Session 7: Sub-Optimal Climb and Plane Change

In this lecture we present another application of a low thrust maneuver, but this time
including some optimality notions. From an initially circular orbit of radius r, the objective
is to climb while changing the inclination of the orbit. This would be required, for instance,
when starting from a circular LEO after launch from KSC (N28.5◦ ) and then transferring to
GEO, on the equatorial plane. The generic configuration of such maneuver is illustrated in
the figure below.

dL 
L

F
α
α θ

F line of nodes

Plane change using impulsive maneuvers requires thrusting exactly over the line of nodes.
The thrust vector FF should be normal to the original plane to change the inclination, i.
Thrusting at some angle from the normal will change the orbit to an elliptical one by raising
the apoaxis. Using this strategy for low thrust propulsion is highly inefficient, since it would
require a substantial amount of time to obtain a noticeable change in the orbital altitude and
inclination. Instead, thrust is distributed around the line of nodes as shown in the figure. In
this case, firing occurs inside arcs defined by the angle θ and there will be some particular
thrust angle distribution α = α(θ) that would optimize the trajectory for a particular net
change in orbital radius Δr and inclination Δi.
To analyze this problem, we start by calculating the rate of change of the magnitude of the
orbital angular momentum,

dL di
= L = (F sin α)r cos θ (1)
dt dt
Since the angular momentum for a circular orbit is given by,

√
L = mh = m µr (2)

and the angular velocity of the orbit is,

�
dθ µ
= (3)
dt r3
then we can write an equation for the rate of change of the orbital inclination with respect
to θ (true anomaly),

di F r2
= sin α cos θ (4)
dθ mµ

1
The inclination and time increment (one period) after one orbit will then be,

2π
F r2 r3
(Δi)1 = sin α cos θ dθ and (Δt)1 = 2π (5)
mµ 0 µ

So, the rate of change of inclination over one orbit will be,
�
r 2π
di 1 F r
= sin α cos θ dθ (6)
dt 2π m µ 0

To perform a similar averaging for the rate of change of the radius, we start from a power
balance,

�
r
d µ µ dr F r3
m − = (F cos α) → =2 cos α (7)
dt 2r r dt m µ

The average over one orbit is then,

2π
dr 1F r3
= cos α dθ (8)
dt πm µ 0

These equations could be solved numerically for a particular profile α = α(θ), or incorporated
into an optimization process to figure out what profile minimizes some cost function.
Instead of dealing with this general problem, Edelbaum (1961, 1973) just kept |α| constant
during each orbit, then optimized |α|(r). In this case,
⎧ π π
⎨ +α
⎪ for − <θ<
2 2
α(θ) = π 3π (9)
⎩ −α
⎪ for <θ<
2 2
From Eq. (4), but now (using α ≡ constant),

π/2
1 2
(sin α cos θ)θ = sin α cos θ dθ = sin α
π −π/2 π

therefore,

di 2 F r2
= sin α (10)
dθ πm µ

Similarly, we still have from Eqs. (5) and (7),

dr F r3
= 2 cos α (11)
dθ mµ
which needs no averaging. Dividing Eq. (10) by Eq. (11) and dropping the averaging sign,

2
 di tan α
= (12)
dr πr
We also have,

dΔv F/m
=
dr dr/dt

which combined with Eq. (7) results in,

dΔv 1 µ/r3
= (13)
dr 2 cos α
To optimize α(r) we look to minimize dΔv/dr for a given di/dr. We introduce the following
cost function, or Hamiltonian,

1 µ/r3 di tan α(r)

H = +λ − (14)
2 cos α(r) dr πr

The control variable is α(r), the state variable is i and the independent variable (replacing
time) is r. The optimality and “transversality” conditions are,

∂H
=0
∂α and [λδi]rr21 = 0 (15)
dλ ∂H
=
dr ∂i
with the second being satisfied automatically, because i is prescribed at both ends. Since H
does not depend explicitly on i,

dλ
=0 → λ ≡ constant (16)
dr
1 √
And using = 1 + tan2 α, the first of the conditions in Eq. (15) is,
cos α
1 µ/r3 tan α
�
r
λ 2λ r
√ − =0 → sin α = (17)
2 1 + tan2 α πr π µ

Pending determination of λ, Eq. (17) indicates that the thrust tilt amplitude α(r) increases
over the mission, so that most of the plane-change activity is deferred to the last part of the
climb, when the orbital velocity is lower.
To find λ, use the constraint on the total Δi, from Eq. (12),
Z r2
tan α
Δi = dr (18)
r1 πr

From Eq. (17) we solve for r, so that,

3
  π 2 dr d(sin α) dα
r=µ sin2 α → =2 =2 (19)
2λ r sin α tan α
Hence,
Z α2
2 2
Δi = dα → Δi = (α2 − α1 ) (20)
α1 π π
A separate relationship between α1 and α2 comes from Eq. (17),
�
r
sin α1 r1 v2
= = (21)
sin α2 r2 v1
where v1 and v2 denote the initial and final orbital velocities. Combining Eqs. (20) and (21),

sin α1 + π2 Δi
� �
v1 π  π  v
1
= → cos Δi + cot α1 sin Δi =
sin α1 v2 2 2 v2
we obtain,

v1 π  π 
− cos Δi sin Δi
v π 2 2
cot α1 = 2 or sin α1 = s  2 (22)
sin Δi v1 v1 π 
2 1+ − 2 cos Δi
v2 v2 2

The most important quantity is the optimized Δv. From Eq. (13),

r2 r2
Z r
� Z r
�
1 µ dr 1 µ dr/r
Δv = =
2 r1 r3 cos α 2 r1 r cos α
and combining with Eq. (19),

Z α2    Z α2
1 2λ 1 2dα 2λ dα 2λ
Δv = = 2 = (cot α1 − cot α2 ) (23)
2 α1 π sin α cos α tan α π α1 sin α π

From Eq. (17) we have 2λ/π = v1 sin α1 , and using,

π  v
2
cos Δi −
2 v1
cot α2 = π 
sin Δi
2
we calculate the optimized Δv,

 π � �  π  v ��
v1
π  2
v1 sin Δi − cos Δi − cos Δi −
2 v2 2 2 v1
Δv = s  2 π 
v1 v1 π  sin Δi
1+ − 2 cos Δi 2
v2 v2 2

4
and simplifying,
�
r π 
Δv = v12 + v22 − 2v1 v2 cos Δi (24)
2
Geometrically, Δv appears as the vector difference of the final and initial velocities, except
the angle between Fv1 and Fv2 is not the actual Δi, but π2 Δi. The extra factor reflects the
inefficiency associated with thrusting through the full π/2 in each out-of-plane direction.

Example
Consider a LEO (400 Km) to GEO, with:

• Δi = 28.5◦

• v1 = 7673 m/s

• v2 = 3072 m/s

From Eq. (24),

�
r π 
Δv = 2 2
7673 + 3072 − 2(7673)(3072) cos ◦
28.5 = 5903 m/s
2
This is noticeably worse than the true optimum Δv = 5768 m/s calculated for the case when
α = α(θ) is also modulated as tan α = cos θ.
The initial and final tilt angles are,
π 
◦
sin 28.5 × 3072
sin α1 = 2 = 0.3665 α1 = 21.5◦
5903
π
α2 = α1 + Δi α2 = 66.3◦
2
These are smaller than the peak values α1max = 30.5◦ and α2max = 72.2◦ in this optimal case,
but, of course, they are applied for the whole half-orbit.

5
MIT OpenCourseWare
http://ocw.mit.edu

16.522 Space Propulsion

Spring 2015

For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.