October,
1971.
ATTITUDE STABILITY AND
PERFORMANCE OF A
DUAL-SPIN SATELLITE WITH LARGE
FLEXIBLE APPENDAGES
by
D. B. Cherchas
,
9
'U
4ATTITUDE STABILITY AND PERFORMANCE OF A DUAL-SPIN SATELLITE WITH LARGE
FLEXIBLE APPENDAGES
by
D. B. Cherchas
Submitted September, 1971.
ACKNOWLEDGEMENT
The author wishes to express appreciation to Dr. P. Co Hughes for his expert guidance throughout the course of this study.
The cooperation and interest of Mr. G. J. Cloutier and Mro R. Hornidge of the Hughes Aircraft Company in providing sate1lite configurationsand parameters is gratefully acknow1edged.
Appreciation is a1so expressed to Mrs. B. Wadde1l for typing the manu-script and to Mr. N. K. Phung for drawing the mathematica1 symbo1s and figures.
Financia1 support has been provided by a 1967 NRC Science Scholarship, NRC Grant No. A4183 and USAF Grant No. AFOSR 68-1490.
SUMMARY
The attitude stability criteria and nutation decay times of a dual-spin'
satellite with a large flexible solar array are found in terms of significant
satellite parameters. The equations of motion are developed from Hamilton~
prin-ciple and ~hen linearized, with solar array natural modes utilized as array degrees
of freedom. It is shown that with proper mass balancing, the equations for spin
variables can be decoupled from those for transverse variables. The linearized
equations are numerically studied through the system eigenvalues thus yielding
direct statements of stability and nutation damping time. These eigenvalue studies
are done first assuming the array to be rigid and then with array flexibility in-cluded. A complementary analysis for the solar array develops practical
informa-tion regarding array natural mode frequencies and shapes and boom buckling con~
siderations •
Numerical studies of the satellite attitude stability and nutation
damp-ing time assumdamp-ing the solar array is rigid indicate that the ratio of spin inertia
to the geometrie mean of the combined transverse inertias is a significant ratio.
The value of unity marks the minimum value of this ratio before energy
dissi-pation in the despun section is required for stability. It is shown that
stab-ility is unachievable below a certain value of this inertia ratio for a given
damper mass. Upper and lower stability boundaries exist for the nutation damper
damping constant and the damper performance is best at high nutation frequencies.
Results of the same studies aliowing array flexibility but na array damping show
little ~hange in stability boundaries and putation damping times from the rigid
array results except when the nutation frequency becomes close to an array natural
frequency. In this circumstance, stability criteria and performance are signifi
-cantly and undesirably affected. _I~is shown thatthe ~b~trary use of ~ery smal~
-amou~ts of array damping in the flexibility studies could lead to many important
I
.
Ilo 111. IV.v
.
TABLE OF CONTENTS Notation INTRGDOCTION 1.1 Dua1-Spin Sate11ites 1.2 Spin Stabilization 1.3 Research NeededSOLAR ARRAY ANALYSIS
2.1 Analysis Requirements
2.2 Array Description
2.3 Related Analyses
.2.4 Differential Equations
2.5 Linearized Equations with Boundary Conditions
2.6 Natural Vibrations
2.7 Results and Discussion
EQUATIONS OF SATELLITE MarION
ABour
THE-~.C.ENTRE 'OF3MASS3.1 Deve10pment of Equations
3.2 Generalized Coordinates
3.3 Kinetic Energy
3.4 P0tential Energy
3.5 Generalized Forces
3.6 Linear Equations
3.7 Discussion
STABILITY ANB PERFORMANCE ANALYSIS
4.1 Stability Conditions and Nutation Decay Time
4.2 Description of Eigenvalue Study
4.3 Case Studies
4.4 Motion of Coro~lete1y Rigid Dual-Spin Systems
CQNCLUSIONS
5.1 Solar Array
5.2 Dual-Spin Satellites without Flexible Appendages
5.3 Dual-Spin Sate11ites with Large F1exib1e Appendages on
Despun Portion
5.4 Despin Control and Unbalance
REFERENCES FIGURES
APPENDIX A~ Attitude Motion of a Single Spinning Body with a Two
Degree of Freedom Nllitation Damper
PAGE 1 1 1
6
6
6
7 77
9
1319
20 20 20 22 23 23 24 32 33 33 33 34 3840
40
40
41 41 42APPENDIX B: Satellite Kinetic Energy with Large Flexible Solar Array
APPENDIX C: Component Angular and Centroidal Veloeities APPENDIX D: Solar Array Mass Integrals
APPENDIX E: Solar Radiation and Gravity-Gradient Torques i J . '
A B b d d -xp d s E. 1
E
s e N0rATIONy coefficient matrix in decoupled nutation variable equations
antenna reference axes
satellite combined moment of inertia in ~ direction
through total center of mass
satellite combined moment of inertia in ~2 direction
through total center of mass
y coefficie~t matrix in decoupled nutation variable equations
solar array sheet half width
nutation damper bob wire length nutation damper damping coefficient
vector from reference point on array mount to mass element on flexible array
vector from center of array end piece to mass element on end piece
distance along ~l from center of array mount to base of array boom 1
solar array symmetrie mode shape Young's modulus of solar array sheet
solar array width
antenna centroidal body axes; ~l' ~2' ~3
axes fixed to base of nutation damper bob wire fuel centroidal bodyaxes
total satellite centroidal inertial attitude axes
array mount centroidal body axes
rotor centroidal body fixed axes
axes fixed at mounting point of array boom 1;
~l~ ~2' ~3
I I I aU, a22, a33, 1a12,la13,123 1 1'll, etc. , I pll, etc. , 1 rl1, etc. , I s J b J s Kd
t
M
M.
l. m m a md mf m p m r m spa m xp N Q. J q. J9t
Rf rdforce field acting on array cross-piece at sheet attachment
moments and products of inertia of antenna in
~
moments and products of inertia ofmoments and products of inertia of moments and products of inertia of total spin inertia
=
1r33 + 1f33 flexural stiffness of array boomfuel in
~
array mount rotor in~
flexural stiffness of sheet in lateral plane nutation damper spring stiffness
solar array length
in
~
decoupled nutation variable equations system matrix solar array torsional mode shape
total satellite mass antenna mass
damper bob mass fuel mass
array mount mass rotor mass
solar array mass array end piece mass
eigenvalue of satellite nutational mode generalized force
generalized coordinate satellite inertia ratio radius of fuel tank
r f r p r r ~l' T T e -t t s U
V
.
l W xal'
13-13
2, )')'3
5 1, E . , al ~2' ~3a
2,a
3
~3'
~:t 5 2 Ebidistance, along spin axis, from antepna reference point
to fuel mass centre
distance, along spin axis, from antenna reference point to array mount reference point
distance, along spin axis, from antenna reference point
to rotor referenee point
solar array reference axes
satellite kinetic energy about mass centre
forcejwidth across array sheet
time
array sheet thickness
satellite potential energy function
solar array lateral mode shape
distance from centre line of sheet, measured
perpendiu-lar to centre line, to sheet mass element, positive in
-
~3 dii'ectionposition vector of array mass element as seen in
~
position of antenna mass centre relative to antenna
reference point
position of array mount mass centre relative array
mount reference point
decoupled nutation variable equations state vector where lT =
(à
1a
28
1,82.'
Gai. • Ébi •)lat. ' Pb;'"
~àt
• ;>bt Jai •
a2 • &1 ,8
2 ' Eat ' Ebi' ).Lat' P-bi ' Jat' "bi) Euler angles taking~-9i
generalized coordinate for rotor spin
àrray lateral mode functions
array sheet massjlength
rotation taking
~--~
rotation angles taking ~A~9b
I-lai, I-lbi I-l v . ,vb' al. l. v s 'f3 IT l.J •. kt Pb Pbp Pf Pxp 0-S
.
0-angles taking%
~~
angle'; taking
~ ~~,
to zero th order; also used modal amplifier time function in array analysis array torsional mode generalized coordinates fuel viscosityarray lateral mode generalized coordinates array sheet Poisson ratio
rotation angle taking
~ ~ ~
solar array mass integrals solar array boom mass/length array base-piece mass/length fuel mass/volumearray end-piece mass/length array sheet mass/area
rotor nominal spin rate
frequency of solar array natural mode angular velocity of jth body in satellite
fuel motion variables transformed from ~1'~2'~3
I. IN'I'RODUCTION
1.1 Dual-Spin Satellites
<
The abili ty to confidently design and build a satelli te that meet.s the the mission requirements for attitude stability and controllis of fundamental importance in ensuring mission success. The classification of attitude control systems ranges from the passive systems such as gravity-gradient,stabilized space-craft to cQmple~, I . fully active systems such as OAO.
A class of currentcommunications satellites might wel~ be placed in the semi-active category. These are the dual-spin variety, cemprised of a rapidly spinning rotor intended to giv~ gyroscopic stiffness, and one or two ,slowly spinn-ing bodies rotating ab out the rotor spin axis and acting typically as communi ca-tions antennae or solar energy gathering surfaces. The attitude of thespin,axis
and rotation rates of the bodies'can be adjusted by mass expulsion or momentum wheel control. Figure 1 indicates .a prototype· (courtesy of Hughes A~rcraft
Company) of such a satellite.
It is the intention of this analysis to find and study the torque-free stability 'criteria and nutation damping time of such a dual~spin spacecraft.
1.2 Spin Stabilization
The ccmcept of spin: stabilization is of fundament al importance ,in under-standing dual-spin spacecraft attitude motion, so a review of bot:q. thebasic pri' n-cipals and advanced techniques of gyroscopic stabilization is in order.
a) single rigid body
The Euler equations of motion for a single rigid body in·a torque-free environment are Il ,J;2,I 3 principalmoments of
l
1w
1 W 2 W 3( 1
2 -1
3 ) inertia of body -I~W2. W3 w1 (
I~
-
I, )
(1.1) ~'W2'W3 comp?nents of body1
3w?,
W1 lU 2 (1
1 -1
2 )angular velocity in
i1,12.,
-
i
a
where, referring to Fig. 2, axes
i
,
~,1
are body fixed principal.axes. It is clear that there are three immediates
olut~ons
to these equations, i .e., pure spin about any of the three axes. However, if we examine the stability of rotation close to a principal, axis, say1
3, by assuming <.03) Wl , <.02 and dif:ferentiating and intersubstituting the first two equations of (1.1) we see that
,
(L2) Therefore, for stabIe .periodic solutions for W
l and W2' the mass must be distributed such that 1
3) 11, 12 or 13
<
11, 12. Rotation about the axis of inte r-mediate inertia is unstable.b) stngle body wit~ energy dissipation
A heuristic ,analysis of a single body witll internal energy dissipation* induced by nutational motion cau be done from an 'energy-sink' approach, i,e"
by recognizing that the kinetic energy decreases, but ,the total angular momentum remains constant, A spinning body which is stabIe in this situation then is one which decreases its nutational amplitude as the kinetic energy ~ecreases,
An
energy-sink argument for a single "spinning body, is, as follows ., For a fixedmagnitude of angular momentum h, the kinetic energy T of a body spinning about an axis with moment of inertia I is
T
=i
h
22 1 (1,3)
Therefore, the axis of maximumr, I is the axis of mJ.nJ.mum T 0 Thus, rotation about a body axis near the ax~s of maximum I would decay to pure spin about the maximum inertia axis .when
T<O
.
Note that in pure spin, L e., when nutation and precess.ion no longer exist,it i s assumed that the energy dissipation disappears. Reference 2 is one of the early analyses identifying the aboyestateà 'major-axis' rule·for stability. An energy-sink analysis of course mu~t
be complemente~ with a more rigorous stability analysiso Appendix A analyzes the motion of a spinning body with a two degree of freedom internal damper, The
requir~ents for nutational stability are shöwn to be the same as disclosed by the energy~sink. argument, The nutational instability of Explorer I in spin about i ts minor axis of inertia (the hoped for axis of stabili zation) is an expensi ve illustration of the major axis rule. The energy dissipation occurred in small flexible antennas.
c) dual-spin systems
A dual-spin,satellite (see Fig. 3) is also capableof being spin stabi-lized. The gyrosc9pic stiffness is developed by revolving the bodies about their common spin axis in such a way that a significant amount of angular momentum exists
along the spin axis. A practical realization of this is one body rapidly spinning and one or more bodies almost c9mpletely despun for solar power gathering or
communications purposes, It ,is the combination of the simple spin stabilization mechanism and a despun platform that is the essence of duai-spin desirability. Since .it is known that for a single spinning body any internal energy dissi pa-tion will cause a nutation buildup unless the spin is about the maximum moment of inertia axis the question of stability in the presence of internal energy dissipation for dual-spin systems must be investigated.
At this point, a note on meeting inertial geometry constraints in satel-lite manufacture is in order. The inertia ratio of a satellite is a function of its geometry and mass density distribution. For· a spin-stabilized body, the axis of spin is generally an axis of revolution i,e., allmoments of inertia per-pendicular to this axis tbrough the centre of mass are the same, To make maxi -mum use .of space in the launch vehicle compartment, the satelli te is mounted for launch with the spin axis 'along the rocket centerline. To accommodate larger satellites, it is relatively easy and inexpensive to make the compartment longer along the centre line but,much moreexpensive to 'increase the payload compartmen~ diameter. Thus, satellites which have a narrow cylindrical shape (minimum moment of inertia) can be much less expensive to launch than dis~ shaped (maximum moment * herein energy dissipation shal.). refer only to dissipation induce.d by- nutational
of inertia spacecraft). Also, having to make a rotor of a certain diameter for
no functional reason except to satisfy inertia ratio requirements is an
undesir-able design restriction.
A considerable amount .of research into dual-spin stabili ty theory has
been developed in recent years. The most significant advance was made ip
1964-1965 when it was noted independently by 1orillo3 , and Landon and Stewart4 , that
by ensuring sufficient energy dissipation takes place on the despun bodie(s),
spin stabilization can be maintained despite rotor internal energy dissipation
at any ratio of rotor spin inertia to satellite transverse inertiao These
analyses assume both bodies to be bodies of revolution about the common spin axis.
As in the single spinning bodyanalysis, the stability conditions have been
developed from an energy-sink approach and by stability analysis of the .satellite equations of motion with an explicit damper model. Reference 3 develops the
cri teria from the equations of motion while Ref . .
4
is an energy-sink development 0An energy-sink analysis similar to that of Ref. 5 is given below.
Referring to Fig.3, the angular momentum h and ki~etic energy
the dual-spin system can be expressed through I
where
h
2 (1.0.1)2 +
(C.aIl)~
+(Aw
r)2
2
T
Aw~
1"Hl~
"I-C.Q~I
I
=
moment of inertia about spin axis of body I C=
moment of inertia about spin.axis of body 11A
=
total satellite transverse inertia through satellite centre of massW
T
=
satellite transverse angular velocityn
I=
spin rate of body Inu
=
spin rate of body II T of (104) (I. 5)Following the energy sink concept , it is recognized that h2 will be a
constant and
T
will be negative in the presence of energy dissipation. Combiningthe expressions for
T
andh
2 we have2 • 2
fL
2 •0
=
I .0,1UI
+ C aU 11"1-A
WTWT ( 1.6).
AWTWT
1-0.
16.
1 "I-C.{111.ö.
U (I. 7)T
=
"I-.
.
.
.
.
T
=
Tl +Tu
= - IÀ.1 O,I
- C
À.u
n.
u
( I.8).
.
where TI and TIl represent the rates of energy dissipation in bodi~s land 11 respectively and
Rewriting,
thus, from the
.
TI - ~l statement.
cnu
= for T, Eq c (1.7), andTu
~1I substi tuting (I.9 j from Eq. (1.9) we have( Tl
+TIl )
À1 Àll
=
(LIO)where
TI' T
rr (
0, À T ) 0Now, utilizing Lyapunov stability theory, since
Y2
AWT 2
is a positive
definite function, if A~T
W
T is negative definite, the nutational,motion isasymp~
totically stabIe about W
T
=
O.or.
or
Thus; for asymptoticstability
ÀT
(Tl
oTT
n )<
0ÀI Àu
Therefore, the system is stabIe when
À1 ' Àn )
À
1) 0,
Àu<Q
and À 1<
Q , À. 11>
0 and 0TI
') Àl Tl<
TIl ÀI Àll.
Tu
ÀIIFor the case of one body despun i.e.,
n
I
=
0, the condition becomes+
<
0( LIl)
( L12)
(L13)
Thus, if body I has no dissipation i.e., TI
=
0 then(ciA
-
1» 0 i.e.,spin about maximum inertia axis is required for stab~lity. The presence of a
significant amount of dissipation in body I can modify this requirement
signifi-cantly, allowing values of
cl
A <. L The energy -sink analysis of Ref.4
is similarto the preceeding development except that damping is allowed in only one body.
A more definitive argument can be.advanced by writing the satellite
equations of.motion with an explicit damper model. References 3 and 5 give
analyses of this type with dashpots as the damper modeIs. The stability
require-ments developed through these analyses are the same as yielded in the
energy-sink approach.
The development of the dual-spin theory led to a symposium
6
devot~d
to the subject. Some of the papers are described below. A persistent problem
in.dual-spin equations is the presence of periodic coefficients in the equations
of motion. These coefficients appear because of the large rotation rate.between
the despun and spinning parts. If damping appears in only one of two symmètric
bodies, the periodicity can be avoided. However, for a general configuration,
an initial development of the equations yields periodic coefficients. Mingori!
handled the periodicity by utilizing Floguet theory which essentially gives the
conditions for stability of a linear set of equations if the system matrix is
8
this can be a lengthy and cestly precedure. Barba, Heeker and Leliakev were
able_ te remeve the periedic ceefficients by a suitable variable .transfermatien.
The transfermatien is pessible because ef the symmetry ef the roter damping medel
abeut the spin axis. ~e, remaining censtant ceefficient linearized equatiens are
easily handled. Velman develeped a tuIl scale simulatien heweyer, ne deeper
understanding ef the stability criteria is interpreted frem it. In Ref. 10,
Ierille discussed the perfermance ef the Hughes Aircraft dual-spin cenfiguratien. Any ef the papers in this sympesium whiah yield stability criteria.are develeped
enly fer c9nfiguratiens in which beth spinning parts are bedies ef revelutien
abeut the spin axis,
Likinsll developed an analysis fer damping in the platferm enly. This
paper utilized Reuthian analysis to state the stapility criteria.
More recently, Pringle12 applied Lyapunov stability theory te a general
class of dual-spin -satellites. Vigneren13 used the methedef.averaging to develop
appreximate stability ctiteria for a dual-spin system with a dashpet damper in
beth bedies. Cloutierl eptimized platferm damper design; this paper allewed a
platferm with different transverse inertias but assumed no damping in the reter,
hence stability criteria are net developed.
Several dual-spin satellites have been placed inte erbit. The first
dual-spin satellite to be stabilized by spinning abeu~ a minimum mement ef
inertia axis 'was the TACSAT I , a U.S. military cemmunicatiens satelli te launched en Feb
9
,
1
969
.
This satellite exhibited seme occasienal perieds ef marginalstability during which a nutation angle ef abeut ene degree .persistêd for several
days. It is recognized15 that flexibility in the spacecraft spin-despin bearing (and net the spin stabilizatien mechanism) was responsible for this marginally
stable .behavieur . On Jan
26,1971
,
the Intelsat IV was launched. This spacecraftis a large duel-spin cemmunicatiens satellite. Cempar~es frem ten natiens have
been enrel led as majer subcentractors assisting Hughes Aircraft Ce. in the
fabri-cation and testing of the satellite. Ne reperts 6n the nutational stability
have been available.
Dual-spin satellites with large flexible appendages are beceming mere
impertant with the need for greater en-beard pewer. The pewer requirement indi
-cates the usefulness ef large selar cell arrays which are compactly stored during
launch and then deployed when the satellite is in its eperatienal pesitian. The
presence of large flexible appendages on a dual-spin spacecraft necessitates
re-determining the fundament al dual-spin criteria with damping in both spinning
members and, as weIl, in the flexible structure. The effect af flexipility on
the performance of attitude and despin cantral systems alsarequires careful
study.
References
16
and17
appraach the fermulation af the flexible satelliteequatians by using a cambinatian ef discrete variables fer the essentially rigid
campanents and modal defarmatian caardinates far flexible appendages. This allo
-cation ef variables is termed the hybrid ceordinate methadlB. The emphasis in these papers is en the influence ef flexibility en the attitude and despin control
systems. Since na damping is included in the rator, the effect ef flexibility on
the basic dual-spin criteria is not shawn. Also, the use ef smal1 amounts of
viscous damping in the appendage models may serve te mask seme important effects.
1.3 Research Needed
The analysis to date of,dual-spin satel~ites has revealed the stability criteria for a dual-spin system in which both the spinning and despun portions are
bodies of revolution about the spin axis. By limiting the energy dissipation to be present in only one body, analytic stabil.i ty criteria have been obtained. How-ever, when damping is present in both bodies, the periodicity of the system has restricted (due to the computational requirements of Floquet analysis) the stabi -lity analysis to statements of stability at a small number of points in parameter space. Referenc~ 13 has advanced this somewhat by applying the method of avera-ging. It has been indicated, e.g. Ref. 3, that the requirement for an internal
energy dissipation devieeto act to damp out nutation is that the nutation fre-quency be positive in the body containing the dissipation. More explicitly, the transverse angular velocity vector must rotate positively, relative to fixed body axes, about the bodyaxis along which spin is positive. For bodies of revol u-tion, the only cases analyzed, this requirement has been met by the major axis
rule. In systems that have unequal transverse inertias, it must be determined whether the inertia ratios necessary to ensure a positive nutation frequency in· a body containing energy dissipation will ensure the removal of nutational angu-lar mOPlentum.
The studies of dual-spin satellites have not developed a statement of the effect of flexibility on basic dual-spin criteria.
Evidently, basic dual-spin theory can be advancedp Some of the research
requirements which motivated the present contribution are:
a) development of a clear statement of the equations of motion and solution for
a dual-spin system with no damping in either body and no despin motor torques, bearing friction, environmental torques and with one body not necessarily a body of revolution.
b) analysis to determine the stability criteri a of a dual-spin system with representative models of damping in both the spinning and despun members but no large flexible appendages. The despun members should not neces
-sarily be·. bodies of revolution about the spin axis.
c) repeat of b),but with large flexible appendages on the despun damping in the flexible appendages should not be included for of·this study to determine the effects of flexibility alone. tions including appendage damping should also be done.
section. The
the majori ty Some calcul
a-d) a complementary analysis of the solar array will also be required to facili -tate incorporating array deformations into the satellite equations of motion.
The solar array is discussed in more detail in section 11. 11. SOLAR ARRAY ANALYSIS
2.1 Analysis Requirements
Deflection modes of the solar array are required for array degrees of freedom in the satellite equations of motion. The deflection of an array (where
an array here is meant to be one.solar cell sheet with its two supporting booms) relative to its base which is fixed to the rigid components of the satellite is,
vibration modes of an array as it vibrates with a fixed base are ideal for this
purpose in that the natural modes, for small deflections, are orthogonal with
respect to an integral over the entire array mass i.e"
where X. is an array natural mode shape. Thus, werequire the natural modes of a solar~array as it vibrates with a fixed base,
2,2 Array Description
The array analyzed is the Hughes Aircraft Co. FRUSA* variety as depicted
in Fig. 4. The array is stored during launch and initial maneuvers by rolling the
sheet of solar cells onto a cylindrical roller. The supporting booms are the weIl
known STEM** or BISTEM booms,which can be rolled into a compact compartment until
required. When the satellite is in its operational posture, booms and sheet are
extended outward into th~ final configuration shown in Fig,
4
with thesheet heldin tension, Reference.19 explains the array construction in ,more detail, Another array configuration is the T - boom type as described in Ref. 20.
2.3 Related Analyses
Analysis of the dynamics of large roll-out arrays for spacecraft
application is a relativelY new field. Yao 21 has performed a finite element
analysis for the CTS*** satellite array. Hughes22 has analyzed the same array
from a continuum mechanics approach, Coyner and Roas 20 have analyzeä a similar
T - boom roll-up array from fini te element considerations. Hughes Aircraft Co,
have both finite element and continuum mechanica analyses of FRUSA arrays,
al-though no publications of extensive analyses have been made in the open
liter-ature 0
2.4 Differential Equations
The analysis technique is to write the differential equations of motion
of the separate arrayelements i.e., booms, sheet and cross-piece and then to establish the boundary conditions to be met for a periodic vibration. It is
anticipated that there will be three types of modal vibrations possible i.e"
symmetrie, torsional and lateral modes (see Fig.
5
)
.
In the symmetrie mode,both booms and the sheet vibrate in the same direction out of the nominal plane.
The deflection of each boom is the same and the sheet's deflection is not a
func-tion of dist~ce across its width. The torsional mode is a rota~ion about the
array centre line. The deflection of one boom is the negative of the other and
the sheet deflection function is reflected on either side of the centre line, In
the lateral mode, the array deflects only in thenominal plane and both booms
have the,same deflection.
Referring to Fig.
6,
we can write the differential equations for eacharray element as follows:
a) boom differential equation
* Llexible ~olled-~ ~olar ~ray
** extendible,booms manufactured by Spar Aerospace Products Ltd,
The boom is modelled as a thin beam cantilevered at its base. The
differential equation is formed by balancing, at an arbitrary.cross-section, the
boom's elastic resisting moment with the applied moment fr om inertial forces on
the rest of the boom and boom tip forces. This balance is stated by an
integral-differential equation, i.e.,
where T.
=
'G
=
'àx
oS
aI
=
OS
t
1
Pb [(
~(s')
s
+
(~(e)
-
~(S))
x
Pb=
boom mass/lengthiK
a= ot2.
-~(S))x~J
fi
or 2J
b
=
boom flexural stiffness( 11.1)
ds'
~ is the sum of the moments acting at s due to acceleration of a boom
element at-s' and the tip force on the boom. b) s~eet differential equation
(i) symmetric,and torsional modes
The sheet is modelled as a membrane with tension in one direction, along
ox/
dS.
The differential equation which balances the inertial force on a sheetelement with the restoring force on the element due to sheet tension is
( 11.2)
where
0;
=
sheet mass/areaT e
=
sheet. tension, force/width(ii) lateral mode (see Fig.
7)
In the lateral mode, the sheet and booms move only in the nominal array plane. If wrinkling does not occur, the sheet bends as a large flat beam.
Assum-ing the sheet does not wrinkle, the centre lin~ equation can be approximated as
?lX3
Mb
()2.M 1ox
l : : _ _ b- -
(11.3) Ss é)x
2KG
s
1 1 wheree
~(
- x
1) dx'1f
(i-x
1) -p(x~(t)
-
X3(X1))Mb.
~-1
~
o
X2> I=
àt2.x,
+x
1~
=
mass/length of sheetJ
S
=
flexural stiffness of sheet in lateral planeG
=
shear modulus of sheet sK
=
4
t b3
sEquation (II.3)represents.the beam bending equation including the effects of shear. ~ is the bending moment at xl from the inertial forces due to accelera-tion at xl and the end forces on the sheet.
c) cross-piece
The cross-piece must move according to .Newton's laws in translation of its mass centre and rotation about its mass centre. Therefore, the following equa~
tions must be true: For translation
1 2
r
b 3- f -
f
+J ..
f
dw
-b
and.for rotation
where
m
=
cross-piece mass xpPxp
=
cross-piece mass/length2.5 Linearized Equations with Boundary Conditions
(11. 4)
The next step is to assume.the correct boundary co~ditions for tbe three array mode types, i.e., symmetrie, torsional and lateral, and to state the
differential equations for eaeh mode type. Assuming small deflections, the diff-erential equations for the three array vibration types with the appropriate
boundary conditions are: a) symmetrie mode
(i)
boom (either one)(11. 6) Now differentiating Eq. (11.6) twic~ witb respect to xl yields the fourth
order partial differential equation
(11. 7)
with boundary conditions
Xl (0,
t,)
0 ?lx2. (t,t)
0 .::: ()X2.=
t '3Xa (0, t, ) ~ 0J
i/'x,.
(t,
t)
',2. Ox2. (t,t)
1,z
- f
ft ';)(.1 b0
3 + X1 2d
X 1 where fl=
1 f2 1 fl 2=
i
2These boundary conditions describe a cantilever boom base. The tip
con-ditions are found fr om Eq. (11.6) at xl
=
e
and from the first differentiation ofEq. (11.6) at xl
=
t.
(ii) sheet
For tbe symmetrie mode, we use the x
2 component of Eq. (11.2) i.e.,
32.
X2 (x, ,t)
Te
é)\2.(x~,t)
öt
2OS
ox2.
1 (11. 8)with the following boundary conditions x 2 (O,t)
=
0 x 2 (~,t)=
x2(,e
,t)l where(
t
,t)l means x 2(.f"t) x 2 of boom 1 etc.The boundary conditions state that the sheet is unqeflected at Xl
=
0and follows the boom motion at Xl
=
t(the sheet and booms are connected at Xl=
t
by the cross-piece).
(iii) cross-piece
In the symmetrie; mode, the cross-piece translates , without rotation, in
the x
2 direction. The vector translational equation, Eq. (11.4), has two
com-ponents. -2 fl,2 - T 2b
=
0 1 e b -2 fl,2 -T
J
O)l.z(t,
t) d ... 2 e --b,(h,
(11.10)ten-•
the booms and the component of sheet tension in the .x
2 direction act to accelerate .
the cross-pieceo
b) torsi onal mode
(i) boom 1
The boom equation and boundary conditions are the same as Eq. (II.7) i.e.,
o
lb
?J4-X 2 (x.1,t) _
f
,
'0
2 )(.2. ()I., ,t)
~ 4-dX
2 oX, 1 ( IL ll)with boundary conditions
()2X2
U.,
t)
- 0ox2-
,
Xz(o,t)=
0 ë)x2 (o,t)=
0 OX , Jb ~\2 (.t,t)=
ox?>
_ f' +ox,(t,t)
f1
2.ox,
.,
1 whereBoom 2 will have a deflection of opposite sign, same magnitude. (ii) sheet
For the tor.sional mode, Eq. (II.2), is again applicable i.e.,
Te.
'0\2 (X" IN,
t)o-s ox;
( IL12)with boundary conditions x 2(O,w,t)
=
0 x 2(e,w,t)=
x2
(
~,t)1
+~
(1 +~W)(
X2(t,t)' -
X2.ei,
t)1) where x 2(i , t)1=
-x2(t,t )2The boundary condition at xl
=
t
states that the end deflection of thesheet follows the motion of the cross-piece as the booms deflect in opposite directions.
(iii) cross-piece
In the torsional mode, the cross-piece rotates about its mass centre. The only component of the translational equation, Eq. (II.4), is
1
-2 f - T 2b
=
01 e
and the rotational equation is
(n .14)
Note h0W, in theintegral term of Eq. (II.14), the sheet tension adds
torsional stiffness to the array.
c). lateral.mode
(i) boom (either one)
The boom equation and b0undary conditions are the same as Eq. (II.7)
ex-cept st~ted in the x
3
direction i.e.~I
o4X~
(x"t) _f'
O"X3(X "t)
0\3 (X,,t)
b ax4
1 ax"
~
Pb
2~
-0
1 \with boundary conditions
where X 3 (0, t)
=
0, OX?I(o,t)
o
oX 1J
~~X3(.ft,
t) box
l 1 -e for boom 2 ~2X~(.f,t)
é)x~_
f~
+ dlt3(i,
t)f1
oX 1 1 (ii) sheetThe sheet bending equation
a2.x~
( x, ,
t) Mb _ ()2 Mb 1ox~
Js
é)X~
KG
so
M""-l
tt
()'X3(X;,t)(x'_X )dx' +fel-x,) -
p(x
3ct,t)-
Xl(xot))
bat
2 1 1 1 X, (IL15) ( IL16) . t•
can be reduced to a differential equation by differentiating twice wit4 ~espect
to xl' yielding
a
4x.3(x"
t)=
_1r
-1
è)2x!,
(XI' t) + P~2X3 (x,
,t)
1
~x{
Js
l
at'
ox~
J
with boundary conditions
X 3( 0, t ) = 0 02)(?> (
t,
t) Ox.,
2__ 1_(_ '(
iXl
ct
t) KGSot'
-f ...
P
aX3«(t) ()X, + àx?> (0, t)ox,
+ Pà2X~
Ct,
t) ) () X2 1o
O~ )(>
(t..
t) p ~l X ~ (t,t)
~X, àV -äxt
The boundary eonditions at xl
=
t
eome, from Eq. (II .17) at' Xl=
e
and i ts first derivative at Xl=
t.
(iii) eross-pieee
In the lateral mode, the eross-pieee translates in the x
3 direetion. The two eqmponents of the translationa1 equation, Eq. (11.4), are
where
2.6
Natural Vibrations P=
2 T b e (11.18) (11.19)For the purpose of deseribing boom defleetions, we are interested in the array natura1 modes, i.e., in vibrations whieh are periodie in the three modal direetions • For any array moti,on, i t is neeessary that the differentia1 equations and boundary eonditions derived above be satisfied. However, to deseribe periodie motion, we assume separati0n of variables in the defleetionsi.e.~
where
X2 ("'"
t) '"
E(x,) ~(t) (symmetrie mode)=
M
(x,,'1.1)
~(t) (torsiona.;I.. mode) X 3(x,
,t) = V(x.\)~(t) (lateral mode)..
~ (11.20) (11.21) (11. 22)E, M, and
V
are the mode shape funetions for the symmetrie, torsional and lateral modes respeeti vely. èp (t) deseribes the amplitude of the ,mode.Now, substituting these forms for x
2 and x3 into Eqs. (11.7-11.19), we have for the modal differential equations
(i) boom
The mode shape eCluation f'ound from ECl.
(11.7)
iswith the boundary conditions
E(O)
=
0 E'(O)=
0 and À is def'ined by..
~ (ii) sheet E"(l)=
0 J b E 111 (t )
~
= -
f
~
+ E' ( t)f
~ ~
The mode shape eCluation f'ound from Eq.
(11
.
8)
is(11.23)
(rr.2
4
)
E"
+À. E
=
0
(11
.
25
)
o with boundary conditions
E( 0) = 0 E '
Cl )
~
=
x2 (
t,
t) 1and ~o is def'ined f'rom
..
~
o
(iii) cross-piece
The translation eCluations, f'rom Eqs.
(11.9)
and(11.10)
are1 ,..-f 1
=
Teb- 2
f
~
-
Te
Jb
EI ( t)~
d
w
-b•
b) torsional mode (i) boom(11. 26)
(
11. 27
)
(11.28
)
One boom the same as Eq.
(11.23),
the other boom, same magnitude oppositesign.
(ii) sheet
The sheet mode shape equat10n is the same as Eq.
(11.25)
except f'or the(iii) cross-piece.
(i) boom
The rotationa1 equations are
1 - f = Tb 1 e b
ef~
-
T~
f
VIM'(t,
w)~
dw
-b c) lateral mode (n .29) (11.30)The boom mode shape equation for the lateral mode is the same as Eq.
(II.23) with E-V and subscript 2-3.
(ii) sheet
The mode shape equation is, from Eq. (11.17),
where
v""
+ BV" + CV=
0 B=
C=
KGs J s "P 1 + -KGs-
a
2 '[ IJ's 1 +.-L
KGswith the fo11owing boundary conditions
V(o)
=
0yl/(t)~=
1l-'(V(t)~ +PVII(t)~]
KGs
(n .31)
VI(O)
=
0V"'(e)
~
=~l-f
~ +?V/(t)~1+_1
KG
f'(v(Oi -
pv/ll(t)~l
s
L
(iii) cross-piece
The trans1ationa1 equations in terms of mode shape are, from Eqs. (11.18)
and (n .19)
b
Te.
f
V/(O
~
dw-b (11.32)
The equations for mode shape Q~ the boom or sheet are seen to be, ordi-nary linear differential equations of either seeond or fourth order.
The general solution form of the seeond order sheet equation i.e.,
Eq. (II.25) is
1 +
HCÀ~
X1thecoefficients G and H to be determined from boundary conditions.
(I1.33)
The general solution form of the fourth order boom equation, , i .e., Eq. (II.23) is
(I1.34 )
where
The general solution form of the fourth order sheet equation i.e., Eq. (II.3l) is
v
=
l
S ~t){' + V! C~a)l.1 -+- t-lsh~3)t,1 + Q Ch~3"'1 (I1.35)where
~2
=
l
~
(
1 + (1 -~~)l)J
~?l
•
l
~
((
1 -~)a
1 )1
-
1) a
Now by writing the boundary eonditions for the three modes (i.e., the boundary conditions in Eqs. (II.23 - (II.3l) ) in terms of the general solution
forms just.given, the eondition for a periodic so~ution to exist can be expressed.
The fact tbat a set of linear homogeneous algebraie equations for the eoeffieients of the general solutionforms must have a non-trivial solution is utilized.
This in turn means that the determinant of a certain matrix must be zero. For all
three modes, the algebraic equations for the coefficients in the general solution ean be redueed to ones involving only two unknowns.
a) symmetrie mode
The mode shape is of the form
E ... (II .36)
for each boom and
A
1=
À~2.
S À,at
+À~
À3sn
À!It
A -
.À~ A2. ::À.~
C À2.f.
+ À2.3 ehÀ3~
A2.
and A, G are solutions of[Ds(~)]
[;]
-
[~]
( 11.38)Therefore, for a non-trivia1 solution for A and G we require À such that
ID
(À)\=
0 wheres
\ D
s (À,) \ =1 -
À 3 SÀ~
t
l ( -
J
b À~
+-Te.
b
~
2.) (
A
2 cÀ,
2.t
+ 1\ 1 S À?.t )+
Jb>'~ (-X~À,
eh
À3
e
+ À\À.~
sh
À~i)
+Teb
(-oh,
À, ehÀ~.e
.
.
+ À
1 À.3
sh
À~t
))
+(À.
31.
2 SÀ,2.l -
Áa
À2sn
À.~ ~
(11.39)+ À
~
A., (- c
À2.
~
Teh
À. 3t )) (
Te.
b
À~
c
>-.to.f. -
S À~ ~
f1.p
e
J
b À /(2
f
b)} } / À~
Á
2\ D s (,,) \ is zero when i ts numerator is zero, providing the denominator ~3 Jt 2 '1= O. The advantage of Eq. (11.39) as the form for
ID
(~)\ is that the roots of the numerator can be used for /D(À)\
=
0 where thesnumerator invo1ves no possib1ezero divisors. s.
b) torsional mode
The mode shape for the torsiona1 mode is
(II.40)
for a boom and
(11. 41)
for the sheet.
For the re1ated determinant to be zero we require
\ DT
p.)\ '"
0 where \ DT ("-) \= {-
Àl S À!
t
l (-
J
b
À'.
+T,b
À.)(
11..
c~,t
+1'., ""t)
+J
bÀ3 (- À2. À.,eh
À!It
+A~
X23 shÀ~~)
+ Teb(-Ä
2 }..2eh
À~t
+
Ä~
h.
ash
À~
.f. )
J
+(À
~
A 2. S À~
t -
Ä 2. À 2 S h À!It
(11.42) +>.~À\
(-C>-'z!
+en
X~t))(1 ;~~~ X!
cÀ
1
0i -
s>.ttrltpeJbÀ
)\1
À?,Àa
c) lateral mode
The lateral mode shape can bestated as
(II.43) for ei ther boom.
and
v
=
L(ShX1
+~r
e~2)1.,
-~: sh~~x.~
-~r
eh~aX1)
for the sheet.
(II.44)
For the related determinant to be zero, we require \
DL., (
À') \ ...
0where \ DL (À.) \
=
l
s~2.~
+~r (C~2t
-
eh~3
i) -
~
sn
~3
e ][
Jb (-À
3 2cÀ/, -
À~
1\sÀJ and wher.e _À2À~
eh À3t
+AÀ~ shÀ~t)
+t
(À2 cÀ2.t
+À~.ÀsÀ2.t
- À2chÀsi
+ À3
ÀsnÀ?,l)
+~(SÀ2t -Ac~at- (~:)ShÀ3t
+ .i\ehÀ3t)]
+ :ISr
(1
+:G )
l-
p!
Cf2
e
+~r
(~:
s
P2
t -
p~
sn
~2 ~
) -f~?2 ch~3
e]
2L
s
(II.45)-(is -
~:: )(~2CPtt
-
pr(~2SPzt
+f3sn~3e)
-
P2
ch
P3
t)]
xlSÀzt -
ACÀ2t -
~: ShÀ3~
+A
ch~3t1
-~~ S~2.t
-P2P3Sh~3t
- K
1sP2t
+K,-i:-
sh~?,l
~~Cf2i
+p~chp"&t
+K1C~2,e
-
K,ch~3i
The nat~ral frequencies of the solar arr~ in the three types of modal
vibrations are found numerically by starting ~ at zero and increasing it ,until the ,value of the critical determinant changes sign. The À atlthis point represents
the existence of a natural vibration of frequency SJ,
=
(J
b
'AI
fb
)V2.
The mode shapesare now available to withi~ a constant multiplier
i To complete their specification, the mode shapes are later normalized so that
J
EI dm z:: 1 etc. All parts of thecomputer calculations in which roundoff error can be critical are performed in
2.7
Results and Discussion\
Figures
8
to11
indicate typical results of the calculations. Theten-.sion in the substrate* can be varied for a given array without changing other array
parameters. Therefore the dynamic characteristics of the array can easily be
modi-fied by changing the tension. From a design standpoint, the variation of natural
frequencies with this parameter is an important behavioural pattern. The array parameters selected below are representative of a Hughes FRUSA.
l
=
18
.
2
ft. ~=
0
.
25
')
lb/ft.2 e=
6
ft. E=
7
x10
6
psi s fb=
0.1949
lb/ft.Y
s=
0
.
4
J b3.32
x105
lb-in.2
=
A plot of the first three modal frequencies versus sheet tension for the symmetric and torsional modes is shown in Fig.
8
.
As an interesting compari-son, Fig.9
shows the frequency versus tension plot for the sheet stretchedbe-tween two fixed points. The symmetric and torsional mode frequencies are close
to those of the sheet alone. Figure
10,
a plot of mode sha~es at Te=
3.12
lb./ft. also indicates the closeness to pure sheet modes in the symmetric and
torsional vibrations.
Figure
11
shows that for lateral vibrations, thefrequencies arecon-siderably higher than in the symmetric or torsional modes and that they are
not strongly dependent on tension. The first mode, see Fig.
10,
is primaritythat of the booms alone, while mode 2 is a combination of both sheet and boom motionin closer relative proportions. Calculations made using a boom stiffness and
sheet tension four times those of the array just described (allother parameters the same) gave frequencies roughly twice as large, as would be expected.
It is clear that increasing the tension to too high a value will re sult in bOGm buckling. An indication of some Euler critical load values would be useful here. The Euler critical loads (see Ref.
23)
in terms of sheet tensionfor various boundary conditions are:
a) pinned-pinned c) T
=
24
.
95
lb/ft e fixed-fixed1
T=
99.4
lb/ft e b) d) pinned-fixed1
T =6.24
lb/ft e sliding-fixed~
T =49.75
lb/ft. eThese are the statie critical loads for common boundary conditions. The appropriate static critical buckling load for the array is case a). It is note-worthy that the first radical frequency drop in Fig.
8
occurs at approximately T=
6.3
lb/ft. i.e., very close to the Euler load for case b). Therefore, fore
practical buckling considerations it would be wisest to consider case b) as the critical load if Euler loads are being used as a basis for design. Case b) represents the closest of the four cases to the actual buckling conditions in
that the booms are cantilever mounted and have the tip force transmitted to them from the sheet which deflects as the booms deflect.
111. EQ,UATIONS OF SATELLITE MOTION
ABour
THE CENTRE OF MASS 3.1 Development of EquationsTh~ equations of satellite motion about the combined mass centre are written in ~erms of the kinetic energy T, pote4tial energy U and generalized coordinates q. from Lagrange's equations.
J
(11101)
An important feature of the analysis is the modelling of energy dissi-pating devices within the spacecraft. The sources of energy dissipation are an
antenna-mounted nutation damper, fuel slosh in the rotor (allother damping
mecha-nisms within the rotor have an effect at least an order of magnitude below that of the fuel slosh, see Ref. 10) and structural damping in the flexible appendage. The nutation damper modelled is a pendulum type damper (see Fig. 12) consisting of
a mass attached to the end of a flexible wire, the mass being free to move through a viscous fluid. This type of damper is considered to be an excellent nutation
damper and is in current use. See Ref.
6
for a detailed discussion of this andother nutation dampers.
The energy dissipation due to fuel slosh occurs primarily in the fuel boundary layer as indicated in Fig. 13. The fuel model is that of a rigid bulk cont.ained in a ~pherical cavity located on the rotor spin axis. Section
3.5
describes the generalized force representation of the boundary layer damping. Energy dissipation in the flexible solar array is very complex~ coming
from, among other things, friction due to boom seam connections and structural
damping within the array material. A linear viscous representation of damping
in structures is not a good model of the actual damping mechanism. However~ it is used to represent the presence of non-viscous damping in many situations be-cause it is a particularly easy form of damping to incorporate into the equations of motion. In actual fact, structural damping in the array booms is better
represented by using complex number theory. See Ref.
24
for a further discussion of this point. The frictional damping in the boom seam is not well representedby viscous damping in that it is not a linear function of deflection velocity and could be a non-linear mechanism changing with the degree of seam contact. In
this study, some calculations will be done with the commonly used viscous
model-ling but only for the purpose of showing how the presence of an array damping
model will affect the stability results. 3.2 Generalized Coordinates
i .e., discrete coordinates representing the motions of rigid parts of
thesatel-lite and modal deform~tion coordinates representing motion of the flexible appen-dages relative to a reference point on the rigid satellite ·body.
The flexible appendage (solar array) deformation modes are the array
natural modes as found in section 11 normalized so that
J
X:
t.. dm=
1.,
X·
"
being a general mode shape. arra~Figure 14 shows the reference axes used to contain the mathematical
simulation. The coordinates relate the axis sets as follows:
Axes Transformations Generalized Coordinates
~
~~
a
2ai
3, 2, 1 Euler Anglesa
3 ',
~
•
g;»
0°2
6
1"
"
"
)~
..
g:p-"
"
"
'f~
0,
0~
..
~
"
"
"
cP?,
1>2.
•
1>1
~
..
~
"
"
"
'(3 0 0~
...
~
"
"
"
0 0,
ÀtThe vector d is described relative to
~s
by array natural modedefor-mations, The deflections of the array relative to the
~
set are described bymodal deformation coordinates i.e.,
X2
=
L
E . (t) EQ' (X1) +L:
}1
ai. (t)M
ai(X 1 ,
w)
Ol. , or =E
E b~ (t)E
b~t){
1 ) +L
)L bi.(t)
M
bi. (X 1 '\11)
(III. 2) X3E
))Qi (t)V
al.
(><1)
(III. 3) orL
y
b~ (t)V
bI ( )( 1 )where E., M" V, are the symmetrie, torsional and lateral mode shapes respectivelyo
l. l. l.
The subscript 'a' refers to the array on the positive dl side and 'bI to that on the negati ve -dl side. E ai and Ebi represent symmetrie mode deflections in the same
direction. ~ai and )Ubi both represent torsional rotations positive along the Si
axis and
Y
.
,
v.
b, are both deflections in the same direction for the lateral mode.
al. l.
These mode shapes .represent deflections of boom, sheet and crosspiece and have a different functional form depending on which element is being represented.
The generalized coordinate ~3 is redescribed by ~3 =,~t + ~ wher~ ~ is
3
.
3
Kinetie EnergyReferring to Fig. 15, the kinetie energy of a group of rigid bodies about the total eentre of mass eau be written as
where
r
=
1 ~m.
p.
~
m
L.."
~_~Z;
over all bodies exeept referenee body m , ~ over all bodies . Therefore," 0 J
or
(III.4)
(III.6)
Now for a system with flexible appendages, the above formulation eau be used by breaking the deforming appendage into infinitesmal elements and ineluding eaeh element as rigid body.
The total kinetie energy for the satellite we are studying ean then be expressed as (see App.
B),
where
m (m-m ) spa spa
2m
~
is over appendage mass~ is over all satelli te parts
J exeept appendage. m 2
=
1 - m spa /m m-m spa (III.7)P
=
r - d- ~ -e
and ~j'
i j
are as stated in App. C.3
.
4
Potential EnergyThe internally stored potential energy is due to the elastic strain energy in the booms and in the deformed tensioned sheet, i.e.,
(111. 8)
where n
=
number of booms Im
=
number of sheets.3.5
Generalized ForcesThe generalized forces originate from the damper spring resistance, the resistance to damper ball motion offered by the damper fluid and the viscous resistance generated in the boundary layer of the liquid propellant. The spring resistance of the damper could equally well have been included as part of the potential energy.
The damper spring and viscous generalized forces are
(III.9) wherè the first terms represent spring resistance and the second terms represent the viscous resistance of the damper fluid.
The fuel viscous resistance generated in the boundary layer of the
fluid in contact with tank wall, causes a torque ~, to appear as the fuel bulk
moves in its cavity. ~ can be found by integrating,over the entire boundary layer,
the force times the appropriate moment arm ab out the fuel mass centre. This
model-ling is similar to th at used in Ref.
8
and yields, in fuel axes, to first order in~l' ~2'
+3
and derivatives • Is, cf 0 0 ~1o
Is,
Cf 0~2.
(111.10 ) o 0 lS~cf;P'!J
where see Ref. 25Pf
=
mass density of fueljJ.
=
viscosity of fuel-Cl
=
oscillation frequency of relative motion of fuel to cavity.=
;
TT R: (4 _
c2) ( 1 _ c2) Y'è~
1l'
R~
(2
+ C 2 ) ( 1 - C 2 / 2moments of area about each axis I =
s2>
c Rf = radius of spherical cavity r
f
=
radius of cylindrical hollowTherefore the corresponding generalized forces are~ to first order
for q j
=
CP" </>, and4>3
respecti vely .( rILll)
Solar radiation and gravity-gradient generalized forces are formulated
in Appendix E.
3.6 Linear Equations
The equations, of motion for the satellite are written fr om Lagrange's
equations on the basis of the derived T, U and Qj and then linearized. The linearization is done by.negl:cting quantities second Grder or higher in~,
a
2,
a3,öl,ó2'~'
1>1'C:P2'
~3'
'1'3 + '(3'Eti,Et,
,'i:f~M~, 'E~i,Vi.and
t:g.eir derivatives.A set of linear equations which allows a more direct investigation of the basic dual-spin criteria is one in which it is assumed that
a) The rotor is perfectly balanced - In the actual manufacture of a
rotor an attempt is made to ~~lance the rotor exactly. Some typical remaining
imbalance figures are 2 x 10 slug-ft2 of unbalance for rotors less than 3000 pounds. See Ref. 10 for further details,
b) Environmental torques are neglected - The torque-free stability
criteria and performance are of significant value in that they represent the
satellitès motion about an inertially fixed satellite angular momentum vector. Environmental torques will change the direction of the angular momentum vector
and as well alter the sate1lites motion about the angu1ar momentum vector.
The fermer wil1 be accounted for by control torques and the latter is negligibleo c) The nominal spin rates of the antenna and array are approximated as zero. - The non-zero nominal angular momenta of the antenna (spin-rate to stay earth-pointing is 7028 x 10-
5
rad./sec) and the ,array (spin rate to staysun-pointing is 2.0 x 10-7 rad./sec) would on1y slightly change the basic nutation
frequency of the satellite and are negligible quantities in the satellite equations.
d) y