Professional Documents
Culture Documents
net/publication/234116148
CITATIONS READS
31 1,752
2 authors:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Andrea Spaggiari on 24 March 2016.
1 Introduction One of the few materials that behaves better than iron in terms of
magnetic permeability and saturation is the so called ml-metal
Magnetorheological fluids are smart and controllable materials
[3], a nickel-iron-molybdenum alloy that has up to ten times the
that are widely used in several devices, such as dampers and
performance of pure iron. However, ml-metal is extremely ex-
clutches [1]. An external magnetic induction field modifies the
pensive; thus, the most commonly used particles in commercial
macroscopic appearance of MR fluids by switching from a viscous
MR suspensions are carbonyl iron (pure iron obtained with a spe-
free-flow liquid when no field is applied to a quasi solid state
cific technology), achieving a maximum sy ¼ 100 kPa. In commer-
when a strong induction is present. The behavior of MR fluids is
cial MR fluids, additives and surfactants are mixed with the fluid
affected by electromagnetic fields, which reversibly modify their
to prevent or limit the deposit of the carbonyl iron particles on the
apparent viscosity. The sudden change in MR behavior (a few
bottom of the inactive device due to gravity forces. Another aspect
milliseconds) when subjected to a magnetic field makes this mate-
that can be used to improve the value of sy is the volume fraction
rial attractive for damping and dissipative devices.
of the particle in suspension; however, it cannot be increased
MR fluids can be used to build integral, silent, and rapid me-
beyond a certain value otherwise the MR fluid would become too
chanical systems, which are enhanced by electronic controls. Sev-
viscous.
eral ferromagnetic materials can be used as magnetic particles
Ginder and Davis [4] used a finite element analysis to study the
dispersed in the carrier fluid (mainly oil or water) in order to pro-
effect of magnetic nonlinearity and saturation of magnetic par-
vide the magnetorheological effect. In fact, when MR fluids are
ticles. They found a limit for the volume fraction at 50% volume,
subjected to any magnetic field, the particles acquire a dipole
which gives a maximum of sy ¼ 210 kPa. In any case for industrial
moment aligned with the flux lines of the field. When this phe-
applications, a value of 50% volume fraction is too high, and the
nomenon is analyzed at a microscale, it can be seen that it leads to
normal volume fraction ranges from 20% to 40%, which corre-
the formation of linear chains of particles, which result, in a mac-
sponds to a typical iron particle content in weight (75–85%) as
roscale, in a solid-like MR aggregate.
reported in LORD Corp. datasheets [5,6]. MR fluids in squeeze
The key parameter for the MR fluids is the yield shear stress
mode have been demonstrated to give a higher force compared to
(sy), which is the maximum stress the fluid can withstand before
shear and flow mode. Unfortunately, squeeze mode enables only
starting to flow. This value is fundamental in MR fluid design
very short strokes to be reached and, thus, cannot be exploited in
because it is directly linked to the maximum power that can be
many applications. Tang et al. [7] reported that the yield stress of
dissipated by an MR device.
MR fluids compressed along the direction of the magnetic field
Clearly, sy is affected by a magnetic field, as shown in Fig. 1(a)
gave an increment of ten times for sy. They proposed the forma-
which shows the technical datasheet supplied by the producer for
tion of thicker and, thus, stronger columns of particles that are
the commercial MR fluid 140-CG, from LORD Corporation [2].
able to sustain the load.
Figure 1(b) reports the saturation of the magnetic particles typical
Zhang et al. [8] designed an apparatus to evaluate the effect of
of MR fluids at high induction flux levels [2]. Thus, no matter
compression on MR fluids in shear mode. Their experimental
how high the magnetic flux, thanks to a smart application of coils
campaign revealed that a very high compression could enhance sy
or permanent magnets, the maximum dissipated power reaches a
by more than 20 times for a given magnetic field, thus, showing a
plateau due to the limitation of sy.
squeeze strengthen effect. The apparatus they used [8], which was
One possible way to enhance the yield stress of MR fluids is,
quite complex and bulky, consisted of a nonmagnetic container
thus, to select a particle material with a larger magnetic saturation.
for the MR fluid compressed through a big bolt, and a metal sheet
1
was used to assess shear stress under several pressure levels. They
Corresponding author
Contributed by the Fluids Engineering Division of ASME for publication in the
also correlated this behavior not only with the magnetic force of
JOURNAL OF FLUIDS ENGINEERING. Manuscript received May 11, 2012; final manuscript the dipoles formed by the ferromagnetic particles but also with
received July 24, 2012; published online August 22, 2012. Assoc. Editor: Shizhi Qian. the friction between the particles. A hybrid tribological-magnetic
Fig. 4 Flux density inside the hydromagnetic system for a current of 2 amp (a),
scale 0–1 T, and flux distribution along the diametral section of the inner channel
(b) with lr 5 5
2.4 Design of Experiment. A statistical method is often con- Bingham model in Fig. 3; thus, only the behavior in proximity to
venient in dealing with problems involving multivariable analysis. zero is needed in order to estimate the yield stress.
The design of experiment procedure, a well-founded statistical
method based on the analysis of variance analyses (ANOVA)
[17], designed for the analysis of experimental tests, can be
3 Results and Discussion
applied to such scientific problems, and an application to MR fluid 3.1 Experimental Results. The experimental curves of the
is shown in Ref. [18]. design of experiment plan are reported in Fig. 6. The solid black
In this experiment, the variables involved are the applied mag- lines reports the experimental curves retrieved from the test
netic induction field (B) and the internal pressure of the MR fluid machine, which are grouped in terms of the four different pressure
(p). Since an interaction between the two variables was found in levels: Fig. 6(a) reports the curves with no internal pressure,
shear mode [8], the experiments were designed to verify the same Figs. 6(b), 6(c), and 6(d) show the curves for 10, 20, and 30 bars,
interaction in flow mode. The most frequent application of the respectively. The curves show a nearly bilinear behavior with
design of experiment is the two-level factorial experiment, which clear pre-yield and post-yield regions. A bilinear regression is
is used mainly when there is no way to predict the most important applied in order to calculate the pull out force.
variables and how they interact. In this work, on the other hand, The experimental points are fitted using a bilinear interpolation,
we adopted the “general factorial” approach, which consists of and the coefficients that give the best fit are retrieved following
varying only two variables over four levels. This method focuses the procedure. First a linear regression of the post-yield experi-
only on two parameters that influence the behavior of the system mental points is applied and the slope is calculated. Secondly the
and is, thus, able to identify the interaction between these varia- bilinear regression (gray lines in Fig. 6) is calculated and the y-
bles more precisely and to provide a more reliable model to intercept is taken as the pull-out force. The slope of the experi-
describe how the system behaves. The magnetic induction ranges mental force is quite steep in the post-yield range. Since the
from 0 to 800 mT, while the internal pressure spans the 0–30 bars experimental tests were performed at a constant speed, the viscos-
range. The pressure is regulated using the external command ity of the MR fluid cannot play a role; thus, the slope must be
screw, and when the desired value is reached, the upper bar is related to a different phenomenon.
moved by the tensile machine. The pulling force is proportional to This elastic-like behavior is not due to the presence of the
the yield stress of the fluid according to Eq. (4). The complete ex- spring (K) because the stiffness is quite different; thus, a possible
perimental plan is reported in Table 1. explanation may be the magnetorheological effect in the induction
The experimental tests were performed following a quasi-static zone.
procedure at a speed of 1 mm/min in order to avoid, or at least When the fluid is pushed through the magnetized area, the fer-
minimize, any dynamic or viscous effects. The head of the univer- romagnetic particles form chains, which block the fluid movement
sal machine was maintained in displacement control and provided until the yield stress value is reached. The chains then start to
a 1 mm ramp. There was no need for a larger displacement move; however, the magnetic attraction causes the accumulation
because the behavior of the MR fluid is well described by the of the particles near the magnetized area. This phenomenon can
be likened to a filter or a sieve, which becomes finer and finer.
The larger the stroke, the higher the force needed to overcome the
Table 1 Experimental plan barrier. This behavior explains the increase in the force and the
apparent elasticity of the system in the post-yield zone. By observ-
Levels I II III IV ing the slopes of the several curves in Fig. 6, this explanation is
confirmed by the flat curves of Fig. 6(a) and the steep curves of
Induction field, B, (mT) 0 200 400 800
Internal pressure, p (bar) 0 10 20 30 Fig. 6(b), reported using the same scale. Thus, the MR effect is
Replicates 3 for each combination strengthened by the pressure in the same way the yield stress is
Experimental points 16 strengthened by compaction of the particles. As can be seen, with-
Grand total 48 out the presence of pressure, the filtering phenomenon is not evi-
dent because the fluid is not forced through the magnetic sieve
and, thus, the slopes are nearly flat. In the case of internal pres- can be obtained by recalling Eq. (4). The complete set of calcu-
sure, the MR fluid must flow through the magnetic sieve, and the lated total shear stresses sTOT is reported in Table 2.
gray lines become steeper and steeper as the pressure grows.
The pull-out force values, reported in Table 2, show that even
without a magnetic field, the system provides quite a high force 3.2 Analysis of Variance on the Total Yield Stress. An
due to the effect of the applied pressure. This is because the fric- analysis of the variance was applied using Design Expert 8.0 [19].
tional force of the sealing, which is negligible at zero pressure, ANOVA calculates the variance of a response by considering a
increases as the internal pressure increases. The total shear stress specific variable and the global variance in the responses. Of the
possible approaches to graphically represent the results, one of the
most popular is the normal plot, which is used to estimate whether
Table 2 Experimental pull-out forces and total stresses sTOT a certain set of data follows a Gaussian distribution or not. If the
retrieved using Eq. (4) data approximates a straight line, the phenomenon is statistically
“normal,” i.e., follows a stochastic law. The variables or the inter-
Pull-out force (N) sTOT (kPa) actions affecting the system’s behavior will then fall outside the
normal distribution line; thus, their effect cannot be ascribed to a
P (bar) B (mT) R1 R2 R3 Average Average stochastic process. The greater the deviation of the point from the
normal line, the larger the confidence interval (i.e., the probability
0 0 7.34 3.47 0.00 3.60 2.55
that the variables are significant). The half normal plot used in
0 200 16.16 15.95 11.88 14.66 10.37
0 400 28.06 26.66 26.42 27.04 19.13 this paper is interpreted in the same way as the normal plot but
0 800 35.28 37.01 32.52 34.93 24.71 allows absolute values of the effects to be considered. Since the
10 0 16.68 13.89 7.48 12.69 8.97 half normal line starts at the origin, this produces a more sensitive
10 200 16.62 30.51 24.88 24.00 16.98 scale with which to detect significant outcomes [17], which are
10 400 44.22 48.11 44.32 45.55 32.22 immediately detected at a glance. The half normal plot of the ex-
10 800 85.58 74.76 81.14 80.50 56.94 perimental values of sTOT is shown in Fig. 7.
20 0 125.51 106.17 99.05 110.24 77.98 The straight line is built thanks to replicates (triangles) and pro-
20 200 178.97 138.56 135.03 150.85 106.71 vides an estimation of the normal distribution of the experimental
20 400 163.59 150.08 150.66 154.78 109.48
error, which by definition has a stochastic distribution. The trian-
20 800 201.46 203.07 201.71 202.08 142.94
30 0 133.94 132.79 137.54 134.76 95.32 gles are an expression of the sum of errors [17], which is calcu-
30 200 206.03 161.02 171.66 179.57 127.02 lated via Design Expert [19]. The point that falls far from the line
30 400 216.85 239.97 240.18 232.34 164.34 represents an effect that does not follow a stochastic law; thus, it
30 800 264.69 267.31 275.71 269.24 190.45 has a physical influence on the system under scrutiny. Since both
points representing the two variables (pressure and magnetic field)
aij (mT1)
i 0 1 2
are far from the error line, ANOVA demonstrates that these varia-
bles influence the process. The interaction between the variables
is also important, which means that there is an enhancement in the
total yield stress due to the pressure.
Fðp; BÞ Fðp; 0Þ
sy ¼ sTOT sFRIC ¼ (8)
A
Table 4 Matrix of coefficients calculated to plot the surface the AISI 304L may have occurred. As reported in the technical lit-
response erature, the behavior of the AISI 304L steel may become ferro-
magnetic due to plastic deformation [22]. The machining of the
aij (mT1)
main cylinder may have led to a little plastic deformation during
j the cutting process. Secondly the MR fluid 140-CG shows a
behavior that is thickness dependent. In Ref. [23], Wereley
i 0 1 2 showed that the behavior of the fluid is similar to the datasheet
value for a thickness of 0.25 mm. The smaller the thickness, the
0 0.956 0.059 3.84-005 higher the yield stress. In the system shown in Fig. 2, the thickness
1 0.549 0.0029 0 is 3.5 mm, which explains the lower values compared to those of
2 0.026 0 0 the manufacturer [1].
Nevertheless the experimental tests show the effects of the in-
ternal pressure on the properties of 140-CG. The yield stress val-
dependence and no magnetic field applied. The response surface ues under the highest level of pressure tested are eight times the
follows the law expressed in Eq. (4), with the coefficient reported values with no applied pressure. Since there is a positive interac-
in Table 4. tion between the applied magnetic field and pressure, the effect
Also for sy, the interaction between the two main variables is a can be improved greatly with a higher pressure level. This may be
positive influence (warping of the surface in Fig. 10). This means explained with the magneto-tribological model proposed by
that the combined presence of the magnetic field and the internal Zhang et al. [8] and is in good accordance with the test carried out
pressure gives a value of sy that is higher than the simple sum of by Tang et al. [7] using aluminum to test the shear properties of
the two variables considered by themselves. the MR fluid.
These are the key outcomes of this work because they clearly This property of the MR fluid may lead to the design of
testify the presence of the squeeze strengthen effect for magneto- enhanced MR-based devices without complex architectural
rheological fluids, not only in shear mode as demonstrated in Ref. changes by relying on the improved performance of the material
[8], but also in flow mode. The strong effect of the pressure on the when subjected to an internal pressure. Even though the perform-
yield stress of an MR fluid confirms the findings of Ref. [21], ances can be highly improved using the pressure as an additional
which highlights that sy is not an intrinsic property of the fluid parameter, a pressurized MR fluid system has two drawbacks as
itself but of the whole system. well. The first one is the higher off state force, which can be detri-
In order to assess and validate the procedure adopted, our mental for some application, and the second one is the increased
results in the case of no internal pressure can be compared with system complexity due to the presence of sealing rings able to
the datasheet provided by the producer [2] and reported in Fig. 1. contain the higher pressure.
For a given induction field B, using Fig. 1(b), it is possible to For example, this behavior could be used to enhance one of the
obtain the values of the magnetizing field H expressed in kA/m, first magnetorheological devices presented in the technical litera-
then enter the diagram reported in Fig. 1(a) and calculate the cor- ture: the bypass damper [24] depicted in Fig. 11. This architecture
responding yield stress value. It transpires that the measured yield comprises a reservoir, which can be equipped with an additional
stresses are lower than the datasheet values with an average differ- pressurizing system to exploit the advantages of the higher yield
ence of around 30%. Table 5 shows the numerical values and the stress of the MR fluids in flow mode under pressure. This kind of
relative errors between the datasheet and the experiments. The dif- system, which is not common in the industrial world because it is
ference between the measured values and the datasheet is quite not compact enough, can be renewed and improved thanks to the
high and the reasons are twofold. First a slight magnetization of squeeze strengthen effect.
DownloadedViewFrom:
publicationhttp://fluidsengineering.asmedigitalcollection.asme.org/
stats on 07/31/2013 Terms of Use: http://asme.org/terms