0 Stimmen dafür0 Stimmen dagegen

0 Aufrufe6 SeitenStatistical Modeling of a Magnetorheological Fluid Damper Using the Design of Experiments Approach

Feb 17, 2019

© © All Rights Reserved

Statistical Modeling of a Magnetorheological Fluid Damper Using the Design of Experiments Approach

© All Rights Reserved

0 Aufrufe

Statistical Modeling of a Magnetorheological Fluid Damper Using the Design of Experiments Approach

© All Rights Reserved

Sie sind auf Seite 1von 6

net/publication/230942191

of experiments approach

DOI: 10.1088/0964-1726/16/4/044

CITATIONS READS

29 144

2 authors:

Shivaram Ac Gangadharan K V

Applied Materials India Pvt Ltd National Institute of Technology Karnataka

6 PUBLICATIONS 30 CITATIONS 96 PUBLICATIONS 385 CITATIONS

Some of the authors of this publication are also working on these related projects:

All content following this page was uploaded by Gangadharan K V on 09 March 2015.

IOP PUBLISHING SMART MATERIALS AND STRUCTURES

Smart Mater. Struct. 16 (2007) 1310–1314 doi:10.1088/0964-1726/16/4/044

rheological fluid damper using the design

of experiments approach

A C Shivaram and K V Gangadharan

Department of Mechanical Engineering, National Institute of Technology Karnataka,

Surathkal, Mangalore 575025, India

Published 5 July 2007

Online at stacks.iop.org/SMS/16/1310

Abstract

In this work, a through-rod-type magneto-rheological (MR) test damper has

been designed and fabricated for experimental study. Various factors, such as

the magnetic field strength, volume fraction of particles in the MR fluid,

shearing gap between piston and cylinder, vibration frequency and amplitude,

have been considered as input factors, and the root-mean-square (RMS)

damping force as the output factor (response variable). These input factors

are varied in two levels (low and high) during the initial phase of

experimentation using 25 factorial design; the motivation is to identify the

most influential factors. In the second phase of experimentation, the response

surface method has been used to identify the modeling equation and to plot

the response surfaces. Further, force versus displacement diagrams have been

plotted at these factor levels; these give an insight into the damping behaviour

of the MR damper.

(Some figures in this article are in colour only in the electronic version)

fraction, limitations on the fluid flow area when shearing

In the recent past, magneto-rhelogical (MR) and electro- happens, effect of vibration parameters like frequency and

rheological (ER) fluids have commanded a lot of research amplitude on the system performance, optimal carrier fluid,

interest because of their ability to change their flow resistance, effect of change in temperature, and possible segregation of

when exposed to magnetic or electric fields, respectively. particles during operation, have to be resolved. In this work,

Winslow (1947) patented his electro-fluid clutch, which was an attempt has been made to provide a better understanding of

designed to transmit mechanical movement or force, in these issues, by performing a series of experiments on an MR

response to an electrical potential. Further, he patented a test damper. The design of experiments (DOE) approach was

field-controlled hydraulic device, (Winslow 1953), marking used to organize the experimental runs, and to arrive at sensible

the invention and usage of MR fluids. Since then, various conclusions based on experimental results.

devices such as dampers, clutches, and brakes, have been

designed (Jolly et al 1998, Namuduri et al 2006, Ulicny et al

2007) and even used commercially in luxury car suspension 2. Experimental approach

system. Various mathematical modeling approaches such as

the Bingham model (Kamath et al 1996), biviscous model An MR fluid is a suspension of magnetically polarizable

(Wereley et al 2004), constitutive model (Choi et al 2005) and particles in an inert fluid medium like silicone oil. The

phenomenological models (Gandhi and Bullough 2005) have viscosity of an MR fluid would mainly depend on the carrying

been proposed in the literature, which could be of assistance medium, when the fluid is not exposed to a magnetic field.

in designing these MR devices. To achieve technological Its compatibility with magnetic material, damper and sealing

maturity and widespread commercial usage of such devices, materials is important from the engineering perspective, and its

Statistical modeling of an MR fluid damper

an MR damper.

Factor Category

Volume fraction Fluid

Shearing gap Device design

Magnetic field strength Controllable

Frequency Working environment

Amplitude Working environment

Temperature Working environment

Carrying medium Fluid

Particle material and size Fluid

Coating material and size Fluid

Figure 1. Damper used for testing.

thermal stability. On exposure to a magnetic field, the particles

get polarized and align along the flux lines, forming chain-like

structures. The formation of such chain-like structures would

induce additional flow resistance, which directly depends on

the magnetic field strength. When there is no magnetic field,

the particles have a tendency to settle down due to gravity. To

avoid settling of the particles, different coating materials like

fumed silica (Iyengar and Foister 2002) and guar gum (Fang

et al 2005) have been tried out, which adds to the complexity

of chain formation. Though these coatings improve the settling

behaviour, they change the dynamic yield strength of the fluid

(Wu et al 2006). The volume fraction of particles in the fluid

is another important factor which influences the yield strength

of the fluid. Apart from these fluid factors, device factors like

Figure 2. Block diagram of the test set-up.

the shearing gap, vibration frequency and amplitude are also

believed to affect the performance of an MR damper. Any

change in shearing gap would not only change the damping

constant in the off condition, but also alter the magnetic flux The motivation behind the second stage of experiments was

distribution. Vibration frequency and amplitude, which can be to determine the modeling function which can predict the

attributed to the working environment, will decide the fluid RMS damping force at the given factor levels, and to plot

shearing rate, thus determining the damping force. Choice the response surfaces, which show how the response variable

of particle material is mainly governed by its permeability, (RMS damping force) varies with change in the considered

manufacturability and cost. Increasing particle size would factors. Such response surfaces would be useful in identifying

aggravate the settling issue; however, Cox et al (2006) have the optimal factor levels during the design stage, in performing

shown that decreasing particle size would actually decrease sensitivity analysis for control purposes, and in operating with

the yield strength. Particle size and shape is another important maximum efficiency in running conditions. Further, force

factor. Even though controlling the particle shape is practically versus displacement diagrams have been plotted, which give

difficult (at micron levels), a few mathematical simulations an insight into the energy dissipation phenomenon in an MR

(Klingenberg et al 2005) have been done to study the shape damper.

and size effects. All these factors have been summarized in

table 1, and the first five have been considered for this initial 3. Experimental set-up

study. A further detailed study will be made involving all the

factors in the near future. The experimental set-up consists of a through rod type MR

Design of experiments (DOE) has been used to organize test damper shown in figure 1, mounted on an electrodynamic

experimental runs, and to arrive at sensible conclusions from shaker. The piston rod was connected to a fixed support

the experimental data. Experimentation was done in two through a force transducer. The damper base displacement was

stages, the first stage using 2n full factorial design and the measured using a non-contact-type laser displacement pick-up.

second stage using the response surface method (Montgomery Both force and displacement signals were digitized and stored

2003). The damper set-up was mounted on an electro-dynamic using a PC-based data-acquisition system running Labview.

shaker and was subjected to sinusoidal displacement input. A block diagram of the set-up is shown in figure 2 and a

In the first stage all factors were varied in two levels (low photograph is shown in figure 3. The MR fluid was prepared in

and high) and the RMS damping force was measured as the house by mixing different proportions of carbonyl iron powder

response variable. The motivation behind these experiments (average size 10 µm) with silicone oil, and stirring it in an

was to identify the factors having significant influence on emulsifier. The damper was filled with freshly stirred fluid just

the damping force. Once the most influential factors were before the experiment, and the testing lasted for a few minutes

identified, they were varied in three levels in the second stage. at a stretch without any noticeable change in temperature.

1311

A C Shivaram and K V Gangadharan

Half-Normal Plot

Figure 3. Photograph of the test set-up.

Figure 4. Half normal plot.

Table 2. Factor levels for 25 factorial design. Table 3. Factorial analysis effects list.

Factors Low High

A—volume fraction (%) 10 40 Term Effect SumSqr % contribution

B—shearing gap (mm) 1.5 2.5 A—volume fraction 1.5248 54.5286 15.4082

C—magneto-motive 0 120 B—shearing gap −0.3139 2.3116 0.6531

force (MMF) C—magneto-motive force 2.5920 157.5810 44.5279

(ampere-turns) D—frequency −1.3466 42.5287 12.0174

D—frequency (Hz) 0.2 8 E—stroke 0.3104 2.6526 0.7495

E—stroke (mm) 1 3 AB −0.1341 0.4217 0.1191

Response variable RMS damping AC 0.5721 7.6773 2.1694

force AD −0.3973 3.4384 0.9716

AE 0.2597 1.3854 0.3914

BC −0.1109 0.2887 0.0815

BD −0.0549 0.1304 0.0368

4. Results and discussion BE −0.2415 1.2641 0.3572

CD −0.7763 14.1357 3.9943

Initially five factors were varied in two levels, force and CE 0.0994 0.3498 0.0988

displacement data were acquired and the RMS damping force DE 0.0556 0.1505 0.0425

was used as the response variable for factorial analysis. These ABC 0.6098 8.7235 2.4650

ABD −0.3487 2.8472 0.8045

factors and their levels are summarized in table 2. Magneto-

motive force was used instead of field strength, as it has a direct

bearing on the field strength and also it is simple to measure As expected, factorial analysis suggests that volume

and control. fraction (A) and MMF (C) have a significant influence on the

To identify the most important factors, a graph showing RMS damping force of an MR damper. Further, it also suggests

least-square estimates of effects and their half-normal that the contribution of frequency (D) could not be neglected.

probabilities (Montgomery 2003) was plotted. Statistics The other two factors, shearing gap (B) and stroke (E), did

suggests that the larger the effect of a particular factor, the not have much effect (within the considered range) on the

larger will be its deviation from a straight line in a half-normal RMS damping force. With these observations as the base, the

plot. Figure 4 shows such a half-normal plot of the above- shearing gap and stroke were held constant and the other three

considered factors and their interactions. From this figure, one factors were considered for the next phase of experimentation.

can observe that factor C (MMF) is most influential, followed The response surface method was used to organize this phase

by factors A (volume fraction) and D (frequency). Along with and face centered central composite design was used to decide

individual factors, their interactions (such as CD, ABC, and the factor levels (Montgomery 2003). The factors considered

AC) also have a considerable influence on the RMS damping and their levels are summarized in table 4.

force. From statistical analysis, the individual effects of these Power transformation has been used to accommodate large

factors and their interactions have been found out, along with differences between minimum and maximum values of the

their percentage contributions. These data are tabulated in response variable. A transformation factor of −1/2 was used

table 3. based on a Box–Cox plot (Montgomery 2003). Figure 5 shows

1312

Statistical modeling of an MR fluid damper

B: MMF B: MMF

C: Frequency A: Volume fraction

Table 4. Factor levels for the response surface method using central composite design.

Factors Low Mid High

A—volume fraction 10% 25% 40%

B—magneto-motive force 0 ampere-turns 60 ampere-turns 120 ampere-turns

C—frequency 4 Hz 6 Hz 8 Hz

Shearing gap Held constant At 2 mm

Stroke Held constant At 3 mm

the response surfaces of damping force−1/2 versus MMF, like a viscous fluid (Tang and Conrad 1996). As a result,

frequency and volume fraction. From this graph one can with increase in frequency, the damper performance resembles

observe that there is a strong nonlinear relationship between more the hysteretic damping of an viscoelastic material. This

the response variable and the three considered factors. As understanding is very important when designing MR dampers

expected, it suggests that high volume fraction and high MMF for different applications like structural (around 0.1–1 Hz),

will result in high damping force; however, it also suggests automotive (around 0.1–10 Hz) or machine vibration isolation

that such a behaviour is frequency dependent. The lower the applications (around 0.1–10 000 Hz). The same can be

frequency, the larger is the RMS damping force. observed from figure 6.

Equation (1) shows the quadratic modeling equation

obtained from the response surface method. It predicts the

RMS damping force of an MR damper within the considered 5. Summary

factor range.

• An experimental study based on the design of experiments

Damping force−1/2 = +0.199 36 + 9.784 82 × 10−4 × A was conducted, to identify factors that have a significant

− 2.083 68 × 10−4 × B − 0.018 513 × C influence on the RMS damping force of an MR damper.

− 1.857 80 × 10−6 × A × B − 3.000 77 × 10−5 × A × C • As expected, certain factors, namely the magnetic field

+ 8.172 17 × 10−6 × B × C − 3.925 44 × 10−5 × A2 strength and volume fraction of particles in the fluid, were

found to have a significant influence on the RMS damping

− 1.079 44 × 10−6 × B2 + 1.933 76 × 10−3 × C2 . (1)

force.

Figure 6 shows force versus displacement diagrams at • Further, a significant correlation between vibration

different factor levels. From these diagrams it can be observed frequency and RMS damping force was observed.

that, as the frequency increases from 0.2 to 8 Hz, the damping

• At low frequency, the damping nature was more like a

nature changes from Coulomb to hysteretic. Similar results

Coulomb damping, which gradually changed to hysteretic

have been obtained by Snyder et al (2001) and Gamota

nature as the frequency increased.

and Filisko (1991) at various frequencies, however only for

the flows which are predominantly in the pre-yield regime • This frequency dependence would have a lot of bearing

(Sprecher et al 1987). For a given shearing gap, as the on the choice of damping materials to be used in different

oscillating frequency increases, the strain rate also increases. applications.

This would result in a significant portion of the strain cycle • A statistical modeling equation has been formulated based

being spent in the post-yield regime. It has been observed on the experimental data, which can predict the RMS

that MR fluid behaviour in the post-yield regime is more damping force of an MR damper.

1313

A C Shivaram and K V Gangadharan

dependence is under consideration.

Acknowledgments

The current work has been sponsored by a research grant,

MHRD, Government of India and TEQIP-NITK. This

assistance is gratefully acknowledged.

References

Choi Y, Cho J, Choi S and Wereley N 2005 Smart Mater. Struct.

14 1025–36

Cox B J, Thamwattana N and Hill J M 2006 Rheol. Acta 45 909–17

Fang C, Zhao B Y, Chen L S, Wu Q, Liu N and Hu K A 2005 Smart

Mater. Struct. 14 N1–5

Gamota D and Filisko F E 1991 J. Rheol. 35 399–425

Gandhi F and Bullough W A 2005 J. Intell. Mater. Syst. Struct.

16 237–48

Iyengar V R and Foister R T 2002 Use of high surface area untreated

fumed silica in MR fluid formulation Technical Report, US

Patent No 6451219

Jolly M R, Bender J W and Carlson J D 1998 Proc. SPIE

3327 262–75

Kamath G M, Wereley N M and Melanie K H 1996 Smart Mater.

Struct. 5 576–90

Klingenberg D, Ulicny J and Smith A 2005 Appl. Phys. Lett.

86 104101

Montgomery D C 2003 Design and Analysis of Experiments 5th edn

(New York: Wiley)

Namuduri C, Browne A L and Ulicny J C 2006 Impact energy

absorber and process Technical Report, US Patent No 6983832

B2

Snyder R, Kamath G and Wereley N 2001 AIAA J. 39 1240–53

Sprecher A F, Carlson J D and Conrad H 1987 Mater. Sci. Eng.

95 187–97

Tang X and Conrad H 1996 J. Rheol. 40 1167–77

Ulicny J C, Klingenberg D J, Smith A L and Golden M 2007

Magnetorheological fluid device Technical Report, US Pub No

2007/0023247

Wereley N M, Lindler J, Rosenfeld N and Choi Y T 2004 Smart

Figure 6. Force–displacement diagrams. Mater. Struct. 13 743–52

Winslow W M 1947 Method and means for translating electrical

impulses into mechanical force Technical Report, US Patent No

2417850

Winslow W M 1953 Field controlled hydraulic device Technical

• Response surfaces have been plotted showing nonlinear Report, US Patent No 2661596

variations of the RMS damping force with field strength, Wu W P, Zhao B Y, Chen L S, Wu Q, Sheng L and Hu K A 2006

volume fraction and frequency. Smart Mater. Struct. 15 N94–8

1314

## Viel mehr als nur Dokumente.

Entdecken, was Scribd alles zu bieten hat, inklusive Bücher und Hörbücher von großen Verlagen.

Jederzeit kündbar.