16.

522, Space Propulsion

Prof. Manuel Martinez-Sanchez
Lecture 4: Re-positioning in Orbits
Suppose we want to move a satellite in a circular orbit to a position apart in the
same orbit, in a time t (assumed to be several orbital times at least). The general
approach is to transfer to a lower (for positive ) or higher (for < 0 ) nearby
orbit, then drift in this faster (or slower) orbit for a certain time, then return to the
original orbit.
The analysis is similar for low and high thrust, because we have radius ratios very
close to 1, so that, in either case the satellite is nearly in orbit even during
thrusting periods, and V's for orbit transfer amount (in magnitude) to the difference
of the beginning and ending orbital speeds. In detail, of course, if done at high thrust
the maneuver involves a two-impulse Hohmann transfer to the drift orbit and one
other two-impulse Hohmann transfer back to the original orbit. For the low-thrust
case, continuous thrusting is used during both
JG legs, with some guidance required to
remove the very slight radial component of v picked up during spiral flight (although
ignored here).
We will do the analysis for the low-thrust case only, then adapt the result for highthrust.
Let be the advance angle relative to a hypothetical satellite remaining in the
original orbit and left undisturbed. The general shape of the maneuver is sketched
below:

Since the orbital angular velocity is

=

r3

its variation with orbit radius is

16.522, Space Propulsion
Prof. Manuel Martinez-Sanchez

Lecture 4
Page 1 of 10

 = -

3 r d ( )

=
2
r
dt

(1)

During thrusting, r is varying according to

d
dt

F
- 2r M v a r

a = M

(2)

or

dr

2r dt

1dr 2ar 1 2
=
rdt

(3)

and so
d ( )

dt

d 2 ( )
dt

=-

3 2ar 1 2
3a
=

2
r

(4)

For integration, we will regard r r0 as a constant (small variations):

d ( )
dt

=-

3a
t + constant
r0

Starting from t = 0, = 0,

(5)
d ( )
dt

= 0,

we obtain
d ( )
dt

=-

3a
3 a 2
t ; = t ( t < t1 )
r0
2 r0

(6)

After (t=t1), we continue to drift at a constant rate

d ( )
dt

=-

3a
t1

r0

and, since we start from

( t1 ) = -

3 a 2
t1 ,
2 r0

the during the coasting phase is

= -

3at1
3 a 2 3a
t1 t1 ( t - t1 ) = 2 r0
r0
r0

t1

t -
2

(7)

At the end of coasting ( t = t -t1 ) , we have then

16.522, Space Propulsion

Prof. Manuel Martinez-Sanchez

Lecture 4
Page 2 of 10

 ( t - t1 ) = -

3at1
r0

t - 2 t1

(8)

and, after the period t1 of reversed thrust, we return to the initial orbit with
d ( )
= 0, and with as in (8), plus a further ( t1 ) .
dt
This is then the total :
= -

3at1
3 3 a 2
t1
t - t1
r0
2 2 r0

= -

3at1
( t - t1 )
r0

(9)

Clearly, the mission (given and t ) can be accomplished with different choices of
thrusting time t1 (but notice that t1 < t 2 in any case). The required thrust/mass

ratio (a) and V=2 a t1 depend on this choice:

-a =

r0
1
3 t1 ( t - t1 )

(10)

2 r0
3 ( t - t1 )

(11)

V =

We can see here that low thrust ends up as a penalty on V , so that the thrusting
time should be selected as short as possible within the available on-board power.

16.522, Space Propulsion

Prof. Manuel Martinez-Sanchez

Lecture 4
Page 3 of 10

In the limit of impulsive thrust, we realize that t1 cannot really be any less than the
Hohmann transfer time ( 1 2 orbit, or t1 =

). A more detailed analysis of this case

confirms that, for the high thrust case, Equations (10) and (11) are indeed valid
with t1 = .
The power per unit mass required is:
a c
P
1 F c
=
=
M 2 M
2
r0 c

1
P
=
M
6 t1 ( t -
t1 )

It is useful to express results in terms of the coasting time,

tc = t - 2t1
t1 =

t - tc

t1 ( t - t1 ) =

t - t1 =

t + tc
,
2

t 2 + tc
2

,
4

We then have
V =

4 r0
3 t + tc

2c r 0

P
=
M 3 t 2 - tc2

coasting reduces V , but increases P M (not much if tc t is small).

Sun-Moon Effects on a Geosynchronous Orbit The N/S Drift
Geosynchronous orbits are in the Equatorial plane, which is inclined 23.44D to the
plane of the Ecliptic (path of Earth around the Sun). Because of this, the tidal or
gravity-gradient forces exerted by the Sun on a geosynchronous spacecraft will
sometimes exert a torque on the orbit. This happens primarily at Solstice (January
and July) when the orbit dips the most out of the Ecliptic:

16.522, Space Propulsion

Prof. Manuel Martinez-Sanchez

Lecture 4
Page 4 of 10

TIDAL TORQUE, f

GEOGRAPHICAL
NORTH

SUMMER
SOLSTICE

SIDEREAL
NORTH
Z
r ~ 23o

G.N.

SUN

S/C IN GEO
y
r
s

EARTH
RES

TIDAL
FORCES

Fixed
Axes:

x: Along Vernal Equinox

y: In Ecliptic, solsticial line
z: North sidereal

VERNAL
EQUINOX

G
G
If r is the relative position vector of the spacecraft, and s is the unit vector from the
Sun, the Tidal Force (Gravity gradient) per unit mass is

JG

f =

s G G G G
3 r.s s r
3

REs

( )

(1)

(This the imbalance

between solar
attraction and centrifugal force, which cancels on
G
G
the Earth, r = 0 ). Of this, the r part gives no torque about the Earths center. To
G G
compute the effect on the orbit, we smear the torque r f about the
geosynchronous orbit, and calculate an orbit-average torque (per unit mass):
JG
1 2 G G
q=
r f d
2 0

(2)

JG 3 1 2 G G G G
r.s r s d
q = 3s
REs 2 0

( )(

(3)

We project onto a set of axis (x, y, z) as defined in the sketch (fixed axis):
x
G
r = y = R G
z

cos

sin cos
sin sin

(4)

where RG is the geosynchronous orbit radius (42, 200 Km) and = 23.44D is the
angle between the equatorial plane and the ecliptic.
16.522, Space Propulsion
Prof. Manuel Martinez-Sanchez

Lecture 4
Page 5 of 10

Also
cos
G

s = sin

(5)

So that
GG
r.s = R G ( cos cos + sin cos sin )

(6)

-sin sin sin

G G

r s = R G cos sin sin

sin cos - cos cos sin

(7)

In performing the averaging of Equation (3), we keep (approximately) constant.

The Earth does not advance much in its orbit in one day. Also, of course, is
constant. When (6) and (7) are multiplied together we obtain sin and cos
quadratic combinations, and we use
1 2
1 2
1 2
1

sin cos d = 0 ;
sin2 d =
cos2 d
=

0
0
0
2
2
2
2

(8)

and therefore the mean torque is

- cos sin
JG 3

2
S
q=
R
sin
sin

cos cos
G
2 R3
sin cos
ES

(9)

This torque will

G precess
G theG geosynchronous orbit by changing its specific angular
2
momentum A = R G G nG ( nG =unit vector along the Geographical North
direction,

2
rad s ).
86400

We are interested in the long-term (secular) effects. For times much longer than
1 year, we can also average the torque over the angle . Noticing that, as before,
2
sin cos = 0 ,

1
sin2 = ,
2

JG
q

sec ular

3 s 2

R G sin cos 0

=
3
4R
0

ES

16.522, Space Propulsion

Prof. Manuel Martinez-Sanchez

(10)

Lecture 4
Page 6 of 10

This shows a non-cancelling effect consisting on a mean torque which acts along the
(negative) vernal equinox direction (as shown in the sketch):

JJG
G
dA
to A , along (-x), and hence
The effect of this torque along (-x) is to add a small
dt
to rotate (precess) the orbit along the axis perpendicular to the vernal direction, as
shown.

In magnitude
JJG
G di
dA
di
= A
= R 2GG
dt
dt
dt

and equating to

JG
q

(11)

from (10),

3 s sin cos
di
=
dt
4 R3
G
sun
ES

(12)

Or, since the angular frequency of the Earths heliocentric motion is

s
,
R3
ES

3
di
=
dt
4
sun

2
S
sin cos
G

16.522, Space Propulsion

Prof. Manuel Martinez-Sanchez

(13)

Lecture 4
Page 7 of 10

Numerically, expressing

S and G

in ( deg yr ),

2
di

360 )
(
3

=
sin23.44D cos 23.44D = 0.27D yr
dt sun
4 (360 365)

The Moon effect is similar, except that the Moons orbital axis which is inclined at
5.15D with respect to sidereal North, precesses slowly about that direction (once
every 18.6 yr.). This means the angle between the Equator and the Moons orbit
varies between M = 23.44 D + 5.15 D = 28.59 D and M = 23.44 D - 5.15 D = 18.29 D .
di
The expression for
is similar to (12);
dt moon

3 M
di
sin m cos m
=
dt
4 R3
moon
EM

(14)

and
M 2
M

= M 3E = m

3
E R
R
ME

EM
EM

where

Mm
1
is the Moon/Earth mass ratio, and
=
ME
81.3

(15)

is the angular velocity of

the Moon about Earth (about 2 28 days).

We then have
2

365

360

3 1
28
D
D
D
di

sin
23.44
cos
23.44
yr
=

=
0.56
dt moon

4 81.3 360 365

(16)

This is for the average M in the 18.6 yr. lunar precession cycle.
At the peak of M ,
di
= 0.65D yr ,
dt
moon
and at the minimum,
di
= 0.46D yr .
dt
moon
Adding these values to the solar rotation (both act in the same direction) yields

16.522, Space Propulsion

Prof. Manuel Martinez-Sanchez

Lecture 4
Page 8 of 10

 di
= 0.83D yr (Maximum=-0.92, Minimum=-0.73)
dt
Total

(17)

This can be easily translated into an equivalent V:

V=vG i;
vG =

E
= 3070 m/s
RG

(Maximum=49, Minimum=39)

V=44 m/s yr

(18)

What is the effect of i if left uncompensated?

Using some spherical trigonometry,

RO
OR TAT
BI ED
T

sin =

i
EQUATORIAL
ORBIT
SOLSTICIAL
AXIS

sin =

sini sin

(19)

1 - sin2i cos2

(1 - cos i) sin cos

(20)

1 - sin2i cos2

and for small rotations i,

 i sin

i2
sin cos
2

(21)
(22)

So, during one day ( 0<<360D ), the orbit describes, as seen from the ground, a
figure 8 in the sky, about its nominal position:

16.522, Space Propulsion

Prof. Manuel Martinez-Sanchez

Lecture 4
Page 9 of 10

The main deviation is (N/S direction), and this is called a N/S drift. Typically,
communications geosynchronous satellites can only tolerate 0.05D 0.1D of such drift
before correction is needed.

16.522, Space Propulsion

Prof. Manuel Martinez-Sanchez

Lecture 4
Page 10 of 10