You are on page 1of 8

Spin Stabilization 1

9 Spin Stabilization
At this point, we have a general statement for the rotational behaviour of a rigid body,
Euler’s equations. In the absence of external torques,
I1 ω̇1 + (I3 − I2 )ω2 ω3 = 0
I2 ω̇2 + (I1 − I3 )ω1 ω3 = 0 (1)
I3 ω̇3 + (I2 − I1 )ω1 ω2 = 0
An analytical solution for the torque-free motion of an axisymmetric body has been pre-
sented.
In this section, we shall study the stability properties of torque-free motion. In other words,
how do the solutions behave as t → ∞? If a spinning body is perturbed slightly, does the
motion remain bounded (stable)? Does it grow without bound (unstable)? Does the motion
always tend to a particular equilibrium (asymptotically stable)?

9.1 Stability
Consider a system of first-order differential equations with prescribed initial conditions:
ẋ = f (x, t) , x(0) = x0 (2)
where    
x1 f1 (x1 , . . . , xn , t)
 .  ..
 ..  ,
 
x=  f (x, t) = 
 . 

xn fn (x1 , . . . , xn , t)
Here, fi connotes a general functional relationship (Note that Eq.(1) has the form of (2) if
we select x = ω = [ω1 ω2 ω3 ]T ).
We shall assume that there is a unique solution of Eq. (2) which is denoted by x(t, x0 ). Let
xe represent an equilibrium solution, i.e.,
x0 = xe , f (xe , t) ≡ 0 , x(t, xe ) ≡ xe
Now consider other solutions which correspond to different initial conditions x(0) = xe +∆x0 .
These other solutions will differ from the original solution by an amount
∆x(t) = x(t, xe + ∆x0 ) − xe
Our definition of stability is originally due to Lyapunov.
Definition 1. The equilibrium solution of (2), x(t, xe ) ≡ xe is L-stable if there exists a
number δ > 0 such that for any preassigned  > 0, one can maintain k ∆x k<  for all t ≥ 0,
by choosing any ∆x0 subject to k ∆x0 k< δ.
The notation k (·) k denotes the Euclidean norm:

√ q
k x k= xT x = x21 + · · · + x2n
Spin Stabilization 2

Definition 2. The equilibrium solution of (2) is asymptotically stable if it is L-stable and


there exists a number δ > 0 such that k ∆x0 k < δ implies that

k ∆x(t) k→ 0 as t → ∞

Definition 3. The equilibrium solution of (2) is globally asymptotically stable if it is L-stable


and
k ∆x(t) k→ 0 as t → ∞
for any initial condition xe + ∆x0 .

The figures should aid in the understanding of these definitions.


Definition 4. The equilibrium solution of (2) is unstable if it is not L-stable.
Stability of Linear Systems
Consider a scalar function of time x(t) and assume that it is governed by a linear, constant-
coefficient, homogeneous differential equation:

dn x dn−1 x d3 x d2 x dx
+ a n−1 + · · · + a 3 + a 2 + a 1 + a0 x(t) = 0 (3)
dtn dtn−1 dt3 dt2 dt
In order to solve this differential equation, we look for solutions of the form

x(t) = x̄eλt , x̄ = constant (4)

Substituting this into (3) gives


h i
λn + an−1 λn−1 + · · · + a3 λ3 + a2 λ2 + a1 λ + a0 x̄eλt = 0
Spin Stabilization 3

For nontrivial x̄, this implies that

λn + an−1 λn−1 + · · · + a3 λ3 + a2 λ2 + a1 λ + a0 = 0 (5)

which is called the characteristic equation. Let us denote the roots of this equation by
{λ1 , . . . , λn }. We shall assume that the roots are distinct (i.e., λi 6= λj , i 6= j). The
solution of (3) can then be written as a linear combination of solutions of the form (4):

x(t) = c1 eλ1 t + . . . cn eλn t (6)

The constants {c1 , . . . , cn } are chosen to match the initial conditions. If a pair of roots
appear in complex-conjugate pairs,

λ = σ ± ω ,  = −1

then the corresponding solutions are

ci eσt cos ωt + ci+1 eσt sin ωt

We now present the following theorems regarding the stability of (3).


Theorem 1. The equilibrium solution x(t) ≡ 0 of Eq. (3) is stable if the roots of the
characteristic equation are distinct and

<e{λi } ≤ 0

Theorem 2. The equilibrium solution x(t) ≡ 0 of Eq. (3) is (globally) asymptotically stable
if the roots of the characteristic equation satisfy

<e{λi } < 0

(The eigenvalues need not be distinct in this case).


Theorem 3. The equilibrium solution x(t) ≡ 0 of Eq. (3) is unstable if any of the roots of
the characteristic equation satisfy
<e{λi } > 0
(The eigenvalues need not be distinct in this case).
Linearization of a Nonlinear System
Assume that f (x, t) = f (x) in (2); we say that the system is autonomous or time invariant.
Expand the solution as x(t) = xe + (t). Therefore,

ẋ = ˙ = f (xe + )

∂f
= f (xe ) + T
 + O{||||2 }
∂x x=xe

Hence, the linearized system is



∂f
˙ = A, A = (7)
∂xT x=xe

Spin Stabilization 4

with the equilibrium x = xe corresponding to  = 0.


To ascertain stability of the linearized system, we again look for solutions of the form
(t) = ¯eλt
leading to
λ¯ = A¯, ¯ 6= 0
Therefore λ is an eigenvalue of A and ¯ is the corresponding eigenvector. Let us denote
the collection of eigenvalues by {λ1 , . . . , λn } and the eigenvectors by {1 , . . . , n }. Assuming
distinct eigenvalues, the latter are linearly independent and the general solution of (7) can
be written as
(t) = c1 1 eλ1 t + . . . cn n eλn t (8)
The constants {c1 , . . . , cn } are chosen to match the initial conditions. Theorems 1, 2, and 3
also apply to the stability of the equilibrium solution (t) = 0. Do these infer anything about
the stability of the underlying nonlinear system? Yes. Asymptotic stability or instability of
the linearization imply the same property of the nonlinear system. However, L-stability and
global asymptotic stability cannot be established by this method.

9.2 Stability of Torque-Free Motion


We begin our study of stability by noting that the “simple spins,” i.e., those that occur
about a principal axis of inertia, are solutions to Euler’s equations, Eq. (1). For example,
ω1 (t) = ω3 (t) = 0 , ω2 (t) ≡ ν (a constant)
is a solution as are spins about the other two axes. Now, let us consider what happens when
this reference motion is perturbed. The angular velocity in this case shall be written as
ω1 = 1 (t)
ω2 = 2 (t) + ν
ω3 = 3 (t)
where 1 , 2 , and 3 are taken to be small relative to ν. Substitution of these relationships
into Euler’s equation for torque-free motion, (1), while neglecting second order terms, gives
I1 ˙1 + (I3 − I2 )ν3 = 0 (9)
I3 ˙3 + (I2 − I1 )ν1 = 0 (10)
I2 ˙2 = 0 (11)
The last equation integrates to 2 (t) ≡ 2 (0) (a constant) which represents a constant per-
turbation to the angular velocity component about the spin axis.
Taking the time derivative of (9) and (10) gives
I1 ¨1 = (I2 − I3 )ν ˙3 (12)
I3 ¨3 = (I1 − I2 )ν ˙1
Spin Stabilization 5

Substituting for ˙1 and ˙3 from (9) and (10) yields:

¨1 = α2 1 (13)
¨3 = α2 3

where
(I2 − I3 )(I1 − I2 ) 2
α2 = ν
I1 I3
The solutions for 1 and 3 are of the form
(
Aeαt + Be−αt , (α 6= 0)
(t) = (14)
A + Bt , (α = 0)

where A and B are constants chosen to match the initial conditions.


There, are three cases to be considered:

I. α2 > 0. In this case, α is real and in (14) there is an exponent with positive
real part. Hence, the solution for  grows without bound and the motion is
unstable.
II. α2 = 0. In this case,  has a component that grows linearly with time.
Hence, the motion is unstable.
III. α2 < 0. In this case, α is imaginary and the solution of (13a,b) is periodic.
Therefore, the motion is stable.

Therefore, for stability we require that

(I2 − I3 )(I2 − I1 ) > 0

which will occur if

I2 > I3 and I2 > I1


or
I2 < I3 and I2 < I1

We summarize these results in the following theorem:


Theorem. The torque-free motion of a rigid body is stable if the nominal spin of the body is
about the axis of maximum or minimum principal moment of inertia. If the nominal spin is
about the axis of the intermediate principal moment of inertia, then the motion is unstable.

9.3 Effects of Internal Energy Dissipation


The Energy Sink Hypothesis
The above theorem is restricted to the case of torque-free motion of a rigid body. Alas,
spacecraft motion is not entirely torque free, nor are space structures entirely rigid. The
Spin Stabilization 6

details of these two effects will vary from situation to situation but in general one expects
the actual motion as well as external torques to produce some deformation of the body.
These structural deformations create a loss of kinetic energy in the form of heat. Because
of this energy dissipation, the kinetic energy is a nonincreasing quantity. The general effect
can be covered through the following:
Energy Sink Hypothesis. A quasi-rigid body will dissipate energy until a state of mini-
mum kinetic energy is reached. For torque-free motion, it is assumed that the magnitude of
the angular momentum is conserved.
From the previous section, this minimum energy state must correspond to either a major
axis or minor axis spin. For a major axis spin, the kinetic energy can be written as
2 h2
T = 12 Imax ωmax = , h = Imax ωmax (15)
2Imax
and for a minor axis spin
2 h2
T = 12 Imin ωmin = , h = Imin ωmin (16)
2Imin
Therefore, for a given value of angular momentum, which is conserved, kinetic energy is
minimized for a major axis spin. A minor axis spin maximizes the kinetic energy and this
quantity will decrease until a major axis spin is obtained. Hence, we have the following:
The Major Axis Rule. Simple spins about the major axis of inertia are asymptotically
stable. Spins about the minor axis (or the intermediate axis) of inertia are unstable.
Axisymmetric Bodies
We shall now apply the heuristic argument to the specific case of torque-free motion of an
axisymmetric body. From the previous section, the kinetic energy for this case can be written
as
T = 1
(I ω 2
2 1 1
+ I2 ω22 + I3 ω32 )
= 1
(I ω 2
2 a a
+ It ωt2 ) (17)

For the angular momentum,


ht = It ωt = h sin γ
ha = Ia ωa = h cos γ
Spin Stabilization 7

where γ is the nutation angle. Using these relationships the angular velocity components
can be expressed in terms of the angular momentum and nutation angle as
h
ωt = sin γ (18)
It
and
h
ωa = cos γ (19)
Ia
Substituting (18) and (19) into (17) gives
h2 Ia
 
T = cos2 γ + sin2 γ (20)
2Ia It
For torque-free motion, one can argue that the magnitude of the angular momentum h will
remain constant. Therefore, the time derivative of the kinetic energy is given by
dT dγ h2 Ia
 
Ṫ = = −2 cos γ sin γ + 2 sin γ cos γ γ̇
dγ dt 2Ia It
2
h sin 2γ Ia − It
 
= γ̇
2Ia It
Using the energy sink hypothesis, Ṫ < 0. This implies that the sign of γ̇ must correspond
to the sign of (It − Ia ).
The nutation angle will decrease if Ia > It which corresponds to a major-axis spin. For
minimum T , we require Ṫ = 0 which corresponds to sin 2γ = 0. Therefore, γ = 0 or
γ = π/2. From (20),
h2 h2
T |γ=0 = , T |γ=π/2 =
2Ia 2It
Therefore, the minimum value of T occurs when γ = 0,
h2
T |min =
2Ia
which is in keeping with the result of the previous section, Eq. (15). When γ = π/2,
T = h2 /(2It ) which is the maximum value of the kinetic energy.
If Ia < It , then the nutation angle will increase until Ṫ = 0 at γ = π/2. In this case, the
minimum value of T is
h2 h2
T |min = <
2It 2Ia
The arguments of this section show that the existence of energy dissipation causes the nu-
tation angle do die away for major axis spins. When γ = 0, β = 0 since from §4, Eq.
(33),
Ia
tan γ = tan β
It
and therefore the angular momentum vector, the angular velocity vector, and the symmetry
axis are aligned.
Minor axis spins are unstable and the nutation angle grows until a major axis spin is obtained
when γ = π/2.
Spin Stabilization 8

9.4 Application to Spin Stabilization


We have seen that the spin axis of an axisymmetric spacecraft tends to remain fixed in
inertial space for a major-axis spin. This supplies the theoretical basis for spin stabilization
wherein the entire spacecraft is spun about its axis of symmetry.
Explorer I
This was the first American satellite and was launched in 1959, two months after the launch
of Sputnik. It was axisymmetric and spin-stabilized about its minor axis of inertia. The
presence of two whip antennas on the spacecraft was a source of energy dissipation. It began
tumbling within hours of its deployment. It was this spacecraft that led to the “discovery”
of the the major-axis rule.
Alouette I
This was the first Canadian satellite and was launched in 1964 for the purpose of conducting
ionosphere studies. Alouette I was spin-stabilized about its major axis of inertia.
Early Bird
This was the first geostationary communications satellite and was launched in 1965. It was
also spin stabilized about its major-axis of inertia.
Although spin stabilization about the major axis was the method of choice in the early-
mid 1960’s, it suffers from a number of drawbacks. The oblate shape of the spacecraft
is incompatible with the prolate nature of launch vehicles. Also, many spacecraft require
certain portions of the structure to remain motionless with respect to the earth.

You might also like