De-Orbiting Satellites in Leo Using Solar Sails D. Romagnoli (1), S. Theil
De-Orbiting Satellites in Leo Using Solar Sails D. Romagnoli (1), S. Theil
De-Orbiting Satellites in Leo Using Solar Sails D. Romagnoli (1), S. Theil
+ =
(1)
being
on perturbati
a
(
(
|
.
|
\
|
(
(
|
.
|
\
|
(
(
|
.
|
\
|
=
z
z
y
z
x
z
E
J
r
r
r
r
r
r
r
r
r
r
R J
a
2
2
2
5
2
2
5 3
5 1
5 1
3
2
(3)
The J
3
term describes the elliptical shape of the Earths equator and may be represented by the
relation:
(
(
|
.
|
\
|
(
(
|
.
|
\
|
(
(
|
.
|
\
|
=
z
z
z
y
z
z z
x
z
z z
E
J
r
r
r
r r
r
r
r
r r
r
r
r
r r
r
R J
a
2
2
2
2
7
3
3
7
5
3
6
7 3
7 3
2
5
3
(4)
Therefore, the total perturbation coming from the non-spherical shape of the Earth may be
approximated by the sum of these three contributions:
3 2
grav J J
a a a
+ = (5)
2.1.2 Third Body Perturbation
Third bodies, such as the Moon and the Sun, affect the motion of spacecrafts in Earth orbit: the Sun,
in fact, is extremely massive, while the Moon is very close. It is important to note that this
perturbation becomes more relevant for high altitude orbits, that is when the atmospheric drag
effect begins to diminish. The expression for this perturbing acceleration may be derived from the
standard three-body problem. Assuming that the mass of the satellite is negligible, one may write:
|
|
.
|
\
|
=
3
3
3
3
sat3
sat3
3 3Body
r
r
r
r
a
(6)
being
3
the gravitational parameter of the third body,
sat3
r
is the position vector of the third body with respect to the Earth
in the inertial frame. Note that equation (6) may lead to numerical errors when the distance from the
satellite to the third body and the distance from the Earth to the third body are very similar: in fact,
the cube of these distances appears at the denominator in equation (6), therefore the numerically
computed difference is extremely small and may introduce numerical errors in the computation.
This problem may be solved using a Taylor series expansion neglecting the small terms [7]:
(
(
|
|
.
|
\
|
3
2
2
3
3 sat
2
3
3 sat
3 sat
3
3
3
Body 3
2
15
3 r
r
r r
r
r r
r r
r
a
(7)
It is important to note that even though equation (7) is numerically stable, it is an approximation
because the series is truncated. However, it represents a good approximation that can be efficiently
used for preliminary analysis.
2.1.3 Atmospheric Drag
Together with the oblateness of the Earth, the atmospheric drag is the most relevant source of
perturbing acceleration for LEO satellites. The relation that describes the acceleration due to the
atmospheric drag is the well known:
rel
rel
rel
D
v
v
v
m
A C
a
2
drag
2
1
= (8)
where is the atmospheric density value,
D
C is the drag coefficient, A is the projected area in the
direction of the velocity vector relative to the atmosphere, m is the total mass of the spacecraft and
rel
v
is the velocity of the spacecraft relative to the atmosphere. Assuming the atmosphere is rotating
with the Earth and neglecting any wind, the relative velocity may be approximated by the relation:
r
dt
r d
v
rel
=
Earth
e (9)
being sec rad 000072921 . 0
Earth
= e the angular velocity of the Earth around its spin axis. The key
point in computing the atmospheric drag is to determine the atmospheric density value. Different
models are available in the literature, based on both an analytical and an empirical approach, which
may lead to remarkably different values of the atmospheric density according to the parameters
used to feed the model. In this paper, the Harris-Priester model has been adopted [8]. Even though
this is basically a static model, the physical properties of the upper atmosphere in a range between
100 km and 1000 km are computed taking into consideration the solar cycle via tabulated values
(Tab. 1). Then, interpolation determines the density at a particular time and altitude. Note that the
coefficients in Table 1 correspond to the indices F10.7 = 140 and Ap = 30, according to the
predicted values for 2014. Thanks to its computational efficiency, this model is often adopted in
early analysis and whenever the simulation requires an estimate of the density value during
different phases of the solar cycle. In equation (8), the projected area A represents the connection
between the attitude dynamics and the orbital dynamics. As a matter of fact, the atmospheric drag
creates a torque acting on the spacecraft that changes its orientation with respect to the inertial
frame and, as a consequence, the projected area changes over time and the atmospheric drag
acceleration varies during the orbital motion. It is therefore very important to properly simulate the
attitude motion in order to accurately predict the acceleration due to the atmospheric drag.
Tab.1 Example of data tables used in the Harris Priester atmospheric model [7]
Heigth
(km)
Minimum
Density (kg/m
3
)
Maximum
Density (kg/m
3
)
Heigth
(km)
Minimum
Density (kg/m
3
)
Maximum
Density (kg/m
3
)
100 4.974 x 10
-7
4.974 x 10
-7
420 1.558 x 10
-12
5.684 x 10
-12
120 2.490 x 10
-8
2.490 x 10
-8
440 1.091 x 10
-12
4.355 x 10
-12
130 8.377 x 10
-9
8.710 x 10
-9
460 7.701 x 10
-13
3.362 x 10
-12
140 3.899 x 10
-9
4.059 x 10
-9
480 5.474 x 10
-13
2.612 x 10
-12
150 2.122 x 10
-9
2.215 x 10
-9
500 3.916 x 10
-13
2.042 x 10
-12
160 1.263 x 10
-9
1.344 x 10
-9
520 2.819 x 10
-13
1.605 x 10
-12
170 8.008 x 10
-10
8.758 x 10
-10
540 2.042 x 10
-13
1.267 x 10
-12
180 5.283 x 10
-10
6.10 x 10
-10
560 1.488 x 10
-13
1.005 x 10
-12
190 3.617 x 10
-10
4.297 x 10
-10
580 1.092 x 10
-13
7.997 x 10
-13
200 2.557 x 10
-10
3.162 x 10
-10
600 8.070 x 10
-14
6.390 x 10
-13
210 1.839 x 10
-10
2.396 x 10
-10
620 6.012 x 10
-14
5.123 x 10
-13
220 1.341 x 10
-10
1.853 x 10
-10
640 4.519 x 10
-14
4.121 x 10
-13
230 9.949 x 10
-11
1.455 x 10
-10
660 3.430 x 10
-14
3.325 x 10
-13
240 7.488 x 10
-11
1.157 x 10
-10
680 2.620 x 10
-14
2.691 x 10
-13
250 5.709 x 10
-11
9.308 x 10
-11
700 2.043 x 10
-14
2.185 x 10
-13
260 4.403 x 10
-11
7.555 x 10
-11
720 1.607 x 10
-14
1.779 x 10
-13
280 2.697 x 10
-11
5.095 x 10
-11
760 1.036 x 10
-14
1.190 x 10
-13
290 2.193 x 10
-11
4.226 x 10
-11
780 8.496 x 10
-15
9.776 x 10
-14
300 1.708 x 10
-11
3.526 x 10
-11
800 7.069 x 10
-15
8.059 x 10
-14
320 1.099 x 10
-11
2.511 x 10
-11
840 4.680 x 10
-15
5.741 x 10
-14
340 7.214 x 10
-12
1.819 x 10
-11
880 3.200 x 10
-15
8.059 x 10
-14
360 4.824 x 10
-12
1.337 x 10
-11
920 2.210 x 10
-15
3.130 x 10
-14
380 3.274 x 10
-12
9.995 x 10
-12
960 1.560 x 10
-15
2.360 x 10
-14
400 2.249 x 10
-12
7.492 x 10
-12
1000 1.150 x 10
-15
1.810 x 10
-14
2.1.4 Solar Radiation Pressure Perturbation
Solar radiation pressure is the disturbing action due to the interaction between the photons coming
from the Sun and the outer surface of the spacecraft. Even though solar sails rely on this interaction
to generate the desired thrust, in low Earth orbit and for the kind of application considered this has
to be considered as a perturbation acting on the spacecraft. The solar radiation perturbation may be
represented with a relation similar to the one used for the atmospheric drag acceleration:
Sun sat
Sun sat
SRP
=
r
r
m
A SRPc
a
SUN R
(10)
being
R
c the reflection coefficient, SRP the current value of the solar radiation pressure and
SUN
A
the projected area in the direction of the incoming radiation from the Sun. A detailed description of
the model adopted for modeling the solar radiation pressure can be found in [9]. It is important to
note that for LEO objects the perturbing acceleration from the solar radiation pressure is several
orders of magnitudes smaller than the other disturbing actions. As a consequence, this perturbation
plays a minor role in the simulation of the orbital evolution of a low Earth orbit satellite. Note
equation (10) shows an intimate connection between the attitude and the orbital dynamics as well:
the computation of the projected area in the direction of the incoming solar radiation, in fact,
requires the knowledge of the current attitude of the spacecraft. As a consequence, it is crucial to
propagate both the attitude and the orbit at the same time in order to properly simulate the problem.
2.2 Attitude Dynamics and Kinematics
Assuming the sailcraft to be a rigid body without internal moving parts, the equations of motion for
the attitude dynamics may be easily derived from the standard rotational equation of motion of a
rigid body about an arbitrary point [10]:
O
M dm R r
=
}
(11)
where r
is the position vector of the infinitesimal mass element dm relative to an arbitrary point O,
R
= (12)
being
}
= dm R r H
O
the angular momentum of the system about the reference point O. The total
angular momentum of the system is related to its mass distribution via the inertia matrix of the rigid
body about a body fixed reference frame with its origin at the center of mass:
e
J H = (13)
Finally, doing the time derivative in equation (12) in the inertial frame, one can prove that the
attitude dynamics equation for a rigid spacecraft may be written as:
O
M J J
= + e e e (14)
with proper initial conditions on the angular velocity in the body fixed frame. The attitude
propagation is completed with the kinematics equation. The most convenient and numerically
efficient way of representing the orientation of a spacecraft is via quaternions. A quaternion q is
formed by four parameters ( )
4 3 2 1
, , , q q q q according to the relation [11]:
4 3 2 1
q kq jq iq + + + q (15)
being i, j and k the hyperimaginary numbers satisfying the following conditions:
j ik ki
i kj jk
k ji ij
k j i
= =
= =
= =
= = = 1
2 2 2
(16)
The quaternion in equation (15) may be also seen as made of a scalar or real part,
4
q , and a vector or
imaginary part,
3 2 1
kq jq iq + + . As a consequence, using the vector-first notation, the quaternion
q may be also represented as:
( )
4
, q q
= q (17)
Given the definition of quaternion in (17), the attitude kinematics equation in quaternion notation
may be written as follow:
q q O =
2
1
(18)
where O is the skew-symmetric matrix computed from the angular velocity vector in the body
fixed frame.
3 Simulations Analysis
The simulation of both the orbital and the attitude dynamics at the same time is a challenging
numerical task for many reasons. The most crucial one, though, is that they have remarkably
different frequencies. As a matter of fact, the orbital motion can be efficiently computed using an
integration time step in the order of some minutes, while the attitude has to be propagated at a much
higher rate, in the order of at least 10 Hz, to have a good representation of the forces depending on
the actual orientation of the spacecraft in case of high angular velocity values. Hence, the
integration time step has to be selected to meet the requirements for the attitude propagation,
resulting in extremely long computational time. Besides, since the simulated time is usually in the
order of years, the second aspect that has to be faced is related to the maximum available memory
to store the simulations data, i.e. state vectors for orbit and attitude, state vectors of the Moon and
the Sun and perturbing accelerations. As a matter of fact, during simulations it turned out that
MATLAB encountered severe problems in managing memory when the saved files became bigger
and bigger thanks to the adopted 10 Hz resolution. Even though sampling the saved data at a lower
rate was the easiest way to try coping with this issue, eventually it has been not possible to run
simulations longer than several months including the complete attitude and orbital dynamics with
high frequency.
The main test case scenario considers the following conditions:
- Orbit: Dawn-Dusk Sun Synchronous Orbit at 620 km altitude
- Total Satellite Mass: 140 kg
- Sail Area: 25 m
2
- Atmospheric Density Model: Harris Priester
This scenario is consistent with the one adopted for the DLR satellite Asteroid Finder (Fig. 3),
which is being presently considered as a candidate mission to test de-orbiting using a solar sail.
Figure 3. Schematic representation of the current design of the Asteroid Finder satellite
However, in order to check the simulation algorithms avoiding the problems associated with
computation time and memory management, a first run has been made with different conditions.
Particularly, a lower 450 km circular equatorial orbit has been selected so that a shorter simulation
time was required. The preliminary simulation showed that the spacecraft in the test scenario re-
enters the atmosphere in approximately three weeks (Fig. 4): note that the vehicle is considered re-
entered into the atmosphere when it reaches an altitude of 150 km above the geoid. However,
further investigations have to be carried out in a more general framework to determine at which
altitude the booms of the solar sail are no longer able to handle the aerodynamic torques, causing
the collapse of the structure and defining the altitude at which the sail is not effective any more.
Since the attitude has been propagated together with the orbital motion, it is possible to plot the
cross section with respect to the atmosphere over time during the simulation (Fig. 5). This
information makes it possible to easily visualize the orientation of the spacecraft with respect to the
atmosphere: it is clear that the aerodynamic torque acting on the spacecraft is strong enough to
provide a passive stabilization, with the sail acting as a parachute and offering a constant area of
about 24 m
2
to the rarefied atmosphere. This leads to the best case scenario, where the drag
provided by the sails membrane is maximized for free without any kind of active attitude control
thanks to the action of the aerodynamic torque. It is important to note that this passive stabilization
might be achieved only if the aerodynamic torque is strong enough. This simulation used an
integration time step of one second, that is the attitude dynamics has been sampled at 1 Hz and the
associated computation time turned out to be approximately 12 hours. As a consequence, it is easy
to imagine that increasing the initial altitude of the spacecraft to the one of the main scenario
remarkably increases the simulation time up to unmanageable values.
Figure 4. Altitude of the spacecraft over the geoid over time
Figure 5. Cross area over time
The simulations performed within the main scenario conditions showed that, as expected, also in
this case the de-orbiting time might be considerably reduced adopting the sail as an aerobraking
device. However, the simulations showed that the desired passive stabilization to maximize the
area exposed to the atmosphere can not be achieved for the Asteroid Finder-like scenario. This is
due to the fact that the available aerodynamic torque is not enough to get rid of the satellites
rotation. As a matter of fact, the atmospheric density is one to two orders of magnitude less at 620
km than at 450 km, resulting in a much smaller drag force according to equation (8). In addition, the
peculiar geometry of the Asteroid Finder satellite leads to a center of pressure very close to the
satellites center of mass. The center of pressure, in fact, is located at about seventy centimeters
from the center of mass in the Y direction of the body fixed frame. As a consequence, the small
lever arm combined with the low atmospheric density value lead to a weak aerodynamic torque and
to the impossibility of a passively stabilized attitude. Figure 6 shows an example of the cross area
over a period of approximately one week: it is clear that the spacecraft tumbles considerably,
resulting in a cross section area within the whole range 0 m
2
and 25 m
2
.
Figure 6. Cross area over time (detailed view of a week timeframe)
Nevertheless, using Tab.1 for computing the atmospheric density, the spacecraft re-enters the
atmosphere after approximately 390 days, which is remarkably less than the 9.82 years expected for
the Asteroid Finder satellite in the same solar conditions (Fig. 7).
Figure 7. Spacecraft altitude over time for a 620 km SSO orbit
Figure 7 shows the typical atmospheric re-entry behavior: the orbital decay starts extremely slow
and constant for most of the time, accelerating remarkably in the last days thanks to the increased
atmospheric density value. Hence, the presence of a 5x5 meters solar sail on board a spacecraft like
Asteroid Finder represents a big improvement in reducing the danger associated with a dead
satellite at the end of its lifetime and allowing to remove a piece of debris from the orbit in a very
efficient and affordable way. However, it is important to note that there is a remarkable uncertainty
in evaluating the satellites life time. As a matter of fact, the de-orbiting time may double when the
solar flux parameter F10.7 is slightly modified, resulting in much different results. (Fig. 8). Hence,
the prediction of the solar activity and of the geomagnetic indexes is extremely important in order
to properly evaluate the orbital life time of a spacecraft.
Figure 8. Orbital decay for the Asteroid Finder satellite with different solar F10.7 fluxes
Besides, it is also interesting to note that such strategy may also be adopted not only for satellites
natively designed to use a solar sail to de-orbit at the end of lifetime, but also for non cooperative
targets. As a matter of fact, the German Space Agency is currently investigating the possibility of
rendezvous and docking with a non cooperative spacecraft: in this framework, the use of a cubesat
with a deployable solar sail and the ability of chasing and catching a non cooperative target may be
successful in removing space debris from LEO or moving GEO satellites to a different orbit without
using any propellant but the atmospheric drag or the solar photon trust, respectively.
4 Conclusions
In this paper, the possibility of using a solar sail for de-orbiting spacecrafts in low Earth orbit has
been considered. A deep theoretical background associated with the simulation of such an
application has been given, as well as a description of the major problems that may occur when
facing this task. This preliminary analysis showed that using a small 5x5 meters solar sail can
reduce the de-orbiting time up to about 390 days compared to the 9.82 years obtained considering
only the external surface of the Asteroid Finder satellite and the same solar flux conditions. Hence,
using a solar sail as an aerobraking device may be useful and represent an efficient way of testing in
space all the procedures and technical solutions for packaging and deploying the sails booms and
membrane within the framework of a more structured solar sailing developing program. Further
investigations have to be carried out to include the structural strength of the sail assembly (boom
plus membrane) into account, in order to better understand when the solar sail is not working any
more as a drag device due to the collapse of the structure caused by the aerodynamic torque. In
addition, the simulation has been performed neglecting any possible lifting force: as a matter of
fact, since the sails membrane is flying with a certain angle of attack with respect to the
atmosphere, the effect of the lift force may extend the lifetime of the spacecraft. However, modeling
both the aerodynamic coefficients is not trivial and more detailed experimental data may be
required to accomplish this task.
References
[1] Geppert U., Biering B., Lura F. and Reihnard R., The 3-Step DLR-ESA Gossamer Road Map
to Solar Sailing, Proceedings of the 2
nd
International Symposium on Solar Sailing, New York,
USA, 2010.
[2] Romagnoli D. and Oehlschlgel T., Solar Sail Attitude Control Using a Combination of a
Feedforward and a Feedback Controller,Proceedings of the 2
nd
International Symposium on Solar
Sailing, New York, USA, 2010.
[3] Scholz C., Romagnoli D. and Dachwald B., Performance Analysis of an Attitude Control
System for Solar Sails Using Sliding Masses, Proceedings of the 2
nd
International Symposium on
Solar Sailing, New York, USA, 2010.
[4] Romagnoli D. and Oehlschlgel T., High Performance Two Degrees of Freedom Attitude
Control for Solar Sails, Accepted by Advances in Space Research Special Issue on Solar Sailing,
2011.
[5] Vallado D. A. and Finkleman D., A Critical Assessment of Satellite Drag and Atmospheric
Density Modeling, AIAA/AAS Astrodynamics Specialist Conference and Exhibit 2008, 18-21
August 2008, Honolulu, Hawaii, USA.
[6] Woodburn J. and Tanygin S., Efficient Numerical Integration of Coupled Orbit and Attitude
Trajectories Using an Encke Type Correction Algorithm, AAS 01-428.
[7] Vallado D. A., Fundamentals of Astrodynamics and Applications Third Edition, Springer
New York, NY.
[8] Montenbruck O. and Gill E., Satellite Orbits: Models, Methods, Applications, Springer, 2005
[9] McInnes C. R., Solar Sailing: Technology, Dynamics and Mission Application, Springer,
1999
[10] Wie B., Space Vehicle Dynamics and Control, AIAA Education Series, 2008
[11] Wertz J.R., Spacecraft Attitude Determination and Control, Kluwer Academic Publishers,
2002