2.5 Leberre

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

STAKE: A TOOL TO OPTIMIZE FULL ELECTRIC STATION-KEEPING

LE BERRE M. (1), GICQUEL A.-H.(2)


(1)
Tel. (+33) (0)5 62 19 53 81, Email michel.leberre@astrium.eads.net
(2)
Tel. (+33) (0)5 62 19 66 99, Email anne-helene.gicquel@astrium.eads.net
Astrium Satellites, 31, Avenue des Cosmonautes, F-31402 TOULOUSE CEDEX 4, FRANCE
Telefax: (33) (0)5 62 19 97 97

1. INTRODUCTION 2.1. Satellite geometry

Electric thrusters are generally mounted on the anti-


Electric propulsion is a promising solution to reduce the nadir face of the spacecraft, symmetrically regarding
mass of propellant needed for Telecom satellites (X,Z) and (Y,Z) planes (excepted for original 2-
operations. It is already used on Astrium Eurostar thrusters configurations) but can also be mounted on
E3000 platform for North-South station-keeping. In the nadir or E/W faces. The thrusters are tilted around Z
frame of an ESA study, Astrium has developed a design axis towards E or W directions, and around X axis from
& analysis tool to optimize full electric station-keeping, N/S direction to avoid the contamination of solar panels
the so-called STAKE (STAtion-Keeping with Electric or other equipments, implying erosion and torques. Tilt
propulsion) tool. It handles station-keeping in angles depend on thrusters plume angle (from 15° to 45°
geostationary orbit with full electric or hybrid depending on thrusters’ type), on the control capacity
propulsion based on an optimizer from ASTOS desired on each direction, and on the advantage to direct
Solutions, GESOP (described in [2]), which is used to thrusters axis through or near the CoG to limit torques.
compute the optimum manoeuvres. For some configurations thrusters can be mounted on
TOM (Thrust Orientation Mechanisms) to follow the
STAKE tool complements the line of Astrium Mission CoG direction during the spacecraft life.
Analysis tools and Flight Dynamics operational
software QUARTZ (described in [1]). STAKE tool allows considering any full electric
configuration including up to 8 electric thrusters (th. 1
The principles and design of STAKE tool are presented, to 8) plus 2 chemical thrusters (th. 9 and 10) for hybrid
and then interesting results obtained with first analyses configurations, as shown on Fig.1.
conducted with STAKE are described.
Thrusters’ positions, directions and performances can be
defined in the tool. Moreover the case of a thruster
2. STAKE TOOL CAPABILITIES failure after a given duration may be simulated.

STAKE tool has two complementary roles. The first one


is the optimization of the Station-Keeping strategies for
a given thrusters configuration. The second one is the
optimization of a thrusters configuration for a given
station-keeping strategy by tuning the thrusters
directions. Moreover STAKE provides a cluster
geometry analysis functionality.

Several options for cost and constraints functions are


available. It is especially possible to control the satellite
angular momentum while fulfilling the station-keeping
requirements in terms of station-keeping box. The
satellite definition is representative and modular: the
thrusters configuration can be defined as well as the
reflectors & solar panels configuration.

Fig.1. Thrusters configuration


STAKE tool allows considering different satellite ⎧Δa = 2.aS .ΔVT /VS
geometric configurations: ⎪
- a unique central body, ⎪Δex = 2.( ΔVT /VS ) cos S + ( ΔVR /VS ) sin S
⎪⎪Δey = 2.( ΔV /V ) sin S - ( ΔV /V ) cos S
- two solar panel configurations (4 or 5 arrays),

T S R S
(2)
- four antenna sizes (or no antenna) for each of the ⎪ Δix = ( Δ VN /V S ).cos S
four reflector positions. ⎪Δiy = ( ΔV /V ).sin S
⎪ N S

⎪⎩ΔS = -2.ΔVR /VS


Predictable triaxiality is modelled with plume effect
time tables available for each satellite geometry and ΔVT , ΔVN and ΔVR are the velocity increment along
depending on TOMs and solar arrays angles.
tangential, normal and radial axes.

Manoeuvre realisation errors such as unpredictable The Station-Keeping cycle pattern is defined by a given
triaxiality are also modelled. number of days, each day being either a day of electric
manoeuvres during which each electric thruster may be
2.2. Station-Keeping strategies actuated once or a day without manoeuvres. In case of
an hybrid configuration chemical thrusters may be
The purpose of Station-Keeping is to control orbital actuated on one day only.
perturbations so as to keep the spacecraft in its
longitude/latitude box while minimizing ergols
consumption and taking into account additional
constraints. The final purpose is to apply to the
spacecraft optimal orbit control manoeuvres on a
temporal duration called cycle.

Inclination vector evolution due to luni-solar gravity


perturbation, eccentricity vector evolution due to solar
radiation pressure and mean longitude evolution due to
the anomalies in the Earth gravitational field have to be
counteracted by orbit control manoeuvres. Fig.2. Station-Keeping cycle

With a full electric propulsion configuration, the control The classical types of control on inclination and
of inclination, eccentricity and semi-major axis are eccentricity (on a point, on a circle or in a circle) are
performed simultaneously. In the case of hybrid available in STAKE tool.
configurations inclination control is realized with
electric thrusters, longitude control with chemical 2.3. Optimization of Station-Keeping manoeuvres
thrusters and eccentricity control with both electric and
chemical. The optimisation problem consists in minimizing a cost
function subject to constraints (linear or non linear)
The equations that give the effect of electric and which shall be null or negative.
chemical manœuvres on adapted keplerian parameters
are recalled respectively in Eqs.1-2. ⎧ min F ( x ) subject to :
⎪ x
⎪ A ⋅ x ≤ B or Aeq ⋅ x = Beq (linear constraints)
⎧Δa = (4.α /VS ωS ).aS .γ T ⎨ (3)
⎪Δex = (2.sinα /V ω ).[2γ ⋅ cosS + γ ⋅ sinS] ⎪C(x) ≤ 0 or Ceq ( x) = 0 (non linear constraints)
⎪ S S T R

⎪⎪Δey = (2.sinα /VS ωS ).[2γ T ⋅ sinS - γ R ⋅ cosS] ⎩lb ≤ x ≤ ub
⎨ (1)
⎪Δix = (2.sinα /VS ωS ).γ N ⋅ cosS
⎪Δiy = (2.sinα /VS ωS ).γ N ⋅ sinS The cost function F shall be linear and continuous. The
⎪ constraints functions may be non-linear and non-
⎪⎩ΔS = - (4.α /VS ωS ).γ R continuous functions.

α is the half thrust arc. γT, γN and γR are the tangential, The Station-Keeping manoeuvres optimization problem
normal and radial accelerations applied during an orbital is solved in STAKE tool using CAMTOS hybrid
arc (S-α,S+α) (with S mean right ascension). VS is the optimizer from GESOP software (ASTOS Solutions).
satellite speed and ωS is the satellite angular velocity.
Note that a ΔS is equivalent to a Δlm as S=lm + The state vector x, given in Eq.4, is defined as the set of
Greenwich sidereal time. thrusts half angles (for electric thrusters) or delta
velocities (for chemical thrusters) and mean right manoeuvres (6th function of Eq.1-2), as explained in
ascensions of thrusts. Eq.5.

⎛ α the1 ⎞
⎜ ⎟
(
Δl drift = − Δl nat + Δl bias + Δl man ) (5)
⎜# ⎟
⎜ α the8 ⎟
⎜ ⎟ Δl nat is the longitude natural drift, Δl bias is an input
⎜ Sthe1 ⎟ bias that can be used to refine the longitude centring,
⎜# ⎟
x=⎜ ⎟ (4) Δl man is the longitude effect of the SK manoeuvres (6th
⎜ Sthe8 ⎟ function of Eq.1-2).
⎜α ⎟
⎜ thc1 ⎟
⎜ α thc 2 ⎟ Δl drift is computed tanks to the semi-major effect of
⎜ ⎟ the SK manoeuvres (1st function of Eq.1-2) and Eq.6
⎜ ΔVthc1 ⎟
⎜ ΔV ⎟ which gives the relation between the semi-major axis
⎝ thc 2 ⎠ modification and the increment of longitude drift rate.

The cost functions useful for a global system analysis 2aS


are available in STAKE tool: Δa mod i = − Δdi (6)
3ωS
- consumed mass of ergol,
- thrusting time,
- delta velocity, Finally the system of five orbital constraints given in
- power consumption, Eq.7 is obtained.
- torque disturbance controlled with TOMs or
modulated thrusts, ⎧C1orb (α thei , Sthei , ΔVthci , Sthci ) = Δl nat + Δl bias
- orbital excursion. ⎪
⎪C2orb (α thei , Sthei , ΔVthci , Sthci ) = −Δex nat
⎪⎪ orb
⎨C3 (α thei , Sthei , ΔVthci , Sthci ) = −Δe y
nat
Different constraints may be taken into account alone or (7)
simultaneously: ⎪ orb
⎪C4 (α thei , Sthei , ΔVthci , Sthci ) = −Δix
nat
- orbital constraints,
⎪ orb
⎪⎩C5 (α thei , Sthei , ΔVthci , Sthci ) = −Δi y
- power, nat

- kinetic momentum controlled with TOMs or


modulated thrusts,
- ergol mass. These constraints have been generalized to the hybrid
configuration case and to any definition of the Station-
Keeping cycle.
2.4. Orbital constraints

The obvious constraints to take into account for


manoeuvre optimization are the “orbital constraints” 3. STAKE TOOL DESIGN
which are mandatory to solve the Station-Keeping
problem. 3.1. Architecture

From Eq.1-2 the effect of the manoeuvres performed by STAKE software is composed of:
all thrusters over the cycle may be expressed as a - a framework allowing the run of activities and the
system of functions of the state vector variables defined management of the database through dedicated
in Eq.4. Then, a constraints system is deduced from this GUIs, the plot edition, the management of archives,
functions system by considering on the left side the - a database (orbit, spacecraft constants),
desired corrections. - Flight Dynamics algorithms (executables),
- GESOP optimizer,
For inclination and eccentricity, the required corrections - interfaces between Flight Dynamics algorithms and
aim at controlling the natural drifts from the beginning the optimizer.
of current SK cycle to the beginning of next SK cycle
(only the secular drift of inclination is corrected). STAKE tool provides 2 main activities:
- Station-Keeping simulation,
For semi-major axis, the correction aims at creating a - Optimization of thrusters’ configuration.
modification of the semi-major axis Δa mod that implies
a drift in longitude to compensate the natural drift of the The Station-Keeping simulation activity allows
longitude plus the longitude effect of the SK performing simulations over a given period (from one
cycle up to the satellite lifetime). For this activity Station-Keeping cycle such as: corrections to perform,
GESOP optimizer is called once per cycle to compute optimized manoeuvres bulletin, cumulated on-time of
the orbit control manoeuvres (taken into account in each thruster of the configuration.
constraint “orbital constraints” or cost function “orbital
excursion”) has shown in Fig.3. Plots of outputs data from SK simulations are available
to control orbital parameters evolution and SK
manoeuvres characteristics as shown in Fig.5-7.

Fig.3. Station-Keeping simulation

The Optimization of thrusters’ configuration allows


optimizing thrusters directions regarding a given
Station-Keeping strategy simulated over a given period. Fig.5. Orbital parameters evolution free/controlled
For this activity the optimization problem is defined
with the cost function being the total cost over a Station-
Keeping simulation and the state vector being the
thrusters’ angles (Rx, Rz) to optimize. In order to limit
the computation time for this activity it is interesting to
consider symmetric configurations. Thus GESOP
optimizer is called at two levels has shown in Fig.4.

Fig.6. Cumulated on-time and consumed mass

Fig.4. Optimization of thrusters’ configuration

3.2. Outputs

STAKE simulations provide all relevant data on each


Fig.7. Thrust right-ascensions and durations
In the case of an optimization of thruster configuration
with two variables (symmetric 4-thruster configuration
for instance) the optimization path of multiple initial
guesses may be plotted over the cost function mapping,
as shown in Fig.8.

Fig.9. Hybrid configuration

Fig.8. Conf. optim. path over cost function mapping


Fig.10. 4-thruster configuration

4. MAIN RESULTS

Preliminary analysis with STAKE tool allowed


comparing Station-Keeping strategies with a classical
hybrid configuration and the two considered full electric
configurations. Furthermore it allowed to analyse the
possibility to take into account new constraints with full
electric configurations and demonstrated the interest of
a preliminary optimization of the configuration with Fig.11. 8-thruster configuration
STAKE tool. Moreover sensitivity to different on-
station longitudes and spacecraft configurations has Tab.1 gives the comparison of the consumed mass over
been analysed. Robustness of the station-keeping 15 years.
strategy on a thruster failure has also been assessed with
the tool. Tab.1. Compared configurations consumed mass

4.1. Thruster configurations analysis

The SK performances have been compared with three


kinds of thrusters’ configurations:
- hybrid configuration (2 electric / 2 chemical),
- 4-thruster symmetric full electric configuration,
- 8-thruster full electric symmetric configuration
(corresponding to the 4-thruster one plus additional The costless configuration is the 8-thruster
thrusters 5,6,7,8). configuration.

The gain of consumed mass with this configuration as


compared to hybrid configuration is of 8.8% which is
explained by the fact that for hybrid one the longitude
control is performed with chemical thrusters 9 and 10
which are less efficient than electric thrusters in term of
mass consumption.

The 4-thruster configuration is more costly than the 8-


thruster one as the first configuration is included in the
second one, restricting degrees of freedom for the
optimization of SK manoeuvres.
Fig.12 gives the plot of compared delta velocities in made on the consumed mass. They are less interesting
satellite frame. as the aim of the Station-Keeping is not to minimize the
orbital excursions but to keep the satellite inside its
control window.

4.3. 8-thruster configuration optimization

The cost saving which may be obtained with a 8-


thruster full electric configuration previously optimized
with STAKE tool has been analysed.

To reduce the computation time the 8-thruster


configuration is optimized symmetrically:
- the two angles of th.1-5 are optimized (4 variables),
- the angles of th.2,3,4,6 are deduced symmetrically,
- th.7-8 are fixed.

A regular mesh of 8 initial guesses is considered to


improve the solution.

Fig.13 shows the symmetric 8-thruster full electric


Fig.12. Compared configurations delta velocities configurations after optimization.
Fig.12 shows that between hybrid configuration and full
electric configurations the cost on ±X axis increases due
to the fact that the global control with electric thrusters
with radial components implies parasitic radial effects
for each manoeuvre. Between 4-thruster and 8-thruster
configurations the cost increases on –Z axis due to the
use of thrusters 5 and 6, and decreases on +Z axis due to
the lower use of thrusters 1,2,3,4.
Fig.13. Optimized 8-thruster configuration
Note that these configurations have not been
preliminary optimized in term of thrusters’ orientations. The main modifications from initial configuration is the
high reduction of Rx angle of th.1,2,3,4 and the slight
4.2. Cost and constraints analysis increase of Rx angle of th.5-6.

Simulations with kinetic momentum constraints show Tab.2 gives the compared consumed mass with initial
that TOMs are more interesting than off-modulations as and optimized configurations, adding TOMs use and
TOMs mean a relative optimization of the configuration kinetic momentum constraint.
at each SK cycle. Simulations performed with the 8-
thruster full electric configuration defined in ch.4.1 gave Tab.2. Compared 8-th. configuration - consumed mass
an over consumption of around 10% with off-
modulations.

The use of TOMs without the kinetic momentum


constraints to improve only the orbital control is
possible in STAKE tool. This option allows a reduction
of the consumed mass by optimizing the configuration
at each cycle (in the limit of TOMs clearance). With the
configurations defined in ch.4.1 the consumption
reduction was around 14% with the hybrid
configuration and around 15% with the 8-thruster full Tab.2 shows that the optimization process reduces the
electric configuration. Station-Keeping cost of around 25.5% without kinetic
momentum control. The gain is slightly improved with
The Torque disturbance and Orbital excursion cost TOMs (from 25.5% to 27.1%).
functions imply important over costs as no limitation is
Note that the case of a non-optimized configuration with which corresponds to the longitude control budget. The
TOMs is less interesting than an optimized longitude control budget is maximum for 120 deg East
configuration without TOMs because angles domain longitude (-2 m/s), along -X axis in order to decrease
allowed for configuration optimization in this analysis is the semi-major axis implying a longitude drift along +X
larger than TOMs clearance. axis counteracting the negative natural longitude
deceleration which is maximum at this longitude.
The kinetic momentum control with TOMs with the
optimized configuration gives a higher mass The sum of delta velocities on +Y and –Y is the same
consumption than without kinetic momentum control for each reference longitude which corresponds to the
(gain reduced from 25.5% to 19.0%) but is still below inclination control.
the non-optimized configuration with TOMs without
kinetic momentum control. The sum of delta velocities on +X and –X and the sum
of delta velocities on +Z and –Z are the same for each
4.4. Station longitude analysis reference longitude which corresponds to the
eccentricity control.
A station longitude analysis has been performed with a
8-thruster full electric configuration and a typical Moreover the analysis show that the 8-thrusters full
geometric configuration (4-part solar panel and 2 small electric configuration is robust to a single thruster
reflectors). failure (th.1 or any among 2,3,4 by symmetry). The
failure of th.1 is compensated by an increased use of
Three relevant station longitudes were considered: th.3 in order to keep the longitude control on -X axis.
- 74.9° East without any acceleration (stable position), The SK performances on inclination and eccentricity
- 120° East, where the negative acceleration is vectors are equivalent but the longitude window is
maximum (-0.00197 deg/day2), slightly exceeded (0.06 deg at maximum).
- 164° East without any acceleration (instable
position). 4.5. Geometric configuration analysis
The longitude dead band was 0.05deg.
A geometric configuration analysis has been performed
The Station Keeping has been analysed over 1 year with a station longitude sensitive to a geometric
starting with an initial orbit consistent with the reference modification (120deg East) with different satellite
longitude. geometries: number and size of the reflectors and size of
the solar panels.
Simulations were performed with triaxiality and orbit
restitution errors. The constraints taken into account Simulations results show that the required wheel
were the orbital constraints and the kinetic momentum unloading per cycle depends on the reflector geometry
control with TOMs. The cost function was the as it is linked to the perturbing torques due to solar
consumed mass of ergol. radiation pressure (SRP) on the reflectors only as the
solar array configuration is symmetric. The most
Tab.3. Compared longitudes - delta velocities asymmetric configuration implies the most important
perturbing torques due to SRP and the most important
wheel unloading. This implies a difference of repartition
between on-times of th.1,2,5 (+X axis torque) and 3,4,6
(-X axis torque).

The number and size of the reflectors and the size of the
solar arrays increase the sum of +Z and -Z delta
velocities corresponding to the eccentricity control. This
is due to the increased value of the SRP area coefficient
which impacts directly the amplitude of the natural
Simulation results show that the consumed mass are eccentricity radius and thus the consumed mass. This
very close between stable and instable position, which is phenomenon counterbalances the wheel unloading
due to the model of natural longitude evolution which is reduction in term of consumed mass, leading to
quasi identical at first order between the two cases. With consumptions nearly similar between the different
the maximum negative acceleration the consumption is geometric configurations (variations of about 0.1%).
slightly higher (less than 1%).

Tab.3 shows that the difference of delta velocities on


+X and –X is different for each reference longitude,
5. WAY FORWARD

A concrete use of the tool is likely to occur in the frame


of the NEOSAT (Next Generation telecom Platform)
development. STAKE tool is therefore a very good
basis for a comprehensive design and mission analysis
tool in the scope of full electric propulsion. It could be
used to support the system engineer in the definition of
the electric thrusters’ configuration

STAKE tool paves the way for the future integration of


electrical platforms specificities into QUARTZ, the
Astrium Flight Dynamics operational software.

6. ACKNOWLEDGEMENTS

Astrium Satellites acknowledges ESA for entrusting us


with the study leading to the development of STAKE
tool and to very interesting results on full electric
Station-Keeping.

7. REFERENCES

1. QUARTZ Flight Dynamics software. Online at


http://www.systema.astrium.eads.net/index.php?op
tion=com_content&task=view&id=31&Itemid=43
2. GESOP software. Online at
http://astos.de/products/gesop

You might also like