-
This content has been downloaded from IOPscience. Please scroll
down to see the full text.
Download details:
IP Address: 140.254.87.149This content was downloaded on
09/06/2014 at 08:57
Please note that terms and conditions apply.
A new magnetorheological damper for seismic control
View the table of contents for this issue, or go to the journal
homepage for more
2013 Smart Mater. Struct. 22 115003
(http://iopscience.iop.org/0964-1726/22/11/115003)
Home Search Collections Journals About Contact us My
IOPscience
-
IOP PUBLISHING SMART MATERIALS AND STRUCTURES
Smart Mater. Struct. 22 (2013) 115003 (12pp)
doi:10.1088/0964-1726/22/11/115003
A new magnetorheological damper forseismic controlYang Ding1,2,
Lu Zhang1, Hai-Tao Zhu1,2 and Zhong-Xian Li1,2
1 School of Civil Engineering, Tianjin University, Tianjin
300072, Peoples Republic of China2 Key Laboratory of Coast Civil
Structure Safety (Tianjin University), Ministry of
Education,Tianjin 300072, Peoples Republic of China
E-mail: [email protected]
Received 5 April 2013, in final form 16 July 2013Published 19
September 2013Online at stacks.iop.org/SMS/22/115003
AbstractThis paper proposes a new MR damper with bidirectional
adjusting damping forces toenhance the fail-safe property of the MR
damper. The structure of the composite magneticcircuits is improved
for the new damper. Four prototype dampers are fabricated and
tested bymagnetic field tests and dynamic tests. The magnetic field
distribution in the damping path andthe dynamic properties of the
dampers with different input currents are obtained. TheGompertz
model is proposed to portray the dynamic behavior of the prototype
dampers. Thestudy shows that, due to the improved structure of
composite magnetic circuits, the prototypedampers can maintain a
medium damping force with zero current input. This behavior
mayensure a better fail-safe property and avoid settlement of MR
fluid compared withconventional MR dampers. Furthermore, the
minimum and maximum output powers of theproposed dampers can be
obtained at the states of the negative peak and positive peak
ofcurrents inputs, respectively. In addition, the dynamic range of
controllable force is wider thanthat of conventional MR dampers.
The analysis further shows that the proposed Gompertzmodel can
precisely portray the nonlinear hysteretic behavior of the proposed
dampers withoutcomplicated function forms.
(Some figures may appear in colour only in the online
journal)
1. Introduction
Over the past few decades, magnetorheological (MR) dampershave
attracted much attention because they can significantlyreduce the
response of structures excited by seismic andwind loadings. MR
dampers possess the excellent inherentcharacteristics of stability
and adaptability. Much researchhas been done on the control
strategy and models for MRdampers. Dyke et al proposed a
clipped-optimal controlstrategy for controlling MR dampers to
reduce structuralresponses due to seismic loadings. The strategy
was basedon the acceleration feedback. Both numerical simulation
andexperiments have been conducted to verify the effectivenessof MR
dampers on the reduction of seismic response [1, 2].After studying
several idealized mechanical models ofcontrollable fluid dampers,
Spencer et al proposed aphenomenological model which can portray
the behaviorof a typical MR damper effectively and precisely
[35].
Yang et al investigated a full-scale MR damper with amaximum
damping force of 200 kN and a dynamic rangeof 10 [68]. They also
proposed both a quasi-static modeland a dynamic model for this
damper. Li and Xu designedand manufactured a double-ended shear
mode combined witha valve mode MRF-04K damper with a maximum
dampingforce at a full magnetic field strength of 20 kN. The
maximumpower required was less than 50 W [9]. Tse and
Changdesigned, manufactured, and tested a small-scale rotary typeof
MR damper [10]. They also developed a simplified andrelatively
accurate inverse dynamic model. This model candirectly relate the
damper force to the input voltage. Chooiand Oyadiji derived a
method for designing, modeling andtesting MR dampers using
analytical flow solutions [11]. Theeffectiveness of this method was
validated by simulation andtest results. The test results came from
a double-tube MRdamper fabricated at the University of Manchester.
Gavin et alstudied the optimal design of MR dampers [12]. They
also
10964-1726/13/115003+12$33.00 c 2013 IOP Publishing Ltd Printed
in the UK & the USA
-
Smart Mater. Struct. 22 (2013) 115003 Y Ding et al
developed an algebraic model utilizing the hyperbolic
tangentfunction.
MR dampers possess the fail-safe property. This propertyis one
of the most important advantages of MR dampers,making them superior
to fully active control devices. Thisis because the power supply of
fully active control devicesmay fail during the excitation of
extreme seismic and windloadings. On the other hand, analytical and
experimentalstudies have already shown that MR dampers can work
witha power supply as passive control viscous dampers do. Evenunder
such situation, MR dampers can still offer considerablecontrol
effects. When working without a power supply, MRdampers cannot
provide the best performance compared withother passive states with
power supply at fixed magnitudemost times. This limitation shows
that there is still an optimalstate for MR dampers to provide the
best passive controleffect. Carlson first theoretically discussed
this issue andwhether it was possible to use a permanent magnet to
biasa MR fluid valve or device at a mid-range condition [13].By
this means, MR dampers can be designed as the optimalpassive state
for achieving better performance without a powersupply. Du et al
and Yan et al utilized a similar structure ofcomposite magnetic
circuit to manufacture a MR damper. It iscalled an inverse MR
damper, and is capable of maintainingthe damping force at its
maximum value due to a magneticfield excited by the incorporated
permanent magnet withoutcurrent input. After that, the current can
be applied to theaccompanying electromagnetic coil to cancel the
magneticfield and decrease the damping force [14, 15]. However,the
test results on the prototype damper showed that thedynamic range
of the damping force provided by the inverseMR damper is poor, even
below 1.5. This dynamic range isobviously of no practical
significance. The results also showedthat the magnetic field
excited by the permanent magnet isdifficult to be reduced. This is
due to the magnetic saturationaccruing at the magnetic core, which
may be the mainlimitation of the composite magnetic structures
proposed bythe researchers above.
In this study, a new prototype MR damper is proposed.The
proposed MR damper is developed with bidirectionaladjusting damping
forces. The structure of the compositemagnetic circuits is improved
for the proposed MR damperto avoid magnetic saturation accruing at
the magnetic core.Therefore, the dynamic range will not be affected
comparedto conventional MR dampers. Four prototype dampers
arefabricated and tested by magnetic field tests and dynamictests.
The magnetic field distribution in the damping pathand the dynamic
properties of the dampers with differentinput currents are obtained
in the tests. In terms of theobtained test results, the Gompertz
model is proposed toportray the dynamic behavior of the prototype
dampers. Thestudy shows that, due to the improved structure of
compositemagnetic circuits, the prototype dampers can maintain
amedium damping force with zero current input. This behaviormay
ensure a better fail-safe property and avoid the settlementof MR
fluid compared with conventional MR dampers. Thecomparison further
shows that the proposed Gompertz modelcan precisely portray the
nonlinear hysteretic behavior of theproposed dampers without
complicated function forms.
Figure 1. Composite magnetic circuit.
2. The improved structure of composite magneticcircuit
Figure 1 shows a schematic diagram of the proposedcomposite
magnetic circuit and figure 2 shows the equivalentmagnetic circuit
with different current inputs. In figure 2,NI and Fm are the
magnetomotive forces induced by theelectromagnetic coil and the
permanent magnet, respectively.Rm and Rw represent the reluctance
of the permanent magnetand the damping path, respectively, while R0
represents thevariable reluctance associated with the magnetic
saturation atthe magnetic core. 81,82,83 and 84 are the magnetic
fluxaccording to states with different currents inputs,
respectively.
As shown in figure 2, the composite magnetic circuit isgoverned
by
Fm = 82R0 +81rw + (81 +82)Rm (1)Fm NI1 = 83(R0 + Rm) (2)Fm + NI2
= 84(Rm + Rw), (3)
where I1 represents the negative peak value of current inputsand
I2 represents the positive peak value of current inputs.(1)(3)
correspond to the working states with current inputsat zero,
negative peak and positive peak, respectively.
Improvements to the structure of the composite magneticcircuit
is proposed herein. One improvement concernsavoiding the
difficulties of demagnetization. The secondarygap proposed by
Carlson [13] incorporated in the compositemagnetic circuit is
removed. Instead, by means of controllingthe thickness of the
magnetic core and utilizing the magneticsaturation, the magnetic
flux can be restricted when passingthrough the magnetic core. When
there is no current input,the magnetic flux of the permanent magnet
is supposed to beproportionally distributed between the circuit
passing throughthe magnetic core and the other circuit passing
throughthe damping path. In this way it is possible to maintain
amedium damping force, as shown in figure 2(a). A secondimprovement
addresses how to obtain the minimum andmaximum output powers of the
dampers. When the damperswork in semi-active mode, the minimum and
maximum outputpower of the dampers can be obtained at the stage of
currentinput being at negative peak and positive peak,
respectively.When negative current is applied to the
electromagnetic coil,the operating point of the permanent magnet
falls along itsoperating line as shown in figure 3. The magnetic
flux passing
2
-
Smart Mater. Struct. 22 (2013) 115003 Y Ding et al
Figure 2. Equivalent magnetic circuit. (a) Zero. (b) Negative
peak. (c) Positive peak.
Figure 3. Operating point motion of permanent magnet.
through the damping path decreases until the left magneticflux
completely passes through the magnetic core as shownin figure 2(b).
In contrast, when positive current is applied tothe electromagnetic
coil, the operating point of the permanentmagnet rises along its
operating line as shown in figure 4.The magnetic flux passing
through the damping path increasesuntil the maximum value is
achieved as shown in figure 2(c).That is, the minimum, medium and
maximum values ofthe magnetic flux passing through the damping path
can beobtained at the stage of currents input being at negative
peak,zero and positive peak values, respectively. The
correspondingminimum, medium and maximum output power of the
dampercan be also obtained.
As shown in figure 3, in order to avoid magnetic
propertyvariation under repeated magnetizing and
demagnetizingoperations, the permanent magnet must possess ideal
lineardemagnetization property to eliminate the possibility of
permanent demagnetization. For this reason, NdFeBpermanent
magnets are chosen. Because of high intrinsiccoactivity of NdFeB
permanent magnets, the thicknessof the permanent should be
restricted to 12 mmfor implementing fully magnetizing and
demagnetizingoperations. This performance can be fulfilled within
limitedturns of electromagnetic coils and limited magnitude
ofcurrents.
In accordance with the above assumption, multiple designschemes
are proposed. As shown in figures 4(a) and (b),magnets with axial
distribution magnetic field are chosenfor schemes 1 and 2. The
permanent magnets are installedoutside the magnetic core in scheme
1, which ensures abigger cross section area of the permanent
magnets. Abigger magnetomotive force of the permanent magnets can
beobtained. In scheme 2, the permanent magnets are installedinside
the magnetic core, which means a smaller crosssection area. A
relatively smaller magnetomotive force of thepermanent magnets can
be produced. Because of the differentstructural arrangement,
schemes 1 and 2 can satisfy differentrequirements of MR dampers
depending on the magnitude ofdamping force. Besides the former two
schemes, permanentmagnets with radial distribution magnetic field
are chosen forscheme 3, as shown in figure 4(c). This scheme is
consideredto be a particularly simple scheme. However, the
fabricationof permanent magnets with radial distribution magnetic
fieldis still complicated and relatively costly.
In this paper scheme 1 is used for the damper, consideringthe
design purpose and process cost. In order to evaluatethe proposed
improved structure of the composite magneticcircuit, finite element
analysis is conducted. Figure 5 showsthe magnetic flux
distributions of a single section of theproposed dampers at the
stage of current input being at zero,negative peak and positive
peak values, respectively. Figure 6
Figure 4. Design schemes. (a) Scheme 1. (b) Scheme 2. (c) Scheme
3.
3
-
Smart Mater. Struct. 22 (2013) 115003 Y Ding et al
Figure 5. Distributions of magnetic flux. (a) Zero. (b) Negative
peak. (c) Positive peak.
Figure 6. Distributions of magnetic induction. (a) Zero. (b)
Negative peak. (c) Positive peak.
Figure 7. Schematic diagram of a MR damper with bidirectional
adjusting damping forces. (a) Overall structure. (b) Local
structure.
shows the induction of the magnetic distributions. As shownin
figures 5 and 6, the simulated results agree well with
theequivalent magnetic circuit diagram shown in figure 2. It
isnoted that when current input is at negative peak value,
themagnetic flux passing through the damping path
completelyvanishes. This means that the demagnetization is
fullyimplemented, which ensures a comparable minimum dampingforce
and a dynamic range compared with conventional MRdampers.
3. Design and fabrication of MR dampers
Following the above conceptual design and theoreticalanalysis,
four prototype dampers with bidirectional adjustingdamping forces
are designed and fabricated. Two conventionalprototype dampers are
also fabricated to provide a compar-ison. Figure 7 shows a
schematic diagram of the proposedMR damper with bidirectional
adjusting damping forces. The
Table 1. Geometry parameters of the prototype dampers.
h(mm)
d(mm)
D(mm)
L(mm)
t(mm)
Turns of theelectromagnetic coils
1 20 66 20 5 500
basic geometry of the proposed damper is consistent withthe
large-scale seismic MR fluid damper fabricated by Yanget al [6] and
Lord Company. The major difference is that apiece of permanent
magnet is inserted in the middle of thepiston and the thickness of
the left magnetic core is reducedto a small value, as shown in
figure 7(b).
By means of finite element simulation on the electro-magnetic
field, most of the geometry parameters can bedetermined. These
geometry parameters are listed in table 1.An illustration of the
geometry parameters is shown infigure 7(b). Two geometry parameters
are very important for
4
-
Smart Mater. Struct. 22 (2013) 115003 Y Ding et al
Figure 8. The curves of B0 versus h1.
Figure 9. The curves of B0 versus h1.
the proposed improved structure of the composite
magneticcircuit. One is the thickness of the magnetic core h1 which
isrelated to the residual magnetic induction B0 in the dampingpath
without current input. The other is the thickness of thepermanent
magnet t1 which is related to the peak value ofdemagnetization
current I1. Both relationships are obtainedby means of finite
element simulation on electromagneticfield. The numerical results
are shown in figures 8 and 9.As shown in figure 8, permanent
magnets with both 1.26 Tand 1.37 T remanence Br are selected and B0
is controlledby h1. As 500 mT is considered to be the saturation
valueof the selected MR fluid (supplied by Chongqing
InstrumentMaterials Research Institute of China, denoted MR-J),
therange of B0 from 200 to 300 mT is considered to beappropriate
for the medium output powers of the damper.Considering the
difficulty of simulation process and theproper value of B0, 1.5 and
3 mm are selected for h1. As shownin figure 9, I1 mainly depends on
t1 and two processes withdifferent h1 and Br are simulated. As a dc
power supply withmaximum output power of 30 V/3 A is used for the
tests, 2 mis selected for t1 ensuring that I1 for both processes
will notexceed the maximum output power of the dc power supply.
According to the simulated results, MR dampers withdifferent Br
and h1 are fabricated. Meanwhile, the proposedimproved structure of
composite magnetic circuit is appliedto dampers with single piston
section and multiple pistonsections. Dampers with different numbers
of piston sectionsn are fabricated. Serial numbers and other kernel
geometryparameters are listed in table 2. Figure 10 is a photo
Figure 10. Photo of prototype dampers.
Figure 11. The equipment setup for magnetic field tests.
Table 2. Serial numbers and kernel geometry parameters.
Serialnumber n h1 (mm) Br (T) t1 (mm)
MR-B1 1 3 1.26 2MR-B2 1 1.5 1.37 2MR-B3 3 3 1.26 2MR-B4 3 1.5
1.37 2
of fabricated prototype dampers with single piston sectionand
multiple piston sections, respectively. Two conventionalprototype
dampers of the same geometry parameters aredesigned to provide a
comparison. The two dampers aredenoted as MR-C1 and MR-C2 with
single piston section andmultiple piston sections,
respectively.
4. Magnetic field tests
In order to examine the effectiveness of the proposedimproved
structure of the composite magnetic circuit, a seriesof magnetic
field tests are conducted utilizing the speciallycustomized testing
device designed by the authors as shown infigure 11. A LZ-610H
teslameter is employed in conjunctionwith a dc power supply with a
maximum output power of30 V/3 A. Because the designed thickness of
the dampingpath for the MR dampers is only 1 mm, a specially
customizedHall probe is adopted with a width of 1 mm and a
thickness
5
-
Smart Mater. Struct. 22 (2013) 115003 Y Ding et al
Figure 12. Comparison between measured results and
predictedresults using a MR-B1 damper.
Figure 13. Comparison between measured results and
predictedresults using a MR-B2 damper.
of 0.32 mm. Considering that the length of the Hall probeis
limited, only the MR dampers fabricated with singlepiston section
are tested, i.e., MR-B1, MR-B2 and MR-C1,respectively. The positive
peak value of the input currents is2.5 A, which is associated with
the maximum output powerof the dc power supply. The negative peak
value of thecurrents input is related to the full demagnetization
of thedamping path. That is, the negative peak value leads to
thedisappearance of magnetic flux passing through the dampingpath.
In order to eliminate the influence of the fabrication erroron the
thickness of the damping path around the piston, fourtesting points
are uniformly distributing around the piston.The average thickness
value of these four testing points iscalculated and chosen as the
final result.
Figures 12 and 13 show the magnetic field test results ofMR-B1
and MR-B2 dampers and the comparison betweenthe measured values and
the predicted values, respectively.As shown in figures 12 and 13,
the measured values ofthe four testing points around the piston are
consistent witheach other. The uniform distribution of the magnetic
fieldin the damping path shows that the fabrication error can
benegligible and the thickness of the damping path around thepiston
is nearly constant. It is also noted that the predictions
Figure 14. Comparison of measured results among a MR-B1damper, a
MR-B2 damper and a MR-C1 damper.
of the magnetic induction in the damping path are slightlylower
than the measured values. This difference may be dueto the fact
that there is flux leakage in the piston shaft. Thepiston shaft is
made of nonmagnetic stainless iron. The shaftbecomes slightly
magnetizable after machining. Figure 14shows a comparison of
measured values among MR-B1damper, MR-B2 damper and MR-C1 damper.
As shown infigure 14, the measured values of magnetic induction
ofMR-B1 and MR-B2 with zero current input achieve 117 mTand 215 mT,
respectively. However, no magnetic field existsin the damping path
of MR-C1 damper. Due to the permanentmagnet of higher remanence and
lower-thickness magneticcore, the residual magnetic induction in
the damping pathof MR-B2 damper is obviously higher than that of
MR-B1damper. This phenomenon indicates that the residual
magneticinduction in the damping path can be changed to be in
thedesired state by controlling the remanence of the
permanentmagnet and the thickness of the magnetic core. Moreover,
themagnetic induction in the damping path of MR-B1 damperand MR-B2
damper reach zero with 1.2 A and 2.5 A negativecurrents input,
respectively. Therefore, full demagnetizationcan be achieved with
limited negative current input and higherresidual magnetic
induction at the state of zero current input.In addition, the
maximum magnetic induction in the dampingpaths of MR-B1 damper and
MR-B2 damper is almost thesame. The maximum induction is still
slightly lower than thatof MR-C1 damper. Regarding magnetic
induction increaserate, the increase rates of MR-B1 damper and
MR-B2 damperare lower than that of MR-C1 damper. There is
biggerreluctance of the permanent magnet installed in MR-B1damper
and MR-B2 damper than that of the pure iron pistonof MR-C1 damper.
Also, all measurements are obtained inair circumstance. This
condition is different from the actualoperation states of the
damping path being full of MR fluid.The actual magnetic induction
will be much higher due to thehigher permeability of MR fluid.
Thus, the improvement issufficient for the maximum output power of
the dampers.
5. Dynamic tests
Laboratory tests are further conducted to study the
dynamicperformance of the proposed MR dampers with
bidirectional
6
-
Smart Mater. Struct. 22 (2013) 115003 Y Ding et al
Table 3. Maximum output capacities and dynamic ranges of a MR-B1
damper, a MR-B2 damper and a MR-C1 damper.
Seriesnumber
Zero current inputF0 (kN)
Positive peak currentsinput Fmax (kN)
Negative peak currentinput Fmin (kN) F0/Fmax (%)
Dynamicrange
MR-B1 1.729 5.789 0.591 29.9 9.8MR-B2 2.465 6.382 0.484 38.6
13.1MR-C1 0.661 6.285 10.5 9.5
Table 4. Maximum output capacities and dynamic ranges of a MR-B3
damper, a MR-B4 damper and a MR-C2 damper.
Seriesnumber
Zero current inputF0 (kN)
Positive peak current inputFmax (kN)
Negative peak currentinput Fmin (kN) F0/Fmax (%)
Dynamicrange
MR-B3 3.893 20.295 1.171 19.2 17.3MR-B4 7.593 18.721 0.882 40.6
21.2MR-C2 2.35 21.332 11.0 9.1
Figure 15. The setup of a dynamical test equipment.
adjusting damping forces. A MTS-810 electro-hydraulic
servotester is adopted. A MTS Teststar data acquisition systemand
dc power supply are used as shown in figure 15. Thedampers are
tested under sinusoidal displacement excitation atthe frequencies
of 0.25 Hz, 0.5 Hz and 1 Hz with magnitudesbeing 5 mm, 10 mm and 15
mm, respectively. Several constantinput currents are adopted. The
positive peak value of theinput current is 2.5 A when the maximum
output power ofthe dc power supply is reached. The negative peak
value ofthe input current depends on the occurrence of the
minimumoutput power of the dampers. Figure 16 shows the peak
valuecurves of damping force versus input current for MR-B1damper,
MR-B2 damper and MR-C1 damper. Figure 17 showsthe peak value curves
of damping force versus input currentfor MR-B3 damper, MR-B4 damper
and MR-C2 damper.Table 3 gives the maximum output power versus
differentcurrent input and the dynamic range for MR-B1 damper,MR-B2
damper and MR-C1 damper. Table 4 presents themaximum output power
versus different current input andthe dynamic range for MR-B3
damper, MR-B4 damper andMR-C2 damper.
Comparison is made in these figures and tables. MR-B1damper and
MR-B3 damper with zero current input maintain20%30% of the maximum
output power when currents inputis at positive peak. MR-B2 damper
and MR-B4 damper withzero current input maintain about 40% of the
maximum
Figure 16. Maximum output capacities of a MR-B1 damper, aMR-B2
damper and a MR-C1 damper.
Figure 17. Maximum output capacities of a MR-B3 damper, aMR-B4
damper and a MR-C2 damper.
output power when current input is at positive peak.
Thisincrease is due to the permanent magnet of higher remanenceand
lower-thickness magnetic core. However, MR-C1 damperand MR-C2
damper, having conventional magnetic structure,only maintain about
10% of the maximum output powerwhen current input is at positive
peak. The comparisonshows that the proposed improved structure of
compositemagnetic circuit is effective. The output power of MR
7
-
Smart Mater. Struct. 22 (2013) 115003 Y Ding et al
Figure 18. Comparison between logistic function and
Gompertzfunction.
dampers without current input is significantly increased.
Thisadvantage can ensure a much better fail-safe property andavoid
the settlement of MR fluid. Moreover, because fulldemagnetization
can be achieved, the proposed MR damperswith bidirectional
adjusting damping forces can reach loweroutput power with current
input being at negative peakthan that of the conventional MR
dampers. The maximumoutput powers of all the tested dampers are
almost the same.Therefore, the proposed MR dampers have a bigger
dynamicrange. In addition, the negative peak of current input
forMR-B2 damper and MR-B4 damper are higher than thoseof MR-B1
damper and MR-B3 damper. This result is inaccordance with that of
the above magnetic field tests. Finally,the maximum output powers
of MR dampers with threepiston sections are almost three times as
large as those ofthe MR dampers with single piston section. This
indicatesthat the proposed improved structure of composite
magneticcircuit can be applicable to the dampers with either
singlepiston section or multiple piston sections. The output
poweris proportional to piston section numbers.
6. Dynamic modeling of MR dampers
MR dampers have inherent hysteretic characteristics, whichhave
to be considered in the dynamic model of MR dampers.For the
stressstrain behavior, the Bingham viscoplasticmodel was proposed
by Shames and Cozzarelli [16].Similarly, Stanway et al proposed a
simplified mechanicalmodel, denoted the Bingham model, describing
the behaviorof electro-rheological (ER) and MR dampers [17, 18].
TheBingham model has a simplified function form, which isapplicable
for the initial design of MR dampers. However,this model is not
adequate for the control analysis dueto its deficiency in
describing the nonlinear hystereticcharacteristics of MR dampers.
Spencer et al proposed aphenomenological model of MR dampers which
is ableto precisely portray the nonlinear hysteretic behavior ofMR
dampers [3]. This phenomenological model considersforcedisplacement
relationship and forcevelocity relation-ship using a complicated
function form. This limitation pre-vents the practical application
of the phenomenological modelto semi-active control systems.
Several different models for
Figure 19. Comparison between dynamic models based on
theGompertz function and the logistic function.
MR dampers have been proposed by researchers [1923]. Liand Li
proposed a double-sigmoid model with a symmetricalsigmoid function
[24]. The corresponding experimentalverification was also
conducted.
This study presents an improved dynamic model denotedthe
Gompertz model. This model is adopted to simulate thedynamic
behavior of MR dampers with bidirectional adjustingdamping forces.
A Gompertz function, named after BenjaminGompertz, is a sigmoid
function. It is a mathematical modelof a time series with the
growth being slowest at the start andthe end of a period. The
left-hand or lower-value asymptote ofthe function can be approached
much more gradually by thecurve than the upper right-hand or
future-value asymptote asshown in figure 18. This differs from the
logistic function. Thelogistic function has symmetrical asymptotes.
The formula ofGompertz function is given by
y(t) = aebect , (4)where a is the upper asymptote, c is the
growth rate, b and care negative numbers and e is Eulers number.
Consideringthe asymmetry of the Gompertz function, it seems tobe
more suitable for dynamic modeling of MR dampers.Figure 19 shows a
comparison between dynamic modelsbased on the Gompertz function and
the logistic function. Theexperimental result comes from the MR-B4
damper responseunder a 15 mm-amplitude sinusoidal displacement
excitationof the frequency of 0.5 Hz. The magnitude of input
current is2.5 A. As shown in figure 19, the dynamic model based on
theGompertz function is superior to the dynamic model based onthe
logistic function, especially in the regions of low velocityand
high velocity.
The proposed Gompertz model is governed by
F = 2fd(0.5 e ln 2axsgn(x)x0 )sgn(x)+ xcd, (5)where fd
represents the controllable Coulomb damping force;cd and control
the viscous damping at large velocitiesand small velocities,
respectively; x and x are pistondisplacements and velocities,
respectively; and xd is the pistonvelocity when the damping force
is zero. A total of fourparameters (fd, cd, , and xd) are needed to
characterize theMR damper. Equation (5) can only simulate the
damper
8
-
Smart Mater. Struct. 22 (2013) 115003 Y Ding et al
Figure 20. The curves of model parameters versus input current
I. (a) fd0. (b) cd0. (c) xd0.
Table 5. Relations between model parameters and input current
I.
I fd0 cd0 xd0
2.5 0.6 0.003 51.5 1.45 0.006 8
0 6 0.018 141.5 14.5 0.027 19.52.5 15.5 0.035 21
response when the applied current, frequency and amplitudeare
held at a constant level. As MR dampers are always usedas
semi-active control devices, the adopted dynamic modelhas
fluctuating current, frequency and amplitude. Thereforethe
parameters are defined as the functions of input current Iand
maximum velocity vmax.
First, the relationships between model parameters andapplied
current I are studied. The relationships between modelparameters
and applied current I are formulated according tothe MR-B4 damper
response. A 15 mm-amplitude sinusoidaldisplacement excitation of
the frequency of 0.5 Hz wasapplied. The magnitude of an input
current is 2.5 A. Table 5presents the results. A logistic function
is used to fit the dataas shown in figure 20. From the test results
and the fittingfunction, is kept constant and the other three
parameters aregiven by
fd0 = Af + Af Bf1+ e(ICf )/Df (6)
cd0 = Ac + Ac Bc1+ e(ICc)/Dc (7)
xd0 = Ax + Ax Bx1+ e(ICx)/Dx , (8)
where Af ,Bf ,Cf ,Df ,Ac,Bc,Cc,Dc,Ax,BxCx and Dx areundefined
parameters depending on the test results and thefitting results.
These parameters are listed in table 6.
Second, the relationships between model parameters andmaximum
velocity vmax are introduced. The model parametersdepending on vmax
are given by
fd = (kf vmax + tf )fd0 (9)cd = (kcvmax + tc)cd0 (10)xd =
(kxvmax + tx)xd0 (11) = kvmax + t, (12)
where kf , tf , kc, tc, kx, tx, k and t are undetermined
parame-ters related to vmax, and their values are obtained though
thelinear regression analysis of test results. The regression
resultsof the MR-B4 damper are listed in table 7.
Finally, a total of 20 parameters (Af ,Bf ,Cf ,Df , kf ,tf
,Ac,Bc,Cc,Dc, kc, tc,Ax,BxCx,Dx, kx, tx, k and t) areneeded to
characterize the MR dampers with bidirectionaladjusting damping
forces. The Gompertz model with inputcurrent I and maximum velocity
vmax is given by
fd = Af + Af Bf1+ e(ICf )/Df (kf vmax + tf ) (13)
cd = Ac + Ac Bc1+ e(ICc)/Dc (kcvmax + tc) (14)
xd = Ax + Ax Bx1+ e(ICx)/Dx (kxvmax + tx) (15)
= kvmax + t. (16)
Table 6. Fitting results of fd0, cd0 and xd0.
Af Bf Cf Df Ac Bc Cc Dc Ax Bx Cx Dx
0.76 15.78 0.31 0.52 0.007 0.052 0.771 1.962 2.23 22.64 0.39
1.15 0.9
Table 7. Fitting results of undetermined parameters related to
vmax.
kf tf kc tc kx tx k t
0.0013 0.8784 0.0141 2.3376 0.0093 0.1123 0.0026 0.6568
9
-
Smart Mater. Struct. 22 (2013) 115003 Y Ding et al
Figure 21. The behavior of a MR-B4 damper under a sinusoidal
excitation of 15 mm-amplitude and 1 Hz-frequency.(a)
Forcedisplacement curve. (b) Forcevelocity curve.
Figure 22. The behavior of a MR-B4 damper under a sinusoidal
excitation of 10 mm-amplitude and 0.5 Hz-frequency.(a)
Forcedisplacement curve and (b) forcevelocity curve.
Table 8. Identification results of a MR-B1 damper.
Af Bf Cf Df Ac Bc Cc Dc Ax Bx Cx Dx kf tf kc tc kx tx k t
0.49 4.90 0.45 0.38 0.0008 0.0104 0.315 0.66 4.79 12.07 0.35
0.40 0.0023 0.78 0.026 3.43 0.0095 0.12 0.0034 0.49
A comparison between the measured and predictedresults of the
MR-B4 damper is plotted in figures 21 and22. The red solid line and
the black dashed line representthe measured results and the
predicted results, respectively.The results in the cases of 15
mm-amplitude sinusoidaldisplacement excitation at 1 Hz and 10
mm-amplitudesinusoidal displacement excitation at 0.5 Hz are
plotted.
The modeling of the MR-B1 damper is conducted usingthe same
method. The identification results of the MR-B1damper are listed in
table 8. The results in the cases of15 mm-amplitude sinusoidal
displacement excitation at 1 Hzand 10 mm-amplitude sinusoidal
displacement excitation at1 Hz are presented in figures 23 and
24.
As shown in figures 2124, the Gompertz model can tracethe
measured results well. The Gompertz model can portraythe nonlinear
dynamic characteristics of the proposed MRdampers precisely using a
relatively simplified function formcompared with the
phenomenological model [3].
7. Conclusions
In this study, four prototype MR dampers with
bidirectionaladjusting damping forces were designed, fabricated
and
tested. An improved structure of the composite magneticcircuit
is proposed. Performance tests, including magneticfield tests and
dynamic tests, were conducted in order toverify the effectiveness
of the proposed improved structure.For practical application, a
Gompertz model is proposed andadopted to describe the nonlinear
hysteretic behavior of theproposed MR damper.
The test results show that, due to the improvedstructure, the
proposed dampers are capable of maintaininga medium damping force
with zero currents input. Thisadvantage may ensure a better
fail-safe property and avoidsettlement of the MR fluid compared
with conventionalMR dampers. Furthermore, the minimum and
maximumoutput powers of the dampers can be obtained when
inputcurrents reach negative peak and positive peak
values,respectively. Moreover, the dynamic range of
controllableforce is even larger than that of conventional MR
dampers.Meanwhile, by controlling the remanence of the
permanentmagnet and the thickness of the magnetic core, the
residualmagnetic induction in the damping path can be changedto
accommodate the actual application requirements. Thestudy further
shows that the proposed improved structure of
10
-
Smart Mater. Struct. 22 (2013) 115003 Y Ding et al
Figure 23. The behavior of a MR-B1 damper under a sinusoidal
excitation of 15 mm-amplitude and 1 Hz-frequency.(a)
Forcedisplacement curve. (b) Forcevelocity curve.
Figure 24. The behavior of a MR-B1 damper under a sinusoidal
excitation of 10 mm-amplitude and 1 Hz-frequency.(a)
Forcedisplacement curve. (b) Forcevelocity curve.
composite magnetic current is applicable to dampers witheither
single piston section or multiple piston sections. Theproposed
Gompertz model can precisely portray the nonlinearhysteretic
behavior of the proposed dampers with a simplefunction form.
Acknowledgments
The present work is jointly supported by the SpecializedResearch
Fund for the Doctoral Program of Higher Educationof China under
Grant No. 20110032110042 and the NationalBasic Research Program of
China (973 Program) under GrantNo. 2011CB013606.
References
[1] Dyke S J, Spencer B F Jr, Sain M K and Carlson J D
1996Modeling and control of magnetorheological dampers forseismic
response reduction Smart Mater. Struct. 5 56575
[2] Dyke S J, Spencer B F Jr, Sain M K and Carlson J D 1998
Anexperimental study of MR dampers for seismic protectionSmart
Mater. Struct. 7 693703
[3] Spencer B F Jr, Dyke S J, Sain M K and Carlson J D
1997Phenomenological model of magnetorheological dampersJ. Eng.
Mech. 123 2308
[4] Spencer B F Jr and Sain M K 1997 Controlling buildings: anew
frontier in feedback IEEE Control Syst. Mag. 17 1935Special Issue
on Emerging Technology
[5] Spencer B F Jr and Soong T T 1999 New applications
anddevelopment of active, semi-active and hybrid controltechniques
for seismic and non-seismic vibration in theUSA Proc. Int.
Post-SMiRT Conf. Seminar on SeismicIsolation, Passive Energy
Dissipation and Active Control ofVibration of Structures (Cheju,
Aug. 1999)
[6] Yang G, Ramallo J C, Spencer B F Jr, Carlson J D andSain M K
2000 Large-scale MR fluid dampers: dynamicperformance
considerations Proc. Int. Conf. Advances inStructure Dynamics (Hong
Kong, 2000) pp 3418
[7] Yang G, Jung H J and Spencer B F Jr 2001 Dynamic model
offull-scale magnetorheological dampers for civil
engineeringapplications Proc. US-Japan Workshop Smart Structures
forImproved Seismic Performance in Urban Regions (Seattle,2001) pp
21324
[8] Yang G, Spencer B F Jr, Carlson J D and Sain M K
2002Large-scale MR fluid dampers: modeling and dynamicperformance
considerations Eng. Struct. 24 30923
[9] Li Z X and Xu L H 2005 Performance tests and hysteresismodel
of MRF-04K damper J. Struct. Eng. 131 13036
[10] Tse T and Chang C C 2004 Shear-mode
rotarymagnetorheological damper for small-scale structuralcontrol
experiments J. Struct. Eng. 130 90411
[11] Chooi W W and Oyadiji S O 2008 Design, modeling andtesting
of magnetorheological (MR) dampers usinganalytical flow solutions
Comput. Struct. 86 47382
[12] Gavin H, Hoagg J and Dobossy M 2001 Optima design of
MRdampers Proc. US-Japan Workshop Smart Structures forImproved
Seismic Performance in Urban Regions (Seattle)pp 22536
11
-
Smart Mater. Struct. 22 (2013) 115003 Y Ding et al
[13] Carlson J D 2002 Permanent-Electromagnet
Systemswww.lord.com
[14] Du X L, Niu X D, Liao W Z and Jia P 2006 Design
andexperimental studies on the inverse controlmagneto-rheological
damper J. Vib. Shock 25 4953(in Chinese)
[15] Yan W M, Ge H J, Dong B and Ji J B 2007 Test
ofreverse-effect MR damper Earthq. Eng. Eng. Vib.27 1716 (in
Chinese)
[16] Shames I H and Cozzarelli F A 1992 Elastic and
InelasticStress Analysis (Englewood Cliffs, NJ: Prentice-Hall)pp
1202
[17] Stanway R, Sproston J L and Stevens N G 1985
Non-linearidentification of an electrorheological vibration
damperIFAC Identification and System Parameter Estimationpp
195200
[18] Stanway R, Sproston J L and Stevens N G 1987
Non-linearmodeling of an electro-rheological vibration damperJ.
Electrostat. 20 16784
[19] Gavin H P, Hanson R D and Filisko F E
1996Electrorheological dampers, part 1: analysis and designJ. Appl.
Mech. 63 66975
[20] Gavin H P, Hanson R D and Filisko F E
1996Electrorheological dampers, part 2: testing and modelingJ.
Appl. Mech. 63 67682
[21] Kamath G M, Hurt M K and Wereley N M 1996 Analysis
andtesting of Bingham plastic behavior in
semi-activeelectrorheological fluid dampers Smart Mater. Struct.5
57690
[22] Wereley N M and Pang L 1998 Nondimensional analysis
ofsemi-active electrorheological and magnetorheologicaldampers
using approximate parallel plate models SmartMater. Struct. 7
73243
[23] Wereley N M, Pang L and Kamath G M 1998 Idealizedhysteresis
modeling of electrorheological andmagnetorheological dampers 1998
J. Intell. Mater. Syst.Struct. 9 6429
[24] Li X L and Li H N 2006 Double-sigmoid model
formagnetorheological damper and corresponding
experimentverification J. Vib. Eng. 19 16872 (in Chinese)
12
A new magnetorheological damper for seismic
controlIntroductionThe improved structure of composite magnetic
circuitDesign and fabrication of MR dampersMagnetic field
testsDynamic testsDynamic modeling of MR
dampersConclusionsAcknowledgmentsReferences