Você está na página 1de 14

Vibration interaction in a multiple ywheel system

Jordan Firth, Jonathan Black


n
Air Force Institute of Technology, 2950 Hobson Way, Wright Patterson AFB, OH 45433, United States
a r t i c l e i n f o
Article history:
Received 14 July 2011
Received in revised form
19 November 2011
Accepted 28 November 2011
Handling Editor: G. Degrande
Available online 2 January 2012
a b s t r a c t
This paper investigates vibration interaction in a multiple ywheel system. Flywheels
can be used for kinetic energy storage in a satellite Integrated Power and Attitude
Control System (IPACS). One hitherto unstudied problem with IPACS is vibration
interaction between multiple unbalanced wheels. This paper uses a linear state-space
dynamics model to study the impact of vibration interaction. Specically, imbalance-
induced vibration inputs in one ywheel rotor are used to cause a resonant whirling
vibration in another rotor. Extra-synchronous resonant vibrations are shown to exist,
but with damping modeled the effect is minimal. Vibration is most severe when both
rotors are spinning in the same direction.
Published by Elsevier Ltd.
1. Introduction
Advanced ywheel technology has the potential to replace chemical batteries for energy storage in satellites. In an
Integrated Power and Attitude Control System (IPACS), ywheels could replace both the batteries and the momentum-
exchange attitude control system, offering signicant weight savings. Flywheels are more complex than batteries or
reaction wheels, however, and the complexity and technology immaturity have to date precluded any use of ywheel
energy storage in space.
Flywheel energy storage for satellites has been studied since Roes proposed the idea in 1961 [1]. The idea of combining
attitude control and energy storage functions into an IPACS began to appear in the 1970s in several technical papers from
the National Aeronautics and Space Administration (NASA)s Langley Research Center. Since then, ywheels have been
continually studied for space applications. NASA commissioned very detailed IPACS technical reports in 1974 and again in
1985 [2,3].
Flywheel safety for aerospace applications was addressed through the DARPA Flywheel Program and the DARPA
Flywheel Safety Program. These programs achieved tip speeds over 1300 m/s (3000 mph), and included investigations
rotor fatigue, composite materials, overspeed, and loss of vacuum. As a result ywheels have been proposed for use in a
variety of terrestrial applications, including transit busses, and an ANSI/AIAA Standard for Flywheel Rotor Assemblies for
Space Systems has been released [46].
NASA had plans to develop a Flywheel Energy Storage System (FESS) for the International Space Station (ISS) in the
early 2000s as a technology demonstration mission. Engineers at Glenn Research Center (GRC), with the help of other
schools and industry partners, developed and built several advanced ywheels to advance the state of the art for advanced,
composite-rotor, magnetically suspended ywheels. This led to a successful IPACS demonstration on an air bearing table in
2004, a rst for high-power, high-speed IPACS [7].
Contents lists available at SciVerse ScienceDirect
journal homepage: www.elsevier.com/locate/jsvi
Journal of Sound and Vibration
0022-460X/$ - see front matter Published by Elsevier Ltd.
doi:10.1016/j.jsv.2011.11.028
n
Corresponding author.
E-mail addresses: jordan.rth@us.af.mil (J. Firth), jonathan.black@at.edu (J. Black).
Journal of Sound and Vibration 331 (2012) 17011714
Unfortunately, the FESS program was canceled due to budget cuts before it could get off the ground. Since then, there
have been no rm plans for satellite IPACS [8].
Despite the lack of any successfully launched system, IPACS research has been highly benecial in developing ywheel
technology. High tensile strength carbon bers enable the creation of light and strong rotors. Actively controlled magnetic
bearings waste no energy due to friction and allow precise control of the rotor. Advanced brushless motor/generators
allow for very efcient energy storage. Good design and robust computer control of the bearings and motors enables
stability throughout the operating regime.
With more recent technology advances, some researchers are beginning to design IPACS for smaller satellites with
demanding mission proles. Lappas et al. discussed this in an article that contains the most recent comprehensive review
of IPACS history and literature [9].
One of the problems facing IPACS designers is keeping satellite vibrations to an acceptable level. To date, the study of
IPACS vibration has been limited; historically, most IPACS research has focused on control algorithms. IPACS models used
for control development assume rigid connections and perfectly balanced single-degree-of-freedom rotors.
Most studies of IPACS vibrations have focused on structural modes or individual rotor vibrations. Park studied
imbalanced rotors connected to an IPACS with linear springs in order to develop control algorithms that would mitigate
imbalance-induced vibration [10].
One unstudied aspect of IPACS vibration is the vibration interaction between multiple connected ywheels. Engineers
must understand the potential problems associated with this interaction in order to successfully design a technology
demonstration mission.
This paper will use a linear model to study the existence of extra-synchronous resonant vibrationresonant vibration of
ywheel whirl at non-spin speeds due to the synchronous imbalance-induced vibration input of another ywheel in the
system. Whirl is a precession resonance present in a rotor/bearing system. Extra-synchronous resonant vibration can be
sub-synchronous or super-synchronous. Sub-synchronous excitation is excitation of a forward whirl mode at a speed below
the speed of the rotor in question, and super-synchronous excitation occurs above the speed of the rotor.
2. Scope
Several assumptions are made in this paper to enable a simpler analysis. The only source of vibration studied is rotor
imbalance. Rotation of the rotor about its long axis is ignored except for its effect on gyroscopic stiffness. Also, the satellite
is assumed to be massive enough that it can be considered xed. In an actual satellite with an IPACS, the satellite body will
be free to rotate and can be controlled by the varying rotation rates of the ywheel rotors. Since the satellite is assumed to
be xed and input forces are small and periodic, small angles are assumed. A spring between the satellite and the support
structure enforces this constraint. This paper is also studying rigid ywheel rotors. There are higher bending modes
associated with exible rotors that are neglected.
Nomenclature
a
c
centripetal acceleration
A state matrix
B input stiffness matrix
c translational damping
C rotational damping
C damping matrix
e ywheel eccentricity
f
c
centripetal force
f
i
frequency of ith vibration mode, in Hz
G gyroscopic stiffness matrix
I appropriate size identity matrix
I moment of inertia
k translational spring stiffness
K spring stiffness matrix
m body mass
M mass matrix
P rotor conguration (I
P
/I
T
)
q state vector
r vector of displacement states
R rotation matrix
t time
u vector of input forces
x x-axis
y y-axis
z z-axis
k rotational spring stiffness
r radial distance between ywheel center of
mass and shaft center
h vector of rotation states
y axial ywheel rotation (short form of y
z
)
x ratio of damping to stiffness
o wheel spin speed (short form of
_
y
z
)
0 matrix of zeros
Subscripts
1 rst ywheel or body
i ith ywheel or body
i, j between ith and jth bodies
n nth ywheel or body
P polar (about y
x
)
T transverse (about y
x
and/or y
y
)
J. Firth, J. Black / Journal of Sound and Vibration 331 (2012) 17011714 1702
3. Model
A linear dynamics model was created to study vibration in an IPACS structure. The model was then integrated
numerically using the DormandPrince method.
3.1. Description
3.1.1. Body arrangement
In the model used in this paper, two ywheels are attached to a central support structure. Each ywheel is composed of
a rigid rotor connected to a housing by two linear springs (magnetic bearings) with three translational degrees of freedoms
(dofs), one at each end. The housing is rigid and is rmly xed to the central support, as shown in Fig. 1. Note that the
second ywheel will mirror the rst, attached to the bottom of the central support.
The ywheel shown in Fig. 1 is simplied for the purposes of this model as shown in Fig. 2(a). The housing is replaced
by a rigid link which is xed to the support structure. The two springs at each end of the rotor are replaced by a single
spring at the center of mass (COM) with translational and transverse stiffness.
For a ywheel of length l with individual magnetic bearing stiffness k
mag
, the model will have linear stiffness
k
model
2k
mag
and transverse rotational stiffness k
T,model
k
mag
l
2
/2 [11]. Damping is similar: c
model
2c
mag
and C
T,model

c
mag
l
2
/2.
The entire system is shown in Fig. 2 connected to the satellite by a spring with rotational stiffness. The satellite is
massive enough that it is assumed to be xed. When damping is modeled, all springs shown in Fig. 2 have both stiffness
and damping.
3.1.2. Input forces
The sources of vibration in this model are rotor imbalances. An unbalanced spinning rotor causes a centripetal force as
follows [12]:
f
c
2eo
2
2mqo
2
(1)
where eccentricity e represents rotor mass m and the distance between the ywheels COM and the center of the shaft, r.
The distance r is shown in Fig. 3. The vectors e and q point from the center of mass to the ywheel shaft.
The rotors will be assumed to be imbalanced in such a way that they cause a purely two-dimensional rotational
vibration (in y
x
, y
y
). This moment-inducing imbalance is shown conceptually in Fig. 4.
Rotor Housing
Central Support
Fig. 1. Arrangement of rotor and housing for the ywheels used in this paper.
Fig. 2. Simplied model: (a) support/rotor connection and (b) model attached to satellite.
J. Firth, J. Black / Journal of Sound and Vibration 331 (2012) 17011714 1703
The rotor imbalances are modeled in this paper as perfectly balanced rotors with an input force which is synchronized
to the rotation of the wheel. The input vector u describing these forces is shown in Section 3.3.
3.2. Body EOM
In this paper, ywheels are modeled as gyroscopic rigid bodies. The matrix equation of motion (EOM) for a gyroscopic
rigid body is as follows [13,14]:
M

qtCGwt
_
qt Kqt ut (2)
where the state vector q[r
T
h
T
]
T
is composed of r[x y z]
T
and h[y
x
y
y
y
z
]
T
describing the ywheel coordinate system
shown in Fig. 1, u is a vector of input forces, M is the mass matrix described in terms of body mass m and directional
moment of inertias (MOIs) I
x,y,z
. If the body is represented as a lumped mass, M is the diagonal matrix Mdiag([m m m I
x
I
y
I
z
]). K and C are the stiffness and damping matrices between the body and external nodes, and, when described in wheel-
aligned coordinates, G is the sparse skew-symmetric gyroscopic matrix shown below:
G
& ^
0 I
z
o 0
I
z
o 0 0
0 0 0
_

_
_

_
Fig. 3. An imbalance causes a centripetal force proportional to r and rotational speed squared.
Fig. 4. The rotors in this model are imbalanced in such a way that they cause a purely rotational vibration.
J. Firth, J. Black / Journal of Sound and Vibration 331 (2012) 17011714 1704
State-space representation is a convenient format for writing and solving linear differential equations, and it is the form that
will be used for the model in this paper. In general, a linear system can be described in state-space according to Eq. (3) [13].
Note that the notational dependence on time will be dropped for convenience
_
q AqBu (3)
Eq. (2) can be represented in state space as shown below in Eq. (4), which also includes an input matrix, B. The input
matrix is used to provide a convenient means of applying different input force combinations
_
q

q
_ _

0 I
M
1
K M
1
CG
_ _
q
_
q
_ _

0
M
1
B
_ _
u (4)
3.3. System EOM
When a system of independent, unconnected, gyroscopic rigid bodies is described in state-space, it takes the form of the
block diagonal matrix shown in Eq. (5). These stiffness and damping terms represent each body being independently
connected to a xed body
_
q
A
1
0
&
0 A
n
_

_
_

_q
B
1
^
B
n
_

_
_

_ (5)
where the state vector q[q
1
T
yq
n
T
]
T
is composed of multiple q
i
[r
i
T
h
i
T
v
i
T
x
i
T
]
T
terms. A
i
and B
i
are dened according to
Eqs. (6) and (7), similar to Eq. (4). R
i
is a rotation matrix. r
i
and h
i
are the position (x, y, z) and rotation (y
x
, y
y
, y
z
) vectors of
the bodies, respectively, and v
i
and x
i
are the corresponding translational and rotational velocities
A
i
R
i

0 I
M
1
i
K
i
M
1
i
C
i
G
i

_ _
R
T
i
(6)
B
i
R
i

0
M
1
i
B
i
_ _
R
T
i
(7)
In a system of connected rigid bodies, the stiffness terms and some of the off-block-diagonal zero terms in Eq. (5) are
replaced by terms representing the stiffness of each connection. These terms are found in a stiffness matrix, which can be
found using Hookes Law: Fkx.
The net spring force on particle i of Fig. 5 due to displacements, is shown as follows:

F
x
2k
x
x
i
y
y,i
d2x
j
(8)
Similarly, the spring force and torque can be found for each particle. These equations are written in matrix form as
follows:
f
i
f
j
_ _

K
i,i
K
i,j
K
j,i
K
j,j
_ _
q
i
q
j
_ _
(9)
K
i,i

k
T
0 0 0 k
T
d 0
0 k
T
0 k
T
d 0 0
0 0 k
P
0 0 0
0 k
T
d 0 k
T
d
2
k
T
0 0
k
T
d 0 0 0 k
T
d
2
k
T
0
0 0 0 0 0 k
P
_

_
_

_
K
i,j

k
T
0 0 0 0 0
0 k
T
0 0 0 0
0 0 k
P
0 0 0
0 k
T
d 0 k
T
0 0
k
T
d 0 0 0 k
T
0
0 0 0 0 0 k
P
_

_
_

_
K
j,i
K
T
i,j
J. Firth, J. Black / Journal of Sound and Vibration 331 (2012) 17011714 1705
K
j,j

k
T
0 0 0 0 0
0 k
T
0 0 0 0
0 0 k
P
0 0 0
0 0 0 k
T
0 0
0 0 0 0 k
T
0
0 0 0 0 0 k
P
_

_
_

_
The components of the stiffness matrix K in Eq. (9) are the same stiffness matrix terms used in the state-space equation
for this model.
Recalling Eq. (3), a state-space system EOM is written as
_
q AqBu
For the complete connected model described in this paper, the A and B matrices are as follows:
AM
1
system
0 I 0 0

K
1,1

C
1,1
G
1
K
1,n
C
1,n
&
0 0 0 I
K
1,n
C
1,n

K
n,n

C
n,n
G
n

_
_

_
(10)
B M
1
system
0
B
1
^
0
B
n
_

_
_

_
(11)
where q and u are dened as in Eq. (5) and (2) and
M
1
system

I 0
0 M
1
1
&
I 0
0 M
1
n
_

_
_

_
Subscripts on stiffness and damping terms are the i, j subscripts found in Eq. (9). The stiffness term for the support
structure,

K
1,1
, also contains the stiffness between the xed satellite bus and the support structure.
The input matrix B links each of the states with an input u
n
. The column and row of a B term determine which input
force is applied to which state, respectively.
Fig. 5. System displacements in x. Bodies i and j are point masses. This diagram is used to determine spring forces between bodies i and j in the x
direction. The ywheel (j) is attached to the end of a rigid link length d extending from the support structure (i).
J. Firth, J. Black / Journal of Sound and Vibration 331 (2012) 17011714 1706
The centripetal force inputs in this model are synchronized to the rotation of each ywheel. The input vector describing
each of these terms is as follows:
u e
1
w
2
1
siny
1
t e
1
w
2
1
cosy
1
t . . . e
n
w
2
n
siny
n
t e
n
w
2
n
cosy
n
t
T
(12)
As Eq. (12) shows, the input vector has two centripetal force terms for each wheel which are 901 out of phase from each
other. The magnitude of the force is e
i
o
i
2
, and the force acts in the direction of the current wheel rotation, y
i
. In the model
this is divided into y
x
and y
y
inputs, which vary periodically with sin y
i
and cos y
i
. This vector is used for all model input.
Different system input cases are created by varying the input matrix B.
3.4. Whirl
Two of the vibration modes of a rigid body rotor/bearing system are forward and backward whirl. Forward whirl is a
precession in the spin direction, and backward whirl is whirl in the direction opposite rotor spin.
Fig. 6. A long rigid rotor constrained by springs in x and y.
Fig. 7. No-spin-normalized forward whirl speeds for rotors of varying length, P.
J. Firth, J. Black / Journal of Sound and Vibration 331 (2012) 17011714 1707
For the rotor/bearing system shown in Fig. 6, eigenanalysis of Eq. (2) reveals the forward (o
3
) and backward (o
4
) whirl
speeds as functions of rotor speed, o, which are as follows [11]:
o
3
o
I
P
2I
T
o

kL
2
2I
T

I
P
2I
T
o
_ _
2

(13)
o
4
o
I
P
2I
T
o

kL
2
2I
T

I
P
2I
T
o
_ _
2

(14)
where k is bearing stiffness, L is rotor length, and I
P
and I
T
are polar and transverse MOIs, respectively.
Forward and backward whirl are speed-dependent for a given rotor. They also depend on rotor conguration, PI
P
I
T
(polar MOI/transverse MOI). For a short coin-like disc or hoop, P2. For an innitely long rotor, P0. When spin speed, o,
is zero, o
3
(0)o
4
(0)o
T

KL
2
=2I
T
_
. As rotor speed increases, the frequencies of the forward and backward whirls
diverge from o
T
, with forward whirl speed increasing and backward whirl speed decreasing. As wheel speed approaches
innity, o
3
approaches P and o
4
approaches zero [11].
Figs. 7 and 8 show the no-spin-normalized forward and backward whirl speeds for a rotor/bearing system. Forward
whirl is always super-synchronous for a short rotor (P41). For a slender rotor (Po1), there is a crossover speed, o
con
where forward whirl speed matches wheel spin speed. When spin speed is higher than o
con
, forward whirl is sub-
synchronous.
4. Extra-synchronous whirl excitation
Whirl speeds for the two rotors can be found with Eqs. (13) and (14). Fig. 9 shows whirl speeds for the slender rotor,
and Fig. 10 shows the whirl speeds for the short rotor. The slender rotor has a critical coning frequency f
con
at which the
resonant forward whirling speed matches the wheel speed [11]. Steady-state operation at or near this speed can be
hazardous because the excitation frequency is the same as the resonant frequency. For this reason, the operating region of
a slender rotor is often set above the crossover point [14]. A short rotor has no self-exciting critical coning speed.
The model described in this paper will be used to study extra-synchronous resonant excitation in a system of two
connected ywheel rotors, looking for both sub- and super-synchronous vibration. In both cases one perfect rotor will spin
at 700 revolutions/second (RPS), and the excitation rotor will spin with a small imbalance at a speed intended to excite
forward whirl in the perfect rotor.
Forward (f
3
) and backward (f
4
) whirl for the slender rotor at a wheel speed of 700 RPS are f
3
E400 Hz and f
4
E11 Hz, as
shown in Fig. 9. For the short rotor, f
3
E890 Hz and f
4
E12 Hz, as shown in Fig. 10.
Figs. 9 and 10 also show that the excitation wheel will also have resonant whirling frequencies: f
3
E242 Hz and
f
4
E18 Hz for the slender rotor, and f
4
E7.9 Hz for the short rotor. f
3
for the short rotor is well above any frequency in the
system, and it is unlikely to be excited.
5. Examples
As described in Section 4, the two rotor system was used to look for extra-synchronous whirl excitation. In both test
cases, one perfect rotor was spun at 700 RPS. The other rotor was given a moment-inducing imbalance and ramped from
350 to 450 RPS (slender rotor, sub-synchronous excitation) or 850950 RPS (short rotor, super-synchronous excitation)
Fig. 8. No-spin-normalized backward whirl speeds for rotors of varying length, P.
J. Firth, J. Black / Journal of Sound and Vibration 331 (2012) 17011714 1708
simulating ywheel charging. These examples were selected to demonstrate the types of possible excitations caused by
standard operation of the rotors. Each was example was run over a range around the expected excitation frequencies to
demonstrate their sensitivity to small changes in spin rate. Note that while the exact conditions at which extra-
synchronous whirl occurs is quite sensitive to physical parameters of the rotors, its existence is not. Most congurations of
rotor systems are susceptible and will need to avoid these conditions operationally. Spectrograms were created from y
x
rotations in the perfect rotor to examine frequency content of the rotors vibration. Support response in all cases was
similar.
The sub- and super-synchronous cases were both studied with and without damping. Magnitudes are included only for
the purpose of comparison between different cases. The actual magnitude of the whirl response is a function of input
imbalance force or moment, and is typically quite small.
5.1. Sub-synchronous whirl excitation
The slender rotor was used to look for sub-synchronous whirl excitation. Sub-synchronous resonance occurs at a higher
frequency than the critical coning frequency f
con
at which the forward whirl frequency matches the rotor speed. In the sub-
synchronous test, a perfect rotor is operated at 700 RPS. The wheel speed o of the excitation rotor varies in time from 350
to 450 RPS. These rotor speeds excite several resonant whirling frequencies, given in Section 5.
5.1.1. Undamped sub-synchronous response
At rst, damping is neglected to see if there is any response. The excitation rotors response is shown in Fig. 11. Note the
strength of the self-induced, sub-synchronous, forward whirling mode. This rotor is well past its critical coning speed for
forward whirl, but it is still responding strongly at the whirling frequency. In fact, at lower rotor speeds the resonant
forward whirling response is stronger than the input disturbance.
Fig. 9. Slender rotor whirl speeds.
Fig. 10. Short rotor whirl speeds.
J. Firth, J. Black / Journal of Sound and Vibration 331 (2012) 17011714 1709
The perfect rotors response over the same time period was examined at multiple speeds. Fig. 12 shows the response at
a baseline of o0 and at o7700 RPS. The perfect rotor experienced a signicant forward whirling motion when it was
rotating in the same direction as the unbalanced rotor.
Two notable vibrations appear in Fig. 12. The rst of these vibrations is at the backward whirl frequency, f
4
E18 Hz, of
the unbalanced rotor. This vibration is visible in subgures (b) and (c) of Fig. 12, but is strongest in the response shown in
subgure (b), in which case the rotors are spinning in the same direction. This excitation indicates that the unbalanced
Fig. 11. Spectrogram of excitation rotors response to its own undamped imbalance. o
excite
varies in time from 350 to 450 RPS.
Fig. 12. Spectrogram of perfect rotors undamped response to sub-synchronous vibration input: (a) o0; (b) o 700 RPS; and (c) o700 RPS. Note
the different scales in subgure (b). Vibration at 50 Hz fades in and out repeatedly (visible on a longer time scale). o
excite
varies in time from 350 to
450 RPS.
J. Firth, J. Black / Journal of Sound and Vibration 331 (2012) 17011714 1710
rotor is exciting its own backward whirl mode through interaction with the other rotor, even though the other rotor is
perfectly balanced.
The other vibration apparent in Fig. 12 is the vibration at 400 Hz that appears near t1 s in Fig. 12(b). This frequency is the
forward whirl speed of the perfect rotor, and the intended excitation frequency of this test conguration. This vibration
appears and is most severe about halfway through the test, when the excitation frequency matches the whirl frequency.
Forward whirl is expected to be at a maximum at that time because it is being excited at its resonant frequency. The
persistence of the whirling motion after the excitation frequency has moved past the resonant frequency shows that whirl
tends to persist after excitation. The same time history appeared when excitation speed was varied backwards, from 450 to
350 RPS. In other words, the appearance and decrescendo of a whirling motion was consistent in time, not in input frequency.
Fig. 13. Spectrogram of excitation rotors response to its own damped imbalance. o
excite
varies in time from 350 to 450 RPS.
Fig. 14. Spectrogram of perfect rotors damped response to sub-synchronous vibration input: (a) o0; (b) o 700 RPS; and (c) o700 RPS. Note the
different scales in subgure (b). o
excite
varies in time from 350 to 450 RPS.
J. Firth, J. Black / Journal of Sound and Vibration 331 (2012) 17011714 1711
A constant frequency vibration at approximately 50 Hz also appears in Figs. 12(a) and (c). The 50 Hz vibration appears
to be worse when the excitation is at 400 Hz, but it is actually worse at approximately 1 s, and occurs repeatedly if the
integration time is longer. These vibrations do not match any of the resonant frequencies of either rotor. They are system
level effects caused by the bus/support springs and the net gyroscopic system stiffness.
Note that the model conguration represented in Fig. 12(b) is not operationally realistic for a two rotor IPACS, but it
demonstrates sub-synchronous excitation. The problem is that both rotors are spinning in the same direction, which does
not allow for high power storage and transfer with a net zero angular momentum change. In reality, IPACS rotors will be
spinning in opposite directions as they are for the case shown in Fig. 12(c). This situation might occur, however, in a bank
of multiple rotors being used primarily for energy storage.
5.1.2. Damped sub-synchronous response
Since sub-synchronous excitation was found to affect a two ywheel system with no damping, the next step was to see
if it remained when damping was included in the model. The same test conditions used in the undamped case were used
to study a more realistic damped ywheel conguration.
Fig. 13 shows the response of the excitation rotor. Note that the resonant whirling response has been attenuated below
the threshold of visibility in this spectrogram and all that remains is the synchronous imbalance-induced vibration. Since
the rotor speed is so far from the critical whirling frequency f
con
, the very small amount of damping present in this model
is enough to eliminate the resonance.
Fig. 14 shows the damped response of the perfect rotor to the same sub-synchronous imbalance input that induced
forward and backward whirl in the undamped case. All resonances have fallen below the detectable threshold, leaving only
the direct input frequency created by the unbalanced rotation of the excitation rotor.
Fig. 15. Spectrogram of perfect rotors undamped response to super-synchronous vibration input: (a) o0; (b) o 700 RPS; and (c) o700 RPS.
o
excite
varies in time from 850 to 950 RPS.
J. Firth, J. Black / Journal of Sound and Vibration 331 (2012) 17011714 1712
From the signicant and benecial effect of even modest damping in this multiple ywheel IPACS model, it appears that
ywheel systems should be safe from the harmful effects of sub-synchronous ywheel vibration in an axially opposed
conguration. Real ywheels necessarily include damping in the magnetic bearings.
5.2. Super-synchronous whirl excitation
The short rotor was used to look for super-synchronous whirl excitation. Ordinarily whirl is not a problem for short
rotors, since their forward whirling frequencies always remain higher than their spin speed. An external vibration input at
the resonant speed could induce vibration in this mode, however. In the super-synchronous test, a perfect rotor is operated
at 700 RPS. The speed of the excitation rotor varies in time from 850 to 950 RPS. These rotor speeds cause several resonant
whirling frequencies, given in Section 4.
5.2.1. Undamped super-synchronous response
Damping is again at rst neglected to see if there would be an excited response. The super-synchronous excitation
response is very similar to the sub-synchronous excitation response.
Spectrograms of the unbalanced rotor and support responses did not reveal any resonances. The only place in the model
with any visible extra-synchronous frequency response is in the perfect rotor. The spectrograms for that rotor are shown in
Fig. 15. Similar to the sub-synchronous case, there is a slight forward whirl response when spin directions are identical,
and there is a slight backward whirl excitation regardless of relative spin directions.
Fig. 15 also reveals a slight response around the backward whirl modes. The separate backward whirl modes of the two
rotors are very close in frequency (approximately 8 and 12 Hz) and the response of both is blending together in this gure.
Fig. 16. Spectrogram of perfect rotors damped response to super-synchronous vibration input: (a) o0; (b) o 700 RPS; and (c) o700 RPS. o
excite
varies in time from 850 to 950 RPS.
J. Firth, J. Black / Journal of Sound and Vibration 331 (2012) 17011714 1713
5.2.2. Damped super-synchronous response
As seen in Fig. 16, resonant vibrations due to a super-synchronous vibration excitation are not apparent when damping
is included in the model, similar to the disappearance of resonant vibration in the sub-synchronous test.
Just as damping mitigated the effects of sub-synchronous vibration excitation, it also mitigates the effects of super-
synchronous excitation.
6. Conclusion
The model developed in this paper was used to investigate potential causes of troublesome extra-synchronous
vibration interaction in an IPACS. Both sub- and super-synchronous vibrations were found to be naturally excited in a
multiple ywheel IPACS. Backward whirl was excited regardless of relative spin direction, but extra-synchronous forward
whirl was only excited when spin directions were identical. While the identical spin direction conguration is not
representative of a simple IPACS, such a conguration would exist in a bank of ywheels such as one intended primarily
for energy storage. Extra-synchronous vibration is very small relative to the inducing imbalance, but it does exist in an
ideal undamped system. Fortunately, it appears that the small amount of damping inherent in real bearings is enough to
mitigate the effects of this vibration without the need for any additional active control.
Disclaimer
The views expressed in this article are those of the authors and do not reect the ofcial policy or position of the Air
Force, Department of Defense or the U.S. Government.
Acknowledgments
The authors would like to thank Ray Beach and the other members of the ywheel team at NASA Glenn Research Center
for their support.
References
[1] J.B. Roes, An Electro-Mechanical Energy Storage System for Space Application, Energy Conversion for Space Power 3, Progress in Astronautics and Rocketry,
Academic Press, New York, 1961, pp. 613622.
[2] F. Biggs, Flywheel Energy Systems, Technical Report SAND74-0113, Sandia Laboratories, 1974.
[3] C.D. Hall, Integrated spacecraft power and attitude control systems using ywheels, Technical Report AFIT/ENY/TR-000, AFIT (October 2001). URL
/http://www.aoe.vt.edu/$cdhall/courses/aoe4065/litrev.pdfS.
[4] R. Thompson, Flywheel Safety, University of Texas at Austin Center for Electromechanics, 2011 URL/http://www.utexas.edu/research/cem/projects/
featured%20presentationp%20pdfs/Flywheel%20Safety%20-%20Richard%20Thompson%20-%20Jan%202011.pdfS.
[5] C.S. Hearn, M.M. Flynn, M.C. Lewis, R.C. Thompson, B.T. Murphy, R.G. Longoria, Low cost ywheel energy storage for a fuel cell powered transit bus,
IEEE Vehicle Power and Propulsion Conference, 2008, pp. 414421, 10.1109/TEC.2010.2045925.
[6] Standard for Space SystemsFlywheel Rotor Assemblies, American Institute of Aeronautics & Astronautics, ANSI/AIAA S-096-2004, 2004, ISBN
9781563477416.
[7] J. Firth, Vibration Interaction in a Multiple Flywheel System, Masters Thesis, Air Force Institute of Technology, March 2011. URL /http://www.dtic.
mil/dtic/tr/fulltext/u2/a539262.pdfS.
[8] R.A. Delventhal, Flywheel energy storage system designed for the international space station, Research and Technology 2001, NASA Glenn Research
Center, 2002, pp. 204205, NASA/TM-2002-211333. URL /www.grc.nasa.gov/WWW/RT/RT2001/index.htmlS.
[9] V. Lappas, D. Richie, C. Hall, J. Fausz, B. Wilson, Survey of technology developments in ywheel attitude control and energy storage systems, Journal
of Guidance, Control, and Dynamics 32 (2).
[10] J. Park, Mimo Active Vibration Control of Magnetically Suspended Flywheels for Satellite IPAC Service, PhD Thesis, Texas A&M University, May 2008.
[11] J.M. Vance, Rotordynamics of Turbomachinery, John Wiley & Sons, Inc., New York, 1988.
[12] A. Bedford, W. Fowler, Engineering Mechanics: Dynamics, Prentice-Hall, 2002.
[13] J.G. Reid, Linear System Fundamentals, McGraw-Hill, Inc., New York, 1983.
[14] J. Vance, F. Zeidan, B. Murphy, Machinery Vibration and Rotordynamics, John Wiley & Sons, 2010.
J. Firth, J. Black / Journal of Sound and Vibration 331 (2012) 17011714 1714

Você também pode gostar