Concepts of Stability

Summer Semester 2018

TU Munich 4th June 2018

roberto.lampariello@dlr.de
Contents

We first introduce general concepts of stability which apply to both

linear and non-linear systems.

We then look at two examples: the mass-damper-spring system and a

rigid body in rotation. We describe the stability properties of these two
dynamic systems.

For the attitude dynamics of a rigid body, we also provide generalities

on the solutions of the equations of motion, we look at the case of a
free gyroscope, we analyze the effect of energy dissipation and of
external orbital disturbances and we introduce internal and external
torques which typically play a significant role in orbit.

Folie 2
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Concepts of stability

Folie 3
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Qualitative Definition of Stability
A dynamic system is said to be stable if starting the system somewhere near its
desired operating point implies that it will stay around the point ever after.

Consider a pendulum: θ=0

stable
equilibrium
point

θ=180
R unstable
equilibrium
point
θ

M
Folie 4
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Equilibrium Point
A dynamic system can usually be represented by a set of differential equations
𝐱𝐱̇ = 𝐟𝐟 𝐱𝐱, 𝑡𝑡 (1)
where f is a nx1 vector function and x is the nx1 state vector. If f is nonlinear,
then the dynamic system is nonlinear. The number of states n is called the
order of the system.

A solution of the equations, x(t), corresponds to a curve in state space,

generally referred to as state trajectory or system trajectory.

It is possible for a system trajectory to correspond to a single point. Such a

point is called an equilibrium point:

A state x* is an equilibrium state (or point) of the system if once x(t) is equal to
x*, it remains equal to x* for all future time.

Mathematically, we have 0= 𝐟𝐟 x∗, 𝑡𝑡 (2)

Folie 5
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Equilibrium Point Example
Equilibrium points can be found by solving the (nonlinear) algebraic equation
(2). Consider the equations of motion of the pendulum in S.4:

𝑀𝑀𝑅𝑅 2 𝜃𝜃̈ + 𝑏𝑏𝜃𝜃̇ + 𝑀𝑀𝑀𝑀𝑀𝑀 sin𝜃𝜃 = 0

where M is the mass, R its length, b the friction coefficient at the hinge and g
the gravity constant. Letting 𝑥𝑥1 = 𝜃𝜃, 𝑥𝑥2 = 𝜃𝜃̇, such that the state vector is
𝐱𝐱 = 𝑥𝑥1 𝑥𝑥2 T, the corresponding state-space equation is

𝑥𝑥1̇ = 𝑥𝑥2
𝑏𝑏 𝑔𝑔
𝑥𝑥2̇ = − 𝑥𝑥 − sin𝑥𝑥1
𝑀𝑀𝑅𝑅2 2 𝑅𝑅
Therefore, the equilibrium points are given by
𝑥𝑥2 = 0, sin 𝑥𝑥1 = 0
which leads to the points (0, [2π], 0) and (π, [2π], 0). Physically, these points
correspond to the pendulum resting exactly at the vertical up and down
positions.

Folie 6
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Stability and Instability
We let BR denote the spherical region (or ball) defined by 𝒙𝒙 < 𝑅𝑅 in state-
space, and SR the sphere itself, defined by 𝒙𝒙 = 𝑅𝑅. We can then define
stability and instability as follows:
The equilibrium state x=0 is said to be stable if for any R>0, there exists r > 0,
such that if 𝒙𝒙(0) < 𝑟𝑟, then 𝒙𝒙(𝑡𝑡) < 𝑅𝑅 for all t≥ 0. Otherwise, the equilibrium
point is unstable.
In words, stability (or Lyapunov stability) means that the system trajectory can
be kept arbitrarily close to the origin by starting sufficiently close to it.

More formally, the origin is stable, if, given that we do not want the state
trajectory x(t) to get out of a ball of arbitrarily defined radius BR , a value r(R)
can be found such that starting the state from within the ball Br at time 0
guarantees that the state will stay within the ball BR thereafter.

Note that the equilibrium point for the given system is in the origin (x=0). Any
equilibrium point can be defined to be in the origin, by applying an appropriate
transformation to (1) in S. 5.
Folie 7
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Stability and Instability
Conversely, an equilibrium point is unstable if there exists at least one ball BR ,
such that for every r > 0, no matter how small, it is always possible for the
system trajectory to start somewhere within the ball Br and eventually leave the
ball BR .

A
B 0 – equilibrium point
C
A – asymptotically stable
B – marginally stable
0 C – unstable
Sr x(0) – initial condition
x(0) S r – sphere with radius r
S R – sphere with radius R

SR

Folie 8
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Asymptotic Stability
In many engineering applications, Lyapunov stability is not enough, since we
want to return to the equilibrium. As such we have the following definition:
An equilibrium point 0 is asymptotically stable if it is stable, and if in addition
there exists some r > 0 such that 𝒙𝒙(0) < 𝑟𝑟 implies that 𝒙𝒙 𝒕𝒕 → 𝟎𝟎 as 𝑡𝑡 → ∞.

A
B 0 – equilibrium point
C
A – asymptotically stable
B – marginally stable
0 C – unstable
Sr x(0) – initial condition
x(0)
S r – sphere with radius r
S R – sphere with radius R

SR

Folie 9
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Marginal Stability
If an equilibrium point 0 is Lyapunov stable but not asymptotically stable, it is
then marginally stable.

A
0 – equilibrium point
B
C A – asymptotically stable
B – marginally stable
C – unstable
0
x(0) – initial condition
Sr
S r – sphere with radius r
x(0)
S R – sphere with radius R

SR

Folie 10
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Stability of a pendulum
Consider a pendulum with an initial state θ R
displaced from the vertical down equilibrium
point 𝟎𝟎𝐝𝐝 .
θ
The initial position of the mass M is located
within Sr . When released from this position, 𝟎𝟎𝐝𝐝
M
the pendulum will begin to swing, and
will settle at the vertical down equilibrium Sr
point 𝟎𝟎𝐝𝐝 . SR
𝟎𝟎𝐝𝐝 – vertical down
It is clear that the trajectory will always be located equilibrium point
within the sphere SR .

Folie 11
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Stability of a pendulum SR
If the pendulum has an initial state θ located at Sr
the vertical up equilibrium point 𝟎𝟎𝐯𝐯 . With no
external input, the pendulum will remain in M 𝟎𝟎𝐯𝐯
this state. θ
Now consider the case that the pendulum has
an initial state θ displaced a small amount from
the vertical up equilibrium point 𝟎𝟎𝐯𝐯 . R

Note that the initial state is within Sr . However, now

𝟎𝟎𝐯𝐯 – vertical up equilibrium
when the pendulum is released from this position, point
the trajectory will very soon leave SR .

Folie 12
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
 Lyapunov Stability for a reference motion
A reference motion y(t) is stable if a neighboring motion y(t)+∆y0(t) can be
made to lie arbitrarily close to y(t) (∆y0(t) < ε) for all t > t0 by sufficiently
reducing their disparity δ at t=t0.

∆y
∆y0
δ ε
δ

Tube cross- Tube cross-

section at section for
t=t0 all t > t0

Folie 13
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Example 1: mass-damper-spring system

Folie 14
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Mass-damper-spring system (1 dof)

Physical system (process) Modeling abstraction

The equation of motion for mass m is given by Newton’s second Law as
mx = −kx − cx + F
where
k is the stiffness coefficient and − kx is the spring force (Hooke’s law)
c (or B) is the damping coefficient and − cx is the damping force
F is an external force (forcing term)
Folie 15
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Damped harmonic motion
With no forcing term, we have
mx + cx + kx = 0
or in the standard form
x + 2 ζω0 x + ω02 x = 0
with ω0 – natural frequency
ζ – damping ratio
c
and ω0 = k m ζ=
2 mk
λt
Solution: letting x = A e , we obtain

(λ 2
)
+ 2 ζω0 λ + ω02 Aeλt = 0
characteristic equation
from which

λ1 , λ 2 = − ζω0 ± ω0 ζ 2 − 1
roots of characteristic equation
Folie 16
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Damped harmonic motion solutions

ζ = 0: λ1 , λ 2 = ± j ω0
Simple
jω 0 t jω 0 t
x(t ) = A1 e + A2 e harmonic
motion
= A sin(ω0 t + α)
-1 < ζ < 1: λ1 , λ 2 = − ζω0 ± ω0 ζ −1
2 x(t ) A e − ζ ω0t
Underdamped
x(t ) = A e − ζ ω0t sin(ωd t + α) t
oscillatory
motion
ωd = ω0 1 − ζ 2
ζ = 1: λ1 , λ 2 = − ω 0 x(t ) overdamped Critically or

x(t ) = (A + Bt ) e − ω0t
overdamped
oscillatory

ζ > 1: (
λ1 , λ 2 = − ω0 ζ ± ζ 2 − 1 ) t
motion

λ1 t λ2 t critically
x(t ) = A1 e + A2 e underdamped damped

Folie 17
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Damped harmonic motion solutions summary

Damping ratio ζ :

ζ ≥ 1: critical damping, no oscillation, asymptotically stable

0 < ζ <1: damping, with oscillation, asymptotically stable

ζ = 0: no damping, with oscillation, marginally stable

−1 < ζ < 0: damping, with oscillation, unstable

ζ ≤ −1 : critical damping, no oscillation, unstable

Folie 18
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Stability conditions
For the given mass-damper-spring system, the point of equilibrium is
x = x = x = 0
Recalling that
x = A e λt
λ1 , λ 2 = − ζω0 ± ω0 ζ 2 − 1
it follows that for stability

Re(λ1 , λ 2 ) ≤ 0

and for asymptotic stability

Re(λ1 , λ 2 ) < 0

Folie 19
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Step response of a mass-damper-spring system

Consider now the equation:

mx + cx + kx = F (t )

F(t)
The steady-state solution is given by the
particular integral for the forcing term F(t).
It is common to study the system response
to the step function:
t
Step function
0 t<0
F(t) =
1 t≥0

x(t)
The particular integral in this case is
x(t) = 1 / ω0
therefore the solution is
x(t ) = A e − ζ ω0t sin(ωd t + α) + 1 ω0
t
steady-state error System response with ω0=1
Folie 20
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2017 | Attitude control
Example 2: Attitude kinematics and dynamics
of a single rigid body

Folie 21
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude control
Attitude equations of motion of a rigid body
Given the angular momentum

Γ = I.ω
where the dyadic I represents the inertia
of the body with respect to its centre of
mass, it follows that

dΓ 2 d (I.ω)
= + ω × I.ω,
dt dt
or

I.ω
 + ω × I.ω = τ - Euler’s rigid body equations

Folie 22
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
The free rigid body – rotational kinematics
When considering the free rigid body motion,
we make the following assumptions:
ω
flexible or moving parts in the body are
neglected
no external or control forces
such that we can analyse the short term
behaviour.

Now consider first an inertial reference frame CM

I
e ={C1, C2, C3} such that C3 is directed along
the angular momentum unit vector Γ̂ .
The body reference frame is often defined
as eb={i, j, k} and is oriented w.r.t eI after the
three Euler angles ψ, ∆, φ.
We can define the angular velocity in the body
frame as ψ, ∆, φ - rotational DoFs
ω = p i + q j+ r k ψ, ∆, φ , p, q, r – rotational state
Folie 23
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
The free rigid body – dynamics and differential
kinematics
The dynamical equations of motion are

I.ω
 + ω × I.ω = 0

which, extended with the kinematical equations, allows us to define the

problem fully. For example, we can define the unknowns as the three
components of the angular velocity in the body frame bω(t) = [ p(t) q(t) r(t) ]T
and the three Euler angles ψ(t), ∆(t), φ(t), such that (see Lesson 1)
ψ   sin φ cos φ 0   p
 ∆  = 1  sin ∆ cos φ
  sin ∆  − sin ∆ sin φ 0   q 
 φ  − cos ∆ sin φ − cos ∆ cos φ sin ∆   r 
Note that the problem consists in solving a sixth-order ODE.

Folie 24
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
The free rigid body motion – generalities

Lobed open cone traced

by major (minor) axis

CM

Top view: ∆* = minimum ∆ Principal axes of inertia i, j, k

∆1 = maximum ∆ of rotating body

Folie 25
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
The free rigid body motion - generalities
The main features of the closed-form solution are:
ψ – precession, ∆ – nutation, φ – spin angle
The angular momentum is constant and provides a fixed direction Γ̂.
The maximum (or minimum) principal axis of inertia describes a cone
around Γ̂ , of variable aperture ∆ and with monotone but variable
velocityψ . The cone has apex in the center of mass.
The aperture ∆ and velocity ψ are generally periodic. Their period and
amplitude is function of the principal moments of inertia and of the initial
conditions of the angular velocity. The period can go to infinity, for which
∆ and ψ are constant.
While the maximum (or minimum) principal axis of inertia rotates
(precession), the body rotates about it with velocity φ, which is also
generally variable and periodic.
The case of the gyroscopic geometry is such that I11=I22<I33 (I11<I22=I33).
For this, ∆=const., ψ =const. and φ =const. The maximum (or
minimum) principal axis of inertia describes a circular cone (see later).

Folie 26
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
The free gyroscope – analytical solution
For I11 = I22 < I33 (gyroscopic geometry) the analytical solution is as follows:
p(t) = P sin (ωφ t + ω0) with P = p02 + q02
q(t) = P cos (ωφ t + ω0) ωφ = - r0 (I33-I11)/I11
r(t) = r0 = const.
with ωφ the angular frequency and ω0 the phase. p0 , q0
and r0 are the initial conditions from which P, ωφ and ω0
can be obtained. The noticeable properties are
cos ∆ = I33 r0 / Γ0mod = const.

ψ = I 33 r0 (I11 cos ∆ ) = const.

  I 33 − I11 
φ = −  r0 = ωφ = const.
 I11 
This corresponds to a regular precession of the axis k around Γ̂.
Note that satellites are generally axial-symmetric, i.e. ideally free gyroscopes.

Folie 27
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
 Simulation example for the free gyroscope

For the following simulation parameters:

I body = diag ([29.2,29.2,38.4])[kg m 2 ]
Spin angular velocity, r0: 0.0698 rad/s = 4 deg/s [Ψ , ∆, Φ ] 0 = [0, 0.1085, 1.5708] [rad]
Body Reference ω Euler angles versus time
0.07 4
p psi
0.06 q 3 delta
r fi

0.05 2
Angular velocity: p,q,r [rad/s]

psi,delta,fi [ rad ]
0.04 1

0.03 0

0.02 -1

0.01 -2

0 -3

-0.01 -4
0 50 100 150 200 250 300 350 400 450 500 0 50 100 150 200 250 300 350 400 450 500
time [s] time [s]

Folie 28
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Pure spins
Euler’s rigid body equations expressed in the principal inertia frame have the
following form:
I11 p −(I 22 − I 33 ) q r = 0
I 22 q −(I 33 − I11 ) r p = 0
I 33 r −(I11 − I 22 ) p q = 0

Three elementary solutions to these equations are the pure spins:

p = constant; q = r ≡ 0 - major-axis spin

q = constant; p = r ≡ 0 - intermediate-axis spin

r = constant; p = q ≡ 0 - minor-axis spin

The stability of these pure spins is a central issue in spacecraft attitude

dynamics.

Folie 29
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Stability of pure spin

With respect to a perturbation ∆y0(t), a pure spin about the major and minor
spin axes of a rigid body, p(t) or r(t), is stable, while a pure spin about the
intermediate spin axis, q(t), is unstable. This result is however only true with no
energy dissipation (see later) and as such ideal.

I33
I11 > I22 > I33
 I11 : stable axis
I22  I22 : unstable axis
 I33 : stable axis
I11

Note instead that the attitude ψ(t), ∆(t), φ(t) is always unstable. A
perturbation in the attitude will always grow indefinitely.

Folie 30
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
The free rigid body – generalities on stability
A pure rotation about one of the principal axes of inertia is an equilibrium
solution ( dΓ dt = 0 ). Any pure rotation about any other axis is impossible.
A pure rotation about the maximum principal axis of inertia is a solution of
stable equilibrium. This means that the effect of a small perturbation gives
rise to small changes in the velocity for any successive time. In the
presence of energy dissipation the disturbance tends to disappear.
A pure rotation about the minimum principal axis of inertia is a solution of
stable equilibrium. However, in the presence of energy dissipation this
rotation is not stable. It will instead precede until the axis of minimum
inertia is at 90 deg from Γ̂ and the rotation is about the maximum axis of
inertia – flat spin.
A pure rotation about the intermediate principal axis of inertia is an
unstable solution, independently of energy dissipation. A small
perturbation will bring the intermediate axis of inertia to drift away from Γ̂.

Folie 31
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
 Example: Intermediate spin axis
I body = diag ([29.2,30,38.4]) kg m 2 [Ψ , ∆, Φ ]0 = [0,1.569, 0.0014] rad
Nominal 𝜔𝜔
�0 body = [0, 0.0698, 0] rad/s - perturbations in p and in r
Sr ≤ 0.0002 rad/s
Perturbed 𝜔𝜔0 body = [0.0001, 0.0698, 0.0001] rad/s
0.08
Body Reference ω phi versus time Euler Angles
psi
velocity: p,q,r [rad/sec]

0.06 delta
p 4 fi
q
0.04 r
[rad/s]

[rad]
2
0.02
velocity

phi(t)
0

Attitude
0

-0.02 -2
Angular
Angular

-0.04
-4

-0.06
-6
-0.08 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
time [s]
time [s]

• First 500 s: constant spin angular velocity

• Unstable solution, with PERIODICY (T=4500s).
Folie 32
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Energy Dissipation
One aspect that is not taken into account by the model of a rigid body is
the dissipation of internal energy due to internal forces and torques
Souces of dissipation:
- Structural damping in flexible elements (e.g., solar panels)
- Drag in liquid sloshing of propellant
All tumbling motions converge to a pure rotation about the axis of
maximum inertia. This phenomenon is called FLAT SPIN

nominal rotation axis

antennas

Explorer I – NASA - 1958

flat spin axis

Folie 33
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Energy Dissipation
A rigid body is a mathematical abstraction. Furthermore, a rotational
motion of the body in general implies a time-dependent force field
within it. An element dm with vector z relative to the cente of mass
experiences an acceleration (see Lesson 1)

a = bd2z / dt2 + bdω / dt x z + 2 ω x bdz / dt + ω x ω x z

Even with the rigid body assumption, after which bd2z / dt2 = bdz / dt =0,
an internal force field results with the remaining terms, which gives rise
to small, but not insignificant, material deformations. Since ω(t) is
generally not constant, the internal force field varies with time.

The material deformations result in energy dissipation. This dissipative

character cannot be neglected, regardless of how small this dissipation
may be.

Folie 34
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Energy sink hypothesis
This hypothesis is useful to describe the flat-spin phenomenon. It states that
during the motion of any real body the kinetic energy will tend to be converted
slowly to heat energy.
For the torque-free motion, the kinetic energy T then, after the energy sink
hypothesis, is such that
T < 0
It will only be T = 0, if the internal force field is constant, i.e. a pure spin.
At the same time, the angular momentum Γ is constant. Therefore the ratio
Γmod
2
( I112 ω12 + I 22
2
ω22 + I 332 ω32 )
I= =
2T ( I11ω12 + I 22ω22 + I 33ω32 )
is slowly increasing. Its physical limit is however when I=I3 (with the usual
assumption I11 < I22 < I33), which is a pure spin about the major axis.
The major-axis spin is then asymptotically stable, in that a disturbance from
the stable pure spin about the major axis of inertia decays to zero with
progressing time.

Folie 35
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
 Torques

CONSERVATIVE:

Gravity Gradient
Magnetic torque

EXTERNAL
NON- CONSERVATIVE:

Drag (Aerodynamic
torque)
TORQUE Solar pressure

NON- CONSERVATIVE:

INTERNAL
Sloshing
Flexible appendages
* Spacecraft Aerodynamics Torques, NASA Space Vehicle Design
Criteria,1971

Folie 36
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Secular effect of external torques on flat-spin axis

The flat-spin axis is, assuming an external torque-free model, inertially fixed
(dissipation due to internal torques).

However, if orbital disturbances are accounted for, there results a substantial

drift of the spin axis about the inertial frame.

Source: Hughes

Folie 37
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Gravity Gradient Torque P

Let points P0 and P belong to a rigid body. P0 is

the center of mass of the body.
The moment of the gravity gradient force is given by P0
[ ]
M
µ
gg = ∫M ρ × G.ρ dm = −
τM 3 ∫ ρ × 3rˆ 0rˆ 0 − μ .ρ dm r
rmod
M

which, after some manipulation, gives

M =
3µ 0
τgg 3 rˆ × ∫M ρ mod
rmod
[
2
]
μ − ρρ dm . rˆ 0
r0

3µ 0
τgggg = 3 rˆ × I.rˆ 0
M
rmod Earth

When is τgg zero? When r̂ and I.rˆ are parallel, i.e. I.rˆ = λrˆ , which implies
0 0 0 0
0
that r̂ has to be parallel to one of the principal axes of inertia of the body. It can
be shown that the only stable of these is the one for the minor principal axis.

Folie 38
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
 Image source: http://lasp.colorado.edu/home/csswe/system/subsystems/adcs/

Magnetic Torque
A spacecraft can possess a net magnetic moment 𝐦𝐦𝑚𝑚 ,
contributed to by permanent magnets and magnetic
loops. When 𝐦𝐦𝑚𝑚 is non-zero, the spacecraft is a
magnetic dipole.

From physics, we know that a magnetic dipole in the

presence of a magnetic field experiences a magnetic
torque given by
𝛕𝛕mt = 𝐦𝐦𝑚𝑚 × 𝐁𝐁
where 𝐁𝐁 is the magnetic flux density. In the neighborhood of the Earth, 𝐁𝐁 is non-zero. The result of
this torque is to align the magnetic dipole to the magnetic field.
𝜇𝜇
The magnetic potential for Earth’s magnetic field is 𝜙𝜙𝑚𝑚 = − 𝑚𝑚2 sin(𝜆𝜆𝑚𝑚 ),
𝑅𝑅
where R is distance from the center of the Earth, 𝜆𝜆𝑚𝑚 the latitude with respect to the equitorial
plane, and 𝜇𝜇𝑚𝑚 = 1 × 1017 Wb.m is the Earth‘s dipole strength.

Earth‘s magnetic flux density can be expressed in the spacecraft body frame by

𝐁𝐁 = 𝐀𝐀bE 𝐁𝐁 E = 𝐀𝐀bE (−𝛻𝛻𝜙𝜙𝑚𝑚 )

with 3 sin 𝜆𝜆𝑚𝑚 cos 𝜆𝜆𝑚𝑚 cos(𝜂𝜂𝑚𝑚 )
𝜇𝜇𝑚𝑚
−𝛻𝛻𝜙𝜙𝑚𝑚 = − 3 3 sin 𝜆𝜆𝑚𝑚 cos 𝜆𝜆𝑚𝑚 sin(𝜂𝜂𝑚𝑚 ) , where 𝜂𝜂𝑚𝑚 is the longitude.
𝑅𝑅
3 sin2 𝜆𝜆𝑚𝑚 − 1
Folie 39
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Example of chaotic dynamics: THE COSMIC POTATO
Hyperion, a moon of Saturn
Irregular shape
Chaotic rotation (after observations with Voyager 1 and Voyager 2)
Chaos results from dynamic model as Euler-equations with only gravity-
gradient torque*
NO axis stability (flat-spin) on long term despite tidal dissipation
Artificial satellites in Earth orbit may show a similar behaviour

(*) J. Wisdom and S.J. Peale ,“ Chaotic Rotation of Hyperion”, ICARUS 58, pp.137-152, 1987.

Folie 40
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
State of the art: Sloshing dynamics
NLR (Dutch Space Agency) built an experimental spacecraft:
SloshSat FLEVO Facility for Liquid Experimentation
and Verification in Orbit
launched 2005
Flat Spin manouvres
Flat–spin is almost completed in t = 830 s

Folie 41
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Summary of possible motions of tumbling satellites

General tumbling motion – due to collision, explosion,

uncontrolled thruster boost, other
– if no energy dissipation on-board (no
flexible appendages nor fuel sloshing,
negligible air drag, no eddy currents)
Flat spin – if energy dissipation present
Drift of flat spin axis – due to orbital disturbances
– up to 90 degrees drift per month, e.g.
TIROS I satellite [Hughes, “Spacecraft
Attitude and Dynamics”]
Chaos of Hyperion, moon of Saturn

Can it be measured from ground? Yes, with Laser Ranging or Radar

technologies.

Folie 42
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics
Literature

Slotine, Li, “Applied Nonlinear Control”, Prentice-Hall Int., 1991.

Hughes, “Spacecraft Attitude Dynamics”, 2004.
Lecture notes of the course in “Space Systems”, Prof. Arduini, University
of Rome.

Folie 43
On-Orbit Dynamics and Robotics | R. Lampariello | TU Munich | Summer Semester 2015 | Attitude dynamics