Professional Documents
Culture Documents
Asme 125 (Vibration & Acoustic) PDF
Asme 125 (Vibration & Acoustic) PDF
Editor
LAWRENCE A. BERGMAN
Assistant to the Editor
LINDA CONWAY
Past Editors
F. EHRICH
T. CONRY
D. J. INMAN
TECHNICAL PAPERS
249
252
257
267
274
282
290
299
307
317
Manager, Journals
JOAN MERANZE
324
Production Coordinator
RAY RAMONAS
328
335
343
351
359
Production Assistant
MARISOL ANDINO
Transactions of the ASME, Journal of Vibration and
Acoustics (ISSN 1048-9002) is published quarterly
(Jan., April, July, Oct.) by The American Society of
Mechanical Engineers, Three Park Avenue, New York, NY
10016. Periodicals postage paid at New York, NY and
additional mailing offices. POSTMASTER: Send address
changes to Transactions of the ASME, Journal of
Vibration and Acoustics, c/o THE AMERICAN SOCIETY OF
MECHANICAL ENGINEERS, 22 Law Drive, Box 2300,
Fairfield, NJ 07007-2300. CHANGES OF ADDRESS must be
received at Society headquarters seven weeks before
they are to be effective. Please send
old label and new address.
STATEMENT from By-Laws. The Society shall not be
responsible for statements or opinions advanced in papers
or ... printed in its publications (B7.1, Par. 3).
COPYRIGHT 2003 by the American Society of Mechanical
Engineers. Authorization to photocopy material for
internal or personal use under circumstances not falling
within the fair use provisions of the Copyright Act is granted
by ASME to libraries and other users registered with
the Copyright Clearance Center (CCC) Transactional
Reporting Service provided that the base fee of $3.00 per
article is paid directly to CCC, 222 Rosewood Dr.,
Danvers, MA 01923. Request for special permission or bulk
copying should be addressed to Reprints/Permission
Department. INDEXED by Applied Mechanics Reviews and
Engineering Information, Inc. Canadian Goods and
Services Tax Registration #126148048
Contents continued
Journal of Vibration and Acoustics
365
The Effect of Actuator and Sensor Placement on the Active Control of Rotor Unbalance
Marty E. Johnson, Luiz P. Nascimento, Mary Kasarda, and Chris R. Fuller
374
384
389
JULY 2003
TECHNICAL BRIEFS
398
405
407
L. Likhterov
A. Berman
Institutes for Applied Research
and Dept. of Biotechnology Engineering,
Ben-Gurion University of the Negev,
Beer-Sheva, 84105, Israel
Spectrum of High-Frequency
Acoustic Noise in Inviscid
Liquid-Linear Approximation for
Spherical Waves
The high-frequency asymptotics of the acoustic noise spectrum is considered for the case
of spherically symmetric waves propagating in an unbounded inviscid liquid. Using the
Kirkwood and Bethe hypothesis regarding kinetic enthalpy, the Euler equations, the equation of state in the Taits form and following linearization allow the kinetic enthalpy and
reduced pressure to be obtained. The Fourier transform yields the spectral density of
acoustic energy which proves to be inversely proportional to the square frequency and
decreases approximately by 6 decibels per octave with increase of a frequency.
DOI: 10.1115/1.1570446
Introduction
t
r
r
r
pB
p 0 B
0
(1)
(3)
2
c
r h
t
r
2
0,
(4)
Initial Equations
1 p
t
r
r
,
t
r
r
(5)
2 1 h
2
.
r
r
c0 t
(6)
c0
p n1/n 1 ,
n1 *
(7)
where c 20 n/ 0 (p 0 B) is the sound speed in undisturbed medium and p pB/p 0 B is the reduced pressure. The den*
sity, , will then be determined as
0 p 1/n .
*
1 h 2
,
r c 20 t
r
p
*
(11)
(12)
(21)
nI h
c 20
n/n1
(22)
(23)
Using Eq. 6 once more, Eq. 13 can be rewritten in the following form:
C2
sin kr t
e .
r
(24)
p 1 sin kr t
e .
k
r
(25)
Acoustic Spectrum
The Fourier transform of an aperiodic function has the form:
p t
L i e itd ,
(26)
(27)
where
L i
1
2
p t e i t dt
(16)
n
c 20
2h 2 h 1 2h 1 h 2 4 h 8 2 2h
2 2
,
r 2 r r c 20 t 2 c 40 t
r c 20 t
r
c0 tr
(14)
d 2 A 2 dA
k 2 A0,
dr 2 r dr
pB
n
1 2 h . . .
p 0 B
c0
2h 2h
1 h 2 4 2 h
2 h
2
.
4
r r c0 tr r r
r
c0 t2
c0 t
2h 2 h 1 2h
0.
r 2 r r c 20 t 2
1 t
e .
r
(20)
2 4 4 2 1 h 2
2 4
,
r
r r
r
c0 t
(19)
(10)
h h 4 2 h 1 h
2
2 2 2.
r
r
c0 tr r r2
c0 t
2
1
1
cos krC 2 sin kr,
r
r
AC 1
(9)
1 h 2
h
2
.
t
r
r
c0 t
(18)
u k 2 u0,
(8)
Determination of Enthalpy
e 1i t dtK
1
,
1i
(28)
where k /c 0 .
Multiplying this equation by r, this equation connected with
the Bessels equation becomes:
rA 2A k 2 rA0.
S L i 2,
(17)
(29)
therefore, with the accuracy up to numerical multiplier, the spectral density changes inversely proportional to the frequency
square:
S 2 .
(30)
Conclusion
The result obtained shows that a linearization of initial equations gives the decrease of the spectral density with the frequency
growth approximately 10 log 226 decibels per octave for
spherical waves propagating in inviscid medium. This may be
found useful for acoustic signal processing on background of an
ambient noise and separation and classification of acoustic
sources.
References
1 Medwin, H., and Clay, C. S., 1998, Fundamentals of Acoustical Oceanography, Academic Press.
2 Updegraff, G. E., and Anderson, V. C., 1991, Bubble Noise and Wavelet
Spills Recorded 1 m Below the Ocean Surface, J. Acoust. Soc. Am., 86, pp.
2264 2279.
Yukio Nakajima
Bridgestone Corporation
3-1-1, Ogawahigashi-Cho, Kodaira-Shi, Tokyo
187-8531, Japan
e-mail: nakaj1-y@bridgestone.co.jp
A theory on pitch noise of a tire has been developed based on the assumption that noise
is approximated by a series of Dirac delta-function. The proposed theory predicts the
existence of special lug shapes where tire noise becomes almost zero. In addition a tire
noise simulation system is developed based on this theory to predict tire noise from CAD
data. The prediction of overall noise of various tires with practical patterns is verified to
be in good agreement with the measurement. DOI: 10.1115/1.1569513
Introduction
y i x g i x i .
g i (x) is the function of lug shape and i are coordinates determined by pitch variation. The peak amplitude of tire noise spectrum is reduced by a pitch variation treatment 7. The spectrum
with pitch variation treatment is widely distributed in a frequency
range like the white noise which has the same spectrum amplitude
in every frequency. The tonal noise, hence, can be improved by a
pitch variation treatment.
Substituting Eq. 3 for Eq. 1 and integrating it with respect to
the meridian direction, tire noise F(t) is obtained. By applying a
Fourier transformation, F(t) can be described by noise amplitude
a n at nth harmonics as follows:
F t
f t,x
w x tt x
i1
y i x
t i x
V
(1)
h t tt i dth t i .
(2)
ae
n
i1
i n t
(4)
2n
.
T
(5)
F t e i n t dt
f t,x e i n t dtdx.
(6)
e
j1
j /V
w x e i n g j x /V dx
(7)
f t,x dx
i1 . . . N
where w(x) is noise amplitude in the meridian direction represented by x, y i (x) is the ith lug position in the circumferential
direction measured by a reference point, and V is the velocity. is
Dirac delta-function defined by
(3)
1
Wx x
W
(8)
where and are noise amplitude at the tire center (x0) and
noise amplitude at the edge of tire footprint (xW), respectively.
2.2 Theory on Magic Angle 12. Simplified tire pattern
is modeled by assuming that the lug is a straight line with the
same lug angle . The shape of the lug is described by
g i x tan x
i1 . . . N .
(9)
Furthermore, assume that noise amplitude is the same in the meridian direction, w(x)1. Substituting Eq. 9 for Eq. 7 and
integrating it, the following equation is obtained:
a n 2
2V
nW
tan d n
sin
n tan
2V
(10)
e j /V
j1
(11)
where L(TV) is the circumference length of a tire. This equation expresses the effect of the lug angle on pitch noise. P n ( )
possesses multiple zeros where noise is never generated. We call
this lug angle the magic angle described by
The ratio of noise with lug angle and noise with zero lug angle
is derived by using Eq. 10 as follows:
P n
a n 2
2V
nW
tan
sin
a n 0 2
n W tan
2V
nW
tan
L
nW
tan
L
sin
(12)
iM tan1
iL
nW
i1 .
(13)
g j x g x x tan 2M
bx
i2
(14)
prediction
a A
n1
n 2
(15)
Summary
noise level and the existence of magic angles, where tire noise is
theoretically zero, by the noise measurement of a grooved tire.
The theory was extended to pitch noise where noise amplitude
has distribution in the meridian direction. In this case, the shape of
the lug in minimizing tire noise was no longer a straight line and
became a complicated shape. The optimized lug shape, called the
magic shape, was obtained by using optimization technique. Furthermore the tire noise prediction system was developed in order
to apply this idea to the practical pattern. The prediction was
verified to be in good agreement with the measurement.
The proposed theory and the prediction system can be applicable not only to a tire but also to a timing belt or gear where gear
or tooth pitch causes noise.
Nomenclature
a n 2
An
bi
f (t,x), F(t):
g i (x)
L
n
N
T
ti
V
w(x)
W
x
yi
iM
References
1 Nakajima, Y., 1980, Theory on Pitch Noise I, Internal Report, Bridgestone
Corporation.
2 Nakajima, Y., 1981, Theory on Pitch Noise II, Internal Report, Bridgestone Corporation.
3 Nakajima, Y., Inoue, Y., and Ogawa, H., 1993, Application of the Boundary
Element Method and Modal Analysis to Tire Acoustics Problems, Tire Sci.
Technol., 21, p. 66.
4 Koike, H., Oshino, Y., and Tachibana, H., 1998, Identification of Source of
Tire/Road Noise in Japanese, Souonseigyo, 22, p. 11.
5 Mukae, T., et al., 1978, Clarification of Mechanism of Tire Noise Generation
in Japanese, Report of JARI, No. 51.
6 JATMA, 1997, On Noise due to Tire and Road, 6th edition in Japanese.
7 Nakajima, Y., and Abe, A., 2000, Application of Genetic Algorithms of Optimization of Tire Pitch Sequences, Japan Journal of Industrial and Applied
Mathematics, 17, pp. 403.
8 Vanderplaats, G. N., 1984, Numerical Optimization Techniques for Engineering Design with Applications, McGraw-Hill.
9 Yoshida, M., and Takayama, M., 1994, Method of Simulating Tire Tread
Noise and Simulator Therefore, USA Patent #4,838,330.
Jeffrey S. Vipperman
Assistant Professor, Associate Mem. ASME,
e-mail: jsv@pitt.edu
Deyu Li
Ilya Avdeev
John Swanson Graduate Fellow,
Student Mem. ASME
Department of Mechanical Engineering,
University of Pittsburgh,
Pittsburgh, PA 15261
Steven A. Lane
Aerospace Engineer,
Air Force Research Laboratory,
Space Vehicles Directorate,
Kirtland AFB, NM 87117
Introduction
Novel composite structures that have application for use as expendable launch vehicle fairings are being investigated. Composites are already offering higher strength, less weight, and larger
payload envelopes than current systems. Two examples of innovative composite construction are the Advanced Grid-Stiffened
AGS 1 and Chamber Core designs 2, which are additional
contenders to replace the conventional machined-aluminum isogrid structures. Both are fabricated of filament-wound graphite
fiber that has been preimpregnated with an epoxy binder. The
improvements in weight 40 60% decrease and strength 1,000%
increase along the hoop direction are accompanied by a worsening of the sound transmission loss TL behavior of the structure,
which is governed by the physical structural properties mass,
stiffness, and damping and the geometry. The interior acoustic
levels experienced during launch are significant enough to damage
the sensitive cargo.
Initial studies of AGS structures have focused on static or buckling strength, the impact of defects, effects of damage or repairs,
and rudimentary dynamic behavior 1. Modeling studies on improving the transmission loss in fairings have been conducted for
Minotaur-sized launch vehicles where passive 3 and feedback
active structural-acoustic control ASAC 4,5 were numerically
evaluated. Two earlier experimental studies focused on specific
types of transmission control for the AGS mock-scale prototype,
including evaluating the effectiveness of hybrid structural and
acoustic control inputs 6 as well as positive position feedback
PPF controllers 7, which do not require overly complicated
system models and can offer some stability guarantees. An active
control study on a full-scale model has shown that actuator control
authority and power limitations may prove difficult 8. A more
recent study 9 briefly examined an isotropic, truncated, conical
shell and investigated the effects of cavity resonances and ring
Contributed by the Technical Committee on Vibration and Sound for publication
in the JOURNAL OF VIBRATION AND ACOUSTICS. Manuscript received Feb. 2002;
Revised Nov. 2002. Associate Editor: R. F. Keltie.
SA
TL10 log10
W inc
1
10 log10
W trans
mn r o pq r dS
(2)
(1)
NRS10 log10
p ext 2
p int 2
(3)
Fig. 3 Cylindrical coordinate system and dimensions that describe the structure
Special
Structural-Acoustic
Frequencies. Certain
structural-acoustic frequencies are important in characterizing the
sound transmission behavior of cylindrical structures 12,18,19,
including the ring frequency, f R , the critical coincidence frequency, f C , and the internal cut-off frequencies, f CO . The ring
frequency occurs when the longitudinal wavelength equals the
circumference and it sets the threshold where the stiffness effects
from the curvature no longer impact the vibrational behavior of a
cylinder. For high frequencies ( f f R ), the vibration and associated acoustic radiation behave much like a flat plate. Transmission
is affected since the internal radiation efficiencies tend to be unity
above f R . Below the ring frequency, the circumferential modes of
the cylindrical structure will occur in pairs rotated 90 of each
other, which are identical for axisymmetric structures. A dip in
noise transmission loss is typically observed in the vicinity of f R
due to the high number of internal coincidence phenomena that
occur 12. Computation of f R is accomplished with the following
formula
f R
cl
,
2R
(4)
where both end radii are used to compute two ring frequencies
and c l E/ (1 2 ) is the longitudinal plate wave speed, where
E is the Youngs modulus, is the density of the structural material, and 12 130.245 is the Poissons ratio for the material. Since the structure is anisotropic, average effective values for
the modulus and thickness were determined with FEA and used to
compute the wave speeds. It was determined from the FEA model
that an average isotropic modulus of E eff47 GPa produces structural natural frequencies that are commensurate with the experimentally observed resonance frequencies. Using the value for
E eff , the two ring frequencies corresponding to the two end radii
(R 1 0.19 m, R 2 0.305 m) are estimated to be: f R1 4,840 Hz
and f R2 3,020 Hz. Data presented in a previous study 9 suggests that truncated conical shells do not show dips at or between
f R1 and f R2 , as expected, but may occur above f R2 .
JULY 2003, Vol. 125 259
f CO (Hz)
(R0.19 m)
f CO (Hz)
(R0.305 m)
10
20
01
30
329
546
686
752
529
876
1,100
1,207
c 20 3
tcl
(5)
k rpq c 0
2
(6)
which is why they were included in the model. Isotropic homogeneous media was used to model the shell and rib composites, and
a parametric study determined that an effective stiffness of 47 GPa
produces similar modal behavior to the real system. Dynamic behavior from the model was evaluated by comparing the first eight
natural frequencies with those measured in the experiments. The
novel beam-shell mixed FEA model proved to be computationally
effective and accurate.
Figure 5 shows the first three mode shapes predicted by the
FEA model. Only the shell is shown in Fig. 5 since the end caps
would obstruct the cylinder modes. Both the end and side view of
the lowest frequency mode 0,3 repeated is shown at the top
two panes of Fig. 5. The next two modes: 1,0 and 0,2 are
shown in the bottom two panes of Fig. 5. Similarity is noted
between the predicted fairing modes and those for a uniform cylinder. Note that the lowest frequency bending mode for a cylindrical shell does not correspond to the lowest indices: (m,n)
(0,1). It is interesting to observe that the low-frequency modes
of the structure are predicted to respond globally as a shell, rather
than a collection of smaller panels created by the rib stiffeners.
This may or may not be true, since the unidirectional composite
Predicted
Resonant Freq.
r,FEA Hz)
Measured
Resonant Freq.
r,Exp Hz)
%Error Between
r,FEA & r,Exp
Measured
Damping
%
0,3
0,3
1,0
0,2
0,4
0,2
0,4
615
618
702
761
776
818
842
612
627
678
699
775
787
794
0.5
1.2
3.2
8.1
0.2
3.8
4.5
0.8
0.7
0.6
0.8
1.1
1.0
0.5
were given in the first three rows of Table 1. Interestingly, although they are significant for sound transmission, many of these
modes appeared insignificant in the acoustic FRFs. The difference
results from the direct excitation by the speaker for the FRF versus excitation via transmission for the NRS measurement. Previous studies focused mainly on the 100 mode as the primary
contributor to noise transmission 6,7.
A curve-fit was performed on an arbitrary selection of 32 of the
260 FRFs so that modal properties could be extracted. Natural
frequencies and damping ratios for the first eight modes are given
in columns 3 and 5, respectively, of Table 3. The predicted mode
shape and natural frequencies from the FEA analysis are given in
columns 1 and 2 of the same table, along with the error between
predicted and measured natural frequencies in column 4, which
Predicted
Resonant Freq.
r,FEA Hz)
Measured
Resonant Freq.
r,Exp Hz)
% Error Between
r,FEA & r,Exp
Measured
Damping
%
100
0102
1102
0202
020
2102
001
1202
313
383
550
617
622
765
787
815
301
385
547
583
613
735
788
840
4.0
0.5
0.5
5.6
1.4
3.8
0.1
3.1
0.29
0.23
0.18
0.22
0.27
0.14
0.20
0.33
to modal index n for the structure and p for the acoustic cavity.
When comparing the modal indices n and p from the first column
of Tables 2 and 3, one observes that (n3p2) at 612 Hz and
(n2p1) at 787 Hz.
Fig. 10 Measured noise reduction spectrum showing acoustic and structural modes
tion by the ring frequency. Referring back to the section on transmission loss, the stiffness-controlled region will begin at 0 Hz and
end at the first resonance, which is an acoustic cavity resonance at
300 Hz. The resonance-controlled region was predicted to continue through approximately twice the first structural resonance
frequency or 26121,224 Hz. Judging from the figure, the
dips in NRS from the resonances are noted to stop around 1,500
Hz and a general upward trend begins until a wide dip begins at
6,000 Hz. The slope in the mass-controlled region is 28 dB per
decade rather than the classical 20 dB/decade mass law. Since f C
was predicted to be 6,230 Hz, this dip is attributed to external
coincidence. The width of the dip is rather large 6,00012,000
Hz, which may be attributed to effects of the stiffeners and/or
ring frequency effects 12. Regardless, the dip is inconsequential
when compared to the levels at the singularities in the NRS that
occur at internal coincidence frequencies and in the region around
where the resonance- and mass- controlled regions meet.
Note that to improve the noise transmission behavior will
largely require the acoustically dominant modes and/or internal
coincidence phenomena to be controlled. Internal absorptive treatments are limited in rocket fairings due to the volume requirements of the payload. The NRS could perhaps also be improved at
612, 787, 794, 845, and 950 Hz by adding damping treatments or
control that targets these structural modes.
Conclusions
The noise transmission behavior for a novel, advanced gridstiffened AGS composite structure has been investigated. This
material is being considered by the US Air Force for constructing
expendable space launch vehicles. A combined numerical/
experimental approach was used to examine the physical mechanisms of noise transmission by the structure. Separate analyses
were performed on the structure and the acoustic cavity since the
coupling between the two was found to be very light. Finite element analysis permitted a smeared or average value for the composite elastic modulus to be determined, which in turn was used to
estimate the ring and critical frequencies for the structure, since
they influence transmission loss. Mode shapes were also predicted
by FEA, which allow the modal pressure distributions to be examined and which also permit investigation of the internal coincidence phenomena. Experimental modal analyses on the structure
and cavity were also performed. Tap tests at various locations
around the structure provided frequency response functions
FRFs that were curve-fit to obtain resonant frequencies and
damping ratios. A speaker was used to drive the acoustic cavity
with a white input while the pressure response was measured inside the cavity. Again, the measured FRFs were fitted to obtain
resonant frequencies and damping ratios for the acoustic modes.
Comparison of measured and predicted natural frequencies permitted corroboration of the FEA models. The resonances for the
structure and especially the acoustic cavity were found to significantly affect the noise transmission into the structure.
Finally, direct measurement of sound transmission was conducted based upon the noise reduction spectrum NRS, which is
computed from the ratio of the spatial averages of acoustic pressures measured along the internal and external surfaces of the
shell. Classic stiffness, resonance, and coincidence controlled regions were identified, which agreed very well with the predictions. The NRS was dominated by the sharp dips that were noted
at most of the acoustic resonant frequencies. Internal coincidence
was found to play a significant role as well, since negative NRS
values amplification was noted at these frequencies. The internal
cut-off frequencies are good estimates of the internal coincidence
frequencies below the ring frequency. These frequencies were determined from the FEA results by identifying the axial-zero modes
(k z 0). External coincidence and possibly ring frequency effects
Acknowledgments
The work was sponsored by the Air Force Research Laboratory
Space Vehicles Directorate AFRL/VS. The POC for this effort is
Dr. Steven A. Lane. The authors also acknowledge ANSYS, Inc.
for the John Swanson Fellowship. Thanks also to David J. Belasco, Jr. for help with data collection and reduction.
References
1 Huybrechts, Steven, and Tsai, Stephen W., 1996, Analysis and Behavior of
Grid Structures, Comp. Sci. Technol. 569, pp. 10011015.
2 George, Tommy J., Herman Shen, M. H., Huybrechts, Steven M., Meink, Troy
E., and Wegner, Peter M., 2001, Optimal Design of Composite Chambercore
Structures, Compos. Struct., 523 4, pp. 277286, MayJune.
3 Griffin, Steven, Denoyer, Keith K., and Das, Alok, 1999, Passive Vibroacoustic Isolation for Payload Containers, J. Intell. Mater. Syst. Struct., 10008305, pp. 83 87.
4 Griffin, Steven, Hansen, Collin, and Cazzolato, Ben, 1999, Feasibility of
Feedback Control of Transmitted Sound into a Launch Vehicle Fairing Using
Structural Sensing and Proof Mass Actuators, AIAA 991529, pp. 2592
2602.
5 Griffin, Steven, Hansen, Collin, and Cazzolato, Ben, 1999, Feedback Control
of Structurally Radiated Sound into Enclosed Spaces Using Structural Sensing, JASA, 1065, pp. 26212628.
6 Griffin, Steven, Denoyer, Keith K., and Sciuli, Dino, 1998, Hybrid Structural/
Acoustic Control of a Sub-Scale Payload Fairing, SPIE Conference on Smart
Structures and Integrated Systems, Vol. 3329 of 0277-786X, pp. 237243.
7 Griffin, Steven, and Denoyer, Keith K., 1998, Experimental Sensor and Actuator Location Procedure for Control of Dynamically Complex Smart Structures, SPIE Conference on Smart Structures and Integrated Systems, Vol.
3329 of 0277-786X, pp. 717723.
8 Lane, Steven A., Kemp, J. D., Griffin, Steven, and Clark, R. L., 2001, Active
Acoustic Control of a Rocket Fairing Using Spatially Weighted Transducer
Arrays, AIAA J., 381, 112119.
9 Neple, Pascale, el Hafidi, Ali, and Lesueur, Claude, 2001, On Sound Transmission Through a Truncated Isotropic Conical Shell: Experimental and Preliminary Numerical Results, Internoise 2001, The Hague, The Netherlands,
August 2730.
10 Vipperman, Jeffrey S., Li, Deyu, Avdeev, Ilya, and Lane, Steven A., 2001,
Characterization and Control of Sound Radiation in a Complex Fairing Structure, Internoise 2001, The Hague, The Netherlands, August 2730.
11 Denoyer, Keith K., Griffin, Steven, and Das, Alok, 1998, Passive Vibroacoustic Isolation for Reusable Launch Vehicle Payload Containers, AIAA/ASME/
ASCE/AHS/ASC Structures, Structural Dynamics & Materials Conference,
AIAA-98-1977, pp. 2248 2256, Longbeach, CA, April 2023.
12 Fahy, F., 1985, Sound and Structural Vibration, Academic Press, Inc., New
York.
13 White, Pritchard H., 1966, Sound Transmission Through a Finite, Closed
Cylindrical Shell, JASA, 505, pp. 1124 1130.
14 Manning, J. E., and Maidanik, G., 1964, Radiation Properties of Cylindrical
Shells, JASA, 369, pp. 16911698.
15 Cheng, L., 1994, Fluid-structural Coupling of a Plate-ended Cylindrical
Shell: Vibration and Internal Sound Field, JSV, 174, pp. 641 654.
16 Koval, L. R., 1976, On Sound Transmission into a Thin Cylindrical Shell
Under Flight Conditions, JSV, 48, pp. 265275.
17 Koval, L. R., 1978, Effects of Cavity Resonances on Sound Transmission
into a Thin Cylindrical Shell, JSV, 59, pp. 2333.
18 Blake, William K., 1986, Mechanics of Flow-Induced Sound and Vibration,
Vol. 1, General Concepts and Elementary Sources, Academic Press, Inc., London.
19 Blake, William K., 1986 Mechanics of Flow-Induced Sound and Vibration,
Vol. 2, Complex Flow-Structure Interactions. Academic Press, Inc., London.
20 Blevins, Robert D., 1979, Formulas for Natural Frequency and Mode Shape,
Van Nostrand Reinhold Co., New York.
21 ANSYS, Inc., 2001, Canonsburg, PA 15317.
Jianhui Luo
Hae Chang Gea
Department of Mechanical and Aerospace
Engineering,
Rutgers, The State University of New Jersey,
Piscataway, NJ 08855
Introduction
introduces the modal analysis approach to calculating the frequency response for the coupled acoustic-structural system; Section 3 discusses the sensitivity analysis of frequency response for
this system. In Section 4, topology optimization problem is formulated and the solution procedure of identifying the optimal configurations of stiffeners for the reduction of the interior sound
level is described. In Section 5, four numerical examples of an
acoustic box under external excitations are presented. Conclusion
remarks and future work are discussed in the final section.
In this section, frequency response of coupled acousticstructural systems is derived from the modal analysis method.
Consider that a coupled acoustic-structural system consists of an
enclosure surrounded by an elastic body and the elastic body is
subjected to both force and displacement boundary conditions as
shown in Fig. 1. Using the finite element method to discretize the
coupled system yields the following equations 8
M ss
M as
M aa
K ss
u e
p e
0
K sa
K aa
ue
fs
pe
0
(1)
where the upper equation refers to the structural system and the
lower equation refers to the acoustic system. In these equations,
u e is the vector representing displacement components at the grid
points of the structural finite element model, p e is the vector representing sound pressures at the grid points of the acoustic finite
element model. f s represents the external forces applied to the
structure, such as mechanical excitations. M ss and K ss are structural mass and stiffness matrices, M aa and K aa are acoustic mass
and stiffness matrices. M as and K sa are acoustic-structural couT
. The
pling matrices, and they have the property of K sa M as
structural and acoustic equations of motion are coupled through
the matrix M as which transforms the structural accelerations to
acoustic excitations of the interior cavity, and through the matrix
K sa which transforms the acoustic pressures to loads acting on the
structure.
If the external force has a harmonic form of f s e i t , then the
structural and acoustic responses can be expressed as u e e i t and
fs
sT f s
(8)
3
p e e i t , where is the excitation frequency. Introducing these expressions to Eq. 1 yields the frequency response equation as
K ss 2 M ss
K sa
2 M as
K aa 2 M aa
ue
fs
pe
0
(2)
There are two different approaches to computing frequency response in Eq. 2: the direct method and the modal analysis
method. The direct method is a rather straightforward approach
that solves Eq. 2 directly but it is very computationally expensive. On the other hand, the modal analysis method is extremely
attractive because both structural and acoustic modal density are
relatively low under the low excitation frequencies. Therefore, the
modal analysis method is adopted in this study.
Since the mass and stiffness matrices are unsymmetric in the
coupled acoustic-structural system, the left eigenvectors are not
the same as the right ones. Using the modal analysis method, the
right eigenvalue problem can be formulated as
K ss
K sa
K aa
M ss
s
a
M as
0
M aa
s
a
(3)
where is the eigenvalue matrix; s and a are the right eigenvector matrices corresponding to the structural and acoustic fields
respectively. Similarly, the left eigenvalue problem is represented
as
sT
Ta
K ss
K sa
K aa
sT
Ta
M ss
M as
M aa
(4)
s and
a denote the left eigenvector matrices of the
where
structural and acoustic fields. The M-orthonormal condition of the
coupled system is stated as follows:
sT
Ta
M ss
M as
M aa
s
I
a
a
a
(6)
ue
pe
(7)
When an optimization problem is solved by mathematical programming algorithms, we must determine the effect resulting from
a small perturbation in the current design on the objective and
constraint functions. This is known as the sensitivity analysis. In
this study, frequency response sensitivities are the derivatives of
structural response and the sound pressure with respect to the
structural modification design variables. With the correct frequency response sensitivity information, one can use various optimization methods to arrive at a modified structure with reduced
interior sound level. The frequency response sensitivity,
u e ,p e T , can be derived by differentiating Eq. 2 as
K ss 2 M ss
K sa
M as
K aa M aa
2 M ss
K ss
u e
p e
0
0
0
uc
pe
(10)
and M ss
are the derivatives of the structural stiffness
where K ss
and mass matrices with respect to the design variable, respectively.
2 M ss
u e , Eq. 10
If we define a pseudo load g s K ss
can be rewritten as
K ss 2 M ss
K sa
M as
K aa 2 M aa
u e
gs
0
p e
(11)
(5)
(9)
where i , p i and i are the i th component of , p and . Combining Eq. 9 with Eq. 7, the response of the coupled acousticstructural system is recovered.
pi
i 2
T
2 I
gs
sT g s
(12)
qi
i 2
(13)
Solution Procedures
In this section, the solution procedures of identifying the optimal configuration of stiffeners placement for the reduction of the
interior sound level are discussed.
Transactions of the ASME
The overall interior sound level for the acoustic field is evaluated as
S PL
1
m
10 log
10 p i
i1
/p 0 2
(14)
1
1 0
2 M as u e d
(15)
1
1 0
1
m
i1
i1
10 log10 p i /p 0 2 d
(16)
i 0 1
i 0 1
c 1i 1 0
1c 1i 0 1 0 0
c 1i 1 0
1c 1i 0 1 0 0
(17)
(18)
with
1 1 0
3 1 0
(19)
2 45 0
15 1 0
(20)
0
0
1 W
(22)
v 1c
Subject To:
And, the overall interior sound level from Eq. 14 can be defined
as
SPL
(21)
f x f xk
b x d b x d
i
k
i
ri
ri
(23)
Numerical Examples
Structural
frequency
Hz
Acoustic
frequency
Hz
Coupled
frequency
Hz
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
114.95
130.72
179.50
195.32
195.36
231.06
242.49
257.62
291.94
302.71
323.13
335.41
387.39
389.28
411.79
418.62
443.03
462.87
484.05
494.91
0.00
344.05
430.06
-
0.00
117.02
129.92
178.67
194.99
196.09
239.07
241.66
254.44
291.35
302.71
304.21
330.40
370.61
382.70
386.75
411.72
418.62
443.02
443.33
452.73
482.94
495.40
Weight
Kg
initial
1
2
3
4
5
6
7
8
9
10
91.89
89.62
80.87
77.05
75.64
74.33
72.08
70.88
70.52
70.26
68.97
5.5836
5.5836
5.5836
5.5836
5.5836
5.5836
5.5836
5.5836
5.5836
5.5836
5.5836
Structural frequency
Hz
Coupled frequency
Hz
1
2
3
4
5
6
7
8
9
10
70.18
120.74
120.87
124.97
138.17
147.54
177.84
185.15
190.37
0.00
72.31
120.27
120.53
125.10
146.12
148.25
177.51
184.02
187.38
eners are also mostly located in the external force applied region.
Table 4 shows the trend of interior SPL reduction throughout the
iterations. The corresponding uncoupled structural frequencies
and coupled system frequencies in the range of 0 and 200 Hz are
listed in Table 5. The final design also experiences changes in
natural frequencies and mode shapes from the initial design.
5.3 Case 3. In the third case, a unit harmonic excitation
with higher frequency is applied at the center of the top plate. The
excitation frequency is chosen as 150 Hz. We found the objective function reduced from 2.811064e-5 to 1.548597e-5 and the
overall interior sound decreased from 80.13 dB to 67.03 dB after
ten iterations. The iteration history is showed in Fig. 8 and the
trend of SPL reduction is showed in Table 6. Fig. 9 shows the
optimal stiffener location. Unlike the first two cases, in this case
the locations of stiffeners are spread out to different region of the
box. This is due to the fact the excitation frequency is in the
middle of several resonance frequencies of the system. To minimize the acoustic excitation, the add-on stiffeners try to reduce the
effects from all these resonance modes as much as possible. The
natural frequencies below 200 Hz of the final design are also
listed in Table 7 for reference.
5.4 Case 4. In the last example, the stiffener design under a
band of frequency excitation is studied. A unit harmonic excitation
is applied at the center of the top plate with a band of low excitation frequencies ranged from 20 to 100 Hz. A modified objective
function, Eq. 15, is used here. Since the excitation frequency
range falls below any resonance frequency, the optimal stiffener
location turns out to be very similar to that of the first case as Fig.
4. We found the objective function is reduced from 3.106249e-7 to
1.842361e-7 and the overall interior sound defined in Eq. 16 is
decreased from 93.53 dB to 75.16 dB after ten iterations. Fig. 10
shows the interior SPL comparison between the initial design and
the final design within the exciting frequency range.
Iteration
number
Weight
Kg
initial
1
2
3
4
5
6
7
8
9
10
76.73
72.20
61.03
58.29
57.33
56.62
56.90
55.69
53.92
52.12
51.92
5.5836
5.5836
5.5836
5.5836
5.5836
5.5836
5.5836
5.5836
5.5836
5.5836
5.5836
Structural frequency
Hz
Coupled frequency
Hz
1
2
3
4
5
6
104.94
117.86
158.14
167.27
188.17
0.00
107.27
117.13
157.36
167.43
195.37
Weight
Kg
initial
1
2
3
4
5
6
7
8
9
10
80.13
79.59
77.39
74.91
75.88
73.40
71.20
70.46
68.99
67.63
67.03
5.5836
5.5836
5.5836
5.5836
5.5836
5.5836
5.5836
5.5836
5.5836
5.5836
5.5836
In this paper, optimal stiffener design for interior sound reduction of coupled acoustic-structural system is studied. Forced frequency response and its sensitivity of this coupled system were
solved by the modal analysis method. Using a topology optimization based approach, the optimal stiffener placement problem was
converted into an optimal material distribution problem. Instead
of using the overall interior sound level directly, the norm of
acoustic excitation was used as the objective function and from
our numerical examples, it showed the acoustic excitation is a
good indicator for sound reduction. At low frequency excitations,
the optimal stiffeners might be applied to the region near the
external force source, however when the excitation frequency is
among several resonance frequencies, stiffener might be placed
throughout the structure in order to compensate the resonance
modes involved. In this work, a conservative system with no
damping effect is considered, therefore the proposed approach is
only valid for single frequency excitation or a band of low
frequency excitations without any resonance frequency included.
More general method to handle damping is currently under
investigation.
References
Fig. 9 Optimal configuration of the stiffeners placement in
case 3
Structural frequency
Hz
Coupled frequency
Hz
1
2
3
4
5
6
88.77
115.09
161.26
166.55
173.40
0.00
90.39
115.62
160.88
166.00
173.79
1 Hagiwara, W., Kozukue, W., and Ma, Z. D., 1993, The Development of
Eigenmode Sensitivity Analysis Methods for Coupled Acoustic-Structural Systems and Their Application to Reduction of Vehicle Interior Noise, Finite
Elem. Anal. Design, 14, pp. 235248.
2 Huff, J. E., Jr., and Bernhard, R. J., 1995, Acoustic Shape Optimization
Using Parametric Finite Elements, ASME 1995 Design Engineering Technical Conferences, DE-Vol. 84-2, pp. 577584.
3 Wodtke, H. W., and Koopmann, G. H., 1995, Quieting Plate Modes with
Optimally sized Point MassesA Volume Velocity Approach, ASME 1995
Design Engineering Technical Conferences, DE-Vol. 84-2, pp. 647 654.
4 Constans, E., and Belegundu, A., 1996, Minimizing Radiated Sound Power
from Vibrating Shells, The 6th AIAA/NASA/ISSMO Symposium on Multidisciplinary Analysis and Optimization, Bellevue, WA, Paper 96-4111, pp. 1106
1116.
5 Luo, J. H., and Gea, H. C., 1997, Modal Sensitivity Analysis of Coupled
Acoustic-Structural Systems, J. Sound Vib., 119, pp. 545550.
6 Gea, H., 1996, Topology Optimization: A New Micro-Structure Based Design Domain Method, Comput. Struct., 615, pp. 781788.
7 Chickermane, H., and Gea, H. C., 1996, A New Local Function Approximation Method for Structural Optimization Problems, Int. J. Numer. Methods
Eng., 39, pp. 829 846.
8 Craggs, A., 1971, The Transient Response of a Coupled Plate-Acoustic System Using Plate and Acoustic Finite Elements, J. Sound Vib., 15, pp. 509
528.
9 Diaz, A. R., and Kikuchi, N., 1992, Solutions to Shape and Topology Eigenvalue Optimization Problems Using a Homogenization Method, Int. J. Numer. Methods Eng., 35, pp. 14871502.
10 Bendse, M. P., and Kikuchi, N., 1988, Generating Optimal Topologies in
Structural Design Using A Homogenization Method, Comput. Methods Appl.
Mech. Eng., 71, pp. 197224.
Yuji Ohue
Associate Professor,
Department of Intelligent Mechanical
Systems Engineering,
Kagawa University,
Hayashi, Takamatsu, 761-0396, Japan
e-mail: ohue@eng.kagawa-u.ac.jp
Akira Yoshida
Professor
Mem. ASME,
Department of Mechanical Engineering,
Okayaya University,
Tsushima-naka, Okayama, 700-8530, Japan
e-mail: akira-y@mech.okayama-n.ac.jp
Introduction
Generally, the dynamic signals in the field of engineering problems such as vibration, sound and so on, have been analyzed using
the fast Fourier transform FFT. The FFT has been the most
common method to analyze the frequency properties of the signals. In the FFT analysis, the signal as the function of time is
converted to the power spectra in a frequency domain. However,
except for a special case, the frequency components of the most
signals encountered in the engineering problem change with time.
Based on the FFT alone, it is hard to investigate whether the
frequency components of the signals vary with time or not, even
though the phase of the Fourier transform relates to time shifting.
The FFT analysis is no more adequate for those applications.
Therefore, it is significant and important to adopt the timefrequency analysis for those signals varying with time 13.
It is possible by using the time-frequency analysis to investigate
how the frequency components of the signal vary with time. The
time-frequency analysis can provide more beneficial information
about the frequency compared with the FFT. During the last two
decades, a new mathematical technique for the time-frequency
analysis, which has been called the wavelet transform WT,
has been extensively developed 1,2. The WT has some features different from those of the short-time fast Fourier transform or the Wigner-Ville distribution. Applications of the WT
are actively studied in a variety of the fields of engineering
science 4 7.
Gears have been employed in automobiles, motorcycles, industrial robots and so on. The gears are an important machine element to transmit the power and to change the speed of the equipment. The FFT is generally used for analyzing the gear dynamics.
However, it is important to evaluate the gear dynamic performance more precisely using WT. Therefore, in order to evaluate
the difference in the gear dynamics due to the gear materials, the
dynamic performances of sintered and steel gears were measured
at gear rotational speeds of 1600 rpm to 10,000 rpm, using a
power circulating gear testing machine. The gear dynamics concerned with tooth root strain, vibration acceleration of gear box
and sound near gear box, were analyzed in the time-frequency
domain by the continuous WT, and the signals of the gear dynamics were decomposed and reconstructed by the discrete WT. The
validity of the new evaluation method by the WT is discussed.
Table 1 and Fig. 1 show the specification and the shapes of gear
pair, respectively. The module and the pressure angle of the gear
pair are 5 mm and 20 deg., respectively. The contact ratio of the
gear pair is 1.246. The gear has 16 teeth and a face width of 6
mm. The materials of the test gears are sintered and steel JIS;
SCM440 ones. Those test gears were induction-hardened before
grinding of tooth surface. The material of the mating pinion is
steel JIS: SCM415. The chemical composition of the sintered
material is almost the same as that of the steel SCM440. Many
pores existed in the sintered great. Generally, the sintered materials have higher damping ratio due to the pores, compared with the
steel ones. The density of the sintered gear in this study is 6.8
g/cm3. The Youngs modulus and Poissons ratio of the sintered
gear are 152 GPa and 0.25, and those of the steel gear are 206
mm
deg.
Gear
5
20
15
0.571
mm
mm
mm
16
0.560
90.71
94.60
82.55
18
6
1.246
Class 1
Class 1
Grinding
*JIS B 1702
Wavelet Transform
Continuous Wavelet Transform. The continuous wavelet
transform CWT of a function f (t) is defined as follows.
W f b,a a 1/2
GPa and 0.3, respectively. Figure 2 shows the tooth profile errors
of test gear pairs. The tooth surfaces of both gears and pinion were
finally ground. The tooth profile errors of both sintered and steel
gears and almost the same. The accuracy of the test gear pairs are
in class 1 according to Japanese industrial standard JIS B 1702.
The test apparatus shown in Fig. 3 is a power circulating type
gear testing machine with a center distance of 82.55 mm. The test
gear was driven at an interval of 200 rpm within a range of rotational speeds n 2 of 1600 rpm to 10,000 rpm. The loading torque
was statically applied to the gear shaft by a lever loading with
dead weights and a torsional coupling. The torsional torque was
taken as a scale for loading. In this experiment, the value of the
torque was 30 Nm, which corresponds to a load per unit face
width of 142 N/mm in the normal direction to the tooth surface.
The gear pair was lubricated with a gear oil by pressure feeding at
a flow rate of 750 ml/min and an oil temperature of 313 K.
The dynamic tooth root strain was measured using a dynamic
strain amplifier and a wire strain gage with a gage length of 0.3
mm, which was bonded on a compression side of the root fillet of
the gear. The vibration acceleration of the gear box was detected
by a piezo-electric pickup through an amplifier. The pickup has a
maximum response frequency of 20 kHz. The sound pressure near
the gear box was detected by a condenser type microphone having
a maximum response frequency of 12.5 kHz. The microphone was
located at a distance of 300 mm from the side of the gear box.
These signals were stored into a data recorder with a synchronous
signal obtained by a phototransistor and two disks with slits which
were fixed on both shifts of the gear and the pinion. These stored
signals were analyzed using a computer with an A/D converter
board.
f t
tb
dt
a
(1)
Where, the (t) indicates the conjugate of a mother wavelet function (t). The a and b indicate the parameters on frequency and
time. The Gabor function defined by Eq. 2 was adopted as the
mother wavelet function (t) in this study
t 1/4
1/2
exp
1 p
2
i p t
(2)
(3)
Where, Re(W f ) and Im(W f ) are the real and the imaginary
parts in Eq. 1. To express the wavelet map, that is the intensity
map in a time frequency domain, the intensity calculated by the
CWT was normalized by the maximum intensity in the map.
Discrete Wavelet Transform. When the coordinates b, a of
the CWT shown in Eq. 1 are discretized to the coordinates
(2 j k,2 j ) using two integers j and k, the discrete wavelet transform DWT is defined as follows.
d kj 2 j/2
f t 2 j tk dt
(4)
g t
j
(5)
Fig. 3 Power circulating gear testing machine and measurement system of dynamic performance of
gear pair
f j t f j1 t g j1 t g j1 t g j2 t g j3 t
(12)
g j t
d
k
j
j
k 2 tk
(6)
c
k
j
j
k 2 tk
(7)
t
t
p 2tk
k
q 2tk
k
c kj
d kj1
p
l
j
12k c l
b 12k c l j
j1
q k2l d l j1
k2l c l
k0
m
k
m1
tk
(13)
Figure 5 shows the scaling function (t) and the mother wavelet
function (t) based on the function N 4 (t), that is the cardinal
B-spline function in order of 4. The function (t) is equal to the
function N 4 (t). From this figure, it is understood that the function
(t) is very smooth and is very similar to sinusoidal functions.
The sequences a k , b k , p k and q k are given in the reference 1.
(10)
(11)
Where, a k and b k are the sequences to decompose the function f j (t) at a level j. Therefore, the decomposition and the reconstruction on the function f (t) are concluded into the calculations
using the sequences a k , b k , p k and q k . The functions
g j (t) and f j (t) at a level j are able to be found by using Eqs. 6
and 7. The function f j (t) decomposed into the function g j (t) on
the wavelet component is satisfied with the following relation.
276 Vol. 125, JULY 2003
1
N m t
m1 !
(8)
root stress smax , has been used. In this study, the tooth root
stress at a rotational speed n 2 of 6.6 rpm was adopted as the static
stress smax .
Figure 7 shows the dynamic load ratio dmax / smax plotted
against the rotational speed n 2 of gear. The frequency f z indicates
the tooth mesh frequency. The natural frequency of the gear pair
including the shafts, which support the gear pair, was calculated
using a program given in reference 8. The equation of motion
for the calculation of the natural frequency of the gear pair is
given by
2
M x Dx K t,x xW
K t,x e t
i
(14)
where, M is the equivalent inertia mass of gear pair along the line
of action, D is the damping coefficient between gear pair including the shafts, K is the stiffness of gear pair, W is the static load,
e i is the tooth profile error and x is the relative displacement along
the line of action. Subscript i indicates the pinion (i1) and the
gear (i2). Then, the natural frequency f e is calculated using the
following equation.
1
f e
2
1
Tz
Tz K t,x
dt
(15)
Where, T z is the tooth mesh period. In this experiment, the calculated value of f e was approximately 2.56 kHz. The gear box was
hit by a hammer to measure its natural frequency in the horizontal
direction of the gear box using the FFT. As the result, the natural
frequency of the gear box was about 5.0 kHz. From Fig. 7, for
both the sintered and steel gears, the ratio dmax / smax increases
with fluctuation as n 2 increases. The ratio dmax / smax has three
peak values at n 2 4200 rpm ( f z 1.12 kHz, 6000 rpm ( f z
1.60 kHz and 9000 rpm ( f z 2.40 kHz, since the tooth mesh
frequency and its harmonics coincide with both the natural frequencies of the gear pair and the gear box. Namely, the gear pair
for both the sintered and steel gears became the resonance conditions at their speeds. The ratio dmax / smax of the sintered gear is
mostly smaller than that of the steel one against n 2 . Therefore, it
is obvious that the sintered gear is superior to the steel one from
the viewpoint of the damping characteristics.
dynamic load and used only the maximum value of the gear dynamic load. Thus, the information of the gear dynamic load in the
time-frequency domain is important to evaluate the great performance in detail. In this section, the dynamic tooth root strains are
analyzed in the time-frequency domain using the WT.
Figure 8 shows the waveform of the tooth root strain top of the
figure, its WT map lower left of the figure and the analyzed
result using the FFT lower right of the figure at a rotational
speed n 2 of 1800 rpm. The WT maps of both the gears were
obtained using the CWT. The WT map shows the intensity of the
tooth root strain in the time-frequency domain. In the WT map,
not only the frequency components but also the change of the
intensity of each frequency in the tooth root strain with the time
can be provided more precisely, compared with the FFT analysis.
In the WT maps of the sintered and steel gears, each intensity in
two regions above and below f z 480 Hz indicates each peak. The
contact time of one tooth pair at n 2 1800 rpm is approximately
2.6 ms. Suppose that the wave of the tooth root strain is roughly a
sine wave with a period of 5.2 ms, the frequency of the sine wave
is 192 Hz. It can be considered that the frequency components
below f z depend on the contact time of the one tooth pair. On the
other hand, it can be considered that the frequency components
above f z are caused by the torsional vibration of the gear pair. The
relative intensities at the region above f z of the sintered gear are
JULY 2003, Vol. 125 277
lower than those of the steel one. It can be understood that two
main frequency components above and below f z exist in the dynamic tooth root strain.
Figure 9 shows the relationship between the two main frequencies contained in the tooth root strain of the steel gear and the
rotational speed n 2 . The tooth root strain consists of two main
frequencies above and below f z at each rotational speed of gear.
Especially, the relation between the frequency f low below f z and
n 2 is given by a following relation.
f low
fz
1 z 2n 2
1
1 a 60
2 a 2 a T z
(16)
1
Tr
Tr
v t 2 dt
(17)
Acknowledgments
Conclusion
In order to evaluate the difference in the gear dynamics with
different gear materials, the gear dynamics concerning tooth root
strain, vibration and sound were measured using a power circulating gear testing machine in a range of rotational speeds of 1600 to
10,000 rpm. The gear materials employed were steel and sintered
ones. The gear dynamics were analyzed in a time-frequency domain by the continuous and discrete wavelet transforms.
1. The dynamic tooth root strain indicated indirectly the condition of the torsional vibration of the gear pair. The value of
the tooth root strain reconstructed at the region below the tooth
mesh frequency f z depended on the Youngs modulus of the gear
material. On the other hand, the amplitude of the strain reconstructed at the region above f z depended on the damping ratio of
the gear pair.
2. The behavior of the vibration and the sound of the gear box
against the rotational speed of the gear could be also divided
obviously to two different behaviors above and below f z . The
Nomenclature
D damping coefficient between gear pair
K(t,x) stiffness of gear pair
M equivalent inertia mass of gear pair along line
of action
N m (t) cardinal B-spline function in order of m
Tr one revolution period of gear
Tz tooth mesh period
W static load of gear pair
(W f )(b,a) wavelet transform
a frequency parameter
a k sequence for decomposition by discrete wavelet transform
b time parameter
b k sequence for decomposition by discrete wavelet transform
c (kj ) sequence at level j
d k ( j ) sequence at level j(W f )(2 j k, 2 j )
e(t,x) tooth profile error
f (t) signal
f j (t) signal at level j
g j (t) wavelet component of signal f (t) at level j
f e natural frequency of gear pair
f low f z /2 a
f z tooth mesh frequency z 2 n 2 /60
j level
k integer for time parameter
m order of cardinal B-spline function
n 2 rotational speed of gear
p k , q k two-scale sequences for reconstruction by discrete wavelet transform
t time
v (t) signal of vibration acceleration or sound pressure
x relative displacement along line of action
z 2 number of gear teeth
a contact ratio
(t) scaling function
(t) mother wavelet function
dmax maximum dynamic tooth root stress
smax maximum static tooth root stress
constant for Gabot function
p center of angular frequency
Transactions of the ASME
References
1
2
3
4
6 Mori, K., Kasashima, N., Yoshioka, T., and Ueno, Y., 1996, Prediction of
Spalling on a Ball Bearing by Applying the Discrete Wavelet Transform to
Vibration Signals, Wear 195, pp. 162168.
7 Yoshida, A., Ohue, Y., and Ishikawa, H., 2000, Diagnosis of Tooth Surface
Failure by Wavelet Transform of Dynamic Characteristics, Tribol. Int., 33,
273279.
8 JSME Research Group on Accuracy and Design in Gears, 1977, Rep. Fujita,
K., Research Report on Accuracy and Design in Gears, RC-SC38, JSME, in
Japanese.
J. Antoni
Lecturer,
Roberval UMR CNRS 6066,
University of Technology of Compie`gne,
France
R. B. Randall
Professor,
School of Mechanical and Manufacturing
Engineering,
The University of New South Wales,
Sydney 2052, Australia
Introduction
ing of the spectral content of the envelope of the resulting vibration signal. This model was later refined by Ho & Randall who
pointed out that actual rolling element bearings experience some
random slip in their operation so that the train of impacts is
slightly random instead of periodic 4. However small these effects, Ho showed that the resonance characteristic is no longer
sampled in the Fourier domain but rather resembles a continuous
spectral density where all the harmonics tend to smear over each
other. Hos model resulted in a significantly better description of
bearing vibration spectra as observed in the real world, and was
next used by Randall, Antoni & Chobsaard to show that bearing
signals are quasi-cyclostationaryi.e. their statistics have quasiperiodicity 5. Incidentally, this observation offered an elegant
way for justifying the envelope analysis method from the theory
of cyclostationary processes. In a following paper, Antoni & Randall refined their results after specifying that signals from localized faults are not exactly quasi-cyclostationary since the random
slips are non-stationary in their nature 6. However they concluded that the bearing signals could still be treated as pseudocyclostationary as a first approximation.
The purpose of this paper is to attempt a complete treatment of
the stochastic modeling of bearing vibrations as produced by localized faults, putting together a number of unpublished results
and putting the heuristic considerations of 6 on a firmer mathematical foundation. The aim is twofold. Firstly, it is to provide
the mechanical community with a model that has proven very
satisfactory in describing actual vibration signals and, in particular, their spectra and envelope spectraincluding some typical
features that have never been explained elsewhere. Secondly, it is
to demonstrate how this model permits a proper formalization of
the envelope analysis technique as classically used in the diagnostics of rolling element bearings. The paper is organized as follows.
A first section addresses the accurate modeling of the nonstationary impacting process as generated by a localized defect on the
inner race, the outer race, or on a rolling element. The spectral
characteristics of this process are then derived from the theory of
regular point processes and important results are deduced concerning the nature of spectral harmonics. In a second section these
results are used to investigate the spectral properties of the resulting vibration signal after the impacts have propagated through the
system, i.e. as measured on the housing. In particular, the general
spectral signature due to a localized defect is found and its mani-
festations in a number of spectral indicators the Fourier transform, the power spectral density, the spectral correlation density,
the Fourier transform of the squared signal and the envelope spectrum are investigated in detail. The relative effectiveness of these
spectral indicators in diagnostics is finally discussed in the light of
the new results.
2.2 Analytical Forms of the Product Densities. The product density of degree one in Eq. 3 can be expanded into
2.1 Regular Point Process. At the outset, consider the process generated by the repetition of impact forces when a defect in
one surface strikes a mating surface. We shall refer to it as the
impacting process F(t). For a localized defect, each impact may
be well described by a Dirac delta function (t) provided the
measured signal is sampled at a rate well below the impact spectral bandwidth 2,3,4. At this stage it is assumed that all impacts
have equal magnitudes; magnitudes and signs of impacts will be
accounted for later in the text by modulating the impacting process with a suitable time-varying function.
Without loss of generality, the reference time t0 is chosen to
coincide with an arbitrary impact which defines the point from
which the process is starting to be observed. Hence,
F t
tT
i0
where T 0 0
(1)
(2)
n0 in N
E dN t dN t f 2 t, dtd ,
Journal of Vibration and Acoustics
(3)
(4)
f 1 t dt
P tT tdt/T 0 t dt
i
i0
i0
(5)
f 2 t, dtd
P tT tdt,t T t d /T
i
i0 ji
(6)
f 2 t, dtd
i0
i t dt
d,
k1
(7)
(8)
In short Eqs. 5 and 8 give the explicit solutions for first two
moments of the impacting process from which those of the vibration signal will later be derived. For the physical process under
consideration, it is noteworthy that the product density of degree
one f 1 suffices to describe it at least up to the second order because f 2 factorizes into a product of f 1 terms, thus assigning to
the instantaneous mean rate of impacts a major role in this paper.
As a matter of fact, the exact shape of f 1 (t) is worthy of further
investigation. In view of Eq. 5, the first peak in f 1 (t) happens to
be the probability density function 1 (t) of the first time of occurrence T 1 , the second peak the probability density function of
T 2 and so on. Therefore the ith peak is the first one convolved
with itself i times, i.e.
(9)
Then, under mild conditions, the bandwidth of the ith peak as
measured by its standard deviation is i with the standard
deviation of 1 (t). As the peaks slowly enlarge, their amplitudes
decrease accordingly so as to maintain a unit area. This is illustrated in Fig. 2. In the limit, the peaks completely vanish and f 1 (t)
tends to the constant value f 1 ()1/T, that is the mean overall
rate of occurrence.1 However the rate of convergence is extremely
small: considering a percentage of random fluctuation of x/100
/T, then two peaks completely overlap when their band1
An heuristic proof to this result is that, in the limit, the area under each probability density function is still unity whereas the mean paving is one probability
density function per T units of time.
1
2
f 1 t e j t dt 1 * 1
(10)
exp
1 2 2
j t
2
(11)
Fig. 4 Double Fourier transform modulus of the product density of degree two for T 130
1
42
R2
f 2 t, e j t e j dtd
(12)
(13)
This defines a continuous spectral density with marked ridges running along the -variable and centered on all k/T, kZ. For
large values of , F 2 ( , ) ultimately tends to F 1 ( ) and thus
resembles a pattern of parallel and horizontal ridges along the
-axis rapidly falling off on each side of 0, as illustrated in
Fig. 4. The presence of these parallel ridges distinctively characterizes the second-order spectral signature of a random train of
impact forces and consequently, that of a faulty rolling element
bearing. Moreover, the distance between the ridges indicates the
mean rate of occurrence of the fault, thus enabling its identification in the mechanical system.
2.4 Discussion. At this stage, it is instructive to review
some former models proposed in the literature in light of the derived results. Clearly, for the deterministic model f 1 (t) is a perfectly periodic train of Dirac deltas XT (t) and f 2 (t, ) a twodimensional version of it, viz f 2 (t, )XT (t)XT ( ). The same
applies to their respective Fourier transforms, viz F 1 ( )1/T
X1/T ( ) and F 2 ( , )1/T 2 X1/T ( )X1/T ( ). The limitation of these formulas arises from the experimental evidence that
actual data do not have line spectra especially in the vicinity of
the high frequency resonance where they are usually demodulated. On the other hand, the simplified stochastic model proposed
by Randall & Antoni in 5 leads to f 1 (t)XT (t) * 1 (t), that is
a periodic train of Dirac delta functions low-pass filtered by the
probability function 1 (t). Therein f 2 (t, ) turns out to be periodic and low-pass filtered in the t-variable while transient in the
-variable. These low-pass filter and transient effects give a
better explanation for the continuity of experimental spectra in the
Transactions of the ASME
P2: b(t) has an effective duration shorter than the mean interarrival time T, or equivalently its spectral bandwidth is larger
than the mean rate of impacts 1/T.
Therefore, the overall impulse response of the system is obtained from cascading the amplitude modulation function A(t)
with the time-varying impulse response r(t, ) and finally with the
band-pass filter b(t). This is illustrated in Fig. 5b.
In this procedure, A(t) and g(t, )b(t)r(, ) have some
important properties which will make the computation of the
input-output relationship tractable. Specifically, because A(t) encompasses all the periodic modulations with possible stochastic
effects, it has first and second-order statistics given by:
m A t E A t m A t
In the preceding section, expressions were derived which describe the spectral signature of a train of pulses as produced by a
localized defect. This section now discusses how this spectral signature is transformed after the impacts have propagated through
the system, i.e. as it is likely to be measured on the bearing housing by an accelerometer.
3.1 Response of a Rolling Element Bearing to a Random
Train of Impacts. Following classical models, the vibration signal produced by a faulty rolling element bearing may be viewed
as the response of a linear system driven by the impacting process
F(t) 2,7. For this input-output relationship to be fully comprehensive, we now show that the impulse response of the system
should be time-varying and should also accommodate some degree of stochasticity.
At the outset, the impacting process should be modulated by a
periodic and positive function A(t) to account for the variations in
the impact magnitudes as the defect enters and exits the load zone
2. Some random modulation might be incorporated in A(t) due
to the dependence on the position and the number of the rolling
elements in the load zone at time t, but also due to rolling and slip
on possibly rough surfaces especially after a defect has appeared
and spread to some extent.
Next, let us define r(t, ) the structural response at time t of the
system subjected to an impulse at time . In contrast to a
static structure, the impulse response r(t, ) of a rolling element
bearing is time-varying for a variety of physical reasons, the most
obvious of which being the variations in the transmission path as
the coordinates of the point of impact move w.r.t the location of
the sensor, and the variations in the relative angle between the
impact forces and the axis of the sensor. For a system operating at
constant speed, these variations periodically affect the magnitude,
the sign and the phase of the impulse response r(t, ) with a
period depending on whether the defect lies on the inner race, the
outer race or on a rolling element 2. In addition, r(t, ) might
have some small random fluctuations to account for unpredictable
effects such as contact non-linearities. Figure 5a gives a schematic illustration of how the impacting process A(t)F(t) is transformed into a vibration X(t) after passing through the impulse
response r(t, ).
In practice, the vibration signal X(t) produced by a local fault
cannot be observed totally because it is contaminated by other
vibrations from a multitude of neighboring sources in the system.
Therefore, it is customary to filter it in a frequency band where the
signal-to-noise ratio is maximum so that virtually no other sources
than that stemming from the faulty bearing are measured by the
experimenter. This is usually done by designing a band-pass filter
b(t) around a high-frequency resonance of the structure or the
sensor that is excited by the impacts 1 4. In order to retain the
diagnostic information, the band-pass filter b(t) must have the
following properties:
P1: b(t) is a band-pass filter with central frequency 0 much
higher than the mean rate of impacts 1/T,
Journal of Vibration and Acoustics
and
a k e jt
kZ
R A t, E A t A * t R A t
(14)
2
R Ak e jt
,
kZ
(15)
g t
2
,
g k t e jt ,
kZ
(16)
g t, A dN
jt
kZ
g k t A dN
(17)
in which each Fourier coefficient g k (t) is to be interpreted as a
linear, causal and homogeneous impulse response.
3.2 Spectral Characteristics of the Vibration Response.
From Eq. 17, the spectral characteristics of the band-pass vibration signal can now be derived and applied to a number of potential indicators for use in diagnostics, namely the Fourier transform
of the expected signal, the power spectral density, the spectral
correlation, the Fourier transform of the expected squared signal
and the power spectral density of the squared signal.
(a) Fourier Transform of the Expected Response
From Eqs. 1, 14 and 17, the expected value ensemble
average of the vibration signal is
m Y t E Y t
g t, m A f 1 d
(18)
M Y
1
2
m Y t e j t dt
kZ
Y k k
1
M Y k G k F
F 1 M A * F 1
(19)
a F l
lZ
to overlap with this support, its central radian frequency 0 resonance frequency chosen for demodulation should be such that
0 T100&/x with x/100 /T the percentage of fluctuation.
Or equivalently, with i c the number of peaks in F 1 ( ) before it
dies to 1, 0 should be such that 0 T2 i c . In most instances
this condition would not be satisfied if a good signal-to-noise ratio
were to be maintained, thus justifying the poor performance anticipated from the Fourier transform of the vibration signal.
(b) Power Spectral Density of the Response
From Eqs. 4, 15 and 17, the autocorrelation function of the
vibration response is:
R Y t, E Y t Y * t
h t ,
h * t, R A , f 2 , dd
(20)
R Y lim
R y t, dt
(21)
whose Fourier transform then yields the explicit expression for the
power spectral density:
SY
1
2
R Y e j d
kZ
y k k
S Y k G k 2 F
2
(22)
F 2 Re F * S
2
1
A
T
1
42
R Y t, e j e j t d dt
(23)
SY ,
k,lZ2
S y k y l k, l
2 ,
S Y k Y l , G k G l* F
F 2 ,
(24)
F p, * S
pZ
p
A
where S Ap ( ) is the Fourier transform of R Ap ( ) in Eq. 15. Although involving two frequency variables, the construction of
S Y ( , ) is again similar to that outlined in Eqs. 19 and 22.
Nevertheless, there is now a domain in the frequency plane ,
2
There is a simple relation between the spectral correlation density and the power
spectral density, viz S Y (0, )S Y ( ) ( )
SY , d
1
2
E Y t 2 e j t dt
(25)
p t, A 2 dN
(26)
Fig. 9 Power spectral density of a vibration signal in case of no fault continuous line and an
inner race fault dotted line
Diagnostic skills
ture of the inner race fault shows up in good accordance with Fig.
4, with 1/T71 Hz ball pass frequency on the fault and
10 Hz shaft rotation.
From a theoretical point of view, it is worth mentioning that the
Fourier transform of the squared signal preserves the diagnostic
information by exploiting the non-stationarity of the signal, while
the power spectral density exploits its non-Gaussianity. In fact, the
power spectral density of the squared signal is implicitly a fourthorder stationarized statistic. Strictly speaking, these two indicators have different theoretical justifications and this is supported
by recalling that the former only requires property P2, while the
latter requires the more stringent condition P2P1. However,
both are inclined to provide envelope analysisor squared envelope analysiswith a strong formal justification.
Comparison of the five spectral indicators, which have been
assessed so forth, is summarized in Table 1.
Conclusion
A comprehensive stochastic model has been proposed for describing and simulating the vibration produced by localized faults
in rolling element bearings. Sources of stochasticity were modeled
in both the impacting force processby means of a regular point
processand in the transmission pathby means of a cyclostationary process, thus encompassing a large range of physical situations. These refinements proved very valuable in explaining
some of the actual features observed on experimental data. The
spectral signature of a localized fault was derived analytically and
new results were deduced concerning the nature of spectral harmonics produced by the impacting process. These were shown to
be distributed and equi-spaced by the mean rate of impacts
peaks with a rapid fall-off that could be quantified as a function of
the percentage of stochastic fluctuations. Next, the spectral signature of a defect was shown to duplicate when it propagates
through the structure with shifts equal to the rotation speed of the
defect, thus generating additional families of pseudo-harmonics.
These results finally helped in investigating the effectiveness of a
number of spectral indicators dedicated to the diagnostics of rolling element bearings. From simple considerations on band-pass
and low-pass filtering operations, it was demonstrated that both
the Fourier transform and the power spectral density of the
squared signal are the most relevant indicators, thus bringing new
supports in favor of squared envelope analysis.
References
1 Darlow, M. S., and Badgley, R. H., 1975, Applications for Early Detection of
Rolling Element Bearing Failures Using the High-Frequency Resonance Technique, ASME Paper 75-DET-46.
2 McFadden, P. D., and Smith, J. D., 1984, Model for the Vibration Produced
by a Single Point Defect in a Rolling Element Bearing, J. Sound Vib., 911,
pp. 69 82.
3 McFadden, P. D., and Smith, J. D., 1985, The Vibration Produced by Multiple Point Defects in a Rolling Element Bearing, J. Sound Vib., 982, pp.
69 82.
4 Ho, D., and Randall, R. B., 2000, Optimization of Bearing Diagnostics Techniques Using Simulated and Actual Bearing Fault Signals, Mech. Syst. Signal
Process., 145, pp. 763788.
5 Randall, R. B., Antoni, J., and Chobsaard, S., 2001, The Relationship Between Spectral Correlation and Envelope Analysis in the Diagnostics of Bearing Faults and other Cyclostationary Machine Signals, Mech. Syst. Signal
Process., 155, pp. 945962.
6 Antoni, J., and Randall, R. B., 2002, Differential Diagnosis of Gear and
Bearing Faults, ASME J. Vibr. Acoust., 127, pp. 17.
7 Roberts, J. B., 1966, On the Response of a Simple Oscillator to Random
Impulses, J. Sound Vib., 41, pp. 51 61.
8 Srinivasan, S. K., et al., 1967, Response of Linear Vibratory Systems to
Non-Stationary Stochastic Impulses, J. Sound Vib., 62, pp. 169179.
9 Lin, Y. K., 1965, Nonstationary Excitation and Response in Linear Systems
Treated as Sequences of Random Pulses, J. Acoust. Soc. Am., pp. 453 460.
Nabeel Shabaneh
Graduate Student
Jean W. Zu
Associate Professor
Department of Mechanical & Industrial
Engineering,
University of Toronto,
5 Kings College Road,
Toronto, Ontario, Canada, M5S 3G8
Introduction
and the ring 9. Ji and Zu 10 performed free and forced vibration analysis to calculate natural frequencies for nonlinear rotor
bearing system with cubic nonlinearity. However, incorporating
nonlinear bearing characteristics in rotor-shaft systems with viscoelastic supports is very limited. A preliminary investigation was
reported by Bhattacharyya and Dutt 11. They studied the unbalanced response and stability of a rotor shaft system mounted on
nonlinear rolling element bearings with viscoelastic supports. The
shaft was assumed to be massless with linear elasticity and internal damping. In addition, the nonlinear restoring force of the bearings was linearized by the method of effective linearization, enabling an approximate stability analysis using the Routh Hurwitz
criterion. The effects of gravity on the system and nonlinearity
were discussed. In the above work, the system was discretized and
the restoring force was approximated by a linear relation.
The aim of this research is to investigate the nonlinear dynamic
behavior of a continuous rotor shaft system with viscoelastically
supported bearings. The nonlinearity occurs at the boundaries due
to nonlinear characteristics of the bearings. Timoshenko shaft
model is used for the shaft, Kelvin Voigt model is utilized for the
viscoelastic supports, and typical roller bearings with cubic nonlinearity are employed. In the analysis, free and forced vibration
are developed based on the direct multiple scales method of oneto-one frequency-to-amplitude relationship of nonlinear system
using the third order perturbation expansion. This approach does
not require the selection of an orthogonal basis; however, it requires additional eigenfunction solutions at every level of approximations. Moreover, the direct multiple scales method yields better
results than the discretized one for finite mode truncations and for
systems having quadratic and cubic nonlinearities 12. The results of the nonlinear analysis indicate that a limiting value of the
internal damping coefficient of the shaft exists where the trend of
the frequency-response curve switches. Also, a jump phenomenon
takes place for high values of the nonlinear elastic coefficient of
the bearings.
2
Contributed by the Technical Committee on Vibration and Sound for publication
in the JOURNAL OF VIBRATION AND ACOUSTICS. Manuscript received February
2002; Revised October 2002. Associate Editor: G. T. Flowers.
Equations of Motion
x i y
(1)
Q 1,t
2 u 1,t
1
M d
M d e 2 exp it
2
t2
(9)
1,t 0
(10)
where M d and e are the mass of the disk and the eccentricity of
the rotor, respectively. F l and F n are the linear and nonlinear
terms exerted by the bearings on the shaft and the viscoelastic
supports. Assuming that the bearings possess cubic nonlinear restoring forces in addition to linear damping forces, the associated
forces are
F l F xl iF yl k b1 u v c b
u v
t t
(11)
Au AG u C i u iu 0
(2)
F n F xn iF yn k b3 u x v x 3 i u y v y 3
I s iJ s EI s AG u 0
(3)
Equation 14 can be rewritten in terms of the complex transverse deflections u and v , and their corresponding complex conjugates u and v , as
EI s ,t
l
1 u ,t
Q ,t Q x ,t iQ y ,t AG
,t
l
2v
k v 1i v F l 0,t F n 0,t
t2
4 u 3 3u
2v 3u
v 2 v 3
(13)
Introducing a set of nondimensional quantities given in Appendix A, the nondimensional equations of motion with the associated
boundary conditions can be written as
2u * * 2u *
u*
C i
i * u * 0
t *2
2
t*
(14)
*
2 *
u*
2 *
2
3 *
0
2 i 1 *
t*
t*
2
(15)
M * 0,t 0
(16)
* u * v * c *
Q * 0,t * k b1
b
(5)
where M ( ,t) and Q( ,t) are the transverse bending moment and
the shear force at each cross section along the shaft.
Due to symmetry of the system, only the left half of the rotorshaft system, i.e., 0zl, is considered. Thus, the associated
boundary conditions can be defined as
M 0,t 0
F n k b3 43 u 2u 2uu
v v 2u u 2v 2u vv
v 2v
(4)
(12)
u* v*
k *
b3 W *
t* t*
0,t *
(17)
v * 0,t
k v* 1i v * 0,t
t *2
2
M 2*
k *
b1 u * v * c b*
u* v*
* W*
k b3
t* t*
0,t *
(6)
(18)
(7)
u * 1,t 1
1
M*
F * * 2 exp i * t * (19)
2 d t *2
2
(8)
Q * 1,t
* 1,t 0
(20)
(21)
v * v 1 ,T 0 ,T 2 v 3 ,T 0 ,T 2 . . .
(22)
* 1 ,T 0 ,T 2 3 3 ,T 0 ,T 2 . . .
(23)
C i* 2 C i ,
F * 3 F ,
* n* 2
(24)
1 u1
C i D 0 u 1 i n* u 1 0
2
2
2 1
u1
D 20 1 i 1 *
0
2 3 1
n D 0 1 2
at 0
1
0,
u1
* u 1 v 1 0,
1 k b1
2
*
M*
2 D 0 v 1 k *
v 1i v 1 k b1 u 1 v 1 0
at 1
1 0,
(25)
(27)
M*
d
(28)
2D 2 D 0 u 1 C i D 2 u 1 i u 1
(30)
* W 1 , and
c D 0 u 1 v 1 k b3
2
*
M*
2 D 0 v 3 k *
v 1i v 3 k b1 u 3 v 3
2M 2* D 0 D 2 v 1 c D 0 u 1 v 1 k *
b3 W 1
(31)
at 1
3 0,
M*
u3
d
1
2
D 20 u 3 M *
3
d D 2D 0u 1 2 F *
n
2
exp i n* 2 t *
where
W 1 41
(32)
41 u 31 3u
21v 1 3u
1v 21 v 31 .
3.1 First Order Equations. Noting that the equations of
motion and the associated boundary conditions are linear at order
, the first order approximation of the deflections and deformations can be assumed in the form
u 1 ,t * A T 2 e i *n T 0 Y u ,
1 ,t * A T 2 e i *n T 0 Y ,
v 1 t * A T 2 e i *n T 0 Y v
(33)
Substituting Eq. 33 into Eqs. 25 and 26 leads to a set of
ordinary differential equations given by
a 1 Y u a 2 Y u a 3 Y 0
(34)
b 1 Y b 2 Y b 3 Y u 0
(35)
a 1 1,
b 1 2 ,
a 3 1
b 2 1 1 *
n 3 ,
(36)
b 3 3
(37)
Ae
j1
s j
Be
j1
s j
(38)
where s j ( j1, . . . ,4) are the roots of the quartic auxiliary equation 7. Only four of the arbitrary complex constants A j and B j
( j1, . . . ,4) are independent, since they are related by the
relation
a 1 s 2j a 2
a 3s j
j1, . . . ,4
(39)
3 2u 3
C i D 0 u 3 i n* u 3
3
u3
* u 3 v 3
0, 3
3 k b1
B j jA j ,
u1
1
D 20 u 1 0
at 0
Y u
and
2 3
u3
3 3
2
2D 2 D 0 1 i 1 *
n D 2 1 D 0 1
(26)
Order 3 :
D 20 u 3
D 20 3 i 1 n* D 0 3 2
(40)
*
m 1k 3 s j j k b1
*
m 15k b1
m 2k j s j
m 250
*
m 3k k b1
2
*
m 35M *
2 *
n k *
v 1i k b1
m 4k e s j 3 s j j M d* n* 2 /2
m 450
m 5k j e
m 550
sj
3 ,t * ,T 2 e i *n T 0 ,
v 3 t * v T 2 e i n* T 0
(42)
u
2
1 1 *
n 3
2
(43)
(44)
at 0
0,
u
* u v
k b1
c i *
n A T 2 Y u 0 Y v
3
k * A 2 T 2 A T 2 W 0 ,
4 b3
and
* 1i v k b1
* u v
n* M *
2 v k v
2i n* M 2* A T 2 Y v i n* c A T 2 Y u 0 Y v
* A T 2 A T 2 W 0
k b3
2
(45)
at 1
u 3 2i n* C i A T 2 Y u iC i A T 2 Y u
0,
u
2
*
n u
2
1
F * 2 exp i T 2
2 n
(46)
u 0,t * c i n* A T 2 Y u 0 Y v
1
2
i *
n M*
d A T 2 Y u 1 2 F *
n exp i T 2 (47)
where u and are the solutions of the adjoint homogeneous system of Eqs. 43 and 44 and the boundary condition Eqs. 31
and 32 with a zero right hand side.
It can be seen that the homogeneous adjoint system has an
identical form to the first order approximation given by Eqs. 25
28; hence, the solutions for u , and v should have the same
form as u 1 , 1 and v 1 .
Substituting the solutions of u and into Eq. 47, the solvability condition takes the compact form
b 1 2i *
n C i A T 2 b 1 iC i A T 2
b 2 i n* 2 1 A T 2 b 2 1 *
n A T2
1
2
2
i *
n M*
d A T 2 Y u 1 2 F *
n exp i T 2 Y u 1 (48)
M*
d
i n* M d* A T 2 Y u 1
i *
n 2 1 A T 2 Y 1 *
n A T 2 Y d
3
2
2
* A 2 T 2 A T 2 W 0 Y u 0
c i *
n A T 2 Y u 0 Y v 4 k b3
3
4
Solvability Condition
* A 2 T 2 A T 2 W 0 u
43 k b3
1,t *
i n* 2 1 A T 2 Y 1 n* A T 2 Y
2i n* C i A T 2 Y u iC i A T 2 Y u
(41)
where ( ) / T 2 , A (T 2 ) is the complex conjugate of the coefficient A(T 2 ), and W 0 Y 3u (0)3Y 2u (0)Y v 3Y u (0)Y 2v Y 3v .
Note that the above terms in Eqs. 45 and 46 are collected only
for the primary resonance, i.e., exp(in*T0).
2 u
2 *
n u
k1, . . . ,4
(49)
T 2
(50)
terms and all the damping terms are eliminated. Such frequencies
are called free frequencies of the nonlinear system, and they are
dependent among others on the initial conditions and strictly applicable at steady state. However, these frequencies should not be
confused with the linear system natural frequencies.
(52)
b 3 a b 4 a b 5 a b 6 a 3 b 7 aF n* 2 Y u 1 cos
b 4 a b 3 a b 8 a b 6 a b 9 aF *
n Y u 1 sin
2
(53)
b3 a b6 2
b5
a 1
b4 a b4
b4
(54)
where
c 2 (b 8 b 4 b 5 b 3 ) and c 3 b 6 b 4 b 6 b 3 .
Equations 53 and 54 can be solved in conjunction with the
initial condition aa 0 and 0 at T 2 0 to yield either of the
following three cases:
Case 1: c 2 /c 3 0
c 1 b 24 b 23 ,
a c 2 a 21 e 2c 2 T 2 /c 1 / c 3 a 21 c 3 e 2c 2 T 2 /c 1 c 3
T 2 0
b3 1
ln
b4 2
a 21 e 2c 2 T 2 /c 1
2
a 1 e 2c 2 T 2 /c 1 1
(55)
1 b3
ln
2 b4
a 21 1
a 21
b6
c1
ln a 21 e 2c 2 T 2 /c 1 1
b 4 2c 3 a 21 1
2a 21 c 2
c1
T 2 ln a 21 1
b5
T2
b4
T 2 0
b5
T2
b4
where a 22 (a 20 c 2 /c 3 )/a 20 .
Case 3: c 2 0
T 2 0
b3
2c 3 2
ln 1
a T
2b 4
c1 0 2
c1 b6
2c 3 2
b5
ln 1
a 0 T 2 1
T2
2c 3 b 4
c1
b4
(59)
(60)
* n* 2
nn
(61)
E2.041011 Pa
v 0.3
7750 kg/m3
d s 0.0286 m
c b 400 kg/s
0.68
l0.15 m
0.01
k b3 1107 N/m3
k v 1109 N/m
M 2 0.5 kg
(57)
(62)
(56)
(58)
a a 20 / 12c 3 a 20 T 2 /c 1
k b1 110 N/m
2c 2
b6
T ln a 22
ln a 22 e 2c 2 T 2 /c 1 1 ln a 22
c1 2
2b 4
b 14
1
b 2 4b 13b 15
2b 13 2b 13 14
b3
c1
ln a 22 e 2c 2 T 2 /c 1 1
2c 3 b 4 a 22 1
C i 200 kg/m.s
where a 21 a 20 /(a 20 c 2 /c 3 ).
Case 2: c 2 /c 3 0
a c 2 / c 3 a 22 c 3 e 2c 2 T 2 /c 1 c 3
d d 0.13 m
M d 55 kg
e0.01 m
tic coefficient k b3 of the nonlinear bearing; i.e., for weak nonlinear system. As k b3 increases, the change in the natural frequency
of the nonlinear system increases significantly. The increase of the
frequencies of the nonlinear system and the bend of the curve
towards the higher frequencies are typical trends of hardening
nonlinearities.
Figures 4 and 5 show the effect of the internal viscous damping
coefficient C i of the shaft on the free frequency of the nonlinear
system versus amplitudes. Figure 4 depicts the trend for low C i
values of 100 kg/m.s to 200 kg/m.s, whereas Fig. 5 shows the
trend for higher C i values of 270 kg/m.s to 300 kg/m.s. Figure 4
shows that increasing the internal viscous damping of the shaft
shifts the free frequency of the nonlinear system to higher values
while almost maintaining the curve characteristic. However, this
trend has a limiting value of C i 270 kg/m.s beyond which any
increase in C i will result in a shift of the curve to lower frequencies, as shown in Fig. 5. The limiting value of the internal damp-
Fig. 3 Free frequency of the nonlinear system versus amplitudes for various values of bearing nonlinear elastic coefficient k b 3
Fig. 4 Free frequency of the nonlinear system versus amplitudes for various values of internal viscous damping coefficient C i of the shaft
the multiple scales method starts to deviate from the basic assumption of weak nonlinearity. Thus, the error in predicting the
frequency response curves increases.
Fig. 5 Free frequency of the nonlinear system versus amplitudes for higher
values of internal viscous damping coefficient C i of the shaft
Fig. 6 Frequency response curves for various values of the bearing nonlinear elastic coefficient k b 3
Appendix A
G
t,
l2
u
u * ,
l
C i*
Js
1
,
Is
k v*
l
A G
,
l
M*
d
b 4 2b 1r n* b 1m C i b 2r n* 1 2 b 12r M d* n*
b 5 2b 1r n* 2b 2r n* b 12r M d* n* , b 6 3/16 k *
b3 b 10r ,
b 7 b 11m c n* , b 8 2b 1m n* 2b 2m n* b 12m M d* n* ,
b 9 b 11r c n* ,
b 11Y 2u 0 Y u 0 Y v
b 12Y 2u 1 , and b j b jr ib jm
j1,2,10,11,12 .
*
k b3
l
M ,
Is d
l5
k ,
GI s b3
M*
2
l2
,
G
v
,
l
l
M ,
Is 2
*
k b1
c b*
l3
k ,
GI s b1
l2
I s G
F *
cb ,
eM d
,
Is
W * 43 u * 2u * 2u *u * v * v * 2u * u * 2v * 2u * v *v *
v * 2v * 41 u * 3 3u
* 2v * 3u
*v * 2 v * 3
Journal of Vibration and Acoustics
b 13b 25 b 28 ,
b 142b 6 b 5 b 8 a 2 2 b 5 b 7 b 8 b 9 ,
b 15 b 6 a 2 b 7 2 b 6 a 2 b 9 2
*
l
M *
M 2
,
GI s
u*
l2
Q 3
* ,
GI s
l3
k ,
GI s v
v *
Ci ,
Al 2
3
,
Is
E
,
2
G
Q *
b 3 2b 1m *
n b 1r C i b 2m *
n 1 2 b 12m M *
d *
n ,
Appendix C
b 10W 0 Y u 0 Y 4u 0 3Y 3u 0 Y v 3Y 2u 0 Y 2v Y u 0 Y 3v ,
Acknowledgment
t *
Appendix B
1 2
F *4Y 2 1
a2 n u
References
1 Darlow, M., and Zorzi, E., 1981, Mechanical Design Handbook of Elastomers,
NASA CR3423.
2 Dutt, J. K., and Nakra, B. C., 1992, Stability of Rotor Systems with Viscoelastic Supports, J. Sound Vib., 1531, pp. 8996.
3 Dutt, J. K., and Nakra, B. C., 1993, Vibration Response Reduction of a Rotor
Shaft System Using Viscoelastic Polymeric Supports, ASME J. Vibr. Acoust.,
115, pp. 221223.
4 Dutt, J. K., and Nakra, B. C., 1995, Dynamics of Rotor Shaft System on
Flexible Supports with Gyroscopic Effects, Mech. Res. Commun., 226, pp.
541545.
5 Kulkarni, P., Pannu, S., and Nakra, B. C., 1993, Unbalance Response and
Stability of a Rotating System with Viscoelastically Supported Bearings,
Mech. Mach. Theory, 283, pp. 427 436.
6 Shabaneh, N. H., and Zu, Jean W., 1999, Vibration Analysis of Viscoelastically Supported Rotor-Bearing Systems, Asia-Pacific Vibration Conference
1999 A-PVC99, Singapore, December.
7 Shabaneh, N. H., and Zu, Jean W., 2000, Dynamic Analysis of Rotor-Shaft
Systems with Viscoelastically Supported Bearings, Mech. Mach. Theory,
359, pp. 13131330.
8 Shabaneh, N. H., and Zu, Jean W., 2000, Dynamic and Stability Analysis of
Rotor-Shaft Systems with Viscoelastically Supported Bearings, Trans. Can.
Soc. Mech. Eng., 241B, pp. 179189.
9 Yamamoto, T., Yasuda, K., and Nagasaka, I., 1976, Ultra-Subharmonic Oscillations in a Nonlinear Vibratory System, Bull. JSME, 19138, pp. 1442
1447.
10 Ji, Z., and Zu, J. W., 1998, Method of Multiple Scales for Vibration Analysis
of Rotor-Shaft Systems with Non-Linear Bearing Pedestal Model, J. Sound
Vib., 2182, pp. 293305.
11 Bhattacharyya, K., and Dutt, J. K., 1997, Unbalance Response and Stability
Analysis of Horizontal Rotor Systems Mounted on Nonlinear Rolling Element
Bearings with Viscoelastic Supports, ASME J. Vibr. Acoust., 119, pp. 539
544.
12 Nayfeh, A. H., Nayfeh, J. F., and Mook, D. T., 1992, On Methods for Continuous Systems with Quadratic and Cubic Nonlinearities, Nonlinear Dyn. 3,
pp. 145162.
13 Nayfeh, A. H., 1981, Introduction to Perturbation Techniques, Wiley, New
York.
14 Yamamoto, Toshio, and Ishida, Yukio, 2001, Linear and Nonlinear Rotordynamics: A Modern Treatment with Applications, Wiley Series in Nonlinear
Science.
Animesh Chatterjee
Assistant Professor,
Department of Mechanical Engineering,
Visvesvaraya National Institute of Technology,
Nagpur, India-440011
e-mail: animeshch@rediffmail.com
Nalinaksh S. Vyas
Professor,
Department of Mechanical Engineering,
Indian Institute of Technology,
Kanpur, India-208016
e-mail: vyas@iitk.ac.in
Introduction
(1)
with
x n t
...
h n 1 , . . . , n f t 1 . . . f t
nd1 . . . dn
(2)
JULY 2003, Vol. 125 299
H n 1 , . . . , n
...
where
X 0
h n 1 , . . . , n
j ii
i1
d1 . . . dn
(3)
with
i n 2
A
A
f t A cos t e j t e j t
2
2
(4)
A
2
pqn
C q H np,q e j p,q t
n1
x t
n1
pqn
A
2
H 1 p,q
n
Cq
k3
(7)
p ,q 1
n1C q1H n 1
p i q i n i
n 1 n 2 n 3 n
i1, . . . .N
(9)
n1
(11a)
4
X 3
i 3
A3
i2
q ,q 2
p ,q
p ,q
p ,q
C q 1 H n 1 1 * n 2 C q 2 H n 2 2 * n 3 C q 3 H n 3 3
1
2
3
(13)
(11b)
(11c)
where the higher order kernel transforms are related to the lower
order kernel transforms through nonlinear parameters Chatterjee
and Vyas, 14 as
ni1,i1
C i1 H n2i2
2
X 2
i 2
A2
i2
H 3 , , ,
* n2C q2H n2
n2i2
1
H 1
X
i
A
i2
(6)
X 3 cos 3 t 3 ..
H np,q
(8)
n2i2
H 2 ,
x t X 0 X cos t 1 X 2 cos 2 t 2
and n X n
mx t cx t k 1 x t k 2 x 2 t k 3 x 3 t A cos t (10)
p i q i n i
n 1 n 2 n
C n H n,n
2n
first three response harmonic series, after re-arranging and truncating Eq. 8 to a finite number of terms, k, can be expressed as
C q H np,q e j p,q t
k2
i1
2n
Parameter Estimation
(5)
A
2
2n
For a system with polynomial form of nonlinearity under harmonic excitation given by
(5)
X n
for n1
(12)
(14)
3 H 31 H 1 3
(15)
where
k
i ( ) is computed with the H 1 ( )
Step-III: The series i2
values taken from the best fit curve estimated in Step-I and the
nonlinear parameter, k 3 , estimated in Step-II and substituted in
equation 11a to obtain new estimates of linear parameters.
k
Step-IV: The series i2
i (3 ) is computed and substituted in
Eq. 11c to refine the estimate of the nonlinear parameter k 3 .
Iteration is continued till the estimate of nonlinear parameter, k 3 ,
converges within a specified limit.
Experimental Investigation
(16)
n 1012.2 Hz,
0.01156.
1019 N/m3 . Assuming that both the bearings are identical and
act in parallel, the nonlinear stiffness parameter of each bearing
then becomes 0.8751019 N/m3 . Figure 8b shows the final estimate of the first order kernel transform along with its preliminary estimate. It is seen that the successive iterations have improved the FRF values significantly near and around the natural
frequency. Final estimates of natural frequency and damping become
n 1011.47 Hz,
0.01078.
The linear stiffness parameter, k 1 , of the bearing system is computed from the natural frequency and the equivalent rotor mass
per bearing and is found to be 1.585107 N/m.
Similar exercises are carried out with excitation level of 3 N
Case II and 2 N Case III corresponding to 5% and 3% measurability respectively. The convergence trend in the estimate k 3 is
shown respectively in Fig. 9a, b. The estimates of nonlinear
stiffness parameter of each bearing is found to be 1.4285
1019 N/m3 Case II and 1.7401019 N/m3 Case III respectively. Damping and linear stiffness parameter are found as
Case II:
0.01015
k 1 1.535107 N/m.
Case III
0.01002
k 1 1.586107 N/m
Validation of Estimates
For validation of the experimentally estimated nonlinear stiffness parameter, analytical formulations of Harris 8 and Ragulski
et al. 9 are employed. These formulations are based on Hertzs
theory of elastic contacts and treat the bearings in isolation of the
shaft. Figure 11 shows a typical isolated ball bearing configuration, in which external forces act along x-axis. i is the angle
between load axis i.e., x-axis and the radial direction of a typical
ith ball element. For a displacement x, y of the moving ring
along the respective axes, total elastic force acting in radial direction at the point of contact of ith ball is given by
F i k n gx cos i y sin i 3/2
(17)
(18)
(19)
where g is the radial pre-load between the ball and the races, k n is
a coefficient of proportionality depending on the geometric and
material properties of the bearing. Total restoring force of the
bearing is equal to sum of elastic forces of all the elements,
n
i.e.,
F
i1
xi
(20)
gx cos
i
i1
i1
sin i ;
gx cos
i1
3/2
sin2 i
3/2
i1
(21)
gx cos
i
i1
1/2
gx cos
gx cos
i1
k x F/ x
(22)
3/2
sin i cos i
(24)
sin2 i
3/2
sin2 i cos i
k x K n
AD n1 / Bn 2 sin i cos i
where
304 Vol. 125, JULY 2003
cos i CBn
1/2
(23)
Conclusion
0.8750
1.4285
1.7400
1011.47 Hz
1012.47 Hz
1011.90 Hz
1.585
1.535
1.586
0.01078
0.01015
0.01002
1.20107 4.011019x 2
1.47107 2.181019x 2
1.69107 1.421019x 2
1.89107 1.021019x 2
2.08107 0.611019x 2
k1
k2
k3
m
x n (t)
3( )
i (n )
References
Fig. 12 Comparison of estimates of stiffness parameters 15:
Theoretical values with pre-load 0.2, 0.3, 0.4, 0.5 and 0.6 m
respectively. Harris 8 and Ragulski et al. 9 6,7,8: Present
experimental estimates for cases I, II and III respectively. 9: Experimental estimate of Tiwari 16 10: Experimental estimate of
Khan 13
Acknowledgment
The authors wish to express their thanks to the financial aid
being provided by the Propulsion Panel of Aeronautical Research
and Development Board, Ministry of Defense, Government of India, in carrying out the study.
Nomenclature
H n( 1 , . . . , n)
X(n )
c
f (t)
g
h n( 1 , . . . , n)
1 Bedrosian, E., and Rice, S. O., 1971, The Output Properties of Volterra Systems Nonlinear System with Memory Driven by Harmonic and Gaussian
Input, Proc. IEEE, 5912, pp. 1688 1707.
2 Boyd, S., Tang, Y. S., and Chua, L. O., 1983, Measuring Volterra Kernels,
IEEE Trans. Circuits Syst., CAS-308, pp. 571577.
3 Chua, L. O., and Liao, Y., 1989, Measuring Volterra Kernels II, Int. J. of
Circuit Theory and Applications, 17, pp. 151190.
4 Gifford, S. J., and Tomlinson, G. R., 1989, Recent Advances in the Application of Functional Series to Nonlinear Structures, J. Sound Vib., 1352, pp.
289317.
5 Chatterjee, A., and Vyas, N. S., 2001, Stiffness Nonlinearity Classification
through Structured Response Component Analysis using Volterra Series,
Mech. Syst. Signal Process., 152, pp. 323336.
6 Lee, G. M., 1997, Estimation of Nonlinear System Parameters using Higher
Order Frequency Response Functions, Mech. Syst. Signal Process., 112, pp.
219228.
7 Chatterjee, A., and Vyas, N. S., 2002, Nonlinear Parameter Estimation
through Volterra Series using Method of Recursive Iteration, accepted for
publication in J. Sound Vib.
8 Harris, T. A., 1984, Rolling Bearing Analysis, Wiley, New York.
9 Ragulskis, K. M., Jurkauskas A. Y., Atstupenas, V. V., Vitkute, A. Y., and
Kulvec, A. P., 1974, Vibration in Bearings, Mintis Publishers, Vilnius.
10 Bannister, R. H., 1976, A Theoretical And Experimental Investigation Illustrating the Influence of Nonlinearity and Misalignment on the Eight Film Coefficients, Proc. Inst. Mech. Eng., 190, pp. 271278.
11 Choi, F. K., Braun, M. J., and Hu, Y., 1992, Nonlinear Transient and Frequency Response Analysis of a Hydrodynamic Bearing, ASME J. Tribol.,
114, pp. 448 454.
12 Garibaldi, L., and Tomlinson, G. R., 1988, A Procedure for Identifying Nonlinearity in Rigid Rotors Supported in Hydrodynamic and Ball/Roller Bearing
System, I. Mech. Proc. on Vibrations in Rotating Machinery, 4, pp. 229234.
13 Khan, A. A., and Vyas, N. S., 2001, Application of Volterra and Wiener
Theories for Nonlinear Parameter Estimation in a Rotor-Bearing System,
Nonlinear Dyn., 243, pp. 285304.
14 Chatterjee, A., and Vyas, N. S., 2000, Convergence Analysis of Volterra
Series Response of Nonlinear Systems Subjected to Harmonic Excitations, J.
Sound Vib., 2362, pp. 339358.
15 Ewins, D. J., 1984, Modal Testing: Theory and Practice, Research Studies
Press, England.
16 Tiwari, R., and Vyas, N. S., 1995, Estimation of Nonlinear Stiffness Parameters of Rolling Element Bearings from Random Response of Rotor Bearing
Systems, Journal of Sound Vib. 187 2, pp. 229239.
Y. S. Ho
H. Liu
Department of Mechanical Engineering,
The Hong Kong Polytechnic University,
Hong Kong SAR, P.R. China
L. Yu
Theory of Lubrication and Bearing Institute,
Xian Jiaotong University,
Xian, Shaanxi, P.R. China
Introduction
vestigators deal with the effect of thrust bearing upon the nonlinear stability of T period motion of unbalanced rotor systems.
This paper is concerned with the effect of a TAMB on the
stability and bifurcation of an AMBRS supported by both journal
and thrust active magnetic bearings. The rotor is flexible, and
system equations are formulated by combining the equations of
motion of the rotor and the equations of the decentralized PID
controllers. This is typically a mechanical-electrical system consisting of a large number of linear components such as flexible
shaft elements without disk and bearing and a few of nonlinear
components nonlinear active magnetic bearings. Finite elements
method 14,15, and order reduction techniques are used to reduce
the order of the model 10,11,13. Floquet theory 16 18, shooting method and path-following technique 12,19 are used to
analyses the stability and bifurcation of T periodic motion of the
AMBRS equipped with both journal AMBs and TAMBs. The
effects of the TAMB and the mass eccentricity are discussed
especially.
(1)
where M , G , K R
and Q , f R are the mass matrices,
gyroscope matrices, stiffness matrices, external force vector inS
nn
(8)
T
b KS i 2
kk i
xS x 1 y 1 1 1 x p y p p p T
(2)
MSbb
S
Mbi
S
Mib
MiiS
GSbb
xSb
S
xiS
Gib
QSb
f Sb xSb ,xSb
S
Qi
0
S
Gbi
GiiS
KSbb
xSb
S
xiS
Kib
S
Kbi
KiiS
xSb
xiS
(4)
xSb x 1 y 1 1 1 x m y m m m T
(5)
f Sb f x j f y j 0 0 0 0 M x k M y k T
(6)
where T1 b i
bb
xSb
S
xi
ib
bk pSb
S
ik pk
(10)
1
pSb
bb
pSk
0kb
S
1
bb bk xb
p T2 p
p Sk
Ikk
(11)
TT1 T2
(12)
(13)
(7)
(14)
(15)
with
For reducing the order of the degree-of-freedom of linear components, xS can be written as a linear combination of n c columns:
x T1 p
mz F z
(9)
(3)
Ibb
0ib
q xSb
TT MS TMd
pSk
zT
TT GS TGd
TT KS TKd
Q TT QS 0 T TT f S 0 T FTdex 0 F z T
where Md , Gd and Kd are the mass, damping and stiffness matrices of disks; m is the mass of the rotor; Fdex is unbalance forces
caused by mass eccentricity of disks. The behavior of this nonlinear system depends on the rotating frequency of the shaft and
the eccentricities e explicitly.
The eigenfrequencies cut of Eq. 13 are accurate for the
case QS 0 and GS 0, with cut the highest cut-off eigenfrequency in the reduction, and higher eigenfrequencies will be inaccurate. Because nonlinear systems can generate frequencies
higher than their excitation frequency, cut has to be chosen
higher than the maximum excitation frequency.
2.2 Electromagnetic Force of Journal Bearing. An eightpole journal AMB shown in Fig. 3 is used for analysis. To simTransactions of the ASME
h i 1 c z z r i sin r i cos ,
h o1 c z z r o sin r o cos ,
plify the notations, subscripts indicating bearing numbers are ignored. The forces in this journal magnetic bearing can be
expressed as
f x f r f l xy x/c r f t f b
(16)
f y f t f b xy y/c r f r f l
with
f r
f t
0 N r2 A r I 0x i x 2
4
c r x 2
0 N r2 A r
I 0y i y
c r y 2
f 1
f b
0 N r2 A r I 0x i x 2
4
c r x 2
0 N r2 A r
4
h o2 c z z r o sin r o cos ,
It is assumed that the magnetic flux from the point (r i , ) on the
inner annulus runs entirely back to the point (r o , ) on the outer
annulus as shown in Fig. 5. Therefore the relationship between
radii r i and r o is
r o R 4 r i R 1
(17)
I 0y i y
c r y 2
F z1
R1
F z2
M x1
M x2
R2
R1
M y2
R2
R1
M y1
R1
ro
0 N I 0z i z 2
20
h o1 r o /r i h i 1
where
ro
0 N I 0z i z 2
2
20
h o r o /r i h i 2
cos dr i d
R4
sin dr i d
sin dr i d
R3
R4
R3
R4
ro
0 N I 0z i z 2
20
h o1 r o /r i h i 1
ro
0 N I 0z i z 2
2
20
h o r o /r i h i 2
R3
R4
R3
R4
1
o N I 0z i z 2
r o dr o d
20
h o1 r o /r i h i 1 2
1
0 N I 0z i z 2
r o dr o d
2
20
h o r o /r i h i 2 2
R3
cos dr i d
R3
R4
ro
0 N I 0z i z 2
20
h o1 r o /r i h i 1
(20)
M y M y1 M y2
(19)
M x M x1 M x2
2
0 N I 0z i z 2
r o /r i
r i dr i d
20
h o2 r o /r i h i 2
R2
R2
R 4 R 3
R 2 R 1
F z F z1 F z2
2
r o /r i
0 N I 0z i z 2
1
1 r i dr i d
20
h o r o /r i h i
ro
0 N I 0z i z 2
2
20
h o r o /r i h i 2
R1
R2
R1
R2
with
The forces and moments provided by the thrust magnetic bearing can be expressed as 22
(18)
h i 2 c z z r i sin r i cos ,
ro
0 N I 0z i z 2
20
h o1 r o /r i h i 1
ro
0 N I 0z i z 2
2
20
h o r o /r i h i 2
cos dr o d
cos dr o d
sin dr o d
sin dr o d
(21)
When state variables X(q,q,Ibb ) T are introduced, the corresponding system equations in state space are
q
M1 QGqKq
Aa As Kp xbb Ki xbb Kd xbb
(28)
3
Fig. 5 Magnetic flux path in a thrust magnetic bearing
Method of Solution
Aa
1Ta s
(22)
As
1Ts s
(23)
Kd s
Ki
s
1Td s
(24)
Ki
G s Ga s "Gs s "Gc s Aa As Kp Kd s
s
(25)
(26)
(29)
with
f X,t,
f X,t,
X
X t X tT
q
M1 QGqKq
Aa As Kp xbb Ki xbb Kd xbb
(30)
(31)
with H Xs , Xs t 0 T Xs t 0
and N J is the number of journal AMBs.
2.5 System Equations. Combining the equations for the rotor, the sensors, the amplifiers and the PID controllers, the system
equations are obtained
MqGqKqQ
Ibb Aa As Kp xbb Ki xbb Kd xbb
(27)
For a given s , the corresponding solution Xs can be obtained from Eq. 31 using the iterative Newton process. The Jacobian matrix can be obtained as
Table 1 Parameters of journal and thrust AMBs
Bearing type
Parameter
Value
Journal bearings
radial clearance
bias current
winding number
width
diameter
0.0004 m
4A
57
0.08 m
0.16 m
Thrust bearing
0.0006 m
4A
143
0.1 m
0.04 m
0.065 m
0.08 m
0.1 m
Value
A a 1/
A 2 V/m
kp
k i 1/s
k d s
1
7800
3.8
200
0.01
H
JI
Xs
Xn1 Xn
H X,
Xs
H X,
(34)
n1 n
f t, ,X
f t, ,X
d
S
S
dt
X
(32)
f
d
S
S
dt
X
(33)
(35)
Fig. 9 Hopf T periodic solution with TAMB a Hopf T periodic solution at 70000 revmin b
A amplitude-frequency diagram of x a
Consider TAMB
leading Floquet multiplier
modulus
0.992395i0.0726290
0.995049
0.992400i0.0726408
0.995055
0.992411i0.0726599
0.995067
0.992429i0.0726863
0.995088
0.992444i0.0727040
0.995104
0.992460i0.0727215
0.995121
0.992214i0.0751451
0.995055
0.992219i0.0751573
0.995061
0.992230i0.0751772
0.995074
0.992249i0.0752049
0.995095
0.960353i0.279283
1.00014
0.977981i0.273811
1.01559
Fig. 13 Stable Quasi-periodic motion at 30000 revmin, e 0 22.5 m with TAMB a Stable
Quasi-periodic motion b Poincare maps
Conclusions
Acknowledgment
The work described in this paper was fully supported by a grant
from the Research Grants Council of the Hong Kong Special Administrative Region, China Project No. PolyU5115/98E. It is
also supported by the National Natural Science Foundation, China
Project No. 19990511.
Nomenclature
MS
GS
KS
xS
References
1 Virgin, L. N., Walsh, T. F., and Knight, J. D., 1995, Nonlinear Behavior of a
Magnetic Bearing System, ASME J. Eng. Gas Turbines Power, 1177, pp.
582588.
2 Chinta, M., Palazzolo, A. B., and Kascak, A., 1996, Quasiperiodic Vibration
of a Rotor in a Magnetic Bearing With Geometric Coupling, Proc. 5th International Symposium on Magnetic Bearing, Kanazawa, Japan, pp. 147152.
3 Chinta, M., and Palazzolo, A. B., 1998, Stability and Bifurcation of Rotor
Motion in a Magnetic Bearing, J. Sound Vib., 2145, pp. 793 803.
4 Mittwollen, N., Hegel, T., and Glienicke, J., 1991, Effect of Hydrodynamic
Thrust Bearings on Lateral Shaft Vibration, ASME J. Tribol., 1134, pp.
811 818.
5 Yu, L., and Bhat, R. B., 1995, Coupled Dynamics of a Rotor-Bearing System
Equipped with a Hydrodynamic Thrust Bearing, Shock and Vibration 21,
pp. 114.
6 Lund, J. W., and Nelson, H. B., 1980, Instability Threshold of an Unbalanced
Rigid Rotor in Short Journal Bearings, Proc. Second International Conference on Vibration in Rotating Machinery, Cambridge, UK.
7 Brancati, R., Rocca, E., Rosso, M., and Rosso, R., 1995, Journal Orbits and
Their Stability for Rigid Unbalanced Rotors, ASME J. Tribol., 117, pp. 709
716.
8 Kim, Y. B., and Noah, S. T., 1990, Bifurcation Analysis for a Modified
Jeffcott Rotor with Bearing Clearance, Nonlinear Dynamics, 1, pp. 221241.
9 Choi, S. K., and Noah, S. T., 1994, Mode-Locking and Chaos in a Jeffcott
Rotor with Bearing Clearance, ASME J. Appl. Mech., 61, pp. 131138.
10 Nelson, H. D., Mechan, W. I., Fleming, D. P., and Kascak, A. F., 1983, Nonlinear Analysis of Rotor Bearing System Using Component Mode Synthesis,
ASME paper No. 83-GT-303.
11 Nataraj, C., and Nelson, H. D., 1989, Periodic Solutions in Rotor Dynamic
System With Nonlinear Supports: A General Approach, ASME J. Vibr.
Acoust., 111, pp. 187193.
12 Fey, R. H. B., Van Campen, D. H., and de Kraker, A., 1996, Long Term
Structural Dynamics of Mechanical Systems With Local Nonlinearities,
ASME J. Vibr. Acoust., 118, pp. 147153.
13 Zheng, T., and Hascbe, N., 2000, Nonlinear Dynamic Behaviors of a Complex Rotor-Bearing System, ASME J. Appl. Mech., 67, pp. 485 495.
14 Nelson, H. D., 1980, A Finite Rotating Shaft Element Using Timoshenko
Beam Theory, ASME J. Mech. Des., 102, pp. 793 803.
15 Lalanne, M., and Ferraris, G., 1997, Rotordynamics Prediction in Engineering,
John Wiley, New York.
16 Iooss, G., and Joseph, D. D., 1980, Elementary Stability and Bifurcation
Theory, Springer-Verlag, New York.
17 Parker, T. S., and Chua, L. O., 1989, Practical Numerical Algorithms for
Chaotic System, Springer-Verlag, New York.
18 Seydel, R., 1988, From Equilibrium to Chaos, Practical Bifurcation and Stability Analysis, Elsevier, New York.
19 Sundararajan, P., and Noah, S. T., 1997, Dynamics of Forced Nonlinear Systems Using Shooting/Arclength Continuation Method-Application to Rotor
System, ASME J. Vibr. Acoust., 1191, pp. 1020.
20 Craig, R. R., Jr., 1985, A Review of Time-Domain and Frequency-Domain
Component Modes Synthesis Methods, Combined Experimental/Analytical
Modeling of Dynamic Structural Systems Using Substructure Synthesis, D. R.
Martinez and A. K. Miller, eds., ASCE/ASME, New York, NY, pp. 131.
21 Knight, J. D., Xia, Z., and McCaul, E. B., 1992, Forces in Magnetic Journal
Bearings: Nonlinear Computation and Experimental Measurement, Proc. 3rd
International Symposium on Magnetic Bearing, Alexandria, VA, pp. 441 450.
22 Ho, Y. S., Yu, L., and Liu, H., 1999, Rotor Dynamic Coefficients of a Thrust
Active Magnetic Bearing Considering Runner tilt, Proc. Inst. Mech. Eng.,
Part J: J. Eng. Tribol., 213J6, pp. 451 462.
Chun-Ping Zou
School of Mechanical and Power Engineering,
Shanghai Jiao Tong University,
Shanghai, Peoples Republic of China;
Department of Mechanical and Power
Engineering,
East China Shipbuilding Institute,
Zhenjiang, Peoples Republic of China
Duan-Shi Chen
Hong-Xing Hua
School of Mechanical and Power Engineering,
Shanghai Jiao Tong University,
Shanghai, Peoples Republic of China
Introduction
There are several kinds of methods for torsional vibration calculation of rotating shafting system, namely Holzers method,
transfer matrix method and FEM. The well-known Holzers
method, a simple and systematic approach to calculating the frequencies and mode shapes of systems, was used originally for
analysis of an undamped system. Den Hartog and Li 1 improved
the Holzers method by using complex numbers. Pestel and
Leckie 2 introduced the transfer matrix method with the point
and field transfer matrices. Later, Sankar 3 and Dawson and
Davies 4 improved this approach. Huang and Horng 5 used
complex numbers to extend the transfer matrix method with the
Newton-Raphson technique to analyze the torsional vibration for
damped systems. The method extended by Huang and Horng
eliminates the operation of the inverse matrix because the derivatives of angular displacement and the torque are used directly with
the Newton-Raphson technique to determine the eigenvalues of
torsional vibration system. The major drawbacks of the transfer
matrix approach as noted by Firoozian and Stanway 6 are that
the information obtained is limited to stability assessment and the
prediction of critical speeds, and that inherent numerical instabilities are liable to occur in the analysis of more complex systems.
Li 7 used FEM to analyze the crankshaft torsional vibration of
diesel engine. One of the characteristics of FEM is that a very
large number of degrees of freedom are required. All these methods, namely Holzers method, transfer matrix method and FEM,
are very effective for the calculation of a single-spool shafting
system. However, when these methods are applied to calculate
shafting systems with open-circuit and closed-circuit branches,
they are not very effective.
Modal synthesis method is a modeling method permitting the
representation of a relatively complex structure by a reduced number of degrees of freedom. The concept of substructure synthesis
Contributed by the Technical Committee on Vibration and Sound for publication
in the JOURNAL OF VIBRATION AND ACOUSTICS. Manuscript received November
2001; revised November 2002. Associate Editor: B. Yang.
placement of each substructure is converted into innerdeformation of flexible substructure so that the coordinative
boundary condition between the flexible substructure and other
substructures could be satisfied. The other parts are naturally divided into some substructures and they can be analyzed by FEM.
The lower-frequency normal modes of substructures are retained
and the higher-frequency normal modes are neglected by a frequency truncation criterion. The lower-frequency normal modes
are considered to be the assumed modes of Rayleigh-Ritz analysis
of whole structure. As an example of application of this method,
the analysis of torsional vibration of a cam-type engine shafting
system is carried out both numerically and experimentally.
s1,2,m
s s1 , s2 , si sn
(3)
M si
where
is the ith dominant mass in dominant mass matrix.
The normalized modal matrix from 2 and 3 is
s
s1 s2
,
1
,
2
M s M s
si
,
i
M s
sn
M sn
s1,2,m
(4)
(5)
(6)
(7)
(1)
s1,2,m
(2)
x sk ,x sa s1,2,m
(8)
x s( i )
where
is ith order of normalized modal column vector of sth
second-class substructure; x sk represents the prior l order retained
modes; x sa represents the truncation modes of the order from (l
1) to n; and n is the number of degrees of freedom of sth
substructure.
(9)
P sk
P s P a
s
(10)
P sk
Q s x s P s x sk x sa P a x sk P sk
s
(11)
s1,2,m
(12)
(13)
k T
where P L p k1 ,p k2 ,p m
; K L diagdiag21,k ,diag22,k ,
2
diagm,k]. P L is the matrix of normal modal coordinates of
m second-class substructures where some coordinates are independent of each other. Using boundary connection condition, the
UN-independent coordinates can be eliminated
Q sb x sb P s
s1,2,m
(14)
(15)
x mm P m x 1 1 P 1
x b P L x b1
that is
x b1 P b x b2 P l 0
(16)
(17)
Pb
x b1 1 x b2
P l T P l
I
where
M I T T ,
T
(19)
K l T K L T
T
q a j q a1 j ,q a2 j ,q ai j ,q an j T
is the boundary displacement of substructure a;
q b j q b1 j ,q b2 j ,q bi j , ,q bn j T
q m q m1 ,q m2 , ,q mi ,q mn T
F a j F a1 j ,F a2 j ,F a1 j , ,F an j T
is the boundary force of substructure a;
F b j F b1 j ,F b2 j , ,F bi j ,F bn j T
is the boundary force of substructure b;
F m F m1 ,F m2 , ,F mi , ,F mn T
P b x b1 1 x b2 P l ,
P L P
I
Q e q a j T , q m T , q b j T T
F e F a j T , F m T , F b j T T
where P b and P l respectively represent the matrix of unindependent normal modal coordinates and the matrix of independent normal coordinates among P 1 , P 2 .... P m ; and x b1 is a
square matrix.
The following equation is given from Eq. 17
so
(21)
where
e K e Q e F e
and in matrix form: M e Q
(20)
x 11 P 1 x 22 P 2 ; x 22 P 2 x 33 P 3
i
K ai q ai j q m
F ai j
i
i
i
i
m l q m K ai q a j q m K bi q bi j q m
0
i
i
i
K bi q b j q m F b j
i K i Q i F i ia,b
M i Q
(22)
Q a la P la Q b lb P lb
(23)
Q T1 P
where
Q Q Ta
T
P P la
, Q Te
, Q Te
(24)
, Q Tb T ,
T T
, P lb
,
Tl
la
lb
Q j a j la P la
Q j b j lb P lb
(25)
Q e Q j Ta , Q j Tb T
(26)
P T2q
T
T T
where q P la
, P lb
; T2
jla
jlb
(27)
F T T F * .
For free torsional vibration, the outer vector F 0 obviously.
To solve Eq. 30 and perform transformation of vibration modes
according to 28, the natural frequency and modal shapes of the
total system can be obtained.
Q T 1 T 2 q T q
(28)
(29)
where
M * diag M a , M e , M b ,
K * diag K a , K e , K b ,
and
F * F Ta , F Te , F Tb T
M i , K i , F i ia,b can be obtained from equation 22 .
(30)
where
T T M * T , K
T T K * T
M
and
320 Vol. 125, JULY 2003
shaft (n i 2185 rpm) and outer shaft (n 0 2085 rpm) so that the
fundamental excitation frequency of torsional vibration can be
calculated as Zhao 21:
f 0 mZ n i n 0 /6025 21852085 /60711.67 Hz
In the past, the inner shaft and outer shaft of cam-type engine
were treated separately, and only one of the two natural frequencies of inner shaft and outer shaft was considered as the natural
frequency of whole system so that the results of calculation are
not realistic 22. Owing to the existence of interaction, the inner
shaft and outer shaft should be considered as a whole system in
vibration analysis. The whole system is modeled as two first-class
substructures connected by a flexible substructure. The combination of cylinder-body, outer shaft, front propeller and enginedriven auxiliary machinery is defined as first-class substructure a.
The combination of cam-disc, inner shaft and rear propeller is
defined as first-class substructure b. The piston assembles to link
inner shaft and outer shaft are treated as flexible substructure s in
order to treat the uncoordinated angular displacements. First-class
substructure a is further divided into second-class substructures,
i.e., the combination of cylinder-body, outer shaft and front propeller is defined as second-class substructure c, cooling water
pump as second-class substructure d, fuel pump as second-class
substructure e and generator as second-class substructure f, as
shown in Fig. 5.
Experiment
Substructure c
Substructure a
Substructure b
whole system
404.825
370.310
400.180
198.010
5431.014
1220.826
3577.616
371.843
6296.115
6124.352
5139.323
453.629
10298.600
6675.603
5981.029
1220.560
12164.090
10898.827
6835.063
3576.992
14839.410
12194.780
9408.980
5137.702
Table 2 Natural frequency and error with selected different number of modes retained Hz
Retained No. of modes
Total number 67
Retained number 55
Error %
Retained number 45
Error %
Retained number 35
Error %
Retained number 25
Error %
Retained number 15
Error %
198.010
198.014
0.002
198.101
0.046
198.175
0.083
198.463
0.229
199.016
0.508
371.843
371.848
0.001
372.222
0.102
372.410
0.152
373.589
0.470
374.216
0.638
453.629
453.645
0.003
454.114
0.107
454.983
0.298
455.978
0.518
456.814
0.702
1220.560
1220.748
0.015
1224.303
0.306
1228.936
0.686
1234.871
1.172
1246.213
2.10
3576.992
3577.998
0.028
3583.242
0.175
3588.367
0.318
3597.832
0.583
3604.314
0.764
5137.702
5138.896
0.023
5142.056
0.085
5150.724
0.253
5160.411
0.442
5166.865
0.568
shafts run in opposite directions. Results for the tests are shown in
Fig. 7. It is seen that the component of 34.75 Hz is predominant in
Fig. 7, which is corresponding to the rotating speed of engine
outer shaft where signal gear is fixed on. Many frequencies corresponding to the higher-frequency peaks are the integral times of
34.75 Hz. This is caused by the excitation of unbalanced moment
on outer shaft. Another predominant peak is at about 710 Hz,
which is in accordance with the fundamental excitation frequency
of torsional vibration mentioned above. It is caused by the pulsating output moment of torsion pulsation frequency is f
711.67 Hz). The reason for the phenomenon is that each cylinders work of cam-type engine is uneven and that the moment of
torsion amplitude acted on cam-disc is different.
Calculation
Experiment
Error %
198.010
195.3
1.39
371.843
369.6
0.61
Conclusion
References
1 Den Hartog, J. P., and Li, J. P., 1964, Forced Torsional Vibration with Damping: An Extension of Holzers Method, ASME J. Appl. Mech., 31, pp. 276
280.
2 Pestel, E. C., and Leckie, F. A., 1963, Matrix Method in Elasto Mechanics,
McGraw-Hill, New York.
3 Sankar, S., 1979, On The Torsional Vibration of Branches System Using
Extended Transfer Matrix Method, ASME J. Eng. Ind., Series B, 101, pp.
546 553.
4 Dawson, B., and Davies, M., 1974, An Improved Transfer Matrix Procedure, Int. J. Numer. Methods Eng., 8, pp. 111117.
5 Huang, Y. M., and Horng, C. D., 1999, Analysis of Torsional Vibration Sys-
6
7
8
9
10
11
12
13
14
tems by the Extended Transfer Matrix Method, ASME J. Vibr. Acoust., 121,
pp. 250255.
Firoozian, R., and Stanway, R., 1989, Design and Application of A Finite
Element Package for Modelling Turbomachinery Vibrations, J. Sound Vib.,
134, pp. 115137.
Li, H. Z., 1991, Crankshaft Torsional Vibration Calculation by Finite Element Method, Journal of Internal Combustion Engines, 9, pp. 157162 in
Chinese.
Hurty, W. C., 1960, Vibration of Structural System by Component Mode
Synthesis, J. Eng. Mech. Div., ASCE, 86, pp. 51 69.
Hurty, W. C., 1965, Dynamic Analysis of Structural System Using Component Modes, AIAA J., 3, pp. 678 685.
Craig, Jr., R. R., and Bampton, M. C. C., 1968, Coupling of Substructures for
Dynamic Analysis, AIAA J., 6, pp. 13131319.
Hou, S. N., 1969, Review of Modal Synthesis Techniques and A New Approach, The Shock and Vibration Bulletin, 40, pp. 2539.
Goldman, R. L., 1969, Vibration Analysis by Dynamic Partitioning, AIAA
J., 7, pp. 11521154.
Dowell, E. H., 1972, Free Vibration of an Arbitrary Structure in Terms of
Component Modes, ASME J. Appl. Mech., 39, pp. 727732.
Zhang, H. T., 1990, Free-Interface Mode Synthesis Method for Vibration
15
16
17
18
19
20
21
22
Fusheng Lin
Guang Meng
This paper shows how the dynamics of a rotor in a maneuvering aircraft changes according to the operation of the aircraft. The mathematical model of an unbalanced rotor
system located in the maneuvering aircraft is derived. The dynamic characteristics of the
rotor running at a constant angular speed or a constant acceleration are studied under
the assumptions that the aircraft maneuvers only in a vertical plane and that the pitching
angle and the flight path inclination of the aircraft are equal. The effects of gravity and
unbalance parameter are considered. The results show that the unbalanced response of a
rotor in an aircraft is obviously influenced by the aircrafts flying status.
DOI: 10.1115/1.1576422
e-mail: gmeng@mail.sjtu.edu.cn
State Key Lab of Vibration,
Shock & Noise,
Shanghai Jiao Tong University,
Shanghai 200030, P. R. C.
Introduction
Equations of Motion
In Fig. 1, OXYZ is a space-fixed and stationary coordinate system, and O 1 X 2 Y 2 Z 2 is an aircraft-carried coordinate system parallel to OXYZ. It is supposed that the pitching angle and the flight
path inclination of the aircraft are equal and defined as , i.e., the
axis of the rotor shaft O 1 Z 1 is in the same direction as the longitudinal axis of the fuselage. It is also assumed that O 1 X 1 Z 1 is in
the same vertical plane with OXZ and O 1 X 2 Z 2 , and the aircraft
moves only in the vertical plane. A Jeffcott rotor model is used.
The displacements of the disk center are all measured in a bodyfixed and rotating O 1 reference frame, where and are in
the principal directions of the cross-section of the shaft and
coincides with the Y 1 axis, as shown in Fig. 2. The flexibility of
the rotor is r. The temporary position of the disk center is O 3 . The
position of the mass center of the disk is C. The angle between the
axes O 1 and O 1 Y 1 is . The coordinate O 1 rotates with
respect to O 1 (O 1 Z 1 ) with an angular speed . It is noted that e 0
is a reference eccentricity, m is the equivalent mass of the heavy
disk mounted at the mid-span of a massless elastic shaft, e is the
eccentricity of the disk, is the angle between the orientation of
the eccentricity and the 1 axis, k is the stiffness coefficient of the
shaft, and c is the external viscous damping coefficient of the shaft
and disk.
The dimensionless differential equations of motion of the rotor
system can be written as follows:
1
2
2
1
1
d 2 2 2 1cos 2 d 2 d
2 sin 2 d U 2 sin
2
Ge
2
2
2 cos U sin cos
x cos z d sin x d cos z d sin cos 0
d
2
1
1
1
2
d d U sin d 2 2 2 1cos 2 d 2 d
2 sin 2 d U 2 cos
2
Ge
2
d d U cos
where,
d /e o , d /e o , c k/m, c/2m c , / c ,
G e g/ 2c e o , Ue/e o , x d x/e o , z d z/e o . is the dimensionless viscous damping coefficient, G e is the gravity parameter, U is
the unbalance parameter, and denotes differentiation with respect
to dimensionless time .
If the movement of the aircraft is given, i.e., x( ) and z( ) are
given, then the angle can be calculated by
tg
dx dx dz
dz dt dt
arctg
(1)
x x x d
,
z z z d
x d
z d
Numerical Analysis
Fig. 4 Rotor responses when the aircraft flies in a sine curve 0.8, U 0.5, G e 0.5, z d 0, 0.00002 a Influence
of z d 100 b Influence of z d 50000
and are unaltered. The response when the aircraft moves only in
the horizontal direction and 0 is taken as the basis for comparison. The aircraft first flies in a horizontal direction at a constant velocity. When the free vibration vanishes away and the
steady state amplitude is reached, data of 50 revolutions are taken.
Afterwards, the aircraft maneuvers for several periods. The corresponding period in dimensionless time is 12002 .
If the aircraft maneuvers periodically, its velocity and acceleration will also vary periodically. The amplitude response of the
disk center changes periodically except at the beginning of aircraft
action. In Fig. 4a the amplitude decreases suddenly at the very
beginning of the aircraft maneuver. The vibration response is periodic when the second period of aircraft movement begins, and
the period is the same as that of sine curve (5002 ). For most
part of each period almost 80% the amplitude is less than that
for 0, but for some parts the amplitude is larger than that for
0. The amplitude fluctuates acutely when the aircraft starts to
maneuver. Then the fluctuation decreases. The peak amplitude
value for 100000 is almost 50% greater than that for 0,
and is almost 26% greater than that for 50000 (G e 0.5, U
0.5). In Fig. 4b, larger values of z d correspond to smaller
aircraft maneuvering periods, relatively smaller periods of vibration response and larger peak values. This indicates that quick
maneuvering of the aircraft will cause the vibration response of
the rotor system to change dramatically, although the maneuver
range in the vertical plane is unchanged.
Fig. 5 Influence of the aircraft acceleration in horizontal and vertical directions on rotor accelerating response 0
0.8, x d 0, 10, 50, x d z d 100, G e 1, U 1.0 a z d 0 z d 0 b z d 0 z d 100
Fig. 6 Rotor accelerating response when the aircraft flies in a sine curve 0 0.8, U 0.5, G e 1.5, z d 0 a Influence of z d 100, 0.00002 b Influence of z d 100, z d 0.002
Conclusions
Acknowledgment
The support from China 863 Project No. 2002AA412410
and The Doctor Subjects Research Founding of China University
No. 20020248053 is gratefully acknowledged.
References
1 Lee, A. C., Kang, Y., Tsai, K. L., and Hsiao, K. M., 1992, Transient Analysis
of an Asymmetric Rotor-Bearing System during Acceleration, ASME J. Ind.,
1144, pp. 465 475.
2 Ganesan, R., and Sankar, T. S., 1993, Resonant Oscillations and Stability of
Asymmetric Rotors, Proc. of the 14th Biennial ASME Conference on Mechanical Vibration and Noise, ASME DE, 56, pp. 1922.
3 Spence, A. M., and Cele, R., 1995, Coupled Rotor Fuselage Dynamics and
Aero-Elasticity in Turning Flight, J. Am. Helicopter Soc., 401, pp. 4758.
4 Cao, Y., 1999, Modelling the Unsteady Aerodynamic Forces of a Maneuvering Rotor, Aircraft Engineering and Aerospace Technol., 715, pp. 444 450.
5 Bagai, A., Leishman, J. G., and Park, J., 1999, Aerodynamic Analysis of a
Helicopter in Steady Maneuvering Flight Using a Free-Vortex Rotor Wake
Model, J. Am. Helicopter Soc., 442, pp. 109120.
6 Park, J. S., and Leishman, J. G., 1999, Investigation of Unsteady Aerodynamics on Rotor Wake Effects in Maneuvering Flight, Annual Forum
Proceedings-American Helicopter Society, 1, pp. 467 480.
7 Krothapalli, K. R., Prasad, J. V. R., and Peters, D. A., 2001, Helicopter Rotor
Dynamic Inflow Modelling for Maneuvering Flight, J. Am. Helicopter Soc.,
462, pp. 129139.
Jinhao Qiu
e-mail: qiu@ifs.tohoku.ac.jp
Junji Tani
Taekyu Kwon
Institute of Fluid Science, Tohoku University,
Katahira 2-1-1, Aoba-ku, Sendai 980-8577,
Japan
Introduction
bearing system using the control force generated by the two active
journal gas bearings. The effectiveness of the active pads was
confirmed by experimental results.
LL, UR and UL. The LR and LL sensors are installed in the lower
plane and the UR and UL sensors are installed in upper plane.
Moreover, the LL and UL sensors measure the displacement of the
shaft in the y direction, and the LR and UR sensors measure the
displacement in the x direction of the right-handed coordinate systems shown in Figs. 1 and 3.
The characteristics of the rotor vibration are measured using the
following procedure. Firstly, a 40 V DC voltage is applied to all
four piezoelectric actuators as a bias voltage in the vibration control of the rotor system. Secondly, the thrust bearing is activated
by supplying pressurized air. Thirdly, the valves of the gas supplies for the driving impeller and the braking impeller are turned
on to start the rotation of the shaft. The rotation speed of the shaft,
determined by the relative pressure of the air supplied to the driving impeller and the braking impeller, is raised to 30,000 rpm by
increasing the flow rate of the air to the driving impeller and then
reduced to 0 rpm again by decreasing the flow rate. The vibration
of the shaft is measured when the speed of the rotor reduces from
30,000 rpm to 0 rpm.
Figure 5 shows the waterfall plot of the vibration characteristics
in the y direction measured by the LL sensor. It can be seen that
both the synchronous vibration with the frequency equal to the
rotation speed, and the asychronous vibration self-excited vibration, are induced in the rotor system. The sychronous vibration
reaches the maximum amplitude at 11,600 rpm, which equals the
natural frequency of the first rigid mode of the rotor. The frequency of the self-excited vibration is about 57 Hz, regarless of
the rotational speed, but its amplitude also reaches the maximum
value at around 11,600 rpm. The fluid-related instability in both
the journal bearings and the thrust bearings can induce the selfexcited vibration. A dynamic coupling between the journal bearings and the thrust bearings may also occur in the rotor-bearing
system. The objective of this study is to control the self-excited
vibration of the rotor-bearing system.
In order to control the vibration of the shaft, piezoelectric actuators were embedded in two of the three pivots in each bearing,
as shown in Fig. 3. The piezoelectric actuators are 55
10 mm in size and can generate 6.1 m of displacement at 200
V input voltage. The piezoelectric actuators, which have a wide
frequency bandwidth and can generate a large force output, are
suitable for this kind of application. The positions of the four
actuators are also labeled as LR, LL, UR and UL. The LL and UL
actuators are embedded in the pivots on the y-z plane and the LR
and UR actuators are embedded in the pivots on the plane at 30
degrees clockwise from the x-y plane.
When the shaft rotates, a pressurized gas film is formed between the shaft and each pad due to the viscosity of the gas. The
Journal of Vibration and Acoustics
U s
Ki
1
K p K d s K p 1T d s
Ys
s
T is
(1)
duced. However, the synchronous vibration remained almost unchanged. The small peak at 50 Hz is due to the electromagnetic
noise of the power supplies.
Figure 10 show the rotor frequency response measured by the
four sensors at 11,600 rpm. K d was set to 0.01, and the proportional gains K p of the UL and UR actuators were set to the opti-
Due to the influence of the pressurized air supplied to the driving impeller, the actuators at LL and LR were less effective in the
control of the self-excited vibration than the other two actuators.
Therefore, the gains of these two actuators were fixed at K d
0.01, K i 0.001 and K p 50, while the gains of the other two
actuators at UL and UR were varied to investigate their influence.
Moreover, the integral gain of the two actuators at UL and UR
was also set to 0.001, since it has little influence on the control
effect. Control experiments were performed for different combinations of K p and K d for the actuators at UL and UR. Figure 9
shows the shaft frequency response measured by the sensors at
UR and LR for a rotation speed of 10,000 rpm with gains K p and
K d for the actuators at UL and UR set to 10 and 0.02, respectively.
The asynchronous vibration about 57 Hz was significantly re-
Outer diameter, r 1
Inner diameter, r 0
Diameter of nozzle position circle, r a
Number of nozzles, n
Diameter of nozzles in the lower bearing, r s1
Diameter of nozzles in the upper bearing, r s2
Unloaded clearance of the lower bearing h n1
Unloaded clearance of the upper bearing h n2
Nondimensional axial displacement h n /(h n1 h n2 )
32 mm
16.4 mm
22.6 mm
18
0.6 mm
0.3 mm
23.4 m
16.6 m
0.22
181.35 mm
0.28 kg
8 mm
15.4 mm
110
20 m
0.65
0.5
l1
l2
l3
l4
l5
l6
l7
l8
23.54 mm
26.46 mm
10.84 mm
14.16 mm
90.25 mm
91.1 mm
38.74 mm
6 mm
Fig. 6 Bias displacement of the shaft versus DC voltage applied to the UR actuator
Fig. 8 Block diagram of the control system
mal values of 150 and 100, respectively. Since the rotor speed
equals the resonant frequency of the rotor-bearing system, the
amplitude of the asynchronous vibration increases significantly at
the uncontrolled state. However, the asynchronous vibration is
almost completely suppressed after control. Despite this, little effect could be observed on the synchronous vibration. The other
components of the asynchronous vibration with frequencies of 25
Hz, 79 Hz and 132 Hz were also suppressed.
The influence of the feedback gain on the control results was
further investigated. Fig. 11 shows the control results when the
proportional gains of the actuators at UL and UR were set to 200,
while the rotor speed and the other gains were kept the same as
those in the former case. The asynchronous vibration at a frequency of 57 Hz was almost completely suppressed. However, a
second asynchronous vibration of 93 Hz was excited.
Fig. 10 Power spectra of the shaft at different sensor positions Rotation speed: 11,600 rpm, PID gain: K p of UL act.150, K p
of UR act.100 a UL sensor position b LL sensor position c UR sensor position d LR sensor position
Conclusions
Acknowledgment
The authors thank Dr. Yanagi, and Mr. Tsugawa of Mayekawa
Mfg. Co. Ltd., MYCOM Advanced Technology Laboratory,
Ibaraki, Japan, for their effort and cooperation in the design and
manufacturing of the rotor-bearing system.
References
1 Ino, N., Machida, A., Tsugawa, K., and Hashimoto, H., 1991, Development
of Externally Pressurized Thrust Bearing for High-Expansion-Ratio Helium
Expander, Trans. Jpn. Soc. Mech. Eng., Ser. C, in Japanese 58550, pp.
18851891.
2 Togo, T., 2000, Investigation Report on the Research of Gas Bearings, Japanese Society of Tribologists.
3 Andres, L. S., 1996, Turbulent Flow, Flexure-Pivot Hybrid Bearings for
Cryogenic Applications, ASME J. Tribol., 118, pp. 190200.
4 Qiu, J., Takagi, T., Tani, J., Machida, A., Tsugawa, K., Yanagi, H., and Ino, N.,
1993, Dynamic Characteristics of a Tilting-pad Bearing System for High
Expansion-Ratio Expander, Adv. Cryog. Eng., 39, pp. 909914.
5 Bently, D. E., and Muszynska, A., 1989, Anti-Swirl Arrangements Prevent
Rotor/Seal Instability, Stress, Reliab. Des. 1112, pp. 156 162.
6 Ihara, K., 1987, Gas Expander for Lower Temperature, Journal of Turbine
Machinery in Japanese, 1511, pp. 57 62.
7 Palazzolo, A. B., Lin, R. R., Alexander, R. M., Kascak, A. F., and Montague,
J., 1989, Piezoelectric Pushers for Active Vibration Control of Rotating Machinery, ASME J. of Vibration, Acoustics, Stress, Reliab. Des. 111, pp. 298
305.
8 Palazzolo, A. B., Jagannathan, S., Kascak, A. F., Montague, G. T., and Kiraly,
L. J., 1993, Hybrid Active Vibration Control of Rotorbearing Systems Using
Piezoelectric Actuators, J. Vibr. Acoust., 135, pp. 111119.
9 Horikawa, O., and Shimokohbe, A., 1990, An Active Air Bearing, JSME,
Int. Journal, 331, pp. 55 60.
10 Muszynska, A., Franklin, W. D., and Bently, D. E., 1988, Rotor Active AntiSwirl Control, Stress, Reliab. Des. 1102, pp. 143150.
11 Sun, L., Krodkiewski, J. M., and Cen, Y., 1998, Self-Tuning Adaptive Control of Forced Vibration in Rotor Systems using an Active Journal Bearing, J.
Sound Vib., 2131, pp. 114.
A flex circuit connects the stationary electronic components in a hard disk drive to the
rotating arm that carries the read/write heads and positions them above data tracks on
the disk. Flex circuits are conventionally formed as a laminate of polyimide substrate,
adhesive, and copper conductors. Deformation of a flex circuit is discussed in the context
of the following stages: the initial unstressed shape, configurations in which stresses set
and relax in response to elevated temperature, equilibrium, and small amplitude vibration. The model involves displacements of the flex circuit in the directions tangent and
normal to the local equilibrium shape, and those motions couple with the arms dynamics.
Nonlinearity associated with finite curvature, partial elastic springback, and the arms
geometry and inertia properties are incorporated within the vibration model to predict
system-level natural frequencies, mode shapes, and coupling factors between the circuit
and the arm. Laboratory measurements using noncontact laser interferometry validate the
model with respect to the circuits shape, stiffness, restoring moment, and natural frequencies. The primary degrees of freedom for optimizing flex circuit design are the thicknesses
of the individual layers within the circuit, free length, and the locations and slopes of the
circuits attachment points to the arm and electronics block. The models predictions and
trends developed from a case study in free length are discussed with a view toward
reducing coupling between the circuit and arm in certain vibration modes.
DOI: 10.1115/1.1547661
Introduction
y 0 r sin
(1)
(2)
11.1 mm
31 m
13 m
26 m
114 m
67%
31 mm
2.38 g/m
2.75 GPa
1.03 GPa
115 GPa
0.55
4.78(106 ) Nm2
2.45(104 ) N
Arm
Attachment radius, r
Read/write head radius, R
Attachment angle,
Offset angle,
Angles
Load-unload, LU
Disk OD, OD
Disk ID, ID
Inertia, m
Radius of gyration,
10.6 mm
43.8 mm
95
8
23
32
64
17.2 g
11.9 mm
Electronics Block
24, 11 mm
90
Coordinates, (x L ,y L )
Tangency angle, L
N Tk,
k N/EIk 2
(3)
in which the constitutive relation M EI(kk ) has been embedded. The circuits bending stiffness is given by
(2)
EI
1
1
1
1
E h 3 b E a h 3c 1 b2E a
bh 3 bh h
12 c c
12
12 a 4 a c
h a 2 2E p
1
1
bh 3p bh p h c h p 2h a 2
12
4
(4)
where values for the elastic constants of the conductor E c , polyimide E p , and adhesive E a are listed in Table 1. While Eq. 4 is
specific to the cross-sectional construction of Fig. 3, the treatment
can be adapted for other geometries.
On the basis of measured layer thicknesses and published elastic constants 2 in Table 1, the circuits composite bending stiffness is EI4.78106 Nm2 . The conductor layer contributes
26% to the stiffness, the polyimide layer 71%, and the adhesive
layer only 3%. This value of EI was validated by both static bending and natural frequency measurements conducted with a circuit
segment that was embedded as a cantilever in an epoxy casting. In
the static test, the segment was mounted on a micrometer translation stage, and under specified displacement, the force applied to
its tip was measured. The signal from a planar beam sensor Futek FR-1020 was conditioned and amplified to provide a calibrated and linear force response for loads up to 295 mN. The
JULY 2003, Vol. 125 337
Fig. 4 Predicted
and measured shapes of the flex circuit
in equilibrium with attachment to the arm and electronics
block, and in its natural state. The shapes extracted from photographs of the circuit in the two states are denoted by the
and data points, respectively.
along the circuits length from the arm. While the average strain
over the cross-section is only T/EA1.3106 with
stiffness recorded on this basis was 4.60106 Nm2 . In the second validation test, the value 4.91106 Nm2 was determined
by measuring the segments two lowest natural frequencies and
matching them to the values expected for a cantilever. In the calculations described below, the stiffness value determined from Eq.
4 was used.
The circuits equilibrium curvature is determined by integrating
Eqs. 3, and its shape is found subsequently from the kinematic
relations
x cos ,
y sin ,
k s
(5)
EAE c h c b 2E a h a bE a h c b 1 2E p h p b
(6)
M o T 0 r sin N 0 r cos M 0
(7)
which increases in Fig. 7 from zero at equilibrium to the maximum value 0.35 Nmm at the disks ID. Multiple measurements
of the torque were made for one disk drive at four different slew
angles, and those results are also shown in Fig. 7. The indicated
variation of M o is representative of such measurements and captures hysteresis in the circuit, and friction in the pivot bearing and
Fig. 6 Variation of the flex circuits static shape for arm positions which range between the disks outer and inner diameters. For each arm position, the locations of the circuits endpoints are denoted by .
Fig. 7 Predicted
and measured dependencies of the static restoring moment for arm positions between the disks outer and inner diameters
(8)
(9)
TT T 1
*
kk k 1
(10)
*
about their equilibrium values, denoted by () . Here 1 is a
*
dimensionless scaling parameter used in the linearization, and the
first-order corrections are written 3,4
N 1 EI v ,ss k u ,s ,s
(11)
*
T 1 EA u ,s k v
(12)
*
k 1 v ,s k u ,s
(13)
*
in terms of the circuits tangential and normal displacements. Here
the comma-subscript notation signifies partial differentiation. The
equations of motion become
Au ,tt T 1,s N k 1 k N 1 0
*
*
A v ,tt N 1,s T k 1 k T 1 0
*
*
where A is the circuits mass-per-unit-length.
Vibration of the flex circuit and arm couple through
(14)
(15)
Length ratio
Length mm
Equilibrium
angle deg
Stiffness
mNmm/deg
70%
80%
90%
100%
110%
120%
130%
21.7
24.8
27.9
31.0
34.1
37.2
40.3
41.3
33.0
28.4
26.4
29.3
36.9
44.2
63.6
40.0
20.2
10.2
5.6
5.0
5.8
(17)
v 0 r cos 1 ,
(18)
k v dse
sr sin 1
*
(19)
R 1
max u 2 v 2
(20)
eter measured changes in the two light path lengths through the
interference fringes generated by superposition of coherent beams
that reflected from i a stationary reference surface and ii the
moving flex circuit. The target beam was directed onto the flex
circuit by a right-angle prism located on the concave side of the
flex circuit. To ensure that sufficient light was returned by the flex
circuit into the optical head, a small patch of retroreflective tape
was attached to the circuit at the measurement point. Particles
within the retroreflective medium ensured that a portion of the
incident light was returned into the source optical fiber regardless
of the flex circuits potentially large displacement or slope. As
the arm or circuit was impacted, the displacement or velocity
signal was captured on an digital oscilloscope HP 54600A, and
its frequency content was characterized by using an dynamic signal analyzer HP 35665A. Peaks in the autocorrelation record
provided the natural frequencies. With this technique, vibration
measurements were readily made with a strong signal-to-noise
ratio, and with displacement resolution and bandwidth exceeding
the tests requirement.
In the spectrum for one disk drive shown in Fig. 11, the natural
frequencies of modes two, three, and four were measured at 356
Hz, 844 Hz, and 1.23 kHz. Several of the flex circuits torsion
modes were also present in the illustrated 1.6 kHz frequency
range, but their content in Fig. 11 was suppressed by judicious
placement of the impact and measurement points relative to the
torsion modes nodes. Despite its layered construction and attachments to the arm and electronics block, the flex circuit presents a
damping ratio of only 1.1% in the second mode as indicated by
the time record for ring-down in Fig. 12. By contrast, the mechanism is highly damped in the fundamental sway mode, with a
measured frequency at 3.08 Hz. Figure 13 depicts the arm and flex
circuits transient response in that mode, and just over one cycle
of motion occurred following impact.
Figures 14 and 15 show trends for the natural frequencies and
coupling ratios which are predicted in a parameter study of free
length L. For each length, the tension, shear force, and curvature
were determined on the basis of the equilibrium configurations
shown in Fig. 8. The natural frequencies for modes two through
six decrease monotonically in Fig. 14 as L is examined over a
range 30% below, and 30% above, the nominal value. Also for
these modes, Fig. 15 depicts the behavior of the displacement
ratio . The fundamental mode is dominated by the arms sway
Fig. 14 Dependence of the natural frequencies in modes two through six on the flex
circuits free length
Fig. 15 Dependence of the displacement ratios in modes two through six on the
flex circuits free length
with the flex circuit responding nearly statically, and for that
mode increases gradually within the range 357 412%. In the
even modes two, four, and six, is relatively insensitive to
design changes in L. On the other hand, has a zero crossing for
the odd modes three and five at nearly the same free length (L
26 mm) which corresponds to some 83% of the baseline value.
At that design point, the natural frequencies have increased relative to their values at L31 mm in Fig. 14, and the static stiffness
has likewise grown in Table 2. However, the modal displacement
ratios can be reduced significantly or precisely driven to zero in
modes three and five. That trade-off may be desirable in certain
applications. In short, the equilibrium and vibration model can be
used to advantage for optimizing flex circuit designs with respect
to their vibration and load transmission performance.
Acknowledgment
This work was supported by a grant from IBM Corporation.
The author appreciates the assistance of Matthew Brake in conducting the static stiffness and natural frequency measurements.
Kapton is a registered trademark of E.I. du Pont de Nemours
and Company.
Summary
References
1 Love, A. E. H., 1944, A Treatise on the Mathematical Theory of Elasticity,
Fourth Edition, Dover Publications, New York.
2 Anonymous, 1996, General Specifications, Bulletin GS-96-7, DuPont Films,
E.I. du Pont de Nemours and Company.
3 Perkins, N. C., 1990, Planar Vibration of an Elastica Arch: Theory and Experiment, ASME J. Vibr. Acoust., 112, pp. 374 379.
4 Brush, D. O., and Almroth, B. O., 1975, Buckling of Bars, Plates, and Shells,
McGraw-Hill, New York.
5 Wickert, J. A., 1992, Non-linear Vibration of a Traveling Tensioned Beam,
Int. J. Non-Linear Mech., 27, pp. 503517.
M. Saigo
Research Manager,
Mechanical Engineering Laboratory,
Agency of Industrial Science and Technology,
1-2 Namiki, Tsukuba, Ibaraki 305-8564, Japan
e-mail: saigo@mel.go.jp
K. Tani
Professor,
Gifu University,
1-1 Yanagito, Gifu 501-1193, Japan
e-mail: tani@info.gifu-u.ac.jp
H. Usui
Nippon Steel Corporation,
2-6-3 Otemati, Chiyodaku, Tokyo 100-8071,
Japan
e-mail: usui@tekkai.nsc.co.jp
Introduction
Recently, traveling-wave control has been studied as an alternative to mode-based vibration control by several researchers.
Wave control has several advantages over mode-based vibration
control; it has no control and observation spillovers that may occur in the vibration control when there are some imperfections in
system modeling or inaccuracy in sensor locations; it has a better
control performance than vibration control at low frequencies; and
it is basically a local control method to which we have paid attention in our studies. The last feature means that we can suppress the
vibration of a system using no more than the information about
the dynamic states of the element nearest the actuator. This is
quite advantageous to the system whose parameters are changeable during control operations.
The studies of traveling-wave control include Vaughan 1, Von
Flotow 2,3, Millar 4, Mace 5, Fujii 6, Tanaka 7 and Utsumi 8. These studies of traveling-wave control of elastic beams
or strings have used the theoretical solution expressed in the form
of traveling wave and derived the nonreflecting condition of
waves at the control point. On the other hand, OConnor 9
treated a mass-and-spring system and presented a method of the
wave-absorption in a discrete vibration system. In these studies,
few attempts have been made to apply the wave control strategy
to practical vibrating systems to show the above-mentioned advantages of traveling-wave control.
In the previous paper 10, we presented a new practical wave
control strategy that is easy to build in a control computer with
on-line calculation of the imaginary wave-propagating system. We
applied the presented method to the vibration suppression of a
multiple-pendulum system and showed the effectiveness of the
method experimentally. There, vibration control by controlling the
support movement was addressed and no support movement for
traveling toward a given target position was considered.
In the present paper, the wave control method presented in the
previous paper is expanded to consider the combination of traveling control and vibration control. In this case, the control system
has to achieve a system displacement to a desired target position
Contributed by the Technical Committee on Vibration and Sound for publication
in the JOURNAL OF VIBRATION AND ACOUSTICS. Manuscript received May 2000;
Revised Jan. 2002. Associate Editor: R. L. Clark.
Equation of Motion
2.1 Multiple-Pendulum System. Figure 1 shows a traveling multiple rigid-pendulum system and a traveling wire-and-load
system. The equations of motion of a traveling multiple-pendulum
system of n degrees of freedom DOF are obtained using the
Lagranges equation of motion. The kinetic energy T k and the
potential energy U k of the k-th pendulum are expressed as
T k m k y 0
jk1
l j j
/2I k 2k /2m k y 0
jk1
U k m k g h k 1 2k /2
jk1
l j 1 2j /2
l j j k h k
k1
I k l 2k
i1
h k m k
k1
m i k l k
ik1
j1
j1
lj
i1
k1
m i h j m j j l k
i1
mi
k1
l i i g l k
i1
m i h k m k k 0
(1)
q 1 p 1 J 1 / r 1 m 1 1 q 2 p 1 2 g m 1 r 2 m 2 p 1 1 2
0
q k1 p k1 k1 q k p k J k l k 1r k1 m k1 p k1 k
k2
q k1 p k k1 g
k1
j1
j1
m j r k1 m k1 p k1 k1
r k1 m k1 p k k1 0
m j r k m k p k p k1 k
j1
mj
k1,n
(2)
q n1 p n1 n1 J n l n 1r n1 m n1 p n1 n
n2
gp n1
j1
Control Strategy
n1
m j r n1 m n1 n1
j1
mj
r n m n n y 0
r k h k /l k ,
J k I k /l k ,
p k 1/ 1r k m k r k1 m k1 ,
q k J k h k m k
From the above equation, we see the term of the support movement appears explicitly only in the equation of the uppermost
pendulum.
2.2 Wire-and-Load System. The wire-and-load system
shown in Fig. 1 has a small rigid pendulum between the support
and the wire, and there is a load at the bottom of the wire. The
wire length is fixed. Assuming the equation of the wire is expressed by that of a dangling string and applying the finite difference method to the equation of motion, we obtain a system of
equations similar to that of a multiple simple-pendulum system
see Appendix. In the following numerical simulation, we will
treat the wire-and-load system as a non-homogenous multiplependulum system consisting of the uppermost rigid pendulum and
a large-DOF series of simple pendulums, among them the lowest
having a mass equal to that of the load.
344 Vol. 125, JULY 2003
Fig. 2 N-DOF imaginary system for traveling pendulum system y c : traveling command, x 0 : vibration control a Nontraveling Imaginary System NTIS, b Traveling Imaginary
System TIS
geous first to move the system near the destination with less vibration and then to control the final position accurately. It is an
easy job for the control computer, given the present and target
positions, to cancel the error of the final position.
When the real pendulum system is connected to the imaginary
system, the acceleration of the lowest end of the imaginary system
is represented as
x 0
i 2 /lh g 0 h/l g 1 0
1h/l
(3)
Experiment
pendulum are 60 mm, 40 mm and 15 mm, respectively. The diameter of the wire is 1 mm. The length of the wire and the weight
of the load can be changed.
The computation of the control is conducted by a DSP
TMS320C30 for the rigid-pendulum system and a personal computer with 200 MHz CPU for the wire-and-load system. The sampling period of A/D conversion is 0.1 ms for the rigid-pendulum
system and 2 ms for the wire-and-load system. A 10-DOF system
of simple pendulums has been used as the imaginary system.
4.1.2 Experimental Results. The following system movement pattern is used as the traveling command in the experiments;
the acceleration y c is 4.26 m/s2 for the time period between 0 s
and 0.0352 s and 0.01883 m/s2 between 0.0352 s and 8 s. Using
this acceleration pattern the pendulum system should travel the
distance of 0.60 m in 8 seconds.
a The case of the multiple-pendulum system
Figures 5 show the effects of the different initializing methods,
the PI and VI methods, for the 3-DOF traveling rigid-pendulum
system with 3. In these figures, the curve rising to the righthand side is the position of the support and the vibration waveform is the angle of the uppermost pendulum. From these figures,
we can confirm that our method is effective for a traveling pendulum system as well as for a nontraveling system. Figures 6
show the control performance of the PI method and the VI
method on the 1-DOF pendulum system three pendulums are
connected rigidly. Both initializing methods have excellent vibration suppressing effects. Similarly, the control performance on the
non-homogeneous 2-DOF pendulum system the lowest and the
middle pendulums are connected rigidly is well confirmed figures are not shown. The control program for the 2-DOF system is
the same as that for the 3-DOF system, for the uppermost pendulum and the total mass of the system are the same.
b The case of the wire-and-load system
Figures 7 show the effects of the different initializing methods,
the PI and VI methods, for the traveling wire-and-load system
with the wire length l0.5 m, the load weight w12.3 N and
3. In these figures, the curve rising to the right-hand side is the
position of the support and the vibration waveform is the angle of
the uppermost pendulum. From these figures we can see that the
VI method has quite an excellent damping performance Fig.
7c. The PI method is accurate in positioning the pendulum system at the traveling destination, but the vibration control performance is not so good. The performance of the VI method is not so
good figure is not shown. The vibration waveform shown in Fig.
7c resembles well that of the 1-DOF rigid pendulum shown in
Fig. 6b. This means the dynamic characteristics of the experimental wire-and-load system is similar to that of the 1-DOF rigid346 Vol. 125, JULY 2003
pendulum system and the vibration of the wire is practically negligible. Figures 8 show the effect of the wire length l and the load
weight w on the control performance of the VI method when
1. We can see the control performance is independent of the
change of the wire length and the load weight. Besides, the control performances in Fig. 8 for 1 are better than that in Fig.
7c for 3, which confirms that the quick wave propagation in
the imaginary system brings about a better performance in vibration suppression while the system is stable.
Figures 9 show the results of the operation in which final positioning accuracy is concerned when the VI method is used for
vibration control. As stated earlier, this method can produce a
positioning error, and some measures should be taken when accurate positioning at the target is needed. Figure 9a is an overrun
case where the vibration control is switched to the VI method and
the traveling command is suppressed the instant the support
reaches the exact target position. Figure 9b is a case where the
positioning correction to the exact target position is carried out
using the VI method after the traveling command is over. Either
Fig. 8 Experimental results of wire-and-load system for different wire length and load weight with VI method 1,10. a
l 0.3 m, w 12.3 N b l 0.9 m, w 31.9 N
Fig. 10 Experimental result of nontraveling wire-and-load system for large amplitude with PI and VI methods l 0.3 m, w
12.3 N, 10, 10.
4.2.1 Experimental Apparatus. Figure 12 shows the experimental model of a crane, which has a load suspended by a wire
and a pulley. One end of the wire is fixed to the motor shaft for
winding and the other end is fixed to a small rigid pendulum that
is attached to the nut of the ball-screw with free rotation. The
distance between the axis of the rigid pendulum and the wire
return on the returning pulley is equal to the diameter of the pulley. Then, the angle of the rigid pendulum is practically equal to
the swing angle of the load independent of its height when the
small vibrations of the wire can be ignored. This means the sta-
tionary direction of the pendulum is always vertical and the vibration suppression strategy for the wire-and-load system is applied
directly by monitoring one half of the load suspension system.
Our experiment has shown that it is possible to neglect the small
wire vibration in the wire-and-load system as well as in most
practical crane systems. The velocity pattern used of winding-up
and rewinding-down is 0.1 m/s and 0.1 m/s, respectively. The
load moves between the vertical positions of 0.9 m and 0.3 m
during the time period between 0 s and 6 s.
4.2.1 Experimental Results. Figures 13 show the experimental results for the case of winding up the load; a in Fig. 13 is the
case where no vibration control is used, b is the case with wave
control in the PI method, and c is the case with wave control in
the VI method. Figures 14 show the cases of rewinding down the
load with controls similar to those in Fig. 13. The system movement pattern is the same as in the case for the wire-and-load
system. In Fig. 13 we can see the amplitude of the vibration in
winding-up without control becomes larger as the wire length becomes shorter, due to the instability in winding-up of a suspended
load. The contrast in these figures demonstrates the effectiveness
of the stabilization using the wave-absorbing control. In addition,
the wave control method presented has shown an excellent control
performance regardless of the wire length. Figures 13b and
14b with the PI method for a large value of show relatively
good results in the final state of the load, that is, accurate final
position and small vibration. Thus, we can use the PI method to
position the system accurately at the target position if its vibration
suppression performance is acceptable. Even if the VI method is
Transactions of the ASME
ratio of the mass of an imaginary system pendulum to that of the uppermost rigid pendulum
ratio of the length of an imaginary system pendulum to that of the uppermost rigid pendulum
angle of the uppermost rigid pendulum
angle of the lowest imaginary system pendulum
k angle of the k-th rigid pendulum numbered from
the free end
h k distance between the center of gravity of k-th
rigid pendulum and the axis of k-th connecting
pin
I k moment of inertia of k-th rigid pendulum about
the axis of k-th connecting pin
m k mass of k-th rigid pendulum
l k distance between the axes of the k-th and
(k1)-th connecting pins of the rigid pendulum
k angle of the k-th imaginary system pendulum
numbered from the lowest end
n
x 0 k1
l k k n: degrees of freedom of imaginary
system
y 0 horizontal displacement of the support imaginary
or real
y c traveling command
deflection of the wire
i deflection of the i-th finite element of the wire
z coordinates of the wire measured from the lower
free end
Appendix
We assume the equation of motion of the wire is expressed by
that of a dangling string. By balancing the horizontal component
of forces on an infinitely small element z(zdz), the following
equation is obtained as
Fig. 14 Experimental results of crane system for lowering load
total weight17.6 N. a No control, b Control with PI method
10, 10, c Control with VI method 1, 10
used, the final position errors are not very significant. Figures 13
and 14 have shown our wave absorbing method is useful for the
actual crane system.
Concluding Remarks
In this paper we propose a wave control method using a nontraveling imaginary multiple-pendulum system applied to vibration control of the traveling suspended system. We have shown
that the method presented, using little information about the suspended system states, is quite effective for a traveling suspended
system whose dynamics are changeable during operation. The initializing methods with and without shifting the support of the
imaginary system have their respective merits for the application.
The former realizes accurate final positioning while the latter realizes excellent vibration suppression. Combination of methods
can produce excellent overall performances. This control method
can readily be applied to real crane systems that have a moving
pulley and parallel wiring.
Nomenclature
g Gravitational acceleration
( ), ( ) first and second order differentiation with respect
* *
to time
i 2 square of radius of gyration on the supporting
point of the uppermost rigid pendulum
l length of the uppermost rigid pendulum
h distance between the supporting point and the
center of gravity of the uppermost rigid pendulum
0 ratio of the mass of the total real system to that of
the uppermost rigid pendulum
Journal of Vibration and Acoustics
M
2 1 2
z
0
z g t2
z2
(A1)
where is the lateral deflection of wire, z is the coordinate measured from the lowest end, is the mass of string per unit length
and M is the mass of load.
Appling the finite difference method to the above equation using the following approximations,
2
z
i1 2 i i1
z
i i1
,
z
z
ziz
we obtain
M
2 i i1 i1 i1 2i1 i i i1
z i1
g i
(A2)
References
1 Vaughan, D. R., 1968, Application of Distributed Parameter Concepts to
Dynamic Analysis and Control of Bending Vibrations, ASME J. Basic Eng.,
90, pp. 157166.
2 Von Flotow, A. H., 1986, Traveling Wave Control for Large Spacecraft Structures, AIAA J., 9, pp. 462 468.
3 Von Flotow, A. H., 1986, Disturbance Propagation in Structural Networks,
J. Sound Vib., 106, pp. 433 450.
4 Miller, D. W., and von Flotow, A. H., 1989, A Travelling Wave Approach to
Power Flow in Structural Networks, J. Sound Vib., 128, pp. 145162.
5 Mace, B. R., 1984, Wave Reflection and Transmission in Beams, J. Sound
Vib., 97, pp. 237246.
6 Fujii, H., and Ohtsuka, T., 1992, Experiment of a Noncollocated Controller
for Wave Cancellation, AIAA J., 153, pp. 9397.
7 Tanaka, N., and Kikushima, Y., 1992, Active Wave Control of a Flexible
Beam, JSME Int. J., Ser. III, 351, pp. 236 244.
8 Utsumi, M., 1999, Analytical Implementation of Wave-Absorbing Control
for Flexible Beams Using Synchronization Condition, ASME J. Vibr. Acoust.,
121, pp. 468 475.
9 OConnor, W., and Lang, D., 1998, Position Control of Flexible Robot Arms
Using Mechanical Waves, ASME J. Dyn. Syst., Meas., Control, 120, pp.
334 339.
10 Saigo, M., Tanaka, N., and Tani, K., 1998, An Approach to Vibration Control
of Multiple-Pendulum System by Wave Absorption, ASME J. Vibr. Acoust.,
121, pp. 524 533.
J. R. Banerjee
Professor of Structural Dynamics,
School of Engineering
and Mathematical Sciences,
City University, Northampton Square, London
EC1V OHB
e-mail: j.r.banerjee@city.ac.uk
Introduction
Theory
a1
F x1
0
F y1
0
M1
F x2
a2
F y2
0
M2
0
a2
d1
d2
d4
d5
d2
d3
d 5
d6
a1
d4
d 5
d1
d 2
d5
d6
d 2
d3
U1
V1
1
U2
V2
2
(1)
(2)
EA
cot ,
L
a 2
EA
cos ec
L
(3)
with
and
352 Vol. 125, JULY 2003
m 2L 2
EA
d 1 W 3 3 S C h C S h /,
d 2 W 2 2 S S h /,
d 3 W 1 S C h C S h /
d 4 W 3 S S h /,
(5)
d 5 W 2 2 C h C /,
d 6 W 1 S h S /
with
W 1
EI
,
L
W 2
C cos ,
S sin ,
EI
L
W 3
EI
L3
m 2L 4
C h cosh ,
(4)
(6)
(7)
EI
S h sinh (8)
and
1C C h
or
FK
(9)
2.2 Dynamic Stiffness Formulation for a Two Degree-ofFreedom Spring-Mass System Attached to a Beam. Figure 2
shows a rigid mass M having a mass moment of inertia I about
its centroidal axis and mounted at the ends nodes of a BernoulliEuler beam by means of two linearly elastic springs of spring
constant K S1 and K S2 , respectively. The distances of the center of
gravity g of the mass from the location of the springs are D 1 and
D 2 respectively, as shown.
The governing differential equations of motion for the two
degree-of-freedom spring mass system shown in Fig. 2, undergoing free vibration in translation and rotation, are
M
2v M
2 M
t2
f k1 f k2 0
(10)
f k1 D 1 f k2 D 2 0
(11)
t2
(12)
f k2 K S2 v 2 v M D 2 M
(13)
M x,t A M e i t ,
Transactions of the ASME
v 1 x,t V 1 e i t ,
v 2 x,t V 2 e i t
(14)
a1
F x1
0
F y1 F k1
0
M1
F x2
a2
F y2 F k2
0
M2
0
and
f k1 x,t F k1 e i t ,
f k2 x,t F k2 e i t
(15)
K S1 V 1 K S2 V 2 0
K S1 D 1 K S2 D 2 V M I 2 K S1 D 21 K S2 D 22 A M
(17)
(21)
F k2 K S2 V 2 V M D 2 A M
(22)
(23)
(24)
e 2 K S1 K S2 I D 1 D 2 M /Z
(25)
e 3 K S2 M I
2
K S1 I D 21 M /Z
(26)
d2
d3
d 5
d6
a1
d 4 e 2
d 5
d 1 e 3
d 2
d5
d6
d 2
d3
(27)
(28)
(19)
(20)
e2 V1
e3 V2
d5
ZM I 4 2 K S1 I D 21 M K S2 I D 22 M
e1
F k1
F k2
e2
d 4 e 2
FK
or
where
K S1 K S2 D 1 D 2 2
a2
(18)
A M K S1 D 1 M 2 K S2 D 1 D 2 V 1
K S2 K S1 D 1 D 2 D 2 M 2 V 2 /Z
0
d2
U1
V1
1
U2
V2
1
(16)
K S1 D 1 V 1 K S2 D 2 V 2 0
0
d 1 e 1
To illustrate the above theory two examples are given. The first
is taken from Ref. 28 in which a beam clamped at both ends
carries a two degree-of-freedom spring-mass system as shown in
Fig. 3. The data used in the analysis are as follows:
EA3.01010 N,
m1179 kg/m,
L 3 1.4 m,
EI6.25108 Nm2 ,
L 1 1.4 m,
M 200 kg,
D 1 D 2 0.6 m
and
L 2 1.2 m,
I 70.833 kgm2 ,
K S1 K S2 1010 N/m
also with those obtained with the two degree-of-freedom springmass system missing. Except for the second natural frequency the
results agreed completely with those of Ref. 28 in which a
Laplace transformation technique with respect to the spatial variable is used, and is expected to give exact results. The author was
unable to pin-point the exact reason for the small discrepancy
less than 0.04% in the second natural frequency, but this may be
attributed to the accuracy used in the input data of Ref. 28. In
Ref. 28 the built-in end conditions were simulated by using very
large linear and rotational springs of the order of 1020) and are
thought unlikely to alter the results. The mode shapes corresponding to these first three natural frequencies are shown in Fig. 4,
which agree very well with the ones illustrated in Fig. 3 of Ref.
28.
Although the present theory is capable of determining the natural frequencies in axial vibration, these were suppressed by disallowing the axial deformation of the beam, so that the results are
directly comparable with those of Ref. 28. The axial natural
frequencies are uncoupled from the bending ones and they are not
of interest as they will remain unaffected by the presence of the
two degree-of-freedom spring-mass system. If the spring connections are oblique or the two degree-of-freedom system is connected to a framework a coupling would occur in which case the
axial motion of the beam would have to be included in the analysis. The results of Table 1 show that the presence of the two
degree-of-freedom spring-mass system has changed the first three
natural frequencies by 3.43%, 3.97% and 0.05%, respectively. Of
course, any change in the natural frequencies will be dependent on
the parameters of the two degree-of-freedom spring-mass system
and its location on the beam.
The next set of results for this example was obtained to demonstrate the effect of the spring constants K S1 or K S2 ) on the
fundamental natural frequency of the beam. Note that K S1 and
K S2 are equal for the present problem. Figure 5 shows the variation of the fundamental natural frequency ( 1 ) of the beam
against the variation of the spring constant K S1 or K S2 ). For
smaller values of stiffness coefficients the fundamental natural
frequency increases with the increase in spring constant, but beyond a large stiffness constant it approaches a steady constant
value and remains virtually unaltered. This is to be expected and
is in accord with an earlier investigation 28. Other parameters
Present theory
Ref. 28
Beam alone
1
2
3
983.19
2695.1
5499.0
983
2694
5499
1018.1
2806.4
5501.8
Fig. 5 The effect of spring stiffness on the fundamental natural frequency of the clamped-clamped beam carrying a two
degree-of-freedom shown in Fig. 3
Frame with
two DOF system
Frame without
two DOF system
%
difference
1
2
3
176.62
229.40
246.09
224.76
245.70
267.38
21.4
6.63
7.96
5
Fig. 6 A plane frame carrying a two degree-of-freedom spring
mass system
EI4.0106 Nm2
and
m30 kg/m
The length of each member of the frame can be deduced from the
geometry shown. The parameters of the two degree-of-freedom
system connected to one of the members in the frame as shown in
Fig. 6 are
M 1500 kg,
I 75 kgm2 ,
K S2 1.25108 N/m,
K S1 108 N/m,
Conclusions
Appendix
Derivation of the Dynamic Stiffness Matrix of a BernoulliEuler Beam. The axial and bending deformations for a
Bernoulli-Euler beam are uncoupled. Therefore, the derivation of
the dynamic stiffness matrix for each of these two cases can be
carried out separately, and later can be combined together.
Axial Stiffnesses. A uniform Bernoulli-Euler beam of length L
is shown in Fig. 7 in a rectangular right handed Cartesian coordinate system with the X-axis coinciding with the axis of the
beam. The governing differential equation of motion of the beam
in free axial or longitudinal vibration is given by 36,44
EA
2u
2u
t2
m
2
(A1)
D 1 D 2 1.5 m.
a1
F x1
F x2
a2
a2 U1
a1 U2
(A11)
EA
cot ,
L
a 2
EA
cos ec
L
(A12)
(A2)
2 U0
(A3)
d 4V
d4
m 2L 2
EA
(A4)
U C 1 cos C 2 sin
(A5)
dU
EA dU
dx
L d
At x0 i.e. 0 ,
At xL i.e. 1 ,
UU 1
UU 2
and
and
FF x1
and
C1
cos C 2
m 2L 4
EI
(A16)
C 6 cosh
M x M
EI d 2 V
L d
2
EI
L2
(A18)
2 C 3 cos C 4 sin
C 5 cosh C 6 sinh
(A19)
S x S
(A9)
(A15)
1 dV
C 3 sin C 4 cos C 5 sinh
L d L
(A7)
EI d 3 V
L d
3
EI
L3
3 C 3 sin C 4 cos
(A10)
FF x2 (A8)
C1
sin C 2
0
EA
F1
F2
L
sin
4 V0
Substituting Eqs. A7 and A8 into Eqs. A5 and A6, the following matrix relationships can be obtained
1
U1
U2
cos
(A14)
(A6)
(A13)
EA
C 1 sin C 2 cos
L
where
F x F EA
t2
where
v ,t V e i t
d2
2v
m
4
where EI and m are the bending or flexural rigidity and mass per
unit length of the beam respectively, and v (x,t) is the bending or
flexural displacement of the cross-section at a distance x and t is
time.
Assuming harmonic oscillation with angular or circular frequency so that
4v
1 ,
VV 1 ,
SF y1
M M 1
VV 2 ,
2 ,
(A21)
SF y2
and
M M 2
(A22)
Substituting Eqs. A21 and A22 into Eqs. A17A20, the following two matrix equations can be obtained for displacements
and forces, respectively, in terms of the constants C 3 C 6 .
1
V1
0
1
V2
C
2
S /L
/L
/L
C h
S h
C /L
S h /L
C h /L
C3
C4
C5
C6
0
W 3 3
0
W 3 3
F y1
2
2
W 2
0
W 2
0
M1
3
3
3
F y2
W 3 S W 3 C W 3 S h W 3 3 C h
M2
W 2 2 C W 2 2 S W 2 2 C h
W 2 2 S h
C3
C4
C5
C6
(A24)
where
C cos ,
S sin ,
W 1
EI
,
L
C h cosh ,
W 2
EI
L
W 3
S h sinh ,
EI
(A25)
L3
d1
F y1
d2
M1
F y2
d4
M2
d5
d2
d4
d5
d3
d 5
d6
d 5
d1
d 2
d6
d 2
d3
V1
1
V2
2
(A26)
where
d 1 W 3 3 S C h C S h /,
d 2 W 2 2 S S h /,
d 3 W 1 S C h C S h /
d 4 W 3 3 S S h /,
(A27)
d 5 W 2 2 C h C /,
d 6 W 1 S h S /
(A28)
1C C h
(A29)
with
Combination of Axial and Bending Stiffnesses. A simple superposition is now possible to put the axial and bending dynamic
stiffnesses together in order to express the force-displacement relationship of the Bernoulli-Euler beam. Superposing Figs. 8a
and 8b, see Fig. 1, and using Eqs. A11 and A26, one obtains
the dynamic stiffness relationship of a Bernoulli-Euler beam in
free vibration, as follows
a1
F x1
0
F y1
0
M1
F x2
a2
F y2
0
M2
0
a2
d1
d2
d4
d5
d2
d3
d 5
d6
a1
d4
d 5
d1
d 2
d5
d6
d 2
d3
FK
(A31)
References
(A23)
and
or
U1
V1
1
U2
V2
2
(A30)
30
31
32
33
34
35
36
37 Wittrick, W. H., and Williams, F. W., 1971, A General Algorithm for Computing Natural Frequencies of Elastic Structures, Q. J. Mech. Appl. Math., 24,
pp. 263284.
38 Anderson, M. S., Williams, F. W., Banerjee, J. R., Durling, B. J., Herstrom, C.
L., Kennedy D., and Warnaar, D. B., 1986, User Manual BUNVIS-RG: An
Exact Buckling and Vibration Program for Lattice Structures, with Repetitive
Geometry and Substructuring option, NASA Tech. Memo. 87669.
39 Cheng, F. Y., 1970, Vibration of Timoshenko Beams and Frameworks, J.
Struct. Div. ASCE, 96, pp. 551571.
40 Wang, T. M., and Kinsman, T. A., 1971, Vibration of Frame Structures According to the Timoshenko Theory, J. Sound Vib., 14, pp. 215227.
41 Howson, W. P., and Williams, F. W., 1973, Natural Frequencies of Frames
with Axially Loaded Timoshenko Members, J. Sound Vib., 26, pp. 503515.
42 Cheng, F. Y., and Tseng, W. H., 1973, Dynamic Stiffness Matrix of Timoshenko Beam Columns, J. Struct. Div. ASCE, 99, 527549.
43 Banerjee, J. R., 1996, Dynamic Stiffness Formulation for Structural Elements: A General Approach, Comput. Struct., 63, pp. 101103.
44 Clough, R. W., and Penzien, J., 1975, Dynamics of Structures, McGraw-Hill,
Singapore.
This paper addresses the phenomenon of the destabilizing effect of slight damping on
asymmetric linear systems. Previous works had showed that the destabilizing effect, regarded for a long time as a paradox, depends upon the ratio of the damping coefficients. This work extends those results to n-dof systems. In fact, conditions for a general
asymmetric n-dof slightly damped system to be stable are obtained. Also, a useful sufficient condition is carried out. This practical design tool gives optimum damping ratios
and takes into consideration the parameter uncertainties as well. The solution is based on
the eigenvalue sensitivity analysis. Moreover, a formal physical explanation of the destabilizing effect of damping is given. Eventually, the theory is validated by means of a
simple example. DOI: 10.1115/1.1569945
Introduction
(1)
Theory
2.1
(2)
(3)
where i , ui are respectively the ith eigenvalue and the ith right
eigenvector of the system. In the same way, the left eigenvector
problem is given by the equation
viT M i2 C i K 0
(4)
2M i
d i
dui
d i
U i C
0 (5)
ui M i2 C i K
dc
dc
dc
Fig. 1 Mechanical example of a n d.o.f. system. Nonconservative forces that cause the mass matrix andor the damping matrix to be asymmetric are not depicted in figure. A damping
element can be inserted between two masses andor between
one mass and the frame.
viT 2M i
vT1 Uu1
d 1 d 2
T
0.5263
dc
dc
2v1 Mu1
d i
dui
d i
U i C
u vT M i2 C i K
dc
dc i i
dc
viT 2M i
d i
d i
u 0
U i C
dc
dc i
viT U i ui
d i
T
dc
vi 2M i C ui
c0
viT Uui
Re d i /dc 0,
(8)
a small c will move the ith eigenvalue toward left, causing the
system to remain stable. Therefore, for small c, the damping ratio
matrix will have a stabilizing effect only if Re(di /dc)0 i
1, . . . ,n. Note that the eigenvectors of an undamped system are
always real. Therefore, for this reason and for Eq. 7
d i
d i
Re
dc
dc
(9)
5
4
1.5
10
(10)
0.7327
u1
,
0.3825
u4
0.6466
,
v1
0.5148
0.7327
u2
,
0.3825
0.6526
0.5645
0.6466
v2
,
0.5148
v4
0.2400
0.8289
Therefore
360 Vol. 125, JULY 2003
0.2400
v3
,
0.8289
2.2 An Alternative Simpler Way to Express Stability Conditions. Instead of Eq. 7, by expanding matrix U, we can obtain a simplified formulation. For the sake of clarity, we introduce
Fig. 1, where part of a discrete mass-spring-damped system is
represented. In this way one can visualize the physical meaning of
the elements of the matrix U. In other words, Fig. 1 shows the
damping coefficients one can introduce in the mechanical system.
The forces that cause the stiffness matrix and/or the mass matrix
to be asymmetric are not depicted in figure.
We can now study the structure of the damping ratio matrix U
and give its physical interpretation. This provides the designer
with an important tool to select the best combination of damping
elements. Basically, one has two different ways to introduce a
damping element on a mechanical system:
a damping element can be connected between one mass/
inertia and the frame;
a damping element can be connected between two masses/
inertias.
The former is represented in the figure by a physical damping
element attached between the mass m l and the frame. Its numerical damping value is c ll . We call damping element coefficient
the adimensional coefficient ll .
In this case, if c ll was the only damping element on the mechanical system, the matrix U would be
UUll
0.6526
u3
,
0.5645
(12)
(7)
2viT Mui
vT3 Uu3
d 3 d 4
T
0.1162
dc
dc
2v3 Mu3
(6)
(11)
Ull o, p 0
o,pl
Ull o, p ll
opl
0
]
]
0
ll
0
]
]
0
(13)
0
]
ls
0
]
ls
0
0
]
0
ls
ls
(14)
rr
r,t1
rt
Urt
r,t1
rt
Urt
(15)
viT Uui
r1
viT Urr ui
r,t1
rt
i r
viT Urt ui
n
v u
r1
i r rr
r,t1
rt
vi r ui r vi r ui t
vi t ui r vi t ui t rt
(16)
(17)
where
ai
vi 1 ui 1
]
vi n ui n
vi 1 ui 2 vi 1 ui 2 vi 2 ui 1 vi 2 ui 2
]
vi n1 ui n vi n1 ui n vi n ui n1 vi n ui n
T 11 , 22 , . . . , nn , 12 , 13 , . . . 1n , 23 , . . . , n1 n
(18)
By replacing Eq. 17 into 7, one obtains the derivative of the ith
eigenvalue as a function of the damping element coefficient vector
d i
dc
is
aiT
2viT Mui
(19)
Therefore
d i
0sign viT Mui aiT 0
dc
Journal of Vibration and Acoustics
(20)
(21)
(22)
r1
nr
0.4737
0.1969
0.4737
0.1969
0.1566
0.4679
0.1566
0.4679
(23)
(24)
(25)
(26)
The largest solution cone can be found using the cone algorithm
proposed by Wan 10. A modified version of the cone algorithm
is proposed in appendix. Since the largest solution cone is the
largest cone included in S, it gives a set of solutions of system
21. Of course, it does not represent all the possible solutions; it
defines a subset of the solutions of the system of inequalities. In
spite of that, one can note that, while system 21 is made up of r
inequalities, the largest solution cone is defined by means of only
one inequality:
JULY 2003, Vol. 125 361
arccos
T
wsc
sc
cos sc
wsc
wsc
T
wsc
(27)
(28)
T
/ wsc
In fact, when o pt the value of the scalar product wsc
1 of Eq. 27 is maximum.
Not only can the designer choose the best solution, but he also
can analyze the robustness of the solution.
In a practical scenario model errors occur. Therefore, the real
value of defers from the theoretical optimum solution o pt . One
can take into account the uncertainties in the model by means of
the relationship
o pt
(29)
(30)
Using Eq. 30, given the optimum choice for the damping coefficients, one can analyze the effect of the uncertainties as far as
robustness of the system is concerned. In other words, as long as
uncertainties satisfy Eq. 30, the damped system is stable.
Resuming the example 10, suppose now that the damping
element coefficients 11 , 22 have to be set by the designer. The
optimum solution provided by the cone algorithm is
o pt 0.7262 0.6875 T ,
sc 24.3
(31)
(32)
(33)
1
i e i t
u e i t u
2 i
(34)
sider only the steady periodic part of the solution. In the following
we calculate the energy the system exchanges during each cycle.
x
1
i e i i t
u e i i t u
2 i
(35)
2/i
xT Cxdt
(36)
2/i
xT Mxdt
2/i
xT Kxdt
(37)
dW e
dW d
d i
i
dc
dc
dc
(38)
dW e
dW e
c
c
dc
dc
(39)
dW d
d i
d i
i
c W dD i
c
dc
dc
dc
(40)
Conclusions
This paper deals with asymmetric linear systems. These systems can be found in many problems of interest in aeronautics and
control. In many cases, stability is studied by neglecting damping.
Unfortunately, even slight damping can destabilize the system. We
introduced a new criterion for foreseeing the destabilizing effect
of damping. The criterion asserts that damping does not destabilize the system if and only if a system of homogeneous linear
inequalities is satisfied. Moreover, we proved that the system of
inequalities can be replaced with a simpler inequality that gives a
sufficient condition. A simple example shows the theory step by
step. Eventually, a general physical explanation of the destabilizing effect of damping is given, using energy considerations.
Appendix
Description of the Modified Cone Algorithm. The cone algorithm finds a solution cone C(ws , s ), given an angle s . It
starts with an arbitrary axis w0 . The vector is then modified according to the following rule
Transactions of the ASME
i 90 s
if wi ,a
i
wi1 wi a
wi1 wi
(41)
i 90 s
if wi ,a
2M
2/i
xT Mxdt
2/i
xT Kxdt
2/i
ui e
iit
ie
u
i i t
dt
2/i
c0
2/i
dxT
Cxdt
dc
(42)
xT C
2/i
2/i
xT Uxdt
2/i
iT e i i t U ui e i i t u
i e i i t dt
uiT e i i t u
(43)
2/i
i2 MK
ii
i e i i t T i2 MK
ui e i i t u
4
i
dui i t du
e i
e i i t dt
dc
dc
2/i
2/i
i i t
i i i t i t dui
i
e i e
4
dc
i i i t i t T
dui i t
i u 2 MK
e i e
e i
i
i
4
dc
dt
dui
dc
i2 MK ui uiT
dui
dc
It is reminded that the derivative 44 is evaluated at c0. Conjugating Eq. 44 and considering the fact that d i /dcR, and
that the conjugate of the product of two complex numbers equals
the product of their conjugates, it yields
dui
dc
(46)
iT
du
d i
d i
U i C
ui
M i2 C i K ui
dc
dc
dc
iT M i2 C i K
u
(44)
uiT 2M i
d i
dui
U i i ui i2 MK
0
dc
dc
i
i i dui i t du
e i
e i i t
4 dc
dc
i2 MK
dui
d i
U i ui i2 MK
0
dc i
dc
i2 i t i t T
i u Uu e i i t e i i t dt
e i e
i
i
4
d
M i2 cU i K ui 0
dc
2M
2/i
d
M i2 C i K ui 0
dc
xT Uxdt
2M
i
d i
du
U i i ui i2 MK
0
dc
dc
i2 MK ui e i i t e i i t dt
dxT
dt
dc
i uiT Uui
2M
i2
i
du
d i
U i iui i2 MK
dc
dc
2/i
2M
i e i i t dt
ui e i i t u
dui
d i
U i i ui i2 MK
dc
dc
i /dc
Comparing formula 45 and 44, one obtains du
dui /dc
The increment of the energy given by the external forces for
each cycle due to an increment of c is
i i T i t
iT e i i t i2 MK
ui e i u
4
2/i
2M
(45)
dW e
dc
xT MxKx dt
dui
d i
U i i ui i2 MK
dc
dc
dui
0
dc
(47)
iT
d i
dui du
iT i2 MK
u
i2 MK ui
dc
dc
dc
i i uiT Uui i i
d i
dui duiT
uiT i2 MK
dc
dc
dc
i2 MK ui 0
(48)
uiT i2 MK
dui duiT
i2 MK ui
dc
dc
dW e
dW d
d i
i
dc
dc
dc
dui
dc
dui
dc
(49)
d i
dui
uiT i2 MMT KKT
dc
dc
0
Multiplying by i one obtains
i uiT Uui i
d i
iuiT i2 MMT
dc
KKT
dui
0
dc
(50)
i
dc
dc
dc
(51)
(52)
References
1 Gasparetto, A., 1998, A System Theory Approach to Mode Coupling Chatter
in Machining, ASME J. Dyn. Syst., Meas., Control, 120, pp. 545547.
2 Inman, D. J., 1983, Dynamics of Asymmetric Nonconservative Systems,
ASME J. Appl. Mech., 50, pp. 199203.
3 Kounadis, A. N., 1992, On the Paradox of the Destabilizing Effect of Damping in Non-Conservative Systems, Int. J. Non-Linear Mech., 27, pp. 597
609.
4 Semler, C., Alighanbari, H., and Padoussis, M. P., 1998, A Physical Explanation of the Destabilizing Effect of Damping, ASME J. Appl. Mech., 65, pp.
642 648.
5 Bolotin, V. V., and Zhinzher, N. I., 1969, Effects of Damping on Stability of
Elastic Systems Subjected to Non-Conservative Forces, Int. J. Solids Struct.,
16, pp. 965989.
6 Herrmann, G., and Jong, I. C., 1966, On Nonconservative Stability Problems
of Elastic Systems with Slight Damping, ASME J. Appl. Mech., 32, pp.
125133.
7 Herrmann, G., and Jong, I. C., 1965, On the Destabilizing Effect of Damping
in Nonconservative Elastic Systems, ASME J. Appl. Mech., 32, pp. 592597.
8 Nemat-Nasser, S., Prasad, S. N., and Herrmann, G., 1966, Destabilizing Effect of Velocity-Dependent Forces in Nonconservative Continuous Systems,
AIAA J., 4, pp. 1276 1280.
9 Adhikari, S., and Friswell, M. I., 2001, Eigenderivative Analysis of Asymmetric Non-conservative Systems, Int. J. Numer. Methods Eng., 51, pp. 709
733.
10 Wan, J. S., 1994, Cone Algorithm: An Extension of the Perceptron Algorithm, IEEE Trans. Syst. Man Cybern., 24, pp. 15711576.
11 Murty K., 1976, Linear and Combinatorial Programming, Wiley, New York.
Marty E. Johnson
Assistant Professor,
Vibration and Acoustic Labs,
Mechanical Engineering,
Virginia Tech,
Blacksburg, VA 24061-0238
Luiz P. Nascimento
Sao Paulo State University,
Sao Paulo, Brazil
Mary Kasarda
Rotor Dynamics Group,
Mechanical Engineering,
Virginia Tech,
Blacksburg, VA 24061-0238
Chris R. Fuller
Vibration and Acoustic Labs,
Mechanical Engineering,
Virginia Tech,
Blacksburg, VA 24061
Introduction
Theory
(1)
(5)
u x,
a x
n0
(2)
j
2
a n
x j F
2
2
m r n 2 j n n
(3)
where the natural frequency and damping ratio for the nth mode
are given by n and n respectively. The damping ratio is typically low for a steel rotor, on the order of 0.005. Most of the
damping in this system will be provided by the bearings. The total
mass of the rotor is m r lS where S is the cross sectional area of
the rotor, is the density of the rotor material and l is the rotor
length. The natural frequencies of the modes can be calculated
using the equations presented in Appendix B.
By combining Eq. 2 and Eq. 3, point and transfer mobility
terms can be calculated. The transfer mobility T i j is defined as the
velocity of the beam at position x i due to an input force at position
x j . The point mobility is simply the velocity due to a force operating at the same position on the beam i.e. x i x j .
2
j
T i j
xi x j
2
2
m r n0 n 2 j n n
(4)
f r1
frZum f r2
f r3
j m 1 c 1
k1
j
k2
j m 2 c 2
j
jm3
u m1
u m2
u m3
(6)
Where um is the vector of velocities at the three mass locations
and the reaction force is given by the vector fr . The velocity
vector um can be considered as the combination of the velocity
umr due to the reaction forces fr and the velocity umj due to external input forces to the rotor fj . The external input forces in this
model are created by the mass unbalance and by the active control
forces. Using Eq. 5 and Eq. 6 leads to an expression for the
reaction force in terms of the external forces,
umumjumrTmjfjTmmfrTmjfjTmmZum
um ITmmZ 1 Tmjfj
frZumZ ITmmZ
(7)
1
Tmjfj
(8)
The two matrices Tmm and Tmj contain rotor mobilities describing
the velocity at the three mass locations due to forces acting at the
mass locations and the locations of the external forces respectively. The elements in these matrices can be calculated using Eq.
4. Equations 7 and 8 can be used to calculate a new set of
ij that include the reaction force fr of
augmented rotor mobilities T
the masses, dampers and stiffnesses. From this augmented set of
equations the velocity at any point on the rotor due to a force
applied at any other point on the rotor can be calculated. We can
now re-write the mobility equation Eq. 5 to include reaction
forces.
ijfj
uiTijfjTimfrT
ij TijTimZ ITmmZ 1 Tmj
T
(9)
Table 1 Physical properties of rotor and bearings used for the simulation
Physical Properties
Rotor
Bearing 1
Legth
Diameter d
Modulus of
elasticity E
Density
Poissons
ratio
18 in or
457 mm
0.375 in or
9.52 mm
207 GNm2
3
7800 kgm
0.3
Bearing 2
0.25 kg
Mass
0.25 kg
Mass
0.8 kg
Stiffness
k1
Damper
c1
Position
x1
16000 Nm1
Stiffness
k2
Damper
c2
Position
x2
16000 Nm1
Position
xm
6.5 in or
165 mm
50 Nm
4.5 in or
114 mm
The two matrices Tim and Tij contain rotor mobilities describing
the velocity at the observation locations xi due to forces acting at
the mass locations and the locations of the external forces. Using
this augmented set of equations the performance of an active control system can be evaluated.
Active Control Performance. In this section a method for
calculating the control performance of an active control system
will be presented. An active control system using the filtered
X-LMS system will attempt to minimize the sum of the squared
outputs from the error sensors. If a control system has J control
actuators and L error sensors then at a single frequency the L
length vector of errors e can be written in terms of the vector of
primary disturbance forces fp i.e. rotor unbalance and the J
length vector of secondary control forces fc .
epfpTecfc
eT
(10)
The two matrices Tep and Tec contain augmented rotor mobilities
describing the velocity at the error locations xe due to forces acting at the locations xp of the unbalance forces primary disturbance and the locations xc of the control forces. The sum of the
squared error signals can be calculated using the Hermitian transpose or conjugate transpose denoted by H) as eHe and is minimized when the secondary control forces fc are given by 9,11,
fc
opt
H
1 T
H
T
ecTec
ecTepfp
(11)
ui bTipfp
(12)
ui aTipfpTicfc
Unbalance mass
Mass
opt
50 Nm
14.5 in or 368
mm
(13)
Theoretical Analysis
Fig. 6 Velocity along the rotor at 100 Hz with and without control using actuator 1 and a single error sensor e1 , actuator 1
and both error sensors and then both actuators and both error
sensors.
(14)
This transfer function is composed of components from the position sensor, low pass filter, PID filter and power amplifier. The
equivalent stiffness and damping are then given by,
Fig. 7 The mode shapes and natural frequencies for the first
three modes of the rotor for the three different experimental
mass configurations.
K eq K x K i a G ;
(15)
C eq K i b G /
(16)
Table 2 Different Error Sensors and Actuators Locations for Unbalance Control
Cases
Error Sensors
Actuators
S 1 and S 2
V 1 and W 1
S 1 and S 2
V 2 and W 2
S 1 , S 2 , S 3 and S 4
V 1 and W 1
S 1 , S 2 , S 3 and S 4
V 1 , W 1 , V 2 and W 2
S 1 and S 2
V 1 and W 1
S 1 and S 2
V 2 and W 2
S 1 and S 2
V 1 and W 1
S 1 , S 2 and SV 1
V 1 and W 1
S 1 , S 2 and SV 1
V 2 and W 2
Experimental Results
Conclusions
A theoretical and experimental investigation into the effect of
the location and number of sensors and actuators on both global
and local vibration reduction along a rotor using a feedforward
control scheme with magnetic bearing actuators has been completed. Theoretical approaches developed for the active control of
beams have been shown to be useful as simplified models for the
rotor scenario for optimization studies involving sensor and actuator location and number. Also presented is the time-domain filtered X LMS feedforward control strategy, used widely in the
active control of sound and vibration, as an alternative control
methodology to the frequency-domain open loop approaches commonly presented in the literature. This control scheme performed
extremely well with reductions in vibration amplitudes as high as
51.7 dB a 380 fold decrease.
The rotor under analysis consists of a small 9.52 mm diameter
high-speed laboratory rotor supported in active magnetic bearings
with one large disk that can be placed in different locations. Both
experimental results and simulation results are presented in Table
3. In general,
Acknowledgments
The authors would like to acknowledge the National Science
Foundation NSF for supporting the work presented in this paper
SV 1
SW 1
S1
S2
S3
S4
SV 2
SW 2
Global
SV 1 , SW 1
S1 , S2
S3 , S4
SV 1 , W 1
1
2
3
4
5
6
7
8
9
13.2
3.5
10.4
15.2
8.7
3.5
3.8
2.2
25.6
15.7
11.7
14.5
19.8
14.1
6.5
48.8
49.4
23.2
34.8
41.6
8.1
41.2
14.1
30.6
48.1
51.7
28.5
30.5
50.8
16.4
41.7
11.4
29.6
17.8
11.3
20
29.2
18.9
6.2
22
15.8
26.6
29.3
28.1
4.4
6.8
3.4
8.9
7.8
11.2
9.3
14.3
3.4
15.1
9.4
20.3
8.2
16.8
9.8
17.8
19.1
15.1
3.4
7.4
16.1
13.0
16.7
18.0
4.5
22.1
16.8
16.1
23.3
19.9
3.0
11.1
4.8
6.8
17.6
7.2
20.8
References
The first two modes mode of a free-free beam are rigid body
modes with the higher modes given as combinations of sine, cosine and hyperbolic functions 17.
0 x 1
1 x 1
n x cos
(17)
2x
(18)
x
x
0.5 k cosh 0.5
for n2,4,6,
(19)
where
k
and tan tanh 0
2
2
sinh
2
sin
n x sin
x
x
0.5 k sinh 0.5
for n3,5,7,
(20)
where
k
and tan tanh 0
2
2
sinh
2
sin
n k 2n
0 0
(21)
1 0
(22)
EI n0.5 2 2 d
S
4 2
for
n2
(23)
1 Kasarda, M. E. F., 2000, An Overview of Active Magnetic Bearing Technology and Applications, invited review paper, Public Admin. Rev., 322, pp.
9199.
2 Knospe, et al., 1993, Adaptive On-Line Rotor Balancing Using Digital Control, Proceedings of 93 ROMAG, pp. 153164.
3 Knospe et al., 1998, Synthesis Techniques for Robust Adaptive Vibration
Control, 6th International Symposium on Magnetic Bearings, Cambridge,
Mass, p. 183.
4 Rutland et al., 1994, Comparison of Controller Designs for Attenuation of
vibration in a Rotor-Bearing System Under synchronous and Transient conditions, 4th International Symposium on Magnetic Bearings, Zurich, Switzerland, pp. 107112.
5 Matsumura et al., 1994, Elimination of Unbalance Vibration in AMB Systems Using Gain Scheduled H Robust Controllers, 4th International Symposium on Magnetic Bearings, Zurich, Switzerland, pp. 113118.
6 Hisatani et al., 1997, Adaptive Filtering for Unbalance Vibration Suppresion, Proceedings of MAG97, Alexandria, VA, pp. 125130.
7 Hope, R. W., Tessier, L. P., Knospe, C., and Miyaji, T., 1998, Adaptive
Vibration Control of Industrial Turbomachinery, 98-GT-405, International
Gas Turbine & Aeroengine Congress & Exhibition.
8 Piper, G. E., and Calvert, T. E., 1995, Active Fluidborne Noise Control of a
Magnetic Bearing Pump, NCA-Vol. 21, IMECE, Proceedings of the ASME
Noise Control and Acoustics Division, pp. 5576.
9 Fuller, C. R., Elliott, S. J., and Nelson, P. A., 1996, Active Control of Vibration,
Academic Press, London.
10 Brennan, M. J., Elliott, S. J., and Pinnington, R. J., 1995, Strategies for the
Active Control of Flexural Vibration on a Beam, J. Sound Vib., 1864, pp.
657 688.
11 Nelson, P. A., and Elliott, S. J., 1993, Active Control of Sound, Academic
Press.
12 Fuller, C. R., Gibbs, G. P., and Silcox, R. J., 1990, Simultaneous Active
Control of Flexural and Extensional Waves in Beams, J. Intell. Mater. Syst.
Struct., 1, pp. 235247.
13 Guigou, C., and Fuller, C. R., 1993, Active Control of Sound Radiation from
a Simply Supported Beam: Influence of Bending Near-field Waves, J. Acoust.
Soc. Am., 935, pp. 2716 2725.
14 Post, J. T., and Silcox, R. J., 1990, Active Control of the Forced Response of
a Finite Beam, Proc. Of Noise-Con90, Austin, Texas, pp. 197202.
15 Widrow, B., and Sterns, S. D., 1985, Adaptive Signal Processing, Prentice
Hall, Englewood Cliffs, New Jersey.
16 Elliott, S. J., Stothers, I. M., and Nelson, P. A., 1987, A Multiple Error LMS
Algorithm and Its Application to the Active Control of Sound and Vibration,
IEEE Trans. Acoust., Speech, Signal Process., 35, pp. 14231434.
17 Warburton, G. B., 1954, The Vibration of Rectangular Plates, Proc. Inst.
Mech. Eng., 168, p. 371383.
18 Bishop, R. E. D., and Johnson, D. C., 1960, The Mechanisms of Vibration,
Cambridge University Press.
19 Childs, D., 1993, Turbomachinery Rotordynamics Phenomena, Modeling, and
Analysis, John Wiley & Sons, Inc., New York.
20 Allaire, P. E., et al., 1993, Magnetic Bearings, STLE Handbook of Tribology
and Lubrication, Vol. III.
21 Clements, J. R., 2000, The Experimental Testing of an Active Magnetic
Bearing/Rotor System Undergoing Base Excitation, Master Thesis, Virginia
Tech.
Hemanshu R. Pota
e-mail: h-pota@adfa.edu.au
Ian R. Petersen
School of Electrical Engineering,
UNSW at the Australian Defense Force Academy,
Canberra ACT, 2600, Australia
Atul G. Kelkar
Department of Mechanical Engineering,
Iowa State University,
Ames, IA 50011
e-mail: akelkar@isu.edu
Introduction
resents the noise input and it is assumed that the effects of the
uncertainty show up at the sensor output through the disturbance
channel. The minimax LQG method requires that the inputs due to
uncertainty i.e., the output of the (s) block and the noise enter
through the same channel. Figure 3 shows one particular uncertainty model but the minimax LQG method is not restricted to this
model alone and alternative representations are possible.
The block (s) in Fig. 3 is chosen such that
j 1 ,
P 2 s P 2 s / P 1 s s W s .
(1)
P 2 j P 2 j
W j .
P 1 j
(2)
zC 1 xD 1 u,
,
yC 2 xD 2 D 2 w
y 0 0,
(3)
m
J lim
x t Rx t u t Gu t dt,
Y C 2 D 2 D 2 1 C 2
B 2 ID 2 D 2 D 2 1 D 2 B 2 0
1
1
X AB 1 G 1
AB 1 G X R G
R 1/2x
,
G 1/2u
(5)
(6)
and
(4)
where R0 and G0. The minimax LQG control problem involves finding a controller which minimizes the maximum of this
cost function where the maximum is taken over all uncertainties
satisfying the uncertainty constraint 1. If we define a variable
1
R Y
X B 1 G 1
B 1
1
B B X 0,
2 2
(7)
Y C T2 B 2 D T2 D 2 D T2 1
T
W ,tr C 2 Y D 2 B 2 X IY X 1 .
Y R
x AB 1 G 1
B 1 G 1
x
B 1
1
Y X
y C 2
1
B B X x
2 2
Y C 2 B 2 D 2 D 2 D 2 1
1
D B X x
2 2
u G 1
.
B 1 X x
(8)
(9)
Controller Design
Fig. 9 Uncertainty
P 2 j P 1 j
bound
2 j
dashed-20 log 10 W j ,solid-20 log 10 P
with the order of the weighting filter W(s). Typical order of controllers designed in this paper is around eighty. Such a high order
controller may lead to problems of numerical error and excessive
computational load when implemented. Hence, the balanced controller reduction method described in Section 19.1.1 of 23 is
applied to reduce the controller order to around forty. It was found
that with this level approximation, there was very little degradation in the predicted closed loop performance.
Experimental Results
From the figures it can be seen that even a 40th order model is
unable to match the data exactly.
The uncertainty envelope W(s) and the magnitude of the 40th
order transfer function, for the identified models in Figs. 7 and 8,
derived using Yule-Walker method is shown in Fig. 9.
For the case under discussion the controller order is 80 and
balanced model reduction technique 23 was applied in order to
obtain a 40th order approximation to the 80th order controller. The
designed controller is shown in Fig. 10.
5.2 Control of Higher Frequency Modes. The experimental results presented thus far were restricted to the lower frequency
range of 20100 Hz. In this section the minimax LQG control
Fig. 16 Uncertainty
P 2 j P 1 j
bound
actuators and sensors can be used to get rid of deep nulls. In other
words optimal actuator-sensor placement methods should use
modelling error as a measure of optimality.
The influence of the choice of W(s) on the controller can be
used to shape the controller. For example, the frequency band
where W( ) is large, the controller gain will be low. In the
closed-loop response in Fig. 13 it can be seen that there is an
increase in the gain near the deep null around 70 Hz. This is due
2 j
dashed-20 log 10 W j ,solid-20 log 10 P
Fig. 18 Uncertainty
P 2 j P 1 j
bound
Conclusions
2 j
dashed-20 log 10 W j ,solid-20 log 10 P
very impressive and further work in the selection of actuatorsensor placement to obtain a tighter fit between the model and
data and a choice of frequency weighting function can improve
the results significantly.
References
1 Elliott, S. J., 1999, Down with Noise, IEEE Spectrum, June, pp. 54 61.
2 Hu, J.-S., Yu, S.-H., and Hsieh, C.-S., 1998, Application of Model-Matching
Techniques to Feedforward Active Noise Control Design, IEEE Trans. Control Syst. Technol., 61, pp. 33 42.
3 Omoto, A., and Elliott, S. J., 1999, The Effect of Structured Uncertainty in
the Acoustic Plant on Multichannel Feedforward Control Systems, IEEE
Trans. Speech Audio Process., 72, pp. 204 212.
4 Pota, H. R., and Kelkar, A. G., 2001, Modelling and Control of Acoustic
Ducts, ASME J. Vibr. Acoust., 1231, pp. 210.
5 Kelkar, A. G., and Pota, H. R., 2000, Robust Broadband Control of Acoustic
Duct, Proceedings of the 39th IEEE Conference on Decision and Control,
Sydney, Australia, pp. 4485 4490.
6 Hong, J., Akers, J. C., Venugopal, R., Lee, M.-N., Sparks, A. G., Washabaugh,
P. D., and Bernstein, D. S., 1996, Modeling, Identification, and Feedback
Control of Noise in an Acoustic Duct, IEEE Trans. Control Syst. Technol.,
43, pp. 283291.
7 Clark, R. L., and Cole, D. G., 1995, Active Damping of Enclosed Sound
Fields Through Direct Rate Feedback Control, J. Acoust. Soc. Am., 973, pp.
17101716.
8 Erwin, R. S., and Bernstein, D. S., 1997, Discrete-time H 2 /H Control of an
Acoustic Duct: Delta-domain Design and Experimental Results, Conference
on Decision and Control, IEEE, San Deigo, CA, pp. 281282.
9 Petersen, I. R., and Pota, H. R., 2000, Minimax LQG Optimal Control of an
Experimental Acoustic Duct, IEE Control 2000 Conference, Cambridge, UK.
10 Petersen, I. R., 2001, Multivariable Control of Noise in an Acoustic Duct,
Proceedings of the European Control Conference ECC2001, Porto, Portugal.
11 Pota, H. R., and Kelkar, A. G., 2001, On Perfect Acoustic Noise Cancelling
Control, Control and Intelligent Systems, 292, pp. 48 54.
12 Petersen, I. R., Ugrinovski, V., and Savkin, A. V., 2000, Robust Control Design
using H Methods, Springer-Verlag, London.
13 Ugrinovskii, V. A., and Petersen, I. R., 1998, Time-averaged Robust Control
of Stochastic Partially Observed Uncertain Systems, Proceedings of the IEEE
Conference on Decision and Control, IEEE, Tampa, FL.
14 Fang, B., Kelkar, A. G., and Joshi, S. M., 2002, Modelling and Control of
Acoustic-Structure Interaction in 3-D Enclosures, IEEE 2002 Conference on
Decision and Control, Las Vegas, Nevada.
15 Ding, W. P., and Chen, H. L., 2001, A Symmetrical Finite Element Model for
Structure-Acoustic Coupling of an Elastic, Thin-Walled Cavity, J. Sound
Vib., 2433, pp. 547559.
16 Marburg, S., Beer, H.-J., Gier, J., Hardtke, H.-J., Rennert, R., and Perret, F.,
2002, Experimental Verification of Structural-Acoustic Modelling and Design Optimization, J. Sound Vib., 2524, pp. 591 615.
17 Banks, H. T., Brown, D. E., Smith, R. C., Metcalf, V. L., Wang, Y., and Silcox,
R. J., 1994, Noise Control in a 3-D Structural Acoustic System: Numerical
and Experimental Implementation of a PDE-Based Methodology, IEEE Proceedings of the 33rd Conference on Decision and Control, Lake Buena Vista,
FL, pp. 305310.
18 Nelson, P. A., and Elliott, S. J., 1992, Active Control of Sound, Academic
Press, London.
19 Elliott, S., 2001, Signal Processing for Active Control, Academic Press, Sydney.
20 McKelvey, T., Akcay, H., and Ljung, L., 1996, Subspace-based Multivariable
System Identification from Frequency Response Data, IEEE Trans. Autom.
Control, 417, pp. 960979.
21 Friedlander, B., and Porat, B., 1984, The Modified Yule-Walker Method of
ARMA Spectral Estimation, IEEE Trans. Aerosp. Electron. Syst., 202, pp.
158 173.
22 Petersen, I. R., Pota, H. R., and Jahromi, M. R. S., 2002, System Identification, Uncertainty Modelling and Actuator Placement in the Robust Control of
an Acoustic Duct, Proceedings of the Conference on Information, Decision,
and Control, Adelaide, Australia.
23 Zhou, K., Doyle, J. C., and Glover, K., 1996, Robust and Optimal Control,
Prentice Hall, NJ.
Rifat Sipahi
Nejat Olgac*
Mechanical Engineering Department,
University of Connecticut,
Storrs, CT 06269
Introduction
(1)
(2)
x ABxF
(3)
(4)
(5)
one such stable pocket in delay domain. This knowledge is extremely important, as one can imagine. If there are so called, the
secondary stability pockets one may opt to operate within one of
these, still performing a stable control despite the larger time delay in the picture. No existing methodology, however, could offer
a procedure to detect these additional pockets. The Direct Method
DM 22 suggests a unique construct ultimately yielding an explicit function of delay, which declares completely the stability
pockets. The end results clearly offer much stronger revelations
than simply stating the stability margin. The method is the first of
its kind as it handles the general n-dimensional LTI-RTDS with
commensurate time delays and uncovers its unique properties.
Some of these properties are realized in 23 for scalar dynamics
(n1), however the approach therein is quite restrictive to implement on a general LTI-RTDS i.e. n1). For these reasons, the
Direct Method is unique in the dynamic systems and controls
community, opening number of new directions for future research.
We wish to summarize the strong features of the Direct Method
here to better prepare the reader:
i. It does not have to start with a stable system for zero delay.
ii. It is exact, i.e. the limits of the stability pockets are precise.
They are not conservative results, which is a concern for the
peer methodologies 910,18 21,23.
iii. All stable pockets are declared exclusively, in one explicit
function of time delay.
In what follows we briefly review the procedure suggested by
the DM leaving the details to 22. The system in Eq. 5 has a
transcendental characteristic equation:
n
CE s, det sIABe s
a s e
j0
js
0 with 0
(6)
ck
RT
sgn Re
ds
d
s ck i
k
k1, . . . m
0, . . .
(7)
k k0
ck
0, . . .
(8)
0, . . .
k1, . . . m,
NU NU 0
k1
k0
U , k0 RT k
k
(9)
RT
NU
0
0.3071
0.5441
1.1444
1.9766
1.9818
2.0028
2.0052
7.5032
4.3864
7.5032
4.3864
7.5032
3.0446
2.98
1
1
1
1
1
1
1
0
2
4
6
8
10
8
6
for k0 , ck 0
k0 , ck 0
uF
xAxB
rad/s
the Direct Method of stability for LTI-RTDS. Now, we demonstrate how to deploy the DM for the active vibration suppression
applications by way of example.
U , k0
m 110.2,
c 1 2.2,
k 102,
k 114,
30
11
10
11
13.33
12.67
12.67
11
10
20
11
40 12.67 13.33
(10)
k 122,
m 210.2,
m 220.15 kg
c 2 1.9 Ns/m
k 204,
k 212,
k 222 N/m
26.67 12.67
m 120.15,
88
6.67
6.67
0
f t
0
0
0
0
0
0
81
82
28
det sIA 0
det sIABe s 0
(11)
(12)
c1 4.3864
c2 7.5032
c3 3.0446
c4 2.98
2p
c1
2p
2k 0.307
c2
2p
3k 2.003
c3
2p
4k 2.005
c4
1k 0.544
from
RT 1 1
RT 2 1
RT 3 1
RT 4 1
p0, . . .
(13)
x 12 s
16.78 s5 186.52 s4 887.89 s3 2930.75 s2 6516.5 s6155.36
8
F s
s 25 s7 272 s6 1724 s5 7172 s4 20516 s3 39520 s2 46000 s24000
(14)
F s
q 0 s q 1 s e s q 2 s e 2 s
(15)
for delayed feedback uKx(t ), where p i (s), q i (s) are selfevident polynomials of s.
The frequency response features of this system are given in Fig.
4 for various cases. It is clear that the feedback controlled case Eq.
14 registers substantial improvement over the uncontrolled case
for 2.7 rad/s. And the delayed case Eq. 15 is bringing further
improvement except in the frequency range of 7.18.1 rad/s.
In any event, it offers much better vibration suppression than the
uncontrolled passive case.
It is obvious that, Fig. 4 has meaning if and only if the system
is stable, which can only be assured using the Direct Method
presented. Therefore the novel method plays a very critical role in
this operation.
Conclusions
References
1 Olgac, N., 1995, Delayed Resonators as Active Dynamic Absorbers, United
States Patent 5,431,261.
2 Olgac, N., Special Lecture at the 3rd IFAC Workshop on TDS 2001, http://
www.siue.edu/ifacdelay/
3 Seto, K., and Furuishi, Y., 1991, A Study on Active Dynamic Absorber,
ASME Proceedings, Paper DE, Vol. 38, pp. 263270.
4 Seto, K., and Yamashita, S., 1991, Simultaneous Optimum Design Method
for Multiple Dynamic Absorbers to Control Multiple Resonance Peaks, SAE
Transactions, 100, pp. 14811489.
5 Olgac, N., and Hosek, M., 1998, A New Perspective and Analysis for Regenerative Machine Tool Chatter, Int. J. Mach. Tools Manuf., 387, pp. 783
798.
6 Tlusty, J, 1985, Machine Dynamics, R. I. King, ed., Handbook of High
Speed Machining Technology, Chapman and Hall, New York.
7 Olgac, N., and Holm-Hansen, B., 1994, A Novel Active Vibration Absorption
Technique: Delayed Resonator, J. Sound Vib., 176, pp. 93104.
8 Olgac, N., Elmali, H., Hosek, M., and Renzulli, M., 1997, Active Vibration
Control of Disturbed Systems Using Delayed Resonator with Acceleration
Feedback, ASME J. Dyn. Syst., Meas., Control, 119, pp. 380388.
9 Chen, J., Gu, G., and Nett, C. N., 1994, A New Method for Computing Delay
Margins for Stability of Linear Delay Systems, Syst. Control Lett., 26, pp.
107117.
10 Chen, J., 1995, On Computing the Maximal Delay Intervals for Stability of
Linear Delay Systems, IEEE Trans. Autom. Control, 406, pp. 10871092.
11 Hale, J. K., and Verduyn Lunel, S. M., 2001, Effects of Small Delays on
Stability and Control, Operator Theory; Advances and Applications, 122, pp.
275301.
12 Hale, J. K., and Verduyn Lunel, S. M., 2001, Strong Stabilization of Neutral
Functional Differential Equations, IMA J. Math. Control Inform., 19, pp.
119.
13 Hale, J. K., and Verduyn Lunel, S. M., 1993, Introduction to Functional Differential Equations, Springer-Verlag.
14 Hale, J. K., Infante, E. F., and Tsen, F.-S. P., 1985, Stability in Linear Delay
Equations, J. Math. Anal. Appl., 105, pp. 533555.
15 Hertz, D., Jury, E. I., and Zeheb, E., 1984, Simplified Analytic Stability Test
for Systems with Commensurate Time Delays, IEE Proc., 1311, PtD, pp.
5256.
16 Jalili, N., and Olgac, N., 1999, Multiple Delayed Resonator Vibration Absorber for MDOF Mechanical Structures, J. Sound Vib., 2234, pp. 567585.
17 Kolmanovski, V. B., and Nosov, V. R., 1986, Stability of Functional Differential Equations, Academic Press, London, Great Britain.
18 Niculescu, S-I., 2001, Delay Effects on Stability, Springer-Verlag.
19 Zhang, J., Knospe, C. R., and Tsiotras, P., 2001, Stability of Time-delay
Systems: Equivalence Between Lyapunov and Scaled Small-gain Conditions,
IEEE Trans. Autom. Control, 463, pp. 482 486.
20 Gu, K., and Niculescu, S.-I., 2000, Additional Dynamics in Transformed
Time-Delay Systems, IEEE Trans. Autom. Control, 453, pp. 572575.
21 Park, P., 1999, A Delay-Dependent Stability Criterion for Systems with Uncertain Time-Invariant Delays, IEEE Trans. Autom. Control, 44, pp. 876
877.
22 Olgac, N., and Sipahi, R., 2002, An Exact Method for the Stability Analysis
of Time Delayed LTI Systems, IEEE Trans. Autom. Control, 475, pp. 793
797.
23 Cooke, K. L., and van den Driessche, P., 1986, On Zeroes of Some Transcendental Equations, Funkcialaj Ekvacioj, 29, pp. 7790.
24 Thowsen, A., 1981, The Routh-Hurwitz Method for Stability Determination
of Linear Differential-Difference Systems, Int. J. Control, 335, pp. 991
995.
25 Thowsen, A., 1981, An Analytic Stability Test for a Class of Time-Delay
Systems, IEEE Trans. Autom. Control, 263, pp. 735736.
26 Thowsen, A., 1982, Delay-independent Asymptotic Stability of Linear Systems, IEE Proc., 29, pp. 7375.
27 Seto, K., 1995, Structural Modeling and Vibration Control. Internal Report,
Nihon University. Department of Mechanical Engineering, College of Science
and Technology, Nihon University, 1-8-14 Kanda Surugadai Chiyoda-ku, Tokyo, 101-8308, Japan.
Gaetan Kerschen
e-mail: g.kerschen@ulg.ac.be
Jean-Claude Golinval
Vibrations & Identification des Structures,
Department of Aerospace,
Mechanics and Materials,
University of Lie`ge,
Chemin des Chevreuils 1 (B52),
B-4000 Liege, Belgium
Francois M. Hemez
Engineering Science & Applications Division,
ESA-WR, Mail Stop P946,
Los Alamos National Laboratory,
Los Alamos, New Mexico 87545
e-mail: hemez@lanl.gov
Introduction
The importance of diagnosing, identifying and modelling nonlinearity has been recognized for a long time, e.g., for the design
of shock absorbers and engine mounts. The identification of nonlinear systems began in 1979 with the introduction of the restoring
force surface RFS method by Masri and Caughey 1. An
equivalent method, referred to as force-state mapping, was proposed independently by Crawley, Aubert and ODonnel 2,3.
Since then, numerous methods were proposed. It is not our intention to review all the methods available but rather to cite the most
popular techniques that have been considered during the last
twenty years.
The first application of the Hilbert transform was made in the
frequency domain 4. The time-domain Hilbert transform was
also utilized to solve an inverse problem 5,6. The use of the
Volterra series in the field of structural dynamics began in the late
1980s 7. NARMAX models consist of polynomials that include
various linear and nonlinear terms combining the inputs, outputs
and past errors and were introduced by Leontaritis and Billings
8,9. Another area of signal processing that has gained importance in studying nonlinear systems deals with higher-order spectra 10,11. These are a natural extension of the ordinary linear
spectral analysis. For a detailed description of all these techniques, the reader is referred to reference 12.
The development of frequency response function-based approaches has received increasing attention in recent years. The
reverse path technique has been proposed by Rice and Fitzpatrick
13 and applied to simulated and experimental data 14,15. The
conditioned reverse path formulation 16 extends the application
of the reverse path algorithm to systems characterized by nonlinearities away from the location of the applied force. This method
exploits the spectral conditioning techniques introduced by Bendat 17. A related series of papers by Adams and Allemang also
develop the frequency response function-based approaches
18,19.
Finally, it is worth pointing out that there has also been a
Contributed by the Technical Committee on Vibration and Sound for publication
in the JOURNAL OF VIBRATION AND ACOUSTICS. Manuscript received May 2002;
Revised January 2003. Associate Editor: M. I. Friswell.
Model Fitting
Model fitting generally refers to the calibration of model coefficients given a sequence of points (t k ;y k ) in the design space.
It is assumed that a model is available:
yM ;t
(1)
(2)
(3)
k1 . . . m
x k k x T
(4)
k1 . . . N
e Tk e k e T e
(5)
in which case the best, linear, unbiased estimator of the coefficients is provided by:
X T X 1 X T y
(6)
y1
y2
]
yN
;X
x 1,1
x 1,2
x 2,1
x 2,2
x N,1
x N,2
x 1,m
x 2,m
x N,m
(7)
(8)
1 1 T 1
X T W 1
X W ee y
ee XW bb
(9)
(10)
k1 . . . N
y k x Tk 2
(11)
Note that the likelihood function 11 is similar to Eq. 5 previously discussed. Other functions can be used, in particular the
Bayesian objective 8, as well as the many objective functions
commonly used in test-analysis correlation and model updating
33.
Once the likelihood of a particular model has been estimated,
the posterior probabilities of the models effects can be updated
according to the Bayes Theorem that states that the posterior probability PDF( y) is equal to the likelihood function L(y ) multiplied by the prior probability PDF( ) and divided by the probability of the data PDF(y):
PDF y
L y PDF
PDF y
(12)
The main difference between the two is that the Gibbs algorithm
samples one direction of the design space at a time, which makes
for simpler numerical implementation. Figure 4 illustrates the difference between MCMC and Gibbs sampling. It pictures two random walks from the lower left corner (x0;y0) to the upper
right corner (x1;y1). A constraint is enforced that prevents
the 30 points drawn in both sequences from being repeated and
from moving backwards. Pentagram symbols show a sequence of
Gibbs samples while hexagram symbols picture a realization of
the MCMC chain. In the former case, the solution is advanced in
one direction at a time whereas the MCMC chain randomly advances the solution in the two dimensions simultaneously.
4
Fig. 3 Concept of random walk optimization
Numerical Application
(13)
x 1 sin t
x 2 cos t
x 3 sin 2t
x 4 cos 2t
x 5 sin 3t
x 6 cos 3t
(14)
(15)
Effects
Posterior Probability
1
2
3
4
5
x 2 , x 3 , x 4x 5
x 2 , x 3 , x 4 x 5 , x 10
x 2 , x 3 , x 4x 5 , x 3x 4
x 2 , x 3 , x 4 x 5 , x 2 x 10
x 2 , x 3 , x 4x 5 , x 2x 5
52.0%
3.0%
2.0%
2.0%
2.0%
Experimental Application
1
2
3
4
0.5
2.0
4.0
8.0
Mass 1 2.2 kg
volt
volt
volt
volt
Test
Test
Test
Test
1
2
3
4
Mass 2 5.8 kg
Test 5
f x,x
(16)
(17)
i0 j0
i jx
i j
(18)
(19)
100
N 2y k1 . . . N
y k x Tk 2
(20)
where, to comply with notations introduced in Section 2, y k represents the available restoring force data and y is the standard
deviation of data y k . The vector collects the coefficients k l , c l
and k nl , assuming that the exponent is known and equal to 1.5,
and the vector x k collects the corresponding effects in Eq. 19.
The exercise of identifying the most appropriate model form is
now repeated with the Bayesian model screening. First, three effects are defined in agreement with Eq. 19. They are the linear
stiffness x 12 , linear damping x 12 and nonlinear stiffness
x 12 sign(x 12). The corresponding coefficients are denoted by k l ,
c l and k nl , as before. Second, model-forming rules are defined
which are that main effects and linear interactions between the
main effects are allowed. This means that a total of six effects
leading to sixty two different model forms are allowed. Such combinatorial complexity is trivial compared to the example discussed
in Section 4. The complexity here stems from the fact that real
data sets are analyzed with all the risk of erroneous identification
caused by noisy measurements and signal conditioning issues.
Because the exponent is unknown, the Bayesian model
screening is repeated for several assumed values of . The value
that leads to the smallest MSE is retained. Repeating model
screening could become CPU-time intensive if long MCMC
chains are requested for each analysis. For this application, an
initial chain of length 50 is dedicated to burn-in and a chain of
length 300 is requested for the optimization. It has been verified
that requesting more samples does not improve the quality of the
final results. Figure 9 shows the evolution of the MSE as a function of the exponent . The minimum value is obtained for
kl
cl
k nl
Value
6
1.09 10
183.44
8.52 107
1.5
Units
N/m
N.sec/m
N/m1.5
Unitless
Difference*
Value
6
1.12 10 N/m
198.19 N.sec/m
9.07 107 N/m1.5
1.5
2.75%
8.04%
6.46%
0.00%
Effects
1
2
3
4
5
Posterior
Probability
86.0%
7.0%
3.0%
2.6%
0.3%
It was pointed out previously that only four of the five data sets
are considered during the identification. These are Tests 1, 2, 4
and 5. The remaining data set Test 3, 2.2 kg load mass, 4.0 Volt
level is exploited to assess the predictive accuracy of the identified model.
Although it might not yet be the state of the practice in structural system identification, many authors, among whom we cite a
recent discussion of model validation in Reference 36, have emphasized that identified models should be independently validated.
It essentially means that independent experiments or data sets
should be used for model screening and parametric calibration, on
one hand, and model validation and predictive accuracy assessment, on the other hand. The predictive accuracy of a model cannot be objectively assessed over the operational range of interest
as long as the independence between training data and validation
data is not met.
Here, data sets collected during Tests 1, 2, 4 and 5 are used for
model screening and system identification while the data collected
during Test 4 are used for model validation. Essentially, Eq. 19
is evaluated with the coefficients of Table 5 to predict the restoring force. Displacement and velocity time series in Eq. 19 are
estimated from numerical integration of the measured acceleration
signals. The evolution of the predicted restoring force versus time
is then compared to the true restoring force measured during
Test 3. The true restoring force is estimated directly from acceleration measurements, as shown in Eq. 17. In Fig. 11, the two
JULY 2003, Vol. 125 395
References
Fig. 11 Comparison between measured and predicted restoring forces 4 Volt level
Conclusion
Acknowledgments
The authors acknowledge the VTT Technical Research Center
of Finland for sharing the Dynamic properties of resilient
mounts benchmark data. The authors are very grateful to Mandy
Cundy, technical staff member at Los Alamos National Laboratory, for helping them with the numerical implementation. Dr.
Kerschen is supported by a grant from the Belgian National Fund
for Scientific Research FNRS, which is gratefully acknowledged.
396 Vol. 125, JULY 2003
1 Masri, S. F., and Caughey, T. K., 1979, A Non-parametric Identification Technique for Non-linear Dynamic Problems, ASME J. Appl. Mech., 46, pp.
433 447.
2 Crawley, E. F., and Aubert, A. C., 1986, Identification of Nonlinear Structural
Elements by Force-state Mapping, AIAA J., 24, pp. 155162.
3 Crawley, E. F., and ODonnel, 1986, Identification of Nonlinear System Parameters in Joints using the Force-state Mapping Technique, AIAA Paper
86 1013, pp. 659 667.
4 Simon, M., and Tomlinson, G. R., 1984, Use of the Hilbert Transform in
Modal Analysis of Linear and Non-linear Structures, J. Sound Vib., 96, pp.
421 436.
5 Feldman, M., 1994, Nonlinear System Vibration Analysis using the Hilbert
Transform-I. Free Vibration Analysis Method FREEVIB, Mech. Syst. Signal Process., 8, pp. 119127.
6 Feldman, M., 1994, Nonlinear System Vibration Analysis using the Hilbert
Transform-II. Forced Vibration Analysis Method FORCEVIB, Mech. Syst.
Signal Process., 8, pp. 309318.
7 Gifford, S. J., and Tomlinson, G. R., 1989, Recent Advances in the Application of Functional Series to Non-linear Structures, J. Sound Vib., 135, pp.
289317.
8 Leontaritis, I. J., and Billings, S. A., 1985, Input-Output Parametric Models
for Non-linear Systems: Part IDeterministic Non-linear Systems, Int. J.
Control, 41, pp. 303328.
9 Leontaritis, I. J., and Billings, S. A., 1985, Input-Output Parametric Models
for Non-linear Systems: Part IIStochastic Non-linear Systems, Int. J. Control, 41, pp. 329344.
10 Bendat, J. S., and Piersol, A. G., 1980, Engineering Applications of Correlation and Spectral Analysis, John Wiley & Sons, New-York U.S.A..
11 Choi, D., Chang, J., Stearman, R. O., and Powers, E. J., 1984, Bispectral
Interaction of Nonlinear Mode Interactions, Proceedings of the 2nd International Modal Analysis Conference, Orlando, FL, pp. 602 609.
12 Worden, K., and Tomlinson, G. R., 2001. Non-linearity in Structural Dynamics: Detection, Identification and Modelling, Institute of Physics Publishing,
Bristol, Philadelphia PA.
13 Rice, H. J., and Fitzpatrick, J. A., 1991, A Procedure for the Identification of
Linear and Non-linear Multi-degree-of-freedom Systems, J. Sound Vib., 149,
pp. 397 411.
14 Rice, H. J., and Fitzpatrick, J. A., 1991, The Measurement of Non-linear
Damping in Single-degree-of-freedom Systems, ASME J. Vibr. Acoust., 113,
pp. 132140.
15 Esmonde, H., Fitzpatrick, J. A., Rice, H. J., and Axisa, F., 1992, Modelling
and Identification of Non-linear Squeeze Film Dynamics, J. Fluids Struct., 6,
pp. 223248.
16 Richards, C. M., and Singh, R., 1998, Identification of Multi-Degree-ofFreedom Non-linear Systems Under Random Excitations by the Reverse-Path
Spectral Method, J. Sound Vib., 213, pp. 673708.
17 Bendat, J. S., 1990, Nonlinear System Analysis and Identification from Random Data, John Wiley & Sons, New-York U.S.A..
18 Adams, D. E., and Allemang, R. J., 1999, Characterization of Nonlinear
Vibrating Systems using Internal Feeback and Frequency Response Modulation, ASME J. Vibr. Acoust., 121, pp. 495500.
19 Adams, D. E., and Allemang, R. J., 2000, A Frequency Domain Method for
Estimating the Parameters of a Non-Linear Structural Dynamic Model through
Feedback, Mech. Syst. Signal Process., 14, pp. 637 656.
20 Hemez, F. M., and Doebling, S. W., 2001, Review and Assessment of Model
Updating for Non-linear, Transient Dynamics, Mech. Syst. Signal Process.,
15, pp. 4574.
21 Lenaerts, V., Kerschen, G., and Golinval, J.-C., 2001, Proper Orthogonal
Decomposition for Model Updating of Non-linear Mechanical Systems,
Mech. Syst. Signal Process., 15, pp. 31 43.
22 Dippery, K. D., and Smith, S. W., 1998, An Optimal Control Approach to
Nonlinear System Identification, Proceedings of the 16th International Modal
Analysis Conference, Santa Barbara CA, pp. 637 643.
23 Bendat, J. S., 1986, Random Data: Analysis and Measurement Procedures,
John Wiley-Interscience, New York NY, second edition.
24 Chipman, H., Hamada, M., and Wu, C. F. J., 1997, A Bayesian Variable
Selection Approach for Analyzing Designed Experiments With Complex Aliasing, Technometrics, 39, pp. 372381.
25 Wu, C. F. J., and Hamada, M., 2000, Experiments: Planning, Analysis, and
Parameter Design Optimization, Wiley, New-York, NY.
26 George, E. I., and McCulloch, R. E., 1993, Variable Selection via Gibbs
Sampling, Journal of the American Statistical Society, 88, pp. 881 889.
27 Carlin, B. P., and Chib, S., 1995, Bayesian Model Choice via Markov Chain
Monte Carlo, Journal of the Royal Statistical Society Series B, 77, pp. 473
484.
28 Kulczycki, P., 2001, An Algorithm for Bayes Parameter Estimation, ASME
J. Dyn. Syst., Meas., Control, 123, pp. 611 614.
29 Yuen, K. V., Katafygiotis, L. S., Papadimitriou, C., and Mickleborough, N. C.,
2001, Optimal Sensor Placement Methodology for Identification With Unmeasured Excitation, ASME J. Dyn. Syst., Meas., Control, 123, pp. 677 686.
30 Cybenko, G., 1989, Approximation by Superpositions of a Sigmoidal Function, Mathematics of Control, Signals, and Systems, 2, pp. 303314.
31 Walter, E., and Pronzato, L., 1997, Identification of Parametric Models From
Experimental Data, Springer-Verlag, Berlin Germany.
32 Hasselman, T. K., Anderson, M. C., and Wenshui, G., 1998, Principal Components Analysis for Nonlinear Model Correlation, Updating and Uncertainty
Evaluation, Proceedings of the 16th International Modal Analysis Conference, pp. 664 651.
33 Mottershead, J. E., and Friswell, M. I., 1993, Model Updating in Structural
Dynamics: A Survey, J. Sound Vib., 1622, pp. 347375.
34 http://www.ulg.ac.be/ltas-vis/costf3/costf3.html, portal Web site of the
COST-F3 action.
35 Kerschen, G., Lenaerts, V., Marchesiello, S., and Fasana, A., 2001, A Frequency Domain vs. a Time Domain Identification Technique for Non-linear
Parameters Applied to Wire Rope Isolators, ASME J. Dyn. Syst., Meas.,
Control, 123, pp. 645 650.
36 Doebling, S. W., 2002, Structural Dynamics Model Validation: Pushing the
Envelope, Proceedings of the International Conference on Structural Dynamics ModellingTest, Analysis, Correlation and Validation, Madeira Island
Portugal.
Journal of
Vibration
and Acoustics
Technical Briefs
standard model of the DVA. In 1928, Ormondroyd and Den Hartog 2 pointed out the damping of the DVA has an optimum value
so as to minimize the maximum amplitude response of the primary system. Such optimization criterion is now known as H
optimization. In 1932, Hahnkamm 3 derived the optimum tuning
parameter opt of the DVA using special points in which all re-
Toshihiko Asami
Mem. ASME
Professor, Department of Mechanical Engineering,
Himeji Institute of Technology, 2167 Shosha, Himeji,
Hyogo 671-2201, Japan
e-mail: asami@mech.eng.himeji-tech.ac.jp
Osamu Nishihara
Mem. ASME
Associate Professor, Department of Systems Science,
Kyoto University, Yoshida-Honmachi, Sakyo-ku, Kyoto
606-8501, Japan
e-mail: nishihara@i.kyoto-u.ac.jp
3f
x1
x1
xst
f 0 /k 1
x1
x1
1xst
f 0 /k 1
x1
21xst
x1
f 0 /k 1
Relative response
x1
x0
x 1
x1
1x 0
x0
x 1
21 x 0
x1
x0
21 y 1
1 y1
2
x 0
x0
1y 1 1 y 1
x 0
x0
y1
x0
Introduction
Fig. 2 Systems with hysteretic damped DVA a Force excitation system b Motion excitation system
Table 2 Expressions for the transfer functions and definition of the symbols of the viscous
damped systems shown in Fig. 1
No.
Transfer function
N f
D
x1
f 0 /k1
1f
x1
x0
N a
D
y1
x0
N r
D
1 k 1 /m 1
natural frequency
of primary system
2 k 2 /m 2
x1
y 1 x 1 x 0
Expressions
N f ()( 2 2 ) 2 (2 2 ) 2
N a () 1(2 1 ) 2 ( 2 2 ) 2 (2 2 ) 2
N r () (1 ) 2 2 2 4 (1 ) 2 (2 2 ) 2 4
D() (1 2 )( 2 2 ) 2 2 4 1 2 2 2
(2) 2 ( 2 2 ) 1 1(1 ) 2 2 2
/ 1
m 2 /m 1
mass ratio
natural frequency
of absorber
2 /1
natural frequency
ratio
absolute displacement
of primary system
c1
1
2m11
damping ratio
of primary system
relative displacement
between primary system
and foundation
c2
2
2m22
damping ratio
of absorber
Figure 3a shows some frequency response curves of the undamped primary system ( 1 0) with an optimally tuned DVA
JULY 2003, Vol. 125 399
Table 3 Approximate solutions to the H optimization in the design of viscous damped absorber attached to undamped primary
system derived from the fixed-points method
No.
1
Transfer
function
x1
x0
N a
D
x 1
x1
1x 0
x0
x 1
21 x 0
x1
x0
21 y 1
1 y1
2
x 0
x0
1y 1 1 y 1
x 0
x0
y1
x0
N r
D
1
1
1
1
2
2
1
1
2
2
1
1
8 1
1
4 2
2
2
2 1
4 1 2
8 1
43 85
2 1 2
1
2
1
4 8
3
8 1
2424 5 2 4 85
1
1
1
1
Abscissas P and Q
of the fixed points
4 1
2 1
Height of the
fixed points
2 1
Table 4 Exact solutions to the H optimization in the design of viscous damped absorber attached to undamped primary system derived from our method
No.
1
Transfer
function
x1
x0
2
85
21 x 0
21 y 1
x 0
1
r1
r2
16
1
r3
r4
16
3 6480 27 2
x 1
1
85
2
167 6416 26 2
3
1
2r 1 16r 21 4r 1 2
6
p 1 q 1
p 1 q 1
2q 1 p 1
2q 1 p 1
4 q 0
4 q 0
1 2 1 2
1
2r 1
p 1 q 1
p 1 q 1
1
2
1
2 1
1
169 4 43
6
Maximum amplitude
89 4 43
2
2 1
1625 2 6416 26 2
3 3227
1
1r 21
2 1
1r 1
2r 1 16r 21 4r 1 2
1
3 3227
1
2 1 2
6488 33 2
43 c 0
2
c 0 6416 26 2
1
1
1y 1
x 0
y1
x0
2 1623 9 2 2 2 43
x 1
1x 0
1
1
3
8 1
89 4 43
1
1
3
2 1
Table 5 Approximate solutions to the H optimization in the design of hysteretic damped absorber attached to undamped
primary system derived from the fixed-points method
No.
1
Transfer
function
Optimum tuning
opt
N a
D
x1
x0
x 1
x1
1x 0
x0
x 1
21 x 0
21 y 1
x 0
x1
x0
1 y1
2 x0
1y 1 1 y 1
x 0
x0
y1
x0
N r
D
1
1
1
1
1
1
2
2
1
1
2
2
1
1
1
1
2
2
32 2 2
2450 30 2 5 3 2 55 2 85
2
1
2
3 23 2 2 1
6 2 2
2
3 4 8
8
32 2 2
2
Abscissas P and Q
of the fixed points
Height of the
fixed points
2
1
43 85
2 1 2
2
1
2 1
4 8
4 1
2 1
Coefficients
y1
x0
1 y 1
x0
x 0
21 y 1
21 x 0
x 1
x1
1x0
1 2
q1 1 22opt 2 1 22opt
2
2
q1 1 22opt2 1
2a
b b 2 4ac
3
2
1
q3
2 1
3
2
2a
b b 2 4ac
2 1
1
q3
4 1
2 2
Maximum amplitude
3 8
2 2
3 8
p 2
2p 1
2 1 2 12
3
p
2 0
c (c 0 q 3 c 1 q 2 c 2 qc 3 )
b2(b 0 q 3 b 1 q 2 b 2 qb 3 )
aa 0 q 3 a 1 q 2 a 2 qa 3
p1
p 2 2 1 2
p 1 2(1 2 )3p 0
p 0 2 2 (1 ) 2 1/3
Nomenclature
1
1
2 2 2
4
q1 1 22opt 2 1 1 22opt
q1 1 22opt2
1
1
x1
x0
T.F.
No.
Table 6 Exact solutions to the H optimization in the design of hysteretic damped absorber attached to undamped primary system derived from our method
for 1 and 6
ae 0 e 1 e 23
be 3 3e 1 e 2
ce 1 e 2 3e 23
2 33 2 2 3 4 2 1 21 q 1 1 22opt
6 1
1 1 21
6 1 1 22opt
1 1
x1
x0
y1
x0
Expressions
x1
f 0 /k1
1f
for 1 f and 1
for 6
b b 2 4ac
2a
2 33 2 2 3 4 2 1 21 q 1
e 0 4 2 q 21 8 q 1 36 6 2 6 3 4 (1 )(3 2 ) 21 4(36 6 2 6 3 4 ) 2
16 2 (1 )(96 12 2 18 3 3 4 4 5 ) 21 32 4 (1 ) 2 (3 2 ) 41
e 1 (1 ) 2 q 1 2 (3 ) 2 4 2 (1 ) 21 q 1 2 (36 2 )4 2 (1 ) 21
e 2 2(1 ) q 21 q 1 318 6 2 6 3 4 2 2 (1 )(3 ) 21
2 (3 )(915 18 2 6 3 3 4 5 )4 2 (1 )(69 3 2 9 3 7 4 ) 21 8 4 (1 ) 2 (3 ) 41
e 3 12 (1 )(1 ) 2 1 q 1 2(33 2 2 3 )4 2 (1 ) 21
p 0 (3 ) 4 4(1 )(3 )(93 2 2 ) 21 4 2 (1 ) 2 41
6
p 1 (3 ) 3(1 )(3 ) 3 (99 21 2 5 3 ) 21 12 2 (1 ) 2 (453 2 2 3 ) 41 8 4 (1 ) 3 61
q 0 2 p 1 3(1 ) 2 1 3 2 (1 ) 21 (3 ) 3 8 2 21 3
2p0
q1 1/3 q 1/3
0
q0
6 1 1 21
Primary
damping 1
q 1 2 36 2 2 1 3 2 21
for 1 f
6 1
q 1 2 36 2 2 1 3 2 21
Transfer function
Table 7 Exact solutions to the optimization problem in the design of hysteretic damped absorber attached to damped primary system
0.00
0.01
0.02
0.05
0.10
0.15
0.20
0.50
Optimum
tuning opt
Optimum loss
factor 2opt
Maximum
amplitude
0.909091
0.907941
0.906801
0.903437
0.898000
0.892754
0.887677
0.859966
0.393700
0.395643
0.397566
0.403232
0.412355
0.421124
0.429584
0.475493
4.69042
4.51256
4.34740
3.91648
3.36195
2.94908
2.63249
1.68819
Fig. 6 Absolute displacement response of the damped primary system with optimally tuned and damped DVA
Concluding Remarks
The last unsolved problem is the H optimization of the viscous damped dynamic vibration absorber attached to damped primary systems. However, the algebraic approach is probably impossible to this system. We proposes simultaneous equations for
calculating numerically the exact solution 6.
References
1 Frahm, H., 1911, Device for Damping Vibrations of Bodies, U.S. Patent,
No. 989, 958, pp. 3576 3580.
2 Ormondroyd, J., and Den Hartog, J. P., 1928, The Theory of the Dynamic
Vibration Absorber, ASME J. Appl. Mech., 507, pp. 922.
3 Hahnkamm, E., 1932, Die Dampfung von Fundamentschwingungen bei veranderlicher Erregergrequenz, Ing. Arch., 4, pp. 192201, in German.
4 Brock, J. E., 1946, A note on the Damped Vibration Absorber, ASME J.
Appl. Mech., 134, p. A-284.
5 Den Hartog, J. P., 1956, Mechanical Vibrations, 4th ed., McGraw-Hill, New
York.
6 Nishihara, O., and Asami, T., 2002, Closed-Form Solutions to the Exact
Optimizations of Dynamic Vibration Absorbers Minimizations of the Maximum Amplitude Magnification Factors, ASME J. Vibr. Acoust., 1244, pp.
576 582.
7 Korenev, B. G., and Reznikov, L. M., 1993, Dynamic Vibration Absorbers, p.
11, John Wiley & Sons, New York.
where M, G, K, M , G , and K are linear, real, spatial differential or matrix operators, u(x,t) is the system response, x is the
spatial coordinate, t is time, 1 is a small parameter, p(t) is a
periodic function in time, and a comma indicates partial differentiation. The operators M, G, and K are associated with a complex
inner product , and enjoy the properties
M 0
(3)
d n t n
(4)
n1, . . .
d n,t n
n1, . . .
i nd n n
(5)
The complex eigenfunctions of positive definite, linear, gyroscopic systems possess a number of properties that make them
advantageous and efficient in analytic methods. First and foremost, these eigenfunctions can be used to decouple system equations, thus providing exact modal solutions even for continuous
systems 13. These eigenfunctions can also be used in perturbation studies 4. Unfortunately, these eigenfunctions do not appear to offer meaningful advantages for numerical methods 57.
In this Tech Brief, we demonstrate that complex gyroscopic
eigenfunctions are particularly efficient for determining the stability boundaries of parametrically excited systems using the method
of multiple scales. This efficiency is derived from two properties:
decoupling and reduction of order. Our contribution here is one of
presentation and the examples; we make no claim of originality
for the method which has been used in previous parametric stability studies 8 10. Our aim here is to present the analysis in as
general and simple a manner as possible so that the efficiency of
the procedure is obvious and its application is accessible to those
unfamiliar with it. To that end, the procedure is used to quickly
analyze three parametrically excited systems. The efficiency of the
procedure is evident when our brief analyses are compared with
the alternative approaches given in the literature.
(2)
n1,2, . . .
Introduction
K * K
n m M n , m K n , m nm
G * G
u ,t
M * M
K0
Anthony A. Renshaw
Associate Professor, Dept. of Mechanical Engng.
Columbia University, New York, NY 10027
(1)
d n,t G nm
d n K nm
d n,tt M nm
(6)
M n , m , etc.
where M nm
The parametric instability boundaries are determined from Eq.
6 by applying the method of multiple scales. Define the time
scales T 0 t and T 1 t and let d n (t)D n0 (T 0 ,T 1 )
D n1 (T 0 ,T 1 ) . . . . Substitution into Eq. 6 gives the leading
order equation
D m0,T 0 i m D m0 0
(7)
D m0 A m0 T 1 e i m T 0
(8)
whose solution is
A
n
n0 e
inT0
nm
(9)
where
i n G nm
K nm
nm 2n M nm
(10)
A
n
1
i e iT 0
2 m
n0 e
inT0
nm
(12)
k,n
A kn0 e i kn T 0 R kn r 0 R jm r 0 e i jk T 0
(13)
1
i A e i T 1 mn 0
2 n m0
(14)
(21)
The detuning parameter for a two mode secular combination of
terms on the right hand side is
jm kn jk
(22)
1
A m0,T 1 i m A n0 e i T 1 nm 0
2
(23)
(24)
(11)
A n0,T 1
A jm0,T 1 e i jm T 0
i jm
When the first term on the right hand side is the only secular term,
only bounded solutions result. We therefore look for combinations
of two secular terms. Define the detuning parameter by
m n
D jm1,T 0 i jm D jm1
A n0 a n e i T 1
(15)
A jm0 a jm e T 1
A kn0 a kn e i T 1
(25)
(16)
(17)
km e ik R km r
(18)
where (r, ) are polar coordinates and k and m are integers and
m0. This describes a mode with k nodal diameters and m
nodal circles. We do not need to know the actual forms of M, G,
K, R km , and km in order to analyze stability; these are only
needed if explicit values of the frequencies and boundaries are
required.
Here we consider a perturbation comprising a stationary, concentrated spring transversely loading the rotating plate. In the rotating frame of reference, we have p1 and
K
1
rr 0 t
r
(19)
jm K e ik R kn ,e i j R jm R kn r 0 R jm r 0 e i jk t
K kn
(20)
Equation 9 is therefore
406 Vol. 125, JULY 2003
The most important instability produced by the spring is divergence instability of a mode with a fixed number of nodal diameters. For this case, we let jm 0 0, kn 0 , jk, and
R jm (r 0 )R kn (r 0 ). The conditions for instability are then
4 0 R 2jm r 0 0
(27)
0 j
and
0 2 0 R 2jm r 0 j
(28)
K u
q1
References
1
rr 0 q u r 0 , q ,t
r
u r 0 , q1 ,tT
(29)
i jm
q1 k,n
kn r 0 R jm r 0 e
i jk q
A kn0 e i jm T 0 1e 2 i k kn /N
(30)
(31)
whose solution is
A jm0 exp i jm T 1 R 2jm r 0 N 1exp 2 i j jm /N
(32)
Therefore, the response is stable only when
sgn jm sin 2 j jm /N 0
(33)
jm jN
Timothy A. Brungart
e-mail: tab@wt.arl.psu.edu
Eric T. Riggs
e-mail: etr@wt.arl.psu.edu
The Pennsylvania State University, Applied Research
Laboratory, PO Box 30, State College, PA 16804-0030
Concluding Comments
(34)
1 Meirovitch, L., 1975, A Modal Analysis for the Response of Linear Gyroscopic Systems, ASME J. Appl. Mech., 422, pp. 446 450.
2 Wickert, J. A., and Mote, C. D., Jr., 1990, Classical Vibration Analysis of
Axially Moving Continua, ASME J. Appl. Mech., 57, pp. 738 744.
3 Renshaw, A. A., 1997, Modal Decoupling of Systems Described by Three
Linear Operators, ASME J. Appl. Mech., 64, pp. 238 240.
4 Hryniv, R. O., Lancaster, P., and Renshaw, A. A., 1999, A Stability Criterion
for Parameter Dependent Gyroscopic Systems, ASME J. Appl. Mech., 66, pp.
660, 664.
5 Jha, R. K., and Parker, R. G., 2000, Spatial Discretization of Axially Moving
Media Vibration Problems, ASME J. Vibr. Acoust., 122, pp. 290294.
6 Lee, K.-Y., and Renshaw, A. A., 1999, Solution of the Moving Mass Problem
Using Complex Eigenfunction Expansions, ASME J. Appl. Mech., 67, pp.
823 827.
7 Lee, K.-Y., and Renshaw, A. A., 2002, A Numerical Comparison of Alternative Galerkin Methods for Eigenvalue Estimation, J. Sound Vib., 2532, pp.
359372.
8 Parker, R. G., and Lin, Y., 2001, Parametric Instability of Axially Moving
Media Subjected to Multifrequency Tension and Speed Fluctuations, ASME
J. Appl. Mech., 68, pp. 4957.
9 Mockensturm, E. M., Perkins, N. C., and Ulsoy, A. G., 1996, Stability and
Limit Cycles of Parametrically Excited, Axially Moving Strings, ASME J.
Vibr. Acoust., 118, pp. 346 351.
10 Lee, K.-Y., and Renshaw, A. A., 2002, Stability Analysis or Parametrically
Excited Systems Using Spectral Collocation, J. Sound Vib., to appear.
11 Iwan, W. D., and Moeller, T. L., 1976, The Stability of a Spinning Elastic
Disk with a Transverse Load System, ASME J. Appl. Mech., 43, pp. 485
490.
12 Renshaw, A. A., and Mote, Jr., C. D., 1992, Absence of One Nodal Diameter
Critical Speed Modes in an Axisymmetric Rotating Disk, ASME J. Appl.
Mech., 59, pp. 687 688.
13 Mote, Jr., C. D., 1970, Stability of Circular Plates Subjected to Moving
Loads, J. Franklin Inst., 2904, pp. 329344.
14 Renshaw, A. A., and Mote, Jr., C. D., 1996, Local Stability of Gyroscopic
Systems Near Vanishing Eigenvalues, ASME J. Appl. Mech., 63, pp. 116
120.
15 Tian, J. F., and Hutton, S. G., 2001, Cutting-Induced Vibration in Circular
Saws, J. Sound Vib., 2425, pp. 907922.
16 Stone, E., and Askari, A., 2002, Nonlinear Models of Chatter in Drilling
Processes, Dynamical Systems, 171, pp. 65 85.
Introduction
Propulsors used in marine applications are a source of tonal and
broadband vibration and noise. Unsteady forces at blade rate and
harmonic frequencies are generated by the interaction of rotor
blades with mean inflow spatial nonuniformities, due to upstream
obstructions, such as control surfaces, stators and support struts.
Broadband unsteady forces are generated by the interaction of
rotor blades with turbulence, formed in the boundary layers and
wakes of upstream surfaces, that is ingested into the rotor. The
tonal and broadband hydrodynamically-induced rotor unsteady
forces are transmitted to the watercraft through the rotor hub and
drive shaft. This results in watercraft vibration and noise. Blake
1 provides an in-depth analysis of these vibration and noise generating mechanisms.
Vibration and noise control is achieved by reducing the unsteady forces generated at the source and/or by inhibiting the
transmission of the unsteady forces to the watercraft. An example
of the latter type of control is the use of resilient mounts that
isolate the watercraft from vibrating components such as power
generating equipment and pumps 2. When power generating
equipment and pumps are mounted independently, flexible shaft
couplings are often used to accommodate shaft misalignment and
provide isolation of component torsional vibrations 2. Similar
isolation techniques can be incorporated directly into the rotor for
vibration and noise control.
Approach
A vibration isolation mount incorporated into the rotor hub can
attenuate the unsteady forces generated on the rotor blades before
they are transmitted to the shaft and watercraft. Such a mount is
shown schematically in Fig. 1. A single layer of elastomer separates the outer blade ring from the inner hub. In the event that the
rotors axial or torsional load exceeds the mount design specifications or the elastomer fails, mating axial and circumferential stops
are machined into the rotor. The stops are not engaged under
normal operating loads since this would create a shorting path
around the mount.
inder is driven through the base plate with the attached shaker 3.
The transmissibility can also be shown, using the transmissibility
theorem, to equal the absolute value of the complex ratio of the
force transmitted to the cylinder to the force driving the mass 3.
This is illustrated in Fig. 3 where the foundation represents the
cylinder, M is the mounted mass, V 1 and V 2 are the foundation
and mounted mass velocities, respectively, and F 1 and F 2 are the
forces acting on the foundation and mounted mass, respectively.
With a shaker driving the cylinder of the ECTF, the signals
from two phase and amplitude matched accelerometers, mounted
on opposite sides of the elastomer, were measured with a dynamic
signal analyzer. The accelerometer transfer functions were measured over a frequency range of 0 to 2 kHz in 2.5 Hz frequency
bands with 256 ensemble averages. Values of G and G for
mounts fabricated from Hapco Inc. Hapflex 581, 570, 560 and 540
elastomers were determined by fitting Eq. 1 to the measured
transmissibility. Hapflex 540 was found to be the softest of the
four elastomers considered and, as a result, to provide the greatest
isolation capability. Axial load/deflection curves were measured
on an Instron Corporation Series IX Automated Materials Testing
kG
G j
T
,
Z
(1)
kG
kG
j M
,
G
(2)
2rM
.
r c r M
(3)
Results
System with the Hapflex 540 in the ECTF. The data indicated that
Hapflex 540 also provided suitable stiffness, therefore, Hapflex
540 was used for the construction of the isolated rotor.
In order to assess the static stiffness of the mount fabricated
from Hapflex 540, isolated rotor axial load/deflection curves were
also measured on an Instron Corporation Series IX Automated
Materials Testing System. From the curves, the mount stiffness
was determined to be approximately 1056 N/mm, 965 N/mm, and
863 N/mm at loading rates of 12.7 cm/min, 1.27 cm/min, and
0.127 cm/min, respectively. The mating stops machined into the
rotor are designed to engage when the thrust moves the blade row
forward approximately 0.32 cm at speeds just above the maximum speed of operation. Adequate clearance hub-to-blade tip is
available to permit such movement. Rotor isolation is not expected to affect the powering performance of the propulsor since
the blade tip clearance does not change significantly throughout
the rotors axial range of motion. An increase in blade tip clearance has an adverse effect on propulsor efficiency 4. The extent
of the performance degradation has been quantified in terms of
Isolated Rotor Static Transmissibility. The axial transmissibility of the mount embedded in the isolated rotor was measured
in both air and water in a manner similar to that employed with
the ECTF. The experimental set-up is shown in Fig. 4. A shaker
was mounted to the inner hub of the rotor and supported with
elastic shock chords from overhead. The outer blade ring is unconstrained in this set-up, as required by the transmissibility theorem. Two phase and amplitude matched accelerometers were
mounted opposite one-another across the elastomer and oriented
in the direction of the shaker-applied excitation. The measured
transmissibility, and transmissibility predicted from Eq. 1, with
G and G measured with the ECTF, are shown in Fig. 5. The
measured transmissibility for a nonisolated i.e., solid rotor is 1.0
which is also shown in Fig. 5. The frequency is nondimensionalized by the rotor blade passing frequency at the subject operating
condition (1BR).
The in-air measured transmissibility is within 3 dB of the
simple lumped parameter model Eq. 1 prediction up to a frequency of approximately 4BR. The lumped parameter model
tends to under-predict the transmissibility somewhat at higher frequencies. The transmissibility measured in water is within 3 dB of
the Eq. 1 prediction up to a frequency of approximately 1.7
BR. Equation 1 also under-predicts the in-water transmissibility at higher frequencies but to a greater extent than the in-air
results. Water loading is responsible for the reduction in mount
resonance frequency and corresponding modifications to the transmissibility for the in-water measurements compared to the in-air
measurements. It is reasonable to expect that the measured transmissibility will deviate from the lumped parameter-based prediction at frequencies where the rotor begins to exhibit distributed
system behavior. Structural modal analyses of the rotor, performed
in-air and in-water, indicate that the lowest order mode, the first
bending mode, occurs at frequencies of approximately 6.5BR in
air and 2.8BR in water. Thus, we attribute the deviations be-
Fig. 7 Typical shaft unsteady thrust spectrum measured with and without
rotor isolation
BR, the Eq. 1 predictions are well below the statically and
operationally measured transmissibility since the lumped parameter assumption inherent to Eq. 1 is violated. Resonances associated with the dynamometer and shaft assembly add to the difficulty of interpreting comparisons of the measured static
transmissibility to the transmissibility measured under operational
conditions. Even so, the Fig. 8 transmissibility comparisons show
that the isolated rotor attenuates the unsteady forces transmitted to
the shaft up to approximately 15 dB in level compared to an
identical non-isolated rotor.
Conclusions
A vibration isolation mount has been incorporated into the hub
of a rotor used to propel a marine vehicle. The mount transmissibility is predicted reasonably well under static conditions with a
simple lumped parameter transmissibility model at frequencies
below the first bending mode of the rotor blades. Under operational conditions, the mount resonance is eliminated but otherwise
the measured transmissibility is comparable to the static measurements. The isolated rotor reduces the unsteady forces transmitted
to the shaft up to 15 dB in level compared to an identical nonisolated rotor.
Acknowledgment
This work was sponsored by the Office of Naval Research,
Code 333, and monitored by Dr. Kam Ng. Their support is gratefully acknowledged. The contributions of Dr. Jason T. Gomez and
Mr. Stephen J. Plunkett, of the Naval Undersea Warfare Center
Division Newport, in various aspects of this work, are acknowledged as well.
References
1 Blake, W. K., 1986, Mechanics of Flow-Induced Sound and Vibration, Volumes I and II, Academic Press, Inc., Orlando.
2 Skaistis, S., 1988, Noise Control of Hydraulic Machinery, Marcel Dekker,
New York.
3 Snowdon, J. C., 1968, Vibration and Shock in Damped Mechanical Systems,
John Wiley and Sons, New York.
4 Wisler, D. C., 1986, Advanced Compressor and Fan Systems, Fluid Dynamics of Turbomachinery, Gas Turbine Division, American Society of Mechanical Engineers, ASME Turbomachinery Institute, Ames, Iowa.
5 Lauchle, G. C., Billet, M. L., and Deutsch, S., 1989, High Reynolds Number
Liquid Flow Measurements, Lecture Notes in Engineering, 46, Frontiers in
Experimental Fluid Mechanics, M. Gad-el-Hak, ed., Springer-Verlag, Berlin,
pp. 95158.