You are on page 1of 19

JOURNAL OF GUIDANCE, CONTROL, AND DYNAMICS

Vol. 26, No. 1, January–February 2003

Second-Order Relative Motion Equations

Christopher D. Karlgaard¤ and Frederick H. Lutze†


Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061-0203

An approximate solution of second-order relative motion equations is presented. The equations of motion for
a Keplerian orbit in spherical coordinates are expanded in Taylor series form using reference conditions
consistent with that of a circular orbit. Only terms that are linear or quadratic in state variables are kept in the
expansion. The method of multiple scales is employed to obtain an approximatesolution of the resulting nonlinear
differential equations, which are free of false secular terms. This new solution is compared with the previously
known solution of the linear case to show improvement and with numerical integration of the quadratic
differential equation to understand the error incurred by the approximation. In all cases, the comparison is
made by computing the difference of the approximate state (analytical or numerical) from numerical integration
of the full nonlinear Keplerian equations of motion. The results of two test cases show two orders of magnitude
improvement in the second-order analytical solution compared with the previous linear solution over one period of
Downloaded by OLD DOMINION UNIVERSITY on August 27, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.5013

the reference orbit.

41
I. Introduction

T HE traditionaldevelopmentof relativemotionequationsresults
from a linearized model of orbital mechanics.1 This paper de-
velops relative motion equations that result from nonlinear theory.
Relative motion equations are used often in the�eld of orbital
me- chanics because an explicit dependence on time can be
obtained, whereas the exact theory contains a
transcendentalrelationship be- tween the position and velocity of
the satellite to time. There are other ways to obtain
approximations that lead to this explicit time dependence, but a
relative motion approach is arguably the most conceptually
straightforward of these.
Figure 1 shows the geometry of the relative motion problem.
The goal is to�nd the position vectorR, from a reference
point on a known orbitr 0, to a satelliter, as an explicit function
of time. The relative velocityVmust also be known to solve the
problem completely. Note that it is possible for the reference
point to be occupied by another satellite.
Assuming the magnitude ofRandVto be small compared with
that of positionand velocityof the referenceorbit allows for the gov-
erning equationsof motion to be approximatedby using a truncated
series expansion. Often this series is truncated after only
the�rst term, resulting in a linear expression of the dynamics.
There have been several attempts at developing relative motion
equations that take into accountnonlineardynamics.2¡4 The effectsof
nonlinearity and nonzero eccentricityare investigatedin Ref. 5 for a
certain class of relative trajectories. The solutions presented in
Refs. 2–4 result from a straightforward expansion, a method that
constructs a trun- cated Taylor series representation of the exact
solution. Although these solutions are more accurate than the
solutions resulting from the linearized model, the validity region is
limited to short periods of time due to the presence of secular
terms. For a�xed time, so- lutions derived from the
straightforward expansion will converge to the value of the
exact solution as the number of terms in the expansion grows.
For a�xed number of terms, however, the accu- racy is
diminished as time increases and the secular terms begin to
dominate over the solution.

Received 4 December 2001; revision received


° 1 July 2002; accepted for
publication 14 August 2002. Copyright c 2002 by the American Institute
of Aeronautics and Astronautics, Inc. All rights reserved. Copies of this
paper may be made for personal or internal use, on condition that the
copier pay the $10.00 per-copy fee to the Copyright Clearance Center,
Inc., 222 Rose- wood Drive, Danvers, MA 01923; include the code 0731-
5090/03 $10.00 in correspondence with the CCC.
¤
Graduate Assistant, Department of Aerospace and Ocean Engineer-
ing; currently Project Engineer, Analytical Mechanics Associates, Inc.,
303 Butler Farm Road, Suite 104A, Hampton, VA 23666. Member AIAA.

Professor, Department of Aerospace and Ocean Engineering. Associate
Fellow AIAA.
The secular terms resulting from the straightforward expansions
appear in all components of the relative trajectory. A secular drift is
expected in the in-track direction because the two orbits will
generallyhave differentperiods; however, secularterms in the radial and
cross-track directions are not acceptable. A secular trend in the
radial direction is clearly not a valid result for elliptical orbits, and a
secular growth in the cross-track direction implies that the orbital
plane is not�xed, which violates the conservationof angular
momentum in the two-body problem. The secular terms that appear in
the radial and cross-track directions are hereafter referred to as
“false” secular terms for these reasons. False secular terms are also
discussed in Refs. 5 and 6.
Another problem with the preceding nonlinear solutions is the
presence of a constant offset in the cross-track motion. This con-
stant shifts the plane of the relative orbit in such a way that it does
not always contain the center of mass of the central body. This shift is
clearlynot physicallyvalid. This constantis explicitlygiven in the
solutions of London2 and Anthony and Sasaki,3 and a similar con-
stant for the speci�c class of trajectories investigated in Ref. 5
also appears. It is not clear if such a constant is contained in
Kelly’s solution4 because the�nal solution itself is not
presented. This constant offset is also apparent in the numerical
integration of the second-order relative motion equations shown in
Fig. 7 of Ref. 7.
Severaltechniquesexistthatcan be used to rid the solutionof false
secularterms, oneof whichisthe methodof multiplescales, 8;9 which
assumes that the solution will be a function of several timescales,
each of which is independentof the others. This techniqueis a blend of
the Lindstedt–Poincare´ technique,which is based on the observa- tion
that the frequency of a nonlinear oscillation may depend on its
amplitude, and the method of variation of parameters, which allows the
constants of integration in the linear solution to be functions of
time. The method seeks to determine differential equations that
describe how these “time-varying constants” must behave in such a way
that the solution will be free of false secular terms and, thus,
eliminate the issue of divergence.
To applythe methodof multiplescales, a modelof the relativemo- tion
problem must�rst be developed from the full, nonlinear equa- tions
of motion. Traditionally, the relative motion problem has been
approachedusing a rotating rectangularcoordinate system with the
origin at the reference point. An angular coordinate in the in-track
direction provides an increased level of accuracy for large angular
separations, as indicated in the results of Berreen and Crisp 10 and
Melton6 for the case of cylindrical coordinate systems. For a spher-
ically symmetric force�eld such as that encountered in the
two- body problem, however, a spherical coordinate system is a
natural choice.

II. Equations of Motion


The equations of motion for a particle in a Keplerian orbit ex-
pressed the spherical coordinate system shown in Fig. 2, wheree x ,
4 KARLGAARD AND

nondimensional variables, can be written as

1 Or00 ¡ 31Or¡ 21µO D ².¡31 Or2 C 21 Or1µO ¡ 1ÁO 2 C 1µO C


0 0 02

1ÁO /
02

(14)

1µO C 21 Or0 D ².21 Or1Or0 C 21ÁO 1ÁO ¡ 21µO


00 0 0
Fig. 1 Geometry of the rela-
tive motion problem. 1Or0/(15)

1ÁO C 1ÁO D ².¡21µO 1ÁO ¡ 21ÁO 1Or0/(16)


00 0 0

where ( )0 is the derivative with respect to the nondimensional time


variable¿and²is a place-keeping parameter to be used in the
method of multiple scales to ensure proper ordering. It is important
to point out that because the variableµis ignorable, no assumption
is made with regard to the relative magnitude of the variable1
µO;
hence, the approximate equations of motion (14–16) remain valid
for large in-track angular separations, which is an advantage of
the spherical coordinate formulation of this problem. Note that,
if²is set to 0, Eqs. (14–16) reduce to the familiar linear form. Note
Fig. 2 Spherical coordi-
also that the out-of-planemotion is linearly decoupled from the in-
nate system.
plane motion. The couplingof motion is an advantageof the second-
order approximationof the dynamics. After the method of multiple
scales is applied to this problem, the value of²will be set to 1.
Before an attempt is made to�nd a nonlinear solution,
however, it is�rst necessary to review and understand the
linear solution.

ey , ande z represent the reference frame�xed at the center of III. Linear Solution
mass of the primary body with gravitationalparameter¹, can be When²is set to 0, Eqs. (14–16) reduce to
written as
1 Or00 ¡ 31Or¡ 21µO D 0
0
PrD vr (1) (17)

1µO C 21 Or0 D 0
00
µP D ! µ (2) (18)

1ÁO C 1ÁO D 0
00
ÁP D !Á (3) (19)

vPr Dr !
Á
2
Cr !µ 2 cos2 Á ¡ ¹=r 2 (4) These equations are effectively the same equations found in
Ref. 1, although in terms of spherical coordinates rather than rect-
!P µ D ¡2vr !µ =rC 2!µ !Á tanÁ(5)
angular. The solution of these equations by Laplace transforms, for
example, is straightforwardand may be found in Ref. 11. The result
may be be expressed in terms of a state transition matrix,
0 1
1 Or.¿ /
!P D ¡2v ! =r¡ !2 B
sinÁcosÁ(6) 1µO .¿ /
Á r Á µ B 1ÁO .¿ / C
C
Note that the variableµis ignorable in the sense that it does not D
1vOr .¿/
enter into the dynamics, although it is still important to know its B
value to determine the relative position. Equations (1–6) can be ex- @B1!O µ .¿A /
panded in Taylor series form about a reference trajectory to obtain C1!O Á.¿ /
an approximate representationof the dynamics. For a circular refer- C
ence orbit these conditionsare rref Dr 0,Á ref D 0,v rref D 0,! µref D 2 3
4 ¡ 3 cos¿0 0 sin¿2.1 ¡ cos¿/0
! 0,
and! Áref D 0. It is also convenient to introduce the 6 7
6 ¡ 6.¿ ¡ sin¿ /1 0 ¡ 2.1 ¡ cos¿/4 sin¿ ¡ 7
nondimensional variables
3¿0 0 0 cos¿0 0 sin¿
6 7
¿ D ! 0t (7) 3 sin¿0 0 cos¿2 sin¿0
6 7
/=
4 ¡6.1 ¡ cos¿/0 0 ¡2 sin¿4 cos¿ ¡ 3 5
1 OrD .r¡r 0 0
r 0
(8) 0 0 ¡ sin¿0 0 cos¿

1µO D µ ¡ µ0 (9) 0 1
OD 1 Or00
1Á Á (10)
B 1µO00
µ0
0

1vOr D vr =r0!0 (11) B 1ÁO 0


C 0
1 !O µ D .!µ ¡ !0 /=!0 (12) C B 1vOr00
£B C
@C
A
KARLGAARD AND 4
(20)
1 !O Á D !Á =!0 (13) 1!O Á00

and hence, the initial conditions of the problem may be expressed or, alternately, in the complex form
as1 O r.0/ D1 Or00,1 µO .0/ D 1µO0 0 ,1 ÁDO .0/ 1ÁOO 00 ,1 D vr .0/
1vOr00 , 1! .0/
DOµ O 1! ,
µ00 Oand1 !
DÁ .0/ 1!Á00 . 1 Or.¿ / DAe i¿ C AN e¡i¿ CB (21)
The Taylor
O series expansion of equations (1–6), truncated after
terms with quadratic nonlinearity and expressed in the preceding 1µO .¿ / D 2i Aei¿ ¡ 2i AN e ¡i¿ ¡ 3
B¿ CC (22)
2
Downloaded by OLD DOMINION UNIVERSITY on August 27, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.5013
4 KARLGAARD AND

1ÁO .¿ / DDe i¿ C DN e¡i¿ (23) @


C 1 ÁO 0 D 0 (46)
1vOr .¿ / Di Ae i¿ ¡i AN e¡i¿ (24)
2
1ÁO 0
@ T02
@ 1 µO0
i¿ N @ 2 1 Or1 @ 1µO1 @ C2 ¡
¡i¿ 3 Or2
¡ 3 1 Or
1 ¡ 2
Or0 2 D ¡2 3 1
1 0
1 !O µ .¿ / D ¡2 Ae ¡ 2 Ae ¡ @T2 @T @T@ @T
T
2B (25)
0 1 1
1 !O Á .¿/ Di De i¿ ¡i DN e¡i¿ 0 0
(26)
@ 1µO0 � �
C 21Or ¡ 1ÁO 2
2 ´ @2 (47)
C
C ´ 1ÁO0
@ 1µO0
where 0
@ 0
@ T0 @ T0
T0
1
AD aei® (27)
2 @ @ D ¡2 @ 2 1µO @ @ 1 Or0
@ T @0 C @T
2
1µO1 C 12@Or1
T T ¡ 12@Or0
T
@T2 21Or0
BDb (28) 0 0 0 1 1 0

CDc (29)
C @ 1 ÁO 0 @ 1 µO 0 @
(48)
DD 1
dei± (30) ¡ 1 O2r0
2 21ÁO 0
@ T0 @ T0 @ T0
@
O 1µ0 @ 1ÁO
@ 1ÁO
2 @ 1ÁO
2
@1Or0 0
1 0 @ —2 @ @
with AN and D
N being the complex conjugates of A and D and C 1ÁO 1 D ¡ (49)
@T @ T0@ O0 T0 T0 T0
¡2 T1 21Á
2
a2 D The method of multiple scales has replaced one system of non-
¢2 linear ordinary differential equations with a sequence of systems of
¡ C µ00 C (31)
31Or00 21!O 1vO 2 r00 0

d 2 D 1ÁO 2 C linear partial differential equations. The�rst of these partial differ-


00 Á00 (32)
1 !O
2 ential equations is homogeneous, whereas the second system has a
¯
sin® D ¡ 1vO a forcingfunctionthat dependsonly on the solutionto the�rst system.
r00 (33)
Each of these linear systems is effectivelythe same as Eqs. (17–19),
¢¯ except that the constants of integration will no longer be constants
cos® D C a (34)
µ00 but functions of T1. Therefore, the solution of Eqs. (44–46) may be
¡ 21!O
¡ 31 Or00 written as
sin± D ¡ 1 !O Á00 =d (35)
1 Or0 DA .T1/ei T0 C AN . T1 / e¡i T0 CB .T1/(50)
cos± D 1ÁO 00 =d (36)
1µO0 D 2i A.T1/ei T0 ¡ 2i AN .T1/e¡i T0 ¡ 3 B.T1/T0 CC .T1/(51)
2
bD 41 Or00 C µ00 (37)
21!O 1ÁO 0 DD .T1/ei T0 C DN .T1/e¡i T0 (52)
O
cD 1µ00 ¡ 21vOr00 (38)
where
The next section will apply the method of multiple scales to the
second-orderequationsof motion(14–16) to determinean analytical A.T1/ D 1
a.T1/ei®.T1
/
2 (53)
solution that is free of false secular terms. The linear solution found
in this section will be used as the foundation for developing the 1 /
D.T1/ D d.T1/ei±.T1 (54)
nonlinear solution. 2

IV. Nonlinear Solution Six more initial conditions will be required to solve the next
To apply the method of multiple scales to the nonlinear problem system of partial differential equations. Without loss of generality,
(14–16), it is�rst necessaryto de�ne the fast timescaleas D the approach taken here is to pick these initial conditionsto be such
T0 ¿and the slow timescale
D as T1 ²¿. Next, the state that the constantsof integrationare independentof², that is, that the
variablesare expanded as functions of T0 and T1 by constants in the nonlinear problem are de�ned in the same
manner as in the linearproblem. The initial conditionsof the
1 Or.¿ / D 1 Or0 .T0 ; T1/ C ²1 Or1 . T0 ; T1/ C ¢ ¢ ¢ (39) overallproblem (to order²) are
1µO .¿ / D 10µO 0.T ;1 T / C ² 1µ
1
O 0.T ;1 T / C ¢ ¢ ¢ (40) 1 Or.0/ D 1 Or0 .0/ C ²1 Or1 .0/(55)
1ÁO .¿ / D 10ÁO 0.T 1; T / C ² 1Á
1
O0 .T 1; T / C ¢ ¢ ¢ (41)
1µO .0/ D 1µO0 .0/ C ²1µO1 .0/(56)

Note that the derivative is also expanded (to order²) by the chain 1ÁO .0/ D 1Áµ
O 0 .0/ C ²1ÁO1.0/(57) ¶
rule, so that
1 vO .0/ D .0/ C ² .0/ C .0/ (58)

@
KARLGAARD AND
@ 1 O r0 @ 1 O r0 4
@ 1 Or0 @ 1 Or0 @ r
@ T1Or1
@@1 @ T0
1 Or0 D C C
(42) T0 µ ¶
�@ @ ´ @ @
1 Or1 C C 0 (59)
² 1µO0
1µO1

@ 2 1 Or0 � @
1!µ . 0/ D . 0/ ² . . /
@ 1 Or1
2
O0 @ / 0 @ T0
´ O 1µ T0 @ T1
@ 21Or1
@ µ @
1 Or00 C2 (43) D @ C
C @ T0 @ T1 @ 0
D T0 2 O O
@
² 2 T O 0 1Á 0 0 C ¶ 1ÁO0 0 1Á 1 0 (60)

1!Á . . ² . .
@ @ @
Similar expressions exist for derivatives of1 ÁOas well. / / / /
Substituting Eqs. (39–43) into Eqs. (14–16) and equating orders of T0 T1 T0
O
µ and1 For the initial conditions to be independent of², it must be true
²gives the following systems of partial differential equations:
that
@ @ 1 µ O0
2
1 Or0 — 31 Or0 ¡ 2 D0 (44) 1 Or1 .0/ ´ 1 Or10 D 0 (61)
@
@ T02 T0 1µO1 .0/ ´ 1µO10 D 0 (62)
@ 1 Or0
Downloaded by OLD DOMINION UNIVERSITY on August 27, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.5013

@ C2 D0 (45)
@ T0 1ÁO 1 .0/ ´ 1ÁO10 D 0 (63)
2
1µO0
@ T02
4 KARLGAARD AND

@ 1 O r1 ´ ¶ µ
.0/ ´

@ 1 O r0 @A 19
r10 D¡ .0/(64) — 2i 1 ! O T ei T0 C ¡9i AN 2 ¡
@ O
1v C 3 AB¡ 2i i
T0 AN B
µ10 0
@ @ 2
@ 1µO1
.0/ ´ @ T1
1 µO 0 15 17 T1 @ AN
D¡ .0/(65) — i B2 ¡ 2i DN 2 ¡ i AB¡ 5i A2 ¡ 2i D2 ¡ 6i A AN C 5
1!O µ

@ T0 10 4 2 @ T1
@ T1
@ 1 ÁO 0
@ 1 ÁO ´ O D¡ @A C 7O @ B C
1v C
1 !2O i
1
— 3 @C C 2i
@ T0 .0/ 1!Á10 .0/(66)
@ T1 @ @
r10 µ10
@
T1 � T1 T1 µ ¶
´ ¶ @ AN 5 1
Substituting Eqs. (50–52) into Eqs. (47–49) gives C 3 ABC 2i — iA C 2
iD 2
e2i T
T e¡i T
� ´ N 0
0 0

@ 21 O @ @ @ 2 2
— 31 Or1 ¡ 2 O01 D 2i — 7 AB eiT
r1@ T 1µ
@ ´
A
@ 0 µ 1 ¶
0
1
5 3 3
C i AN 2 i i A2 C i AN2 ¡i ABCi AN B
� @N ¡i T 2
2 2iT e ¡2i T0
¡ 2
A C D 2

2 2 2
C ¡2i —7 AN B 0
C . ¡3 A — 2 D / e 0

@ @
e 1 i DN 2 ¡ 2 — N@ @B
T1 C i D2 ¡ A2 — 14 —21vOr
A
@C @B
C .¡3 AN2 ¡ 2 DN 2/e¡2i T0 C 2 15B2 —6 A AN¡ 3 T 2 2 @ @ 10
T1 @ T1
¡ T1
µ

0
@ @ 4 @ @B 2
2
1µO1 T1 T1 (67) C6 T C 9 A2 C 9 AN2 C 15ABC 15 AN BC 3 D2 C 3 DN
@ 2
� ´
@ 1 Or T1
C C 5i AB ei T0 ¶
1 0
N
2 15
@ C B 2 C 12AN A¡ T
A 31!O @ @A @
D 2
@ T02 @ T0 @ T1 µ10 C A 0
2 — 4 @ T1 — 6i @ T1 C 6i @
� ´ T1 (71)
@ AN
C 2 e¡i T0 C .6i A2 C 2i D2/e2i µ
@ 3
AN B
¡ 5i T1 T0 D B DC 3 AD¡ AN D¡A

@B 3 1
1Á 1 DN AN DN ¡ DN B¡ i
C .¡6i AN2 ¡ 2i DN 2 /e¡2i T0 ¡ 2 4 2
µ10
O 4 ¡ 1!O
(68) � ´ ¶
@ 1 @D 1 @ 3 @D
T1 — i C i ¡ i B D¡ T0 ei
� T0
@ ´ DN2 @ 2 @ T1 2 @ T1
2 1
1ÁO C 1ÁO 1 ¡2 @ C 3DB T1
D@ e i T0 3 1
T
@ D2 µ
0
i T1 3
C B DN C 3 AN DN ¡A DN ¡ AN D¡AD¡ D BC 1 !
iO
� ´ 4 4 2
µ10
@ DN
C 2i C 3 DN B e¡i T0 C 6 ADe2i T0 � ´ ¶
@ 1 @ 1 @D 3
C i — i C i B DN C T e¡i T0
T1
C 6 AN DN e¡2i T0 C 2 AN DC DN @ DN
2 A DN 2 @ 2 @ T1 2 @ T1 0
(69) T1
The solution of Eqs. (67–69) is — 2 ADe2i T0 ¡ 2 AN DN e¡2i T0 C 2 (72)
A DC 2 A DN
N
µ
9 23 15 17 5 For the solution to be free of false secular terms in the radial and
1 Or1 D A2 C ABC B2 CD 2 C 3 A AN C AN BC
2 4 8 4 2 out-of-plane directions, it must be true that

AN 2
KARLGAARD AND 4
5 @A 7 @B 3 @ AN @C 1 @ A D ¡ 3i AB (73)
N2
CD ¡ i i C i — — i 1vOr10 @ T1 2
2 @T 1 —2 @ 1 2 @ 1 @ 2 @B
T T T1
µ
— 1!O �
3 9 D0 (74)
@A ¶ AN
2
23 N @ T1
µ10 ¡ e i T0
i ABC @ 2 4 AB @D 3
2 ´ T0
T1 C C

C B2 C DN 2 C 3 A D ¡ i DB (75)
ANC
15 17 5 5 @ @ 2
ABC A CD C AN
2
i
2
T1
8 4 2 2 @ and to satisfy Eq. (65), the condition
T1
@C 15 3 9
7 @ D B2 C 3 A AN C .D2 C DN / C
2
. A2 C AN /
2
B 3 @ @C 1
A 1
C i — i —
2 @ 2 @ T1 @ C i 1vOr10 ¡ 1 !O @ 8 4 4
1
2 T1
T @ AN T1
µ10 3¡ ¢ 3
C AN B ¡ 1 !Oµ10 (76)
Downloaded by OLD DOMINION UNIVERSITY on August 27, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.5013

� ´ ¶ 2 2
ABC
must alsobe enforced.Equati ons(7 3–75) can be in tegratedto obtain
C
3 i AN B¡ T0 e¡iT0 ¡A 2e2i T0 ¡ AN2 e¡2i T0 ¡ 6 A2
@ aDa 0 (77)
2
BDb 0 (78)
15
— 6 AN2 ¡ 10 AB¡ 10AN B¡ 2 D2 ¡ 2 DN 2 ¡ 6 A AN¡ B2
dDd 0 (79)
4
@A (80)
@ @C @B
C 4i 4i
T1 AN C 2 C 2 1 !O µ10 ¡ 7 T0 (70)
@ ¡
µ @ T @ 1 @1 T
T1
19 15 17
O 9i A2 C i ABC i B2 C 2i D2 C i AN BC 5i ® D ¡ 3 b0 T1 C ®0
AN 2
1µ1 D 2
2 4 2
3
± D ¡ 2b0 T1 C ±0 (81)
@A @ AN @C
i D2 i A A i @B
C 2 N C 6 N C 5 ¡ 3 ¡ 2 where a0, b0, d0,® 0, and± 0 are constants given by Eqs. (31–37).
@ T1 C 7 T1 C
r10
@ T1 @ T1 Substituting these results into Eq. (76) gives
1vO @
4 KARLGAARD AND

@C 15 3 3
D b2 C a2 C d2 cos.2± ¡ 3b T ¡ ¯ ¢
3
0 0 1 C a2 b0 [sin.3b0T1 ¡ 2®0/ C sin 2®0]
@/ 0 0 0
8 4 8
T1
9 3 ´
C a2 cos.2® ¡ 3b T / C a b � £ ¡ ¢ ¤
3 3
cos — bT Ca 0 0sin b0 T1 ¡ ®0 C sin® 0 Cc 0 (84)
8
® 2
0 0 0 1 0 0 0 0 1 where is given by Eq. (38). The initial conditions1 r10 ,1 Á10 ,
8 2 2
c0 vO !O
and1 !O µ10 can be found from Eqs. (64–66) as
3
— 1!O µ10 (82) 1v r10 D ¡ 3 2a0b0 sin® 0 (85)
4
If b0 D 0, then O 2 2
¡ ¢
3 1 !O µ10 D ¡.15=2/b0 ¡ 3a0 ¡ 18a0b0 cos® 0
CD a2 C 3 d0 2 cos 0 C 9 a2 3
0 — 4 1 !O µ10
T1 Cc (83)
4 0 8 8 0 cos 0


— 29 a2 cos 2®0 ¡ 32d20 cos 2±0 (86)
or if b0 6D 0, then 0
£ ¤
CD .15=8/b0 C 4 a0 ¡ 4 1 !O T1
2 3 2 3 1 !O 10Á D ¡2 3 d0b0 sin± 0 (87)
µ10¡ ¯ ¢
C 1 d2 b0 [sin.3b0T1 ¡ 2±0/ C sin Substitutingall of these resultsinto Eqs. (39–41) and setting² D 1
0
8 gives the�nal
2±0]

¡ ¢ £¡ ¢ ¤ £¡ ¢ ¤
1 Or.¿ / D ¡ 31 Or00 C 21!O µ00 cos 1 ¡ 61 Or00 ¡ 31!O µ00 ¿ C 1vOr00 sin 1 ¡ 61 Or00 ¡ 31!O µ00 ¿ C 41 Or00 C 21!O µ00
n¡ ¢ ¡ ¢ £ ¡ ¢
C sin.¿ =2/ 121 Or00 1vOr00 C 61vOr00 1 !O µ00 cos.¿=2/ C 61 Or00 1vOr00 C 41vOr00 1 !O µ00 cos 81 Or00 C 4 1!O µ ¡ 1
¤
¿
3
2 00
¡ ¢ £¡ ¢ ¤ ¡ ¢ £
¢ ¤
¡ 41 Or00 1vOr00 C 21vOr00 1 !O µ00
C cos 121 Or00 C 61!O µ ¡ 1 ¿ ¡ 41 Or00 1vOr C 21vOr 1 !O µ cos
C1 ¿
121 Or00 C 61 !O µ 00
1 1
2 00 00 00 00 2
£ ¤ £ ¡ ¢ ¤ £
1ÁO 00 1 !O Á ¡ .15=2/1 Or00 1vOr ¡ 51vOr 1 !O µ
1 1 3
C cos 241 Or00 C 121!O µ ¡ 1 ¿ C 1 Or00 1vOr C 1vOr 1 !O µ
00 00 00 00 00
2 00 2 00 2 00

¤ £ ¡ ¢ ¤ £
— 21 1ÁO 0 0 1!00O Á cos
2
1
241 Or00 C 001 21!O µ C 1 ¿ C 00.57=2/1
2 rOr2
00 C 3 1vO 2 C 00421 Or00 1!
µ00O µ 2C 1 21!
00 O
2
¡ 3 1ÁO 2

¤ ± ² £ ¡ ¢ ¤ ±
²
C 3 1 !O 2 sin.¿ =2/ ¡ 121 Or2 C 141 Or00 1!O µ C 41 !O 2
sin 1
121 Or00 C 61 !O µ ¡ 1 ¿ C 121 Or2 2
2 Á00 00 00 µ00 2 00 00 00 C 4 1 !µO00
C 141 Or00 1!O µ

£ ¡ ¢ ¤ h i
£ sin 2
1
121 Or00 C 6100!O µ C 1 ¿ 4 C r00 1 1v 2
4 O 00 ¡ 1 1ÁO 2 ¡ 31 Or00
¡ 9 14Or2 00 00 1!O µ
µ00 ¡ 14!O
2
C
Á00
1
1 !O
2

£ ¡ ¢ ¤ h i
00 1ÁO C 151
1 5 2 1 2
£ sin 2 00 O µ C 1 ¿
241 Or00 C 121! C .45=4/1
00 4Or2 r¡00 1v
4O C µ C451 !ÁO002 ¡ 1 1 !O
00Or00 1!Oµ00
2

£ ¡ ¢ ¤ ± ² £ ¡ ¢ ¤o
1 2 2 3
£ sin 241 Or00 C 121!O µ ¡ 1 ¿ C 1vO ¡ 91 Or2 ¡ 121 Or00 1!O µ ¡ 41!O sin 81 Or00 C 41 !O µ ¡ 1 ¿ (88)
2 00 r00 00 00 µ00 2 00

£¡ ¢ ¤ ¡ ¢ £¡ ¢ ¤ ¡ ¢
1µO .¿ / D 21vOr00 cos 1 ¡ 61 Or00 ¡ 31 !O µ00 ¿ C 61 Or00 C 41!O µ00 sin 1 ¡ 61 Or00 ¡ 31!O µ00 ¿ ¡ 61 Or00 C 31!O µ00 ¿
CC .¿/

n± ² h
00 O C 841
2 2 2 2 5 2 1
C sin.¿ =2/ Or2 C 31v
57100 r00 O ¡ 31Á 00 Or2001 !O µ C µ241!
00 O C
Á0031!O cos.¿=2/ C .45=2/1
00 2 Or2
r00 ¡ 1v
2 O 00 C
1ÁO 2

i £ ¡ ¢ ¤ ± ²
2 1 2 3 2
C 301 Or00 1!00O µ C 101!
µ00 O 2 ¡ 1Á!
00 O cos
2 00 41 !O µ ¡ 1 ¿
81 Or00 C ¡ 241 Or2 C 281
00 00 µ00 O µ C 81 !O
Or00 1!

£ ¡ ¢ ¤ ± ² £ ¡ ¢ ¤
1 2 1
£ cos 121 Or00 C 61!O µ ¡ 1 ¿ C 241 Or2 C 281 Or00 1!O µ C 81!O cos 121 Or00 C 61 !O µ C 1 ¿ C
2 00 00 µ00 2 00 r00 00
± KARLGAARD AND 4
2
21vO
00 — 181 Or2

² £ ¡ ¢ ¤ h
i
— 241 Or00 1!O µ ¡ 8 1 !O 2
cos
1
241 Or00 C 121!O µ ¡ 1 ¿ C 1
1vO
2
¡ 9 1 Or2 ¡ 1 1ÁO 2 ¡ 61 Or00 1!O
µ00 r00 00 00 µ00 C 1 1 !O 2
µ ¡ 2 1 !O 2 00 2 00 2 2 2 00
2 Á00

£ ¡ ¢ ¤ ¡ ¢ ¡ ¢
1
£ cos 241 Or00 C 121!O µ C 1 ¿ ¡ 241 Or00 1vOr C 121vOr 1 !O µ sin.¿=2/ ¡ 81 Or00 1vOr C 41vOr 1 !O µ
2 00 00 00 00 00 00 00
£ ¡ ¢ ¤ ¡ ¢ £ ¡ ¢ ¤
1 1
£ sin 121 Or00 C 6 1 !O µ ¡ 1 ¿ C 81 Or00 1vOr C 41vOr 1 !O µ sin 121 Or00 C 61 !O µ C 1 ¿
¡ 2 00 2 00
— 31 Or00 1vOr00
C 1ÁO 00 1 !O Á
¢ £ ¡ 00 00 00
¢ £
00
¢ ¤
— 21vOr00 1 !O µ00 sin 241Or00 ¢ ¤ ¡ 1 !O µ00 sin .241 Or00 C 1 21!O µ ¡ 1 ¿
C 121!O µ 00 C 1 ¿ ¡ 121 Or00 1vOr00 C
1
2 00
1
2 81vOr00
¢ £ ¡
C — 1ÁO 0 0 1!O Á00 sin 81 Or00 C ¢ ¤o
¡ C 1 !O —1 ¿ (89)
41!O µ 00
151Or00 1vOr0 101vOr00 µ00 3
Downloaded by OLD DOMINION UNIVERSITY on August 27, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.5013

2
0
4 KARLGAARD AND

£¡ ¢ ¤ £¡ ¢ ¤ n¡
1ÁO .¿ / D 1ÁO 00 cos 1 ¡ 61 Or00 ¡ 31 !O µ00 ¿ C 1 !O µ00 sin 1 ¡ 61 Or00 ¡ 31 !O µ00 ¿ C sin.¿=2/ 121 Or00 1!O Á00 C 61 !O µ00
¢
1 !O Á00
¡ ¢ £ ¡ ¢ ¤ ¡
£ cos.¿ =2/ C 61 Or00 1!O Á00 C 41!O µ00 1 !O Á00 ¡ 21vOr00 1ÁO 00 cos 81 Or00 C 41!O µ ¡ 1 ¿ C 1vOr 1Á00 ¡
31 Or00 1!O Á
¢ £¡ ¢ ¤ ¡
3
2 00 00
O ¢ £ 00
¢ ¤
¡ 21 !O µ00 1 !O Á00
— cos 241OrC 121!O µ ¡ 1 ¿ C 1vOr 1Á00 ¡ 31 Or00 1!O Á ¡ 2 1 !O µ 1 !O Á cos
C1 ¿
241OrC 121!O µ
1
2 00 00
O 00 00 00
1
2 00
¡ ¢ ¡
C 21vOr00 1 !O Á00 ¡ 61 Or00 1ÁO 00 ¡ 41ÁO 0 0 1!O µ00 sin.¿=2/ ¡ 1vOr00 1 !O Á00 C 31 Or00 1ÁO 00
£ ¡ ¢ ¢ ¤
1
£ 21ÁO 00 241
C sin µ00 C 1 21!O µ C 1 ¿
1 !O Or00 ¡
¡
1vOr ¢ £ ¢ ¤
2 00 00 1!O Á00 C 31 Or00 1ÁO 00 C sin
1
.241 Or00 C 1 21!O µ ¡ 1 ¿
21Á¢O 00 1!
£ O ¡µ00
¡ 1 !O Á00 sin 81 Or00 C ¢ ¤o
2 00
C 61 Or00 1ÁO 00 C 41ÁO 00 C —1 ¿ (90)
41!O µ
1 !O µ00 £¡ 21v Or00 ¢ ¤ ¡ 2
3 ¢ £¡ ¢ ¤ ¡
1vOr .¿ / D 1vOr00 cos 1 ¡ 61 Or00 ¡ 31!O µ00 ¿ C 31 Or00 C 21!O µ00 sin 1 ¡ 61 Or00 ¡ 31!O µ00 ¿ C 61vOr00 1 Or00
¢
C 31vOr00 1!O µ00 cos¿
£ ¤ £¡ ¢ ¤ ¡ ¢
1ÁO 00 1 !00O — .27=4/1 Or00 1vOr — 9 1vOr 1!O
1
C cos 121Or00 C — 1 ¿ ¡ 41 Or00 1vOr00 C 1 !O µ00
4
Á £¡ 00 ¢ ¤ ¡ µ00 61 !O µ00 ¢ £ ¡ 21vOr00¢ ¤
£ cos 61 Or00 C 31!O µ00 ¡ 1 ¿ C 61 Or00 00
1vOr00 C 41vOr00 1 !O µ00 cos 2 61 Or00 C 31!O µ00 ¡ 1 ¿ ¡
2
¡ ¢
21 Or00 1vOr00 C 1vOr00 1 !O µ00 ±
¢ £¡ ¢ ¤
£¡
¢ ¤ ¡ cos 121Or00 C 61 !O µ00 C 1 ¿ C .57=4/1 Or
£ cos 61 Or00 C 3 1 O 1 2
C1 ¿ C 1 Or00 1vOr C 00 00
31!O µ00 1 ! O µ00 ¡ 1Á
Á 0 0 1!
1vOr O
4 00 2 00 4
C 3 1vO 2 C 211 Or00 1!O µ C 61!O 2 ¡ 3 1ÁO 2 C ²
3
14!O 2r00 00 µ00 4 00 4 Á00 sin¿ C 00 — 9 1vO 2 C 1 1ÁO 2 C .27=2/1 Or00 1!O µ C 9 1 !O 2

h
.81=8/1 Or2 r00 00 00 µ00
8 8 2
i £¡ ¢ ¤ ² £¡ ¢ ¤
— 81 1 !ÁO 002 sin 121 Or00 C 6100!O µ ¡ 1 ¿ C 00 O µ C 4 1 !O
C 141 Or00 1!00 2
sin 61 Or00 C 3 1 !O µ ¡ 1 ¿ — 91 Or2
± ±
121 Or2 C 1vO 2
² £ ¡ µ00 00 r00 00
¢ ¤ ±
— 121 Or00 1!O µ00 ¡ 41 !O sin 2 61 Or00 C ² £¡ ¢ ¤
—1 ¿ C 61 Or2 C
3 1 !O µ C 21 !O 2 sin 61 Or00 C C1 ¿
2
µ00 00 71 Or00 1!O µ 31 !O µµ00 00
00 00

h i ¢ ¤
£¡
C 8
1
1vr00O 2 ¡ 89 1 Or2
00 ¡
1
8 00 — 3 1 Or0 0 1!O µ ¡ 1 C 81 1 !ÁO 00 sin 121Or00 C C1 ¿ (91)
2
1Á O2 1 !O 2 61!O µ
2 00 2 µ00 00
¢ ¤
¡ ¢ £¡ ¢ ¤ £¡ ¿ ¡ 61 Or00 ¡ 31!O µ00 C .123=8/1Or
1 !O µ .¿ / D 61 Or00 C 41!O µ00 cos 1 ¡ 61 Or00 ¿ ¡ 21vOr00 sin 1 ¡ 61 Or00 ¡ 31!O
2
00
¡ 31!O µ00 µ00
2 00 00 µ00

— 3 1vO 2 C 9 1ÁO 2 C .15=2/1 Or00 1!O µ C


3 2 — 9 1 !O 2
C C3 — 3 1ÁO 2 C 421 Or00 1!O µ C 1 21!O 2
18!O r00 8 00 00 2 µ00
h 2
1vO
.57=2/1 Or2
8 Á00 00 2 r00
i h ² £¡ ¢ ¤
C 23 1 !OÁ002 cos¿ C4 9
r00 O
2
¡ .81=4/1 1 O 2 ¡ 271 Or00001!O µ ¡ µ C 1
1v 00 Or2 4¡ 1Á
2
00 00 41 !O Á00 cos 121 Or00 C 6 100!O µ ¡ 1 ¿
2
91!O
± ²
2
£¡ ¢ ¤ ± ²
— 481 Or2 C 561 Or00 1!O µ C 1 61!O cos 61 Or00 — 1 ¿ ¡ 181 Or2 C
00 µ00 C 6 1 !µO002
C 31 !O µ 211 Or00 1!O µ
00 00 00 00
KARLGAARD AND 4
£ ¡ h i
£ cos 3 21 Or00 C ¢ ¤ 3
£ ¡ ¢ ¤
¿ C .81=8/1 Or2 ¡
9 C 00 C .27=2/1 Or00 1!00O µ 2C 9 1
µ00 !O
2
¡ 3
Á00 cos 6 21Or00 C 00 ¿
8 8
1 !O µ 1ÁO
2 2
1vO 2 1 !O 1 !O µ
00 00 8 r00
h i ±
5 £ ¡ ¢ ¤
C .45=2/1 Or200 ¡
2 r00 C 12 1ÁO002 C 00 C 1 01!µO 00 ¡ 1 !O 2 1 2
cos 2 61 Or00 C —1 ¿ C 121 Or2
1vO 2 301 Or00 1!O µ 31 !O µ
² £¡ ¢ ¤ h i
2 Á00 00 00
00O µ C
C 141 Or00 1! cos 61 Or00 C C1 ¿ C 1
1vO
2 9
¡ 1 Or2 ¡ 1ÁO 2 ¡ 31 Or00 1!O µ ¡ 1 !O
1 2
C 1 !O1 2
2
41 !O 31!O µ
µ00 00 4 r00 00 4 00 4 00
£¡ ¢ ¤ ¢ i
£ cos 121Or00 C C 61vOr00 1 !O µ00 sin¿ ¡ C 1!O — 21 1ÁO 00 1!00O Á
C1 ¿ ¡
µ00
61 !O£¡ ¢ ¤ ¡ h ¢ £¡ 91v Or00 µ00 ¢ ¤
¡
£ sin 121Or00 C 61!O µ00 ¡ 1Or00¿ 1vOr00
121 ¡ 161Or00 1v.27=2/1
Or00 C 81v
Or00Or00 1 !
Or00 O µ00 sin 61 Or00 C 31!O µ00 ¡ 1 ¿ ¡
1v
¡ ¢
61 Or00 1vOr00 C 31vOr00 1 !O µ00
4 00

£ ¡
Downloaded by OLD DOMINION UNIVERSITY on August 27, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.5013

£ sin 3 21Or00 C ¢ ¤ ¤ £ ¡ ¢ ¤ ¡
¿
9
C [.27=4/1 Or00 1vOr00 C 2 00 1!
O — 3 1ÁO 0 0 1!O sin 6 21 Or00 C 1 ! ¿ C 151 Or00 1vOr00
1 !O µ00 ¢ £ ¡ 1vOr ¢ ¤ ¡ ¢
µ00 Á O µ00
C 101vOr00 1 !O µ00 ¡ 1ÁO 00 1 !O Á00 sin 2 61Or00 C 31!O µ00 ¡ 1 ¿ C 41 Or00 1vOr00 C 21vOr00 1 !O µ00
£¡ ¢ ¤
sin 61 Or00 C 31!O µ00 C 1 ¿
¡ ¢ £¡ ¢ ¤
C 1 1ÁO 00 00 — 3 1vOr 00 1
Or00 ¡ 1 !O sin 121Or00 C C1 ¿ (92)
2 2
1!O Á £¡ 1vOr00 µ00 ¢31!
¤ O µ00 £¡ ¢ ¤ ¡
1 !O Á.¿ / D 1 !O Á00 cos 1 ¡ 61 Or00 ¡ 31!O µ00 ¿ ¡ 1ÁO 00 sin 1 ¡ 61 Or00 ¡ 31!O µ00 ¿ C 61 Or00 1!O Á00 C
¢
31!O µ00 1 !O Á00 cos¿
£
1vOr 1ÁO 00 ¡ ¢ £¡ ¢ ¤ ¡ ¢
3 9
C
2 1!00O Á
1 Or00 2 00 — 31!O µ00 1 !O Á00 cos 121Or00 — 1 ¿ ¡ 61 Or00 1!O Á00 C 1 !O Á00
C 61!O µ00 31!O µ00
£¡ ¢ ¤ ¡ ¢ £ ¡ ¢ ¤ ¡
£ cos 61 Or00 C —1 ¿ C 61 Or00 1!O Á00 C 1 !O Á00 ¡ 21vOr00 1ÁO00 cos 2 61Or00 —1 ¿ C 1
1vOr 1ÁO 00

31!O µ00 41 !O µ00 C 31!O µ00


2
— 3 1 Or00 1!O Á ¡ 1 ! ¢ £¡ ¢ ¤ ¢ ¡
— 31 Or00 1ÁO 00 ¡ 21!O 1ÁO 00 1 Or00 1ÁO 00
9
O µ2 00
1 !O Á00 cos 121Or00 C C1 ¿ C 1!O sin¿ ¡
00
¡ Á00 µ00
61!O µ00
¢ £¡ 1vOr00 2

00 ¢ ¤ ¡ ¢ £¡ ¢ ¤
00 — 1 ¿ ¡ 61 Or00 1ÁO00 C 31ÁO 00 1!O µ00 sin 61 Or00 — 1 ¿
µ
31!OO µ00 1ÁO 00
C 61!
C C 1vOr 1 !O Á00 sin 121 Or00 C 31!O µ00
3
2
¢ £ ¡ ¢ ¤ ¡
¡ 1 !O Á00 sin 2 61 Or00 C 31 !O µ00 ¡ 13 ¿ ¡ µ00 O µ 4
1 Or00 1Á 00 00C 1Á 00 1! Á00
C 61 Or00 1ÁO 00 C 41ÁO 00 1 !O µ00 C 2
O O 00

21vOr00 ¢ £¡
µ
¢ ¤
00 00
C 21 1vOr 1!O Á sin 121 Or00 C C 1 ¿ (93)
61 !O 00

00
4 KARLGAARD AND

where
C.¿ / D
h 00 — 23 1vrO002 C µ00 C 901 Or00 1!00O µ
2
.135=2/1 Or2 3 01!O
i
O 2 ¡ 3 1 !OÁ2 ¿ C 1µO0 0 ¡
C 32 1Á00 00 (94)
2 00
21vOr
if 21 Or00 C 1 !O µ00 D Fig. 3 Relative spherical
0 and coordinate system.
h
C.¿ / D .123=8/1 Or2 ¡ 3 1vOr 2 C C 23 1 !µO 002
00 8 00 00

.15=2/1 Or00 1!O µ


i h ¯ ¡ ¢inh¡ ¢
C 98 1Á00O 2 ¡ 89 1 !Oµ002 ¿ C 1 2 21 Or00 C 00 27=8 100Or2
1 !O µ

— 3 1vO 2 C 1 1ÁO 2 C 9 1 Or00 1!O µ i


C 83 1!rO002 8 00 2 00 2 — 1 1 !ÁO 002
µ00 8
¢ ¤ ¡ and the absolute position and velocity are simply Or D er C RO and
£ C µ00 ¿ ¡ 121 Or2 C µ00 v O D e µ C VO .
£ ¡ 1!O 141 Or00 1!O
sin 6 21 Or00
00
C 41!O ¢ £ ¡ ¢ ¤ ¡ Equations (97) and (98) may use either of the linear or nonlinear
1ÁO 00
1
sin 3 21 Or00 C µ00 ¿ C Á00
2 µ00 solutions to give relative position and velocity vectors as functions
1 !O 1 !O
—9 3 ¢ £ ¡ ¢ ¤ of time. Typically for�ight applications, the relative position

Or
2 1v 00 1 !
O
4 00 4 Or00 C 1 ! ¿
cos 6 21 and velocity vectors are measured as
1 Or00 1vOr µ00
O µ00
¡ ¢ £ ¡
C 41 Or00 1vOr00 C 21vOr00 1 !O µ00 cos 3 21 Or00 C RO Dx Cy C z ez (99)
¢ ¤ o er eµ
— 7
1 !O µ00 ¿ —1 1 !O — 1 1ÁO 00
VO Du er C veµ C wez (100)
1 Or00 1vOr 1vOr µ 1 !O Á
4 00 2 00 00 4 00

C 1µO00 ¡ 21vOr00 (95)


Equating Eqs. (97) and (98) to Eqs. (99) and (100) gives the position
if 21 Or00C O 1!µ006D 0. Note the appearanceof a constant term in transformation
the out-of-planedisplacement, Eq. (90), resultingfrom the sine p
squared terms. The method of multiple scales cannot remove this 1 OrD .xC 1/2 Cy 2 C z2 ¡ 1 (101)
constant,
and so it must be part of a mathematically valid solution result- z
sin1 ÁO D (102)
ing from the second-order expansion of the Keplerian equations of
.1 C 1 Or/
motion (14–16), even though its presence is not physically valid.
Similar constants appear in Eqs. (88) and (89); however, these do y
sin1 µO D (103)
not presenta problem becauseconstantradial or angularseparations .1 C 1 Or/cos1
are physically valid solutions. The next section discusses how the
solutions for1 Or,1 µO,1 ÁO ,1 O vr O,1 !µ , and1 !Á may be used ÁO
O xC 1 (104)
to
determine the relative position and velocity vectors. cos1 µO D
.1 C 1 Or/cos1
ÁO
V. Relative Position and Velocity Vectors
To completethe problem, it is necessaryto expressthe solutionin Note that — ¼=2 · 1 ÁO ·¼=2 and so does not require a cosine for
the form of relative position and relative velocity vectors. When the quadrant checks. Once the position information is known, the rela-
coordinate systems shown in Fig. 3 are used, the nondimensional tive velocity transformation is
relative position vector RO may be written as
1vOr Du cos1 ÁOcos1 µO C .v C 1/cos1 ÁOsin1 µO C wsin1
ÁO

RO D .1 C 1 Or/er 2 ¡ er (96) (105)


wheree r ,e , ande Á
µ
are the unit vectors of the coordinate system .v C 1/cos1 µO u sin1 µO
1!Oµ D¡ ¢ ¡¡ ¢ ¡ 1 (106)
2 2 2
withe r2 aligned with the position vector of the satellite and e r ,e µ , 1 C 1 Or cos1 ÁO 1 C 1 Or cos1 ÁO
ande Á are the unit vectors of the coordinatesystem aligned with the
referenceorbit. Vectorse r2 ,e µ2 , ande Á2 may be expressedin terms of 1 u sin1 ÁOc os 1 µO .v C 1/ sin1 ÁOs i n 1 µO wcos1
!O D¡ ¡ C
KARLGAARD AND 4
ÁO
er ,e µ , ande Á by a series of two simple coordinate transformations: Á
1 C 1Or 1C1 1 C 1Or
�rst a rotation of angle1 µOaboute Á to align the coordinate Or
system with an intermediate systeme r1 ,e µ1 , ande Á1 and then a (107)
rotation ¡ of
angle 1ÁOabout thee µ1 axis to align the intermediate system
with The next section outlines the procedure used for comparing the
new solution given in Eqs. (88–93), (97), and (98) to the linear
er2 ,e µ2 , ande Á2 . Following this transformation, the relative position
vector is solution given by Eqs. (20), (97), and (98) and discusses examples
of some typical results.
RO D [.1 C 1Or/cos1 ÁOcos1 µO ¡ 1]er C .1 C 1Or/cos1 ÁOsin1
VI. Results
µ Oe µ
The new second-order relative motion equations (88–93) can be
compared with the�rst-order relative motion equations (20) by
C .1 C 1 Or/sin1 ÁOeÁ (97)
nu- merically integrating the full, nonlinear equations of motion
The derivativesof the unitvectorsalignedwith the referenceorbitare (1–6) for a given set of initial conditions. The approximate
er0 D eµ ,e 0µ D ¡er , ande 0Á D 0; therefore the relative velocity vector equations of motion (14–16) can also be numerically integrated to
is be compared with the analyticalsolution. A state error is
de�ned to be the approx- imate solution (found by either an
VO D [1vOr cos1 ÁOcos1 µO ¡ .1 C 1 Or/1!O Á sin1 ÁOcos1 µO analytical solution or numerical solution of approximate
equations of motion) minus the numerical integration of the full

.1 C 1 Or/1!O µ cos1 ÁOsin1 µO ¡ .1 C 1 Or/cos1 ÁOsin1 µO]e r nonlinear equations of motion and can be plotted as a function
of time for the comparisons.
Downloaded by OLD DOMINION UNIVERSITY on August 27, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.5013

C [1vOr cos1 ÁOsin1 µO ¡ .1 C 1 Or/1!O Á sin1 ÁOsin1 µO The results of this procedure are shown in Figs. 4 and 5. Figure
4 shows the results for the set of initial conditions1 Or.0/ D
C .1 C 1 Or/1!O µ cos1 ÁOcos1 µO C .1 C 1 Or/cos1
1vO r .0/ D 0:01, and1 !O Á.0/ D ¡0:01, with all of the other
0:001,
ÁOcos1 µO initial

1] e C [1vO
r sin1
O C .1 C 1 Or/1!OÁ cos1 Á]
O eÁ (98)
µ Á
4 KARLGAARD AND
Downloaded by OLD DOMINION UNIVERSITY on August 27, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.5013

Fig. 4 Magnitude of relative position and velocity error for¢^r(0) = 0.001,¢^rv(0) = 0.01, and¢^!Á(0) =¡0.01.

Fig. 5 Magnitude of relative position and velocity error for¢^r(0)= ¡0.001,¢ Á^(0) = 0.01, and¢ ^!µ (0)
=¡¡0.001.

conditionstaken to be zero. Figure 5 is for the case1 Or.0/ D typical. See Ref. 12 for many other such examples.
¡0:001,
1ÁO .0/D 0:01, and1 O!µ .0/D ¡ 0:001. Both Figs. 4 and 5 show
the
magnitudeof the relativepositionand velocityerrors as a functionof
time over one period of the reference orbit with logarithmic scaling
of the vertical axis.
For the�rst case, the maximum position error in the linear
solu- tion over the time period shown£is 2:72 10¡3 , and the
maximum velocity error £ is 2:18 10¡3 . For the new nonlinear
solution, the maximum position £ error is 6:04 10¡5 , and the
£
maximum velocity error is 4:36 10¡5 . For the second case, the
£ solution is 1:13 10¡3 , and
maximum position error in the linear
£
the maximum veloc- ity error is 9:41 10¡4 , whereas the
maximum positionand velocity errors for £ the nonlinear solution
£
are 1:40 10¡5 and 9:38 10¡6 , respectively. These results are fairly
KARLGAARD AND 4
VII. Conclusions
This paper has developed a solution of the relative motion
prob- lem that results from a second-order expansion of the
Keplerian equations of motion (1–6). The method of multiple
scales was used to determine a solution that was free of false
secular terms. The new solution has several advantages when
compared with the linear so- lution: 1) improved accuracy, 2)
recoupling of motion in the plane and out of the plane of the
reference orbit, and 3) better estimate of the drift rate between
orbits of differing periods.
A still unresolved problem with this solution is the presence of
the constant offset term in Eq. (90). The constant term does not
satisfy the basic physics of the problem, althoughit seems to appear
on every occasion that nonlinear terms are kept in the expansion
of Eqs. (1–6). Further work should be focused on removing this
constant from the solution.
4 KARLGAARD AND

References
1 Vol. 23, No. 4, 2000, pp. 604–610.
Clohessy, W. H., and Wiltshire, R. S., “Terminal Guidance System for 7
Mitchell, J. W., and Richardson, D. L., “Maintaining Periodic Trajecto-
Satellite Rendezvous,”Journal of the Aerospace Sciences, Vol. 27, No. 5,
1960, pp. 653–658, 674. ries with the First-Order Nonlinear Hill’s Equations,” American Astronau-
2
London, H. S., “Second Approximation to the Solution of the Ren- tical Society, AAS Paper 01-473, July–Aug. 2001.
8
dezvous Equations,”AIAA Journal, Vol. 1, No. 7, 1963, pp. 1691–1693. Nayfeh, A. H.,Introduction to Perturbation Techniques, Wiley, New
3 York, 1981, pp. 122–131, 166–175.
Anthony, M. L., and Sasaki, F. T., “Rendezvous Problem for Nearly 9
Circular Orbits,”AIAA Journal, Vol. 3, No. 7, 1965, pp. 1666–1673. Kevorkian, J., and Cole, J. D.,Multiple Scale and Singular Perturbation
4 Methods, Springer-Verlag, New York, 1996, pp. 267–409.
Kelly, T. J., “An Analytical Approach to the Two-Impulse Optimal 10
Ren- dezvous Problem,” American Astronautical Society, AAS Paper 94- Berreen, T. F., and Crisp, J. D., “An Exact and a New First-Order
156, Feb. 1994. Solu- tion for the Relative Trajectories of a Probe Ejected from a Space
5 Station,” Celestial Mechanics, Vol. 13, No. 1, 1976, pp. 75–88.
Vaddi, S. S., Vadali, S. R., and Alfriend, K. T., “Formation Flying: 11
Ac- comodating Non-Linearity and Eccentricity Perturbations,” American Vallado, D. A.,Fundamentals of Astrodynamics and Applications,
As- McGraw–Hill, New York, 1997, pp. 348–351.
12
tronautical Society, AAS Paper 02-184, Jan. 2002. Karlgaard, C. D., “Second-Order Relative Motion Equations,” M.S.
6
Melton, R. G., “Time-Explicit Representation of Relative Motion Thesis, Dept. of Aerospace and Ocean Engineering, Virginia Polytechnic
Between Elliptical Orbits,”Journal of Guidance, Control, and Dynamics, Inst. and State Univ., Blacksburg, VA, July 2001.
Downloaded by OLD DOMINION UNIVERSITY on August 27, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.5013
KARLGAARD AND 4

This article has been cited by:

1. Luo Yazhong, Zhang Jin, Tang Guojin. 2013. Survey of Orbital Dynamics and Control of Space Rendezvous.
Chinese Journal of Aeronautics . [CrossRef]
2. Kamran Shahid, Pini GurfilTop-Level Control of Disaggregated Satellites: Cluster Maintenance and
Scatter/Re- gather Maneuvers . [Citation] [PDF] [PDF Plus]
3. Yu Ning, Martin E. Avendano, Daniele Mortari. 2011. Sequential Design of Satellite Formations with
Invariant Distances. Journal of Spacecraft and Rockets 48:6, 1025-1032. [Citation] [PDF] [PDF Plus]
4. Sangjin Lee, Sang-Young Park. 2011. Approximate Analytical Solutions to Optimal Reconfiguration Problems
in Perturbed Satellite Relative Motion. Journal of Guidance, Control, and Dynamics 34:4, 1097-1111. [Citation]
[PDF] [PDF Plus]
5. Gang Zhang, Di Zhou. 2010. A second-order solution to the two-point boundary value problem for rendezvous
in eccentric orbits. Celestial Mechanics and Dynamical Astronomy 107:3, 319-336. [CrossRef]
6. K. Uldall Kristiansen, P. L. Palmer, M. Roberts. 2010. Relative motion of satellites exploiting the super-integrability
of Kepler’s problem. Celestial Mechanics and Dynamical Astronomy 106:4, 371-390. [CrossRef]
Downloaded by OLD DOMINION UNIVERSITY on August 27, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.5013

7. Hancheol Cho, Adam Yu. 2009. New Approach to Satellite Formation-Keeping: Exact Solution to the Full
Nonlinear Problem. Journal of Aerospace Engineering 22:4, 445-455. [CrossRef]
8. Igor Beigelman, Pini Gurfil. 2008. Optimal Fuel-Balanced Impulsive Formationkeeping for Perturbed
Spacecraft Orbits. Journal of Guidance, Control, and Dynamics 31:5, 1266-1283. [Citation] [PDF] [PDF Plus]
9. Steven P. Hughes. 2008. General Method for Optimal Guidance of Spacecraft Formations. Journal of
Guidance, Control, and Dynamics 31:2, 414-423. [Citation] [PDF] [PDF Plus]
10. Fanghua Jiang, Junfeng Li, Hexi Baoyin, Yunfeng Gao. 2008. Study on Relative Orbit Geometry of
Spacecraft Formations in Elliptical Reference Orbits. Journal of Guidance, Control, and Dynamics 31:1, 123-
134. [ Citation] [PDF] [PDF Plus]
11. Rajnish Sharma, Prasenjit Sengupta, Srinivas R. Vadali. 2007. Near-Optimal Feedback Rendezvous in Elliptic
Orbits Accounting for Nonlinear Differential Gravity. Journal of Guidance, Control, and Dynamics 30:6, 1803-
1813. [Citation] [PDF] [PDF Plus]
12. M. Halsall, P. L. Palmer. 2007. Modelling natural formations of LEO satellites. Celestial Mechanics and Dynamical
Astronomy 99:2, 105-127. [CrossRef]
13. Igor Beigelman, Pini GurfilGraph-Theory-Based Optimal Impulsive Formationkeeping . [Citation] [PDF]
[PDF Plus]
14. Prasenjit Sengupta, Srinivas R. Vadali. 2007. Relative Motion and the Geometry of Formations in Keplerian Elliptic
Orbits with Arbitrary Eccentricity. Journal of Guidance, Control, and Dynamics 30:4, 953-964. [ Citation]
[PDF] [PDF Plus]
15. E. Imre, P. L. Palmer. 2007. High-Precision, Symplectic Numerical, Relative Orbit Propagation. Journal of
Guidance, Control, and Dynamics 30:4, 965-973. [Citation] [PDF] [PDF Plus]
16. Philip L. Palmer, Egemen Imre. 2007. Relative Motion Between Satellites on Neighbouring Keplerian Orbits.
Journal of Guidance, Control, and Dynamics 30:2, 521-528. [Citation] [PDF] [PDF Plus]
17. Prasenjit Sengupta, Srinivas R. Vadali, Kyle T. Alfriend. 2007. Second-order state transition for relative
motion near perturbed, elliptic orbits. Celestial Mechanics and Dynamical Astronomy 97:2, 101-129. [CrossRef]
18. Prasenjit Sengupta, Rajnish Sharma, Srinivas Rao Vadali. 2006. Periodic Relative Motion Near a Keplerian Elliptic
Orbit with Nonlinear Differential Gravity. Journal of Guidance, Control, and Dynamics 29:5, 1110-1121. [Citation]
[PDF] [PDF Plus]
19. Mark Halsall, Philip PalmerAn Analytic Relative Orbit Model Incorporating J3 . [Citation] [PDF] [PDF Plus]
20. Pini Gurfil, Konstantin V. Kholshevnikov. 2006. Manifolds and Metrics in the Relative Spacecraft Motion Problem.
Journal of Guidance, Control, and Dynamics 29:4, 1004-1010. [Citation] [PDF] [PDF Plus]
21. Pini Gurfil, Konstantin KholshevnikovDistances on the Relative Spacecraft Motion Manifold . [Citation]
[PDF] [PDF Plus]
22. Pini GurfilOptimal Single-Impulse Formationkeeping . [Citation] [PDF] [PDF Plus]
23. Pini Gurfil. 2005. Relative Motion between Elliptic Orbits: Generalized Boundedness Conditions and Optimal
Formationkeeping. Journal of Guidance, Control, and Dynamics 28:4, 761-767. [Citation] [PDF] [PDF Plus]
24. Marco W. Soijer. 2004. Sequential Computation of Total Least-Squares Parameter Estimates. Journal of Guidance,
Control, and Dynamics 27:3, 501-503. [Citation] [PDF] [PDF Plus]
25. Christopher D. Karlgaard, Frederick H. Lutze. 2004. Second-Order Equations for Rendezvous in a Circular Orbit.
Journal of Guidance, Control, and Dynamics 27:3, 499-501. [Citation] [PDF] [PDF Plus]
Downloaded by OLD DOMINION UNIVERSITY on August 27, 2013 | http://arc.aiaa.org | DOI: 10.2514/2.5013

You might also like