 35 times faster and 21

times more accurately than the feed-forward system.

ACKNOWLEDGMENTS

I would like to express my gratitude for the thoughtful and patient guidance of my
advisor, Dr. Clark J. Radcliffe, throughout my graduate program. The assistance of my
master's committee members, Dr. Philip M. FitzSimons, and especially Dr. John L. Lloyd
for the use of his lab, and his patience and conﬁdence in me, are gratefully
acknowledged. Their insightful comments were particularly helpful in placing my work
in perspective.

Thanks are also due to Jeff Hargrove, for his long hours and extensive assistance in
the lab. Sachin Gogate, Brooks Byam, and Jerry Palazzolo deserve special thanks for
answering my endless questions. Thanks can also be given to my dear friends who have
listened and offered support.

This research was funded by the State of Michigan Research Excellance Fund,
administered through the Composite Materials and Structures Center at Michigan State
University.

Finally, this is to my families on both sides of the oceans. A special thank you to
my parents, Orlando and Phyllis Andersland, and brothers, Mark and John, for their

patience, encouragement and dry shoulders.

iii

TABLE OF CONTENTS

LIST OF TABLES .......................................................................................................... v
LIST OF FIGURES ............. p ........................................................................................... vi
NOMENCLATURE ........................................................................................................ vii
INTRODUCTION .......................................................................................................... 1
FEED-FORWARD AND FEEDBACK CONTROL SYSTEMS COMPARED ........... 6
PROTOTYPE CONTROLLED ER FLUID TEST SYSTEM ........................................ .l 1
ER Fluid State Sensor ......................................................................................... 11
ER Fluid Window ............................................................................................... 12
ER Fluid Controller ............................................................................................. 14
System Integration of Equipment ....................................................................... 15
ER Fluid State Tests, Experimental Procedure ................................................... 16
Measured Feed-forward Control Response ......................................................... 18
Measured Feedback Control Response ............................................................... 22
ER State Feedback Model Validation ................................................................. 23
CONCLUSIONS ............................................................................................................. 26
APPENDD( A ................................................................................................................. 27
APPENDD( B ................................................................................................................. 30
REFERENCES ................................................................................................................ 33

iv

LIST OF TABLES

Table 1: Comparison Between Computed and Measured Gains and Time
Constants. With increasing Kp, gains apprached one, and the time
constants decreased. ......................................................................................... 23

Figure 1:

Figure 2:
Figure 3:
Figure 4:

Figure 5:
Figure 6:

Figure 7:
Figure 8:
Figure 9:
Figure 10:
Figure 11:

Figure 12:

Figure 13:

Figure 14:

Figure 15:

LIST OF FIGURES

Particle polarization and single sphere width chain formation with
increasing ﬁeld, E. Particle chaining changes ﬂuid properties. ................... 1
Feed-Forward Control System ..................................................................... 6
Feedback Control System ............................................................................. 7
Laboratory measurement of zeolite ER ﬂuid response using a

prototype optical sensor feed-forward control, pulse train zeolite—ﬂuid
(1% Vol. Fraction, Dry) (after Tabatabai, 1993). ......................................... 8

Experiment Schematic .................................................................................. 11

Conceptual design of an optical ER ﬂuid state sensor when light
penetrates the ER ﬂuid state. El-Light penetrating dispersed particles;

E5-Light penetrating chained particles. ........................................................ 11
ER Fluid State Sensor and Window Setup ................................................... 12
Composite Window Schematic .................................................................... 13
Laboratory Conﬁguration ............................................................................. 15
Experiment Sequence ................................................................................... l6

Feed-forward response of the ER ﬂuid state and ﬁeld. Relatively slow
response by the ER ﬂuid state. ...................................................................... 18

Feedback response of Kp=0.5, of the desired ER ﬂuid state, ER ﬂuid

state, and ﬁeld. The ER ﬂuid state changed only negligibly when the

ﬁeld was lowered. Removal of the ﬁeld forced the ER ﬂuid state to

fall. The temperature change was negligible. .............................................. 19

Feedback response of Kp=5.0, of the desired ER ﬂuid state, ER ﬂuid
state, and ﬁeld. A large positive and negative change in the ﬁeld

forced the ER ﬂuid state to the desired level. ............................................... 21
Enlarged section from Figure 12, feedback response Kp=5.0, ER ﬂuid
state = 0.57. Oscillations due to the noise in the ﬁeld, E. ............................ 24

Frequency response of the predicted gain range of the analytical

model. Shaded area indicates predicted ER ﬂuid state response range.

The ellipse indicates the actual measurable frequency response due to

noise. ............................................................................................................. 25

vi

Figure A1: Feed-forward n'se response to nominal drive voltage of 0.30kV/mm,
displaying current, ﬁeld, temperature, and ER ﬂuid state. Temperature
decreased to a steady state value. ............................................................

Figure A2: Feedback response with Kp=5.0 of current, ER ﬂuid state, ﬁeld, and
temperature. The ER ﬂuid state tracked the changes in ﬁeld strength.
The current exhibited similar behavior to changes in the ﬁeld strength.

Figure A3: Feedback response with Kp=0.5 of current, desired ER ﬂuid state, ER
ﬂuid state, ﬁeld, and temperature. The ﬁeld was lowered, ER ﬂuid
state changes were negligible, until the ﬁeld was removed. The
temperature changes were negligible. .....................................................

Figure A4: Feed-forward rise response to nominal drive voltage 0.30kV/mm, of
current, ER ﬂuid state, ﬁeld, and temperature. .......................................

Figure A5: Feed- forward response of current, ER ﬂuid state, ﬁeld, and
temperature. The ER ﬂuid state was negligibly effected by changes 1n
the 1:1eld, which readily effected the current. The temperature
remained relatively steady. .....................................................................

 

Figure B 1: Front panel and conﬁguration section of the control program. .............

Figure B2: Data reading, control capabilities, and output sections of the control
program. ..................................................................................................

Figure B3: Clear section of the control program. ....................................................

vii

...... 27

...... 28

...... 28

...... 29

...... 29
...... 31

...... 31
...... 32

NOMENCLATURE

Arabic Symbols

C = ER ﬂuid state

C fb = ER ﬂuid state of the feedback system
Cﬁr = ER ﬂuid state of the feed-forward system
C N = output due to input N

CR = output due to input R

D = controller

E = error

G = system or plant (ER ﬂuid)

H = ER ﬂuid state sensor

k = system gain

Keg“ = equivalent gain K

K p = feedback proportional gain

N = external system disturbance
R = desired ER ﬂuid state

s = Laplace transfer variable

T = transfer function

U = input electrical ﬁeld

VIM“): actual measured voltage level
Vdesired= desired voltage level

Vo = nominal drive voltage

W = nominal drive voltage

Y = output from controller

viii

Greek Symbols

AC
A

T

Tequ

ER ﬂuid state signal
wavelength

system time constant

equivalent time constant

ix

INTRODUCTION

As early as the 19th century (Konig 1885; Duff 1896; Quinke 1897), scientists
began studying electrorheological (ER) response, though it was not until research by
Willis M. Winslow that electro-viscous phenomena gained prominent attention. He
introduced the concept of controlling the viscosity of an electro—viscous ﬂuid by the use
of an electric ﬁeld (Winslow 1947, 1949). Flow resistance of these ﬂuids increased with
ﬁeld strength when exposed to AC electric ﬁelds on the order of 4kV/mm. He observed a
ﬁbrous structure composed of particle chains generally aligned with the applied electric
ﬁeld. Winslow hypothesized that the ﬁeld increased the viscosity of the ﬂuid.

An ER ﬂuid consists of ﬁne polarizable particles suspended in a ﬂuid of lower
dielectric constant. The ﬂuid consists of a continuous hydrophobic liquid phase which

contains hydrophilic particles. It is desirable that the density of the particle should be

 

 

 

 

 

 

Figure 1: Particle polarization and single sphere width chain formation with increasing
ﬁeld, E. Particle chaining changes ﬂuid properties.

1

2
matched as closely as possible with that of the oil to ensure good dispersion upon mixing

of the ER ﬂuid (Stangroom 1978, 1983). In the presence of a high voltage electric ﬁeld,
particle chains are formed due to polarization changing ﬂuid properties (Figure 1).
Higher ﬁeld strength causes the chains to pull together tighter and to increase in length
(Klingenberg et al 1989). At very high ﬁeld strengths (e.g. 2000 V/mm), the lowest
energy state for the chains, ER ﬂuid develops into a body centered cubic (Figure l-E6)
(Tao and Sun 1991a, 1991b). Arcing will occur at higher ﬁeld levels. When the chains
are subjected to a shearing force, the particles still attract even though they may be pulled
away from each other (Duclos et al 1988).

Engineers and scientists have identiﬁed possible applications that would utilize the
special properties of ER ﬂuids. The sensitivity of properties of ER ﬂuids to an electric
ﬁeld means that microprocessors can be used to control dynamic mechanical systems
whose response is determined by that of the ER ﬂuid. The systems would be connected
directly to computers in order to control response in a rapid and precise manner. Possible
applications include vehicle suspensions, hydraulic valves and soft clutches.

Development of commercial applications of devices using ER ﬂuids has been
hampered by the fact that (Butters 1995) science and industry have not been able to
quickly and precisely control the ER ﬂuid state to capture the full potential of the ER
controllable properties. Electrorheological ﬂuid applications have not responded
accurately, which suggests that a new control strategy is necessary for successful
commercialization to occur.

Previous studies have focused on varying essential aspects of ER ﬂuids including
ER effect, preparation of an ER ﬂuid, particle temperature range, yield strengths, shear
stress, and the control of systems harnessing ER ﬂuid properties. Despite some progress
in these areas of research, precise control of the variable ER ﬂuid has eluded
investigators. In this investigation we studied the basic problem of control of the state

and the properties of an ER ﬂuid so that all applications could beneﬁt from the results.

3
Many previous studies focused on the development of new ﬂuid types. The basic

ER ﬂuid consists of a suspension of hydrophilic particles in a hydrophobic ﬂuid
(Winslow 1962). Polymeric materials have been extensively investigated for use as
particles and are often chosen for enhanced viscosity performance and speciﬁc
mechanical applications (Stangroom 1977, 1978, 1980, 1984; Block and Kelly 1986).
Environmentally safe ﬂuids that more readily transform from a Newtonian material to a
Bingham plastic have also been examined (Stangroom 1982; Block and Kelly 1986).
Filisko and Armstrong (1988) were able to prepare particles at very high temperatures
(>100 °C @ atmospheric pressure) without releasing water. The resulting ER ﬂuid is
particularly useful in automotive and other high temperature applications.

Other individuals turned their attention to modeling the ER phenomena. A
simulation method was developed to describe structure formation in electrorheological
suspensions (Klingenberg et a1 1989). Tao and Sun (1991a, 1991b) examined the ground
state and the various phases that exist in the ER ﬂuid. The dynamic stress-strain behavior
of an ER ﬂuidwas investigated by Yen and Achorn (1991). Properties were determined
for an ER ﬂuid consisting of 20% vol. zeolite particles, and a model was proposed to
explain the mechanical response in terms of the dielectric mismatch between particles,
carrier ﬂuid, and ﬁeld (Conrad et al 1991).

The issue of control of response in ER ﬂuid applications has continued to perplex
engineers. ER ﬂuids of high volume concentration are usually considered fast to respond
to the application of a ﬁeld. Low particle volume concentrations are typically thought to
be slow. However, even high particle volume fraction ER ﬂuids may be too slow in some
applications where very fast response is necessary. In both cases the precision of
response may be a concern.

Devices such as ER ﬂuid based valves, clutches or hydraulic mounts typically do
not react quickly enough to meet needs of the application even with their enhanced

performance over those of conventional oil ﬂuids. These devices use high particle

4
volume concentrations, but the physical problem requires faster responses (Duclos 1987;

Ushijima et al 1988, Arguelles et al 1973). Stangroom (1983) recognized the importance
that feedback would add to device control system applications. The ER ﬂuid responded
quickly, but the intended control was of the mechanical device rather than of the ﬂuid.
On the other hand, Lloyd and Zhang, (1994) and Zhang and Lloyd, (1992a, 1992b) used
low particle volume ﬂuids to control transport of thermal energy by a feed-forward
control method and found very slow ER ﬂuid response. They reported that response time
was several minutes as compared to fractions of seconds with high particle volume
concentrations. In either way fast, precise response was not possible; this fact prevented
success of application. Control of the ER ﬂuid response controls the success of the
application.

The control level within the ER ﬂuid application is important. Three types of ER
ﬂuid control are possible: ER ﬂuid device, ER ﬂuid properties, and ER ﬂuid state. The
response of ER ﬂuid devices is dependent on the properties of the ER ﬂuid contained in
them. The properties of the ER ﬂuid are dependent on the chaining state of the ER ﬂuid.
The chaining state is dependent on the applied electrical ﬁeld. Past work has
concentrated on device and ﬂuid property control (Arguelles et al 1973; Stangroom 1983;
Duclos 1987; Ushijima et al 1988; Zhang and Lloyd 1992a, 1992b; Lloyd and Zhang
1994). The following work focuses on control of the ER ﬂuid chaining state which alters
ER ﬂuid properties effecting successful device control.

In this study, a comparison is made between feedback and feed-forward control of
low particle volume concentration ER ﬂuid state. Analytical models for the ER ﬂuid and
control systems were developed which predict ER ﬂuid state responses to the application
of a varied DC ﬁeld drive and were compared to experimentally measured responses.
Greater control of the accuracy and speed of ER ﬂuid response was the objective.

Effective and efﬁcient control of an ER ﬂuid is necessary in order to achieve the beneﬁts

5
of polarized ER ﬂuid particles. This will lead directly to commercially successful

applications, which were here-to-fore not possible.

F EED-F ORWARD AND FEEDBACK CONTROL SYSTEMS COMPARED

There are two traditional methods for controlling an ER ﬂuid state response: feed-
forward, and feedback. In feed-forward control, sometimes called "open-loop" control,
the response of a system is controlled by manipulating system input based on knowledge
of the relationship between system input and system output. In feedback control,
sometimes called "closed-loop" control, the desired system output is compared to the
measured output and the difference is used to drive the controlled system. This use of
feedback allows more precise control of system response in the presence of noise and
variations in controlled system behavior than is possible with feed-forward methods.

Successful feed-forward control (Figure 2) requires invariant system behavior and
limited noise to precisely control system response. Though the ER ﬂuid system is
nonlinear, it is assumed the system can be modeled linearly. To achieve a desired output,
an input value is selected using knowledge of the relationship between input and output,
G(s). Feed-forward control design involves inversion of that relationship, D(s)=1/G(s).

CR“) = D(s)G(s)R(s) = G-1(s)G(s)R(s) (1)
The accuracy of the control system is deﬁned by the accuracy of the known relationship
between input and output. The presence of any external disturbance disrupts the
controlled system's accuracy.
C~(S) = G(S)N(S) (2)

Here, the disturbance is modeled as an additional system input. The total system output.

External N(s)
Disturbance¢(Field, Stress, etc.)

_ Control .
Desired Output Controller Action, Y(s) Controlled System Response
(ER Fluid State) (ER Flurd) (ER Fluid State)

Figure 2: Feed-Forward Control System

 

 

 

 

 

 

 

 

 

, Error Response
Desmiezdgutput + E(s gontroéled _ I‘ER Fluid State)
5 =
(ER Fluid State) - ‘ ( ) p CRIS)

 

 

 

Solar Cell 4%
Sensor H(s

 

 

 

 

Figure 3: Feedback Control System

C ﬁv(s) , is the sum of the outputs due to the control (1) and external disturbances (2).
C1345) = G(S)[D(S)R(S) + N(S)!

= R(s) iﬁD(s) = G'l(s) and N(s) = o (3)
Accurate feed-forward control requires both exact knowledge of the controlled system,
G(s), and minimal disturbance, N (s). It is apparent that any changes from the nominal
values of either the controller D(s) or the controlled system, G(s), (Figure 2) for the feed-
forward system will cause proportional errors in the response, C If ( s ).

Feedback control uses measured system output to drive the control system (Figure

3). The actuating error signal is the difference between desired and measured system
output signals. The objective of feed-back control is a design, D(s), which systematically

reduces the actuating error. Proportional control, Kp, is the simplest form of feedback

design between the controller output, Y(s), and the actuating error signal, E(s).

D(s)=Kp—zﬂ

— 15(3) (4)

System error, E(s), for the feedback controlled system is driven by desired output, R(s),

and measured system output, Cfb(s).

E(3)=R(S)-H(S)Cﬂy(5) (5)
The nominal input, W(s), maintains the controlled system at its nominal output value
when system error, E(s), becomes zero. External noise, N(s), nominal input, W(s), and

control action, Y(s), add to form the control system input.

8
Cﬂ,(s)=G(s)[N(s)+W(s)+KpE(s)] (6)
Combining (5) and (6) yields an expression for feedback controlled system output, Cfb(S).
G(s)[W(s)+ KpR(s)+ N(s)]
1+KPG(s)H(s) (7)
=R(s)zﬂKp —>oo and H(s)=1

 

Cfb(5)=

Accurate feedback control requires both accurate gain k, and feedback sensing, H(s).
The primary objective of implementing feedback is to reduce the sensitivity of the
system to parameter variations and disturbances. Applications of feedback to, G(s), of
ER ﬂuids is ideal since it is inherently non-linear, ﬂuid, and hysteresial. The hysteretic,
' non-linear nature of ER ﬂuids (Figure 4) has been previously observed (T abatabai 1993)
at low particle concentrations. Feedback reduces sensitivity to the system by using a high
quality sensor and implementing high gain. I
Feed-forward and feedback control systems are both advantageous for particular,
but different situations. The feed-forward control system (3) is ideal when external
disturbances, N (s), are small and when there is an exact knowledge of the controlled
system response, G(s), because the expense of an external sensor is not required. Past

work has clearly shown that accurate models of ER ﬂuid state are unavailable and

 

 

   

 

 

 

  
   

 

 

 

 

 

“1600' 10.15 5
314001;“), ' - - . § 10.13;
“11200? ....... E 1' .0311 :71-

. : ': ‘D
§>1000‘g—T ’0.09 E:
3.” I; . W
g 800.5 .007g
.5 600 5 0.05 a.
E 400‘s ’0.03 E
2 203‘; . . , . , ’0.01 §
. - - - - '- A 2 ° - o "

0 430 860 1290 1720 2150 2580 3010 "

Time (s)

Figure 4: Laboratory measurement of zeolite ER ﬂuid response using a prototype
Optical sensor feed-forward control, pulse train zeolite-ﬂuid ( 1% Vol.
Fraction, Dry) (after Tabatabai, 1993).

9
elimination of noise is difﬁcult. The feedback control system (7) is ideal when a high

quality sensor, H(s), is available and the expense of implementing high gain Kp, is
justiﬁed. Even when the ER ﬂuid state response model changes dramatically with time,
the feedback controlled ER ﬂuid state responds accurately.

The ER ﬂuid state response can be modeled as a ﬁrst order system where C(s) is the
output ER ﬂuid state, U(s), is the input electrical ﬁeld, k is the system gain, and r is the

system time constant. The ER ﬂuid state gain k, and the associated time constant 1' have

been observed to vary widely (Figure 4).

G(s)= k -&)-

_ j 8
1+t's U(s) ( )

 

Assuming the availability of sensor, H ( s ) 1t 0, and substituting (8) into (7),
Cﬂ,(s)_ ka/(1+rs)

 

 

 

T(s)= _
R(s) 1+[H(s)ka/(1+rs)] 9
()
= ka/(1+H(s)ka) = kequ
1+[r/(1+H(s)ka)]s 1+Tequ5
1'
where tequ = -) 0 as Kp -> oo
1+H(s)ka
kK

1
k = P -> as K —> oo
“I“ 1+H(s)ka H(s) F

For a large Kp, reg“ approaches zero, and the system will have a faster response time.

 

 

A response time is considered fast if it is signiﬁcantly shorter than the previous response
times. For example, if Kp is increased by an order of magnitude and the response time
decreases from 11 minutes to 1 minute, that would be a fast response time with increasing
Kp. Similarly, an increasing Kp forces the steady state error to become small. Steady
state error occurs when the output of a system's steady state does not exactly agree with
the input. Steady state error could be caused by any number of reasons, such as
hysteresis, ampliﬁer drift, or aging or deterioration of system components. Model

predictions of reg“ approaching zero and kequ approaching one with increasing K ,

were demonstrated in the experiment.

10
The effects of the sensor, H(s), are minimized with an increasing Kp, as long as

H(s) remains non—zero. The sensor has a greater inﬂuence on the gain keg“. By

increasing Kp, keg“ = l/H(s) , the inverse of the sensor transfer function. The system

requires a high quality sensor with near unity gain for precise feedback response. The
sensor transfer function, if constant, reduces steady state error because increasing Kp
decreases the steady state error.

The accuracy and speed of feedback control are determined by the response
measurements of kequ and reg" respectively. Feed-forward fails to provide the accuracy
and speed necessary for ER ﬂuid state response systems with variable behavior (Figure 4)
and inevitable external disturbances. In fact, the feed-forward system will cause
proportional errors in the response to changes in the system, G(s), (Figure 2). Feedback
control retains more precise control of system response in the presence of noise and
variation in system behavior. By design, feedback systematically reduces the actuating
error (Figure 3). Transient feed-forward system measurements will determine both the
ER ﬂuid state gain k and time constant, 1'. These calculated values are used as the basis

for the model predicted values of kequ and reg“ with feedback to be compared with

measured feedback response.

PROTOTYPE CONTROLLED ER FLUID TEST SYSTEM
An ER ﬂuid prototype control system was designed and tested using feed-forward
and feedback control of the ER ﬂuid state reSponse. Measurements for accuracy and
speed of feedback control of an ER ﬂuid state response were compared with model
predictions. The experiment (Figure 5) consisted of 4 distinct features: an ER ﬂuid state

sensor, ER ﬂuid window, an ampliﬁer, and a computer controller.

~. ~ ’ Field

 

 

 

 

 

 

 

 

Controller , . High V/mm ER Fluid L Volts
, (Feed-fonlvard . Ampliﬁer V . d I"
I or Feedback) '_ ‘ (0-4000V) (Wm ow)
Nominal
_, Drive Voltage
0.30011kV/
TT Solar Cells
(Sensor)

 

Figure 5: Experiment Schematic

ER Fluid State Sensor

To implement an ER ﬂuid state sensor it must detect and respond to small or large
changes in the ER ﬂuid state in a measurable way. A sensor must be reliable, its results
reproducible. Light will pass through a thin layer of ER ﬂuid having low particle

concentrations. Particle chaining in the ER ﬂuid (Figure 6) permits greater passage of

LlGl-ﬂ'SOUFlCE

 

Figure 6: Conceptual design of an optical ER ﬂuid state sensor when light penetrates the
ER ﬂuid state. El-Light penetrating dispersed particles; E5-Light penetrating
chained particles.

11

12

Reference Solar Cell B .
ER Fl (1
Windgl'lv Detector Solar Cell A

Laser +9» Lm!-> —>'

Beam Splitter

 

 

 

 

 

 

Figure 7: ER Fluid State Sensor and Window Setup

light, so that the extent of chaining can be indirectly measured by the concentration of
light reaching a detector. When little or no electric ﬁeld is applied, the particles are
random and dispersed, (Figure 6-E1) and only a small amount of light passes through the
window. With increased ﬁeld (Figure 6—E5), the particles form parallel chains and the
ER ﬂuid state response allows more light to pass through the window. Two difﬁculties in
choosing an appropriate sensor and light source are the extended time of the experiment,
and the effects of small changes in the surrounding room light on the sensor. A steady
light source capable of functioning for the length of the experiment is essential.

The sensor design consisted of two Archer silicon solar cells and an Edmund
Scientiﬁc beam splitter. A M1650 Toshiba laser diode (70:650nm) provided the light
source. Detector solar cell A, used in coordination with reference solar cell B, permitted
evaluation of the true transmittance regardless of the light beam steadiness (Figure 7). A
ﬁfty-ﬁfty light split to the two solar cells was provided by the beam splitter. A gain of
1000 and 100 was used on solar cells, A and B respectively. This was to compensate for
the nominal transmittance of the window glass. The ratio (A/B) of the ampliﬁed signals
from the detector, A, and reference, B, determined the transmittance T, of the ER ﬂuid
state. The transmittance permitted the determination of the level of chaining in the ER

ﬂuid state.

ER Fluid Window
The ER ﬂuid window must be capable of holding ﬂuid, applying an even ﬁeld

across the window and allowing transmission of light. A well sealed, electric conductive

13

Plated glass windows with electric Sulfur-Cured Styrene Butadiene Rubber
conductive intium tin oxide Insulator

  

   

Silicon Rubber Insulators

Aluminum Holders

Figure 8: Composite Window Schematic

plated clear glass composite window meets these requirements (Figure 8). The seal is
important, to keep the ER ﬂuid in place and to last the length of the experiment. The ER
ﬂuid window was-also insulated, had replaceable sections and permitted the placement of
a type T thermocouple to monitor ﬂuid temperature. Insulating materials require good
surface and bulk dielectric strength and resistivity, as well as chemical and mechanical
stability (Stangroom 1983). The sections permitted easy cleaning and replacement of
damaged layers. The window, two rectangular glass spectrophotometer cells (38mm x
19mm), was insulated from the aluminum frame by silicon rubber and sulfur-cured
styrene butadiene rubber sections. The glass spectrophotometer cells were plated with
electrically conductive intium tin oxide on one side. The conductive coating permitted an
even spread of electric ﬁeld across the windows. To the naked eye the thin coating
looked like an oil across the transparent glass. The coating's thickness was negligible. A
small gap was left at the top of the center, sulfur-cured styrene butadiene rubber insulator,
for insertion of the dispersed ER ﬂuid. Once the window had been set in place, the

thermocouple was placed between one side of the insulation and the glass cell to monitor

l4
temperature. Since several of the characteristic ﬂuid parameters are temperature

dependent (Stangroom 1978), especially the current, temperature was monitored.
Electrorheological ﬂuid was prepared by letting anhydrous crystalline zeolite
particles adsorb water molecules. A moisture content of approximately 17.1% was
measured by therrnogravimetric analysis. This moisture content is consistent with
preparations used by Winslow (1962). The average particle diameter was less than 10
microns of the agglomerate particle size, and the average crystal size was 1.0 to 4.0
microns (UOP, 1990). Particles (speciﬁc density 1.1 kkg/m3) were mixed to a weight
fraction of 3% with phenylmenthyl polysiloxane silicon oil (speciﬁc density 1.11
kkg/m3). At low temperatures the intentional presence of adsorbed water is critical
(Winslow 1962; Stangroom 1977, 1978, 1984) and mechanistically necessary in
achieving the desired changes in viscosity during the presence of an applied electric ﬁeld
(Filisko and Armstrong 1988). To insure good random dispersion of particles, a magnetic
stirrer was used for a period of 5-10 minutes. The solution was then left undisturbed until
the majority of bubbles created from the stirring had risen to the surface. The ER ﬂuid

could then be inserted by syringe into the window arrangement (Figure 8).

ER Fluid Controller

The ER ﬂuid controller was capable of performing both feed-forward and feedback
control designs. The feedback controller was designed for proportional control and
produced a change in output voltage proportional to the measured error in ER ﬂuid state.
User inputs included desired state, nominal drive voltage, and control gain Kp.

The nominal drive voltage provided a standard base above zero to start the two
control approaches. This allowed comparison of the increase or decrease in change of
ﬁeld strength due to the desired ER ﬂuid state. The 0.30kV/mm nominal value was
chosen only because it was a low non-zero voltage that still provided an upper range of

change in electric ﬁeld within equipment safety speciﬁcations. The controller was

15
written in the virtual instrument program language LabVIEW (Appendix B) for ease of

use.

System Integration of Equipment

Arrangement of the ER ﬂuid state control test system is shown in Figure 9.
Direction of light, passage of signals, and transfer of information are displayed between
each equipment item. The Trek 609C-6 high voltage ampliﬁer (O-4000V) ampliﬁed
signals from the controller to the ER ﬂuid. Due to the possibility of voltage surges, an
Archer 10M!) carbon resistor and Archer metal-oxide varister 175VDC were included to
protect equipment. The Hewlett-Packard Vectra 486 was implemented merely for
convenience of data display. The Hewlett-Packard E363OA triple output DC power
supply was chosen over batteries to operate the laser diode, because of the length of the
experiment. .

Accuracy of the ER ﬂuid system was improved by experimental design and
precision equipment. Errors associated with signal transmission from the controller to the

Archer Silicon Solar Cell
Reference (B) 2cmx2cm

Edmund Scientiﬁc
M1650 Toshiba Beam 5 diner-g

Lens ER Fluid Window Detector Solar Cell (A)
um um: 1:650n

a e
- e o
9 o
‘2')
l ! “1;:
‘ ' u . . \.>

 
    

 

         
 

 

a.“

  

—ypeT

 

 

       
 

  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

c] o I:
Hewlett-Packard Thermocoul- Ba“, 5513 2A
E3630A tri le 7 n -

P Trek 609% Fluke Hydra Data Logger 56311 Transducer
outs: sI‘DIC 1 - High Voltage Conditioning Ampliﬁer
p0 PP Y . . . Amplifer

‘0-4000v E E .
Archer Metal
Varister 175VDC
. -
-
x .- +
M . hI Hewlett-Packard Vecta
ac1ntos 1" ‘T— I 486. Data Display
Controller

 

 

 

 

 

Figure 9: Laboratory Conﬁguration

16

0.51

 

 

 

 

 

 

 

 

 

0.5
a 0.4
E
\
>
5
E, 0.3
u.
Al
0'2. Feed-forward
Rise to Nominal
Drive Voltage A4
A3 Feed-forward
0.1 Rise to Nominal
Control removed Drive Voltage
Key ER State and ﬁeld allowed
— Nominal Drive Voltage to relax
0.0 - ._ . Desrred ER Flmd State ,\/ v >
Time

Figure 10: Experiment Sequence

Fluke Hydra data logger were observed to be less than 1%. Errors associated with output
of the laser beam were eliminated by determining the measured state from the ratio of
reference and detector solar cells signals. Observations of the same time sequence of
experimental conditions resulted in measurements which differed by less than 5%

including hysteretic effects.

ER Fluid State Tests, Experimental Procedure

Experiments were run using a systematic approach to maintain consistency between
runs (Figure 10). The feed-forward and feedback runs each included 2 segments, the
control and relaxation periods. At the start of each trial, the nominal drive voltage was
0.30kV/mm. The data logger monitored actual transmission level of both solar cells,
current, DC electric ﬁeld, desired ER ﬂuid state, and temperature °C. The desired ER

ﬂuid state and proportional constant were then entered and the experiment would

17
commence. The ER ﬂuid initially inserted into the window was used for all experiments

in the sequence discussed here.

Each experiment began at steady measured state. A large change in ﬁeld strength
was followed by a small change in ﬁeld strength in each run. Between the feedback and
feed-forward runs the ER ﬂuid state had all ﬁeld removed and was permitted to relax
from the chained state. The period of relaxation of the ER ﬂuid state continued for 1.3
hours. The range of Kp and desired state values were limited to avoid a body—centered
cubic state whereupon the particle chains would become locked and a decrease in DC

ﬁeld level would be unable to relax the chains.

Measured F eed-forward Control Response

In the feed-forward experiment the nominal drive voltage was equal to 0.30kV/mm.
It was increased from a steady value of ER ﬂuid state, as measured by transmittance T, to
0.50kV/mm, a large change in ﬁeld strength (Figure 11). After steady ER ﬂuid state was
again reached, the nominal drive voltage was lowered to 0.4026kV/mm to reﬂect a small
change in ﬁeld strength. With feed-forward, the state sensor was used to provide a
reading of the ER ﬂuid state Cfb. The gain, kequ , and time constant, reg“, for the feed-

forward rise and decrease were determined using
Al’

k =—
equ AR (10)

where
AY = steady ER ﬂuid state - rise to nominal ER ﬂuid state

AR = desired ER ﬂuid state - nominal ER ﬂuid state

By establishing steady state at each level the time constant 1“,“, could be calculated.

The time constant reg“, for the feed-forward and feedback rise and decrease were

determined using

reg“ = to the corresponding time of 63.2% of the steady ER ﬂuid state (11)

The feed-forward rise for kequ and requ, were determined as

18

 

 

keg. = (Ar/AR)
Tequ3+1 rs+1 (17)
_ (0. 068/0. 20008) _ 0.340
2089. 44s + l 2089. 443 + 1

where

AY = 0.311- 0.243 = 0.068

AR = 0.5002 - 0.30012 = 0.20008
and

Tequ

= O.632(O.147) = 0.093 —) 35 min.

The same method was used in calculating kequ and reg" of the feed-forward decrease.

The feed-forward decrease resulted in kequ =-0.09 and Tequ=36 minutes.

Measured Feedback Control Response

The feedback experiment with gain,.K -0.5, set the desired ER state to 0.47kV/mm

with a nominal drive of 0.30kV/mm. After steady state was achieved, the desired ER

state was raised to 0.56kV/mm (Figure 12). The gain, kequ ,

and time constant, “reg“, for

 

 

 

 

 

 

0.35-- --0.6
ER Fluid State
0'3.”- ""0.5A
0 E
6.? 0.25 .. l I“ g
1112 d" ' i
an 0.2-- Fiel _.
1, «0.3
'5 = 0.151. T
E ° .
.2. ..02 1:
E 0.1 -- ' 3»
II.
005-» "'0-1
0 :iHHiHHHHHHHHHHHHHHHH 0

v- V V V (O (O CO ('0 (O CO CO CO (V) (O CO CO

V a) N (D O V 00 N (O O V to N (D C

O o v- v- N N N (O CO V V V In to to

v- N (O V In to N CD 0) 0 v- N ('0 V In

Time(s)

Figure l 1: Feed-forward response of the ER ﬂuid state and ﬁeld.

response by the ER ﬂuid state.

Relatively slow

19

the feedback rise and decrease were determined using (10) where
AY = ER ﬂuid state - steady ER ﬂuid state

13
AR = desired ER ﬂuid state - steady ER ﬂuid state ( )

and (11). The feedback rise for keg“ and requ, were determined as
keqa ___ (Ar/AR)
requs + 1 13 + 1

__ (0. 042/0. 085) _ 0.500
672. 25' + 1 672. 25 + l

 

(14)

 

where
AY = 0.524 -0.481= 0.042

AR=0.566-0.481=0.085

and
reg“ = 0. 632(0. 0447) = 0.0283 —) 11 min.

The same method was used in calculating kequ and reg“ of the feedback decrease. The

feedback decrease resulted in kequ =0.05 and tequ=4.5 minutes. The percent error from

the desired ER ﬂuid state was 7.5% for the rise, and 12% for the decrease. The decrease

 

 

 

 

 

 

0.6 __ Desired ER Fluid State, Fl __ 2 5
4’ ER Fluid State, 0
0 5 l ;- - -- 24.5
° r . u “ 24 A
. I o
T 0.4 -- l ._
g g a" Field,E I 1 23-5 g:
m I .. '0
g E 0.3 v ”11:," '___: 23 c.
> w” . o
5 5 ' Temperature : " 22:5 5
I 0.2 "I' i j 22 E
a m ‘ ' 0
0 0.1 i E 21.5 g-
0 ' 21 .2
20.5
-0.1 20

 

 

Time (s)

Figure 12: Feedback response of Kp=0.5, of the desired ER ﬂuid state, ER ﬂuid state,
and ﬁeld. The ER ﬂuid state changed only negligibly when the ﬁeld was
lowered. Removal of the ﬁeld forced the ER ﬂuid state to fall. The
temperature change was negligible.

20
had a larger difference because the forced maximum drop of ER ﬂuid state would remove

the ﬁeld completely. The time constant 1' for the decrease was misleading. The amount
of change in the rise from the steady state point was negligible, therefore reached steady
state faster, resulting in the faster time constant. The small gain was not strong enough to
force the ER ﬂuid state to the desired level. In addition, even when the ﬁeld was lowered
to the nominal drive of 0.30kV/mm, the gain was not large enough to overcome the
effects of hysteresis and lower the ER ﬂuid state.

The monitored temperature remained relatively constant. Temperature increase
during the experiments was negligible, and remained just above room temperature at
about 22 0C (Figs. 12, Al-AS). This validated that the other readings had not been
adversely affected by temperature.

The feedback experiment with gain, Kp=5 .0, repeated the desired ER state values
(Figure 13). The gain kequ , and time constant 1,4“, for the feedback rise and decrease

were determined using (10) with (13) and (l 1). The feedback rise for kequ and Tequ

were determined as
kequ = ( AY/ AR)
requs + 1 Ts + 1
= (0. 068/0.078) _ 0.872
59.ls+1 59.1s-l-1

 

(15)

 

where
AY = 0.556 — 0.488 = 0.068

AR = 0.566 - 0. 488 = 0.078

and
tequ = 0.632(0067) = 0.043 —> 0.99 min.

The same method was used in calculating kequ and requ of the feedback decrease. The

feedback decrease resulted in kequ =0.85 and requ=l .8 minutes. The percent error from

the desired ER ﬂuid state and the actual ER ﬂuid state was 1.7% on the rise, and 2.7% on
the decrease. Again, the time constant 1' for the decrease was misleading. The amount

of change in the rise from the steady state point was negligible, therefore reached steady

21
state faster, resulting in the faster time constant. The ﬁeld had an initial large ﬁeld error

which then reduces as the ER ﬂuid state increases. A similar negative drop in ﬁeld
strength forced the ER ﬂuid state to the lower desired level. The larger gain tracked the
desired ER ﬂuid state with very small steady state error. The desired level was both
raised and lowered with the precise tracking relatively unchanged. The sensor in the
feedback control system measured the ER ﬂuid state and returned that value to determine
the actuating error.

Changes in DC electric ﬁeld reﬂect the effective feedback control action (Figure
13). The initial impulse of high ﬁeld strength forced the ER ﬂuid state to rise quickly.
The response of ﬂuid toward the desired ER ﬂuid state was faster than if the ﬁeld had
been raised in slow increments. This impulse prevented damage to-the ER ﬂuid or
equipment from too large a ﬁeld imposed continuously on the window to force the ER
ﬂuid state to rise. The change in ﬁeld strength is due to the actuating error increase as the
desired level is altered. The ﬁeld strength change becomes Kp times the actuating error.

As the actuating error decreases the ﬁeld decreases. A ﬁeld variation can be readily

Desired ER Fluid State, R
0.7 -- -- 0.8

 

 

 

 

 

 

EH Fluid State, C
0.6 .I I no.7 A
1‘1 , e
I 0_5 : '1 0.6 g
o E o 4 -1- l "" 0.5 E
6- 1 Field
*- D I L,” P [N “(V .
g 0.3 .. j I W «All det‘wh PI I $ 04 m
C ‘—" 0.3 15‘
2 0 I11- : I A —
2 5 0'2 l l e’ A! 0.2 E
“- 0.1 -- : 1/
I . I 0.1 o
m o . U I I l l T. a
i a) m m in to co m a) o r~ r~ rs dj r~ r~ Toﬂ
-01 sssesaszassessea-m
v- w- v- 1- v- 1- N N N N N
'I'lme(s)

Figure 13: Feedback response of Kp=5.0, of the desired ER ﬂuid state, ER ﬂuid state, and
ﬁeld. A large positive and negative change in the ﬁeld forced the ER ﬂuid
state to the desired level.

22
observed in the feedback response with large gain (Figure 13). These large oscillations

effecting the ﬁeld will be discussed later.

Accuracy and response time were much improved with the added feedback. The
addition of feedback improved precision 5 times with the use of Kp=0.5 and over 21
times with Kp=5.0. As Kp increased, accuracy of the ER ﬂuid state to the desired ER
ﬂuid state was increased. The feedback control system rise with Kp=5.0 was 35 times
faster than feed-forward rise, and 11 times faster than feedback rise with Kp=0.5.

Settling of particles did not compromise either accuracy or response time over the
experimental period of 13 hours 19 min. This agreed with observations by Goldstein
( 1990) and Monkman (1991). Settling did not effect control of the system since the ﬁeld
was frequently engaged (Duclos et a1 1988; Goldstein 1990; Monkman 1991). The
period of relaxation between feedback and feed-forward was of a short enough duration
not to effect settling.

Simultaneous monitoring of the current, DC ﬁeld, and temperature made it possible
to detect possible variations that could occur in the ER ﬂuid. The DC ﬁeld reading was a
conﬁrmation of values entered from the control program during the feed-forward
experiment. During the feedback experiment, DC ﬁeld readings became a monitor of
window conditions. The current, as previously observed, did not follow Ohm's Law
(Scott and Yamaguchi 1983; Stangroom 1983), and reﬂected changes in the control

system ER ﬂuid state as well as external disturbances to the window (Figs. A1-A5).

ER State Feedback Model Validation

The analytical model (3) was used to predict gains and time constants from the
measured feed—forward response. Due to the hysteretic nature of an ER ﬂuid, feed-
forward responses of the gain and time constant can vary considerably over time.
Measurements were taken for feed-forward responses of the gain and time constant at

both the beginning and conclusion of experiments. The range of responses, computed

23

Table 1: Comparison Between Computed and Measured Gains and Time Constants.
With increasing Kp, gains approached one, and the time constant decreased.

 

 

 

 

 

T Measured Computed
System ype k 1: (s) k I (s)
Feed-forward 0.34 - 7.5 350. - 2100. ---- ----
Feedback Kp=0.5 0.51 670. 0.16 - 0.79 1800. - 79.
Feedback Kp=5.0 0.87 59. 0.63 - 0.97 770. - 9.2
Feedback Kp=50.0 ---- ---- 0.94 - 0.997 116. - 0.94

 

 

 

 

 

both the beginning and conclusion of experiments. The range of responses, computed
feedback, and actual feedback are given in Table 1. The computed gain and time
constants were determined from the analytical model (9) using control gain KP and
measured feed-forward values of k and 1'. These measured values of k and I under
feedback control fell into the ranges computed from the analytical model.

The effect of Kp decreases the magnitude and range for both k and 1'. This
decrease in range improves precision by decreasing the steady state error, and decreases
the response time of the ER ﬂuid control ampliﬁer. Since Kp and t are inversely
proportional, the increase in Kp decreases the ratio for gain k and increases the ratio for
1. Increasing Kp forces k —-) 1.0 and 1' —9 0. Though we were unable to measure the
feedback response of Kp=50, the computed range for gain k and time constant 1' appears
to follow the established pattern of the analytical model.

Oscillations in the measured ER ﬂuid state and applied ﬁeld were observed. An
enlarged section from Figure 12 makes the oscillations more apparent (Figure 14). The

source of these oscillations could have been either the state sensor or the electric ﬁeld

(16)

24

where AC is the state sensor signal oscillation. Comparison of the ER ﬂuid state and
ﬁeld shows AE/AC at 5 , therefore the oscillations must be due to variations in the electric
ﬁeld ampliﬁer. By examining the frequency response of the predicted gain range from
the analytical model, a bandwidth was established (Figure 15). This bandwidth indicates
how well the system will track the input. The bandwidth and time constant are inversely
proportional to each other, therefore a large bandwidth corresponds to a fast system
response. The shaded ellipse indicated the actual measured frequency due to noise, and
lies within the bandwidth. This ellipse ranges from 0.07 to 0.09 (rad/sec) and -37. to -40.
dB. Even with the ampliﬁer disturbance, the measured ER ﬂuid state and ﬁeld were in
the range predicted from the analytical model. The disturbance from the ampliﬁer limited
the increase of gain and proportional constant in this experiment because of possible

damage to the equipment at too high electric ﬁelds.

 

 

 

 

0.555 T T 0.3
0.553 .. i1 ER Flurd State, C .. 0.7
o 0.551 .. 0 5 E
6A0.549-. f; " ' ;
on ' . \ x
S .§ 0.547 -- - . -- 0.5 ..
0) g : “"~ 1 11.1
2 0-545 -- : r- 0.4
2 g : 15‘
I; 5 0.543 ”___ . .. 0.3 E
11.1 0.541 -- “-
0.539 -- " 0'2 g
0.537 " 0-1
0.535 . 1 1 0
V (D Q 0 N V (D co 0 N V (D a) C N
B V N 01 (O CO 0 00 I!) N O) CO V '-
v- N N C") V to 1.0 (D N h 00 O) O
Tlme(s)

Figure 14: Enlarged section from Figure 12, feedback response Kp=50, ER ﬂuid
state=0.57. Oscillations due to noise in ﬁeld, E.

 

ER State Gain dB

 

 

 

 

10°

10 '3 1o '2 1o '1

00015 Frequency (rad/sec) 0.1 1

Figure 15: Frequency response of the predicted gain range of the analytical model.
Shaded area indicates predicted ER ﬂuid state response range. The ellipse
indicates the actual measurable frequency response due to noise.

10"

CONCLUSIONS

Feed-forward and feedback control approaches using nonlinear, hysteretic ER
ﬂuids were examined by experimental comparison and the results were compared to an
analytical model of the ER ﬂuid state response. Conventional feed-forward control was
tested and shown to be inefﬁcient in design and an ineffective control approach of the ER
ﬂuid state. It was determined that feedback in comparison to feed-forward more
effectively controlled the ER ﬂuid state. This was demonstrated in 3 ways. First,
analytical models of the ER ﬂuid and control system were deve10ped which predicted
system responses. These predictions improved with an increased proportional constant.
Secondly, an ER ﬂuid state sensor was identiﬁed that effectively detected and measured
chaining within the ER ﬂuid state. Finally, it was demonstrated that feedback control ..
improved precision, and decreased response time. Measured responses ﬁt within the
range predicted by the analytical model. These results provide a vehicle of control for
stiffness, viscosity, and other heat transfer properties to be employed.

With higher accuracy and speed, an increased number of low temperature
applications of ER ﬂuids become available (Hartsock et a1 1991; Goldstein 1990). Future
work should include examination of more sophisticated control algorithms which provide
more accurate and faster response of the ER ﬂuid state as well as improvement in sensor
technology to increase overall quality of the response. Effective feedback control of the
ER ﬂuid state permits utilization of ER properties formerly hampered by the imprecise,
slow, hysteric and nonlinear response of ER ﬂuids. It was demonstrated for the ﬁrst time
that the proportional feedback control system responded 35 times faster and 21 times

more accurately than the feed-forward system.

26

APPENDIX A

APPENDIX A

Graphical representation of the experimental sequence outlined in Figure 10 is as follows
(Figs. Al-A5). Each graph, though described as if starting from time zero, follows one after
the other. Lapses between data collecting were less than 1.5 minutes, and taken during
appropriate steady state intervals.

The simultaneous monitoring of various properties during the experiment helped to
maintain consistency between runs. The temperature fell to a steady state during the initial run
of the experiment (Figure Al). Large spikes in the current (Figures A2 and A4) may be due to
external disturbances. Constants were used in the desired ER ﬂuid state (Figures A2 and A3),
and the feed-forward runs (Figures. A1, A4, and A5). Feed-forward control was used in the

rise from zero to the nominal drive ﬁeld (Figures. A1 and A4).

  
 
  
  

 

 

 
 

 

 

 

0.6 -~ 26
A ER Fluid State, C
g 0.5 -- 25 A
<' O
o 1. db
3 c 0'4 Field, E 24 g
1:
a - 0-3 "' I l 23 V
5 A ! Temperature 2
.2. E 0'2 " I ~22 .3
s a
0" 5 0-1 "' ' Current, I #21 g
I .‘Hhmmum
m 0 f—Lz". : 4 5 i i 1 ‘ﬁ : 20 '-
r~ - In C» co r~ .- m c» co N - m a co Ix
co no a: m or 00 m - r~ V o [N 00 a) to N
-0 1 or ID rx 0 00 to a: - 00 (D a: - v to a: or" 1 9
- v- v- v- N or or N 00 00 c0 00 V
Tlme(s)

Figure A1: Feed-forward rise response to nominal drive voltage of 0.30kV/mm, displaying
currentail ﬁeld, temperature, and ER ﬂuid state. Temperature decreased to a steady
state v ue.

27

28

Desired ER Fluid State, R

 

 

 

 

 

 

 

 

 

 

 

 

     
 

 

 

i -- 25
I ER Fluid State,C ._ 24.5
g. : E; _l_ 24 8
4:9 :3 23.5 a,
.§ .— ' o ’ k "‘ 23 3
a _ .1. H r' ield,E "'
5 - l' a l'l'l 1'1 0‘.- 22.5 2
z A ' V 'l r" 7"
v E ‘ l ‘ ‘l -" - 22 ‘5
I E u y T B
o. i I 21.5 g-
v Current.l_ : 21 ,2
I." l L I i
' V ' 20.5
<- o co 0: 00 v o co m Ix 00 a: to v— Ix :0
vmm~~22322253333
Tlme(s)

Figure A2: Feedback response with Kp=5.0 of current, ER ﬂuid state, ﬁeld, and temperature.
The ER ﬂuid state tracked the changes in ﬁeld strength. The current exhibited
similar behavior to changes in the ﬁeld strength.

Desired ER Fluid State, R

 

 

 

 

 

 

0.6 " K "F 25
5 ER Fl 'dSate, C
0 5 I ;- __ u' -- 24.5
g f ' l «24 A
‘ 0.4 -- I .. °
5 3 Field. E i 23-5 P,
g 0 3 "WM. . .. 23 3
D " ' i L. ___.
g ,.: ‘ l Temperature 5 " 22-5 g
.2. E 0.2 '1 ' 22 3
I: g . I :r I. g
I «21 l‘
m 0 1
C') m a) m N N N N 1- 1- 1- 1- 1- 1- - 20-5
O O O O O O O O O 0* O O O O O 0
.0.1 a: m Ix to I!) <- :0 N v- .. a: co Ix co to .. 20
r- N C‘) V ID CD [N m 0) O '- N (O V
Time(s)

Figure A3: Feedback response with Kp=0.5 of current, desired ER ﬂuid state, ER ﬂuid state,
ﬁeld, and temperature. The ﬁeld was lowered, ER ﬂuid state changes were
negligible until the ﬁeld was removed. The temperature changes were negligible.

 

29

0.35 -- Field, E -- 25

 

 

0.3 -- .. 24

 

 

 

 
 

 

 

 

2 ! ER Fluid State, c A
V. 0'25 d' I '_ ﬂ ___" um... 23 o
3 i a
a: 5 0'2 .1. I 2 '0
E 2 "
'5 0.15 T i T I' g
A em erat ,.,
g E p ure .. 21 a
a E 0.1 “M 3
o ; | Current, I .. G
as, 0.05 T i 20 .5,
u l-
m 0 19
v V V «1- v to co co co do do do 00 c0 00
co N 00 <- o co N 00 <- o «a N 00 v o L
-0.05 00 Ix o <- o) v- to in N no a: e) (D o v 18
1- 1- v- N N N O) (O CO V V ID In
Time (s)

Figure A4: Feed-forward rise response to nominal drive voltage 0.30kV/mm, of current, ER
ﬂuid state, ﬁeld, and temperature.

    
 

 

 

 

 

 

 

 

 

 

 

A | T 25
g 1.2 -- | Current, \ -- 24.5 8
. g m. l l "" 24
g .. | . T 23.5 3
D _ 0.8 -- I ! T 23
S I Temperature L w 22 5 o
5 A 0.6 . ' S
o E ' a 22 §
; 0.4- Field, E/v - 21-5 3
8 1— E
v , T 21
0-2 ' ER Fl 'dStt c '3
u, 0' ae, T 20.5
0 d-l'lI-l+l-H-H-+-H-H-l-H-l-l-H-+l+H-l-l+i-H-H-l-++HJ- 20
V ‘0 (D O P C') In (D m C N V (D N O) v- (0
N V N O) CO CO C N 10 N O) CO (0 O Q I!)
O) O) 0) (D 0 co m N N N (D (D (D (0 ID ID
'- N (O V In 0 N (D O) C P N CO V I!)
Time(s)

Figure A5: Feed-forward response of current, ER ﬂuid state, ﬁeld, and temperature. The ER
ﬂuid state was negligibly effected by changes in the ﬁeld, which readily effected the
current. The temperature remained relatively steady.

APPENDIX B

APPENDIX B

Experimental Control Systems

The feedback and feed-forward control systems were written in the virtual
instrument (VI) program language LabVIEW (Appendix B, Program 1). Programs were
set up in 3 stages: (1) conﬁguration, (2) control, and (3) clear and error identiﬁer. This
gives easy access and clarity to the program.

The conﬁguration of the program includes input of channels, the device, a NB-
MIO-16L card, and the addition of default values for other available VI options (Figure
Bl). Output channels are provided to feed the measured output to the ER ﬂuid system
and desired transmittance to a Fluke Hydra Data Logger Model 2625A. Initially the
program was tested with simulated ER ﬂuids. The results were observed on an -
oscilloscope for accuracy of the feedback control system.

The second or control stage is the heart of the program (Figure B2). It contains the
formula node which utilizes entered values from the front user panel and calculates the
resulting measured output voltages. The use of a continuous buffer and internal timing of
the Macintosh 11x provides a shorter, faster and more efﬁcient program.

The clear and error identiﬁer section describes errors to the user and clears values of
the buffer after each run (Figure B3). The error messages aid the user in two ways. First,
it can assist in identifying the origin of error, and second, it suggests why the program
may have discontinued running. The clear VI prevents previous data from mixing in with

the current run.

30

003*“ HOW“
w Output OIW Error
.

Into New“ Ratio Ref. Voltage

Elam-l

K, It I“

menu-IA “MI!

I ml 8|“
lﬁ [El E

 

L

Figure B1: Front panel and conﬁguration section of the control program.

 

 

     
   
    
 

 

 

 
 

.lE‘J 9:" Ill ' 12M

-IIII]-l-l-l-lel-l-l-l-l-l-l-lsl-IIl-I-I- all1II-l-lil-l-l-]II-l-l-I!l-l-l!l-J-I-l-l-l-

 

 

 

 

       

 

      

 

 

 

 

 

 

 

 

' 7 , , ,
- m 7, m“ o-(doxrod-x); I
httl‘alizt El [31‘]: m odot-(e-Qlﬂdolt; H
the mm B [gill m odotf'(o¢ot+2'odotl)/3 ; EMI
l -istors T , “- ’t (mu )‘int 1;
a El 33!] El :I:O*Kr;tt «Oklahoma- m' a
I /*nI3‘/ "F
I5 m m a
a tamed R“ m m: D
l ‘ I:
[a l .. r m
I‘ 1 Hal h
1 , l _ m
, , l ._ a
m E ' l n m hi
. - a
|_-'1?U_“7‘ [E ti
a —l@ ‘E
.J i
n _ F ‘3
_ II
III 1. “-5113 a n

 

 

 

 

  

IIIEFI-I'llj-IW-I'IiI-JI‘IFI-FI!III!IiIIJiI-J-Iilﬂ!III'-III3J!ISII

 

 

Figure B2: Data reading, control capabilities, and output sections of the control program.

31

 

   

 

Li,

   
  
    

; I: ‘ PIn-Conuleslred Zout (a) limit niognL

.lWﬁlihll-Y 73--

Wuuununnuuunnnnnnnnnu

 

   

  

 

 

”a: r in" 1 J
:31 5”“ :ﬁ 9

v ,
.3...4-...-.~...‘T.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure B3: Clear section of the control program.

32

LIST OF REFERENCES

LIST OF REFERENCES

Arguelles, J ., Martin, H. R., and Pick, R. J ., 1973, "Some Experiments with
Electrosensitive Fluids," Proceedings of the 3rd International ﬂuid Power Symposium,
May.

Block, H., and Kelly, J. P., 1986, "Electrorheological Fluids," European Patent
0,191,585.

Butters, R., March, 1995, Personal Communication, Lord Corporation, Carey,
North Carolina.

Conrad, H., Sprecher, A. F., Choi, Y., and Chen, Y., 1991, "The temperature
dependence of the electrical properties and strength of electrorheological fluids," The
Society of Rheology, Inc., May, pp. 1393-1410.

 

Duclos, T. G., 1987, "An Externally Tunable Hydraulic Mount which Uses Electro-
Rheological Fluid," Proceedings of the 1987 Noise and Vibration Conference, April.

Duclos, T. G., Acker, D. N ., and Carlson, J. D., 1988, "Fluids That Thicken
Electrically," Machine Design, January, pp. 42-45.

Duff, A. W., 1896, "The Viscosity of Polarized Dielectrics," Physical Review , Vol.
4, pp. 23-38.

Filisko, F. E., and Armstrong, W. E., 1988, U. 8.. Patent 4,744,914.

Goldstein, G. 1990, "Electrorheological Fluids: Applications Begin to Gel,"
Mechanical Engineering, October, pp. 48-52.

Hartsock, D. L., Novak, R. F., and Chaundy, G. J ., 1991, "ER ﬂuid requirements for
automotive devices," The Society of Rheology, October, pp. 1303-1326.

Klingenberg, D. J ., van Swol, F. and Zukoski, C. F., 1989, "Dynamic Simulation of
Electrorheological Suspensions," Proceedings of the ACS Division of Polymeric
Materials Science and Engineering , September, Vol. 61, pp. 154-155.

Konig, W. 1885, Annals of Physics, Vol. 25, p. 618.

Lloyd, J. R., and Zhang, C., 1994, "A Discussion on the use of Electrorheological
ﬂuids in the control of Heat Transfer Processes," Paper presented at the 1994 Heat and
Mass Transfer Conference, Bombay, India, January.

Monkman, G. J ., 1991, "Addition of solid structures to electrorheological ﬂuids,"
The Society of Rheology, Inc. , October, pp. 1385-1392.

33

34

Quinke, G., 1897, "Die Klebrigkeit isolirender Fliissigkeiten im constanten
electrischen Felde; Viscous Behavior of Isolated Flows in Constant Electric Fields,"

Annals of Physics and Chemistry, Vol. 62, pp. 1-13.

Scott, D. Y., and Yamaguchi, J. 1983, "Automotive engineering,"International
Viewpoints, Vol. 91, No. 11, pp. 61-66.

Stangroom, J. E., 1977, "Electric Field Responsive Fluids," U. S. Patent 4,033,892.
Stangroom, J. E., 1978, "Electric Field Responsive Fluids," U. S. Patent 4,129,513.

Stangroom, J. E., 1980, "Improvements in Relating to Electric Field Responsive
Fluids," U. K. Patent 1,570,234.

Stangroom, J. E., 1982, "Electroviscous Fluids," International Patent WO 82/04442.

Stangroom, J. E., 1983, "Electrorheological Fluids," Physics Technology, Vol. 14,
pp. 290-296.

Stangroom, J. E., 1984, "Electric Field Responsive Fluids," U. 8. Patent 4,483,788.

Tabatabai, Stefan, 1993, "Fakultiit Fiir Maschinenwesen," Studienarbeit at Michigan
State University.

Tao, R., and Sun, J. M., 1991a, "Three-Dimensional Structure of Induced
Electrorheological Solid," Physical Review Letters, July 15, Vol. 67, No. 3, pp. 398-401.

Tao, R., and Sun, J. M., 1991b, "Ground state of electrorheological ﬂuids from
Monte Carlo simulations," Physical Review Letters, November 15, Vol. 44, No. 6, pp. 44-
47.

UOP Molecular Sieve Adsorbents, 1990, "Adsorbents Product Information,"
August.

Ushijima, T. Takano, K. and Noguchi, T. 1988, "Rheological characteristics of ER
Fluids and Their Application to Anti-Vibration Devices with Control Mechanism for
Automobiles," Passenger Car Meeting and Exposition, October/November, Society of
Automotive Engineers, Inc.

Winslow, W. M., 1947, U. S. Patent 2,417,850.

Winslow, W. M., 1949, "Induced Fibration of Suspensions," Journal of Applied
Physics, No. 20, pp. 1137-1140.

Winslow, W. M., 1962, U. S. Patent 3,047,507.

Yen, W. S., and Achom, P. J., 1991, "A study of the dynamic behavior of an
electrorheological ﬂuid," The Society of Rheology, pp. 1375-1384.

Zhang, C., and Lloyd, J. R., 1992a, "Control of Radiation Heat Transfer through a
Composite Window Featuring ER Fluid: A Conceptual Investigation: Paper presented at
the Second Minsk International Heat and Mass Transfer Form, May.

35

Zhang, C., and Lloyd, J. R., 1992b, "Measurements of Radiation Heat Transfer in
Electrorheological Fluid Based composite Materials," Paper presented at the National
Heat Transfer Conference.

Zhang, C., and Lloyd, J. R., 1993 "Enhancement of Conductive Heat Transfer
through an Electrorheological Fluid Based Composite Medium," Presented at the 1993
National Heat Transfer Conference, August, Atlanta, Georgia.

HICHIGSRNTTE UNIV

I III III II III ||I9|I9|I:9: