Deutsche Telekom AG, Zentrale
Satellite Systems Operation
Postfach 10 00 03, 64296 Darmstadt
Tel. +49-6151/83-6469
E-mail:
marcus@dasite.consat.de
antimo@dasite.consat.de
Deutsche Telekom owns and operates three geostationary telecommunication and direct broadcasting satellites. The Telekom Flight Dynamics Group took charge of the stationkeeping activities from DLR in April 1994. This Paper presents our operational satellite experiences. Examples for regular operations and repositioning are given. A computer demonstration of automation tools like Kalman Filter and Operation Scheduler is planned. Deutsche Telekom already offers satellite operational services in remote mode from its redundant nodes of Usingen Satellite Control Center and Darmstadt Flight Dynamics Center. Moreover Deutsche Telekom plans to play in the near future a significant operational role in one of the most advanced MEO telecom systems.
In June 1991 Telekom decided to carry out the Flight Dynamics tasks with its own staff. Following a worldwide ITT the Telesat Canada Flight Dynamics System was selected 1992. The system was installed in November 1993 at the ground station at Usingen and in Darmstadt. Since December 1993 the Telekom Group shadowed the DLR state estimation and maneuver planning. In April 1994, three months before the end of DLR support, Telekom successfully took over the Flight Dynamic Activities. Since that Telekom succesfully performs the flight dynamics activities.
The FDS structure satisfies Telekom requirements in the following fields:
The Telekom Flight Dynamics Team is split into two groups. At the ground station, located in Usingen (60 km north of Darmstadt), three non-fulltime orbital analysts perform the everyday activities. The scientific tasks are carried out by a group of two astrodynamicists, located in Darmstadt. In case of emergency the group in Darmstadt can immediately take over all the activities.
FDS Group Usingen FDS Group Darmstadt Orbit Determination Maneuver Strategy Definition Maneuver Planning Performance Monitoring/Reporting Maneuver Calibration Emergency Support Predictions Mission Planning S.W. Updates
The FDS is designed in such a way that a single flight dynamics analyst can cope with the operational workload for the entire fleet. Because of safety reasons a second non fulltime analyst has to check the maneuver planning and a third analyst is needed for backup.
Software: Is based on a Fortran core with up to 20 years of
operational use. The implementation of the applications
in the X11-Windows system is done in C.
Operating The operating system is based on HP UX 9.x , which is
System: overlaid with X11-Windows and HP VUE. (A update to HP
UX
10 is in preparation).
Hardware: Server HP 715 80 MHz 128 Mbyte RAM
Workstation HP 710 50 MHz 64 Mbyte RAM
Network: IEEE 802.3 EtherNet LAN in Usingen and Darmstadt
connected via ISDN routers. (shown in Figure 1)
The indications of time needed to carry out the FDS tasks in this article refer to the harware used.
The Flight Dynamics
System (FDS) is an off-the shelf hardware and software system
aimed to monitor and control the orbit and attitude activities
throughout the satellite lifetime. The Telekom version was qualified
for a mixed fleet of up to ten spinning and three-axis stabilized
satellites, although only four three-axis stabilized satellites
have been operated so far. the FDS consists of about 30 applications.
Acces to the applications is given by the FDS Shell (shown in Figure 2).
This basic step can be performed in two different ways.
The Kalman Filter
( shown in Figure 3 ) monitors
in real time the satellite position without human intervention.
Incoming tracking data are automatically passed to the filter
and duly processed. The tracking data are smoothed, the old state
is propagated to the tracking data epoch, the new state is determined
and the residuals are computed within five seconds from data reception.
In general the Kalman Filter application is iconized in the windows
environment. In the iconized state the filter displays the real
time mini-graph of the range residuals. Slant range deviations
(predicted minus observed) as small as few meters are detected
in real time. Because of the fast converging capability of the
filter a very reliable state can be already obtained after two
tracking passes (6 hours) following the maneuver. Slant range
deviations larger than 4 sigma (16 to 32 m) trigger a warning
to the spacecraft controller & FDS analyst. At this time the
analyst can sleft the real time estimation of the diverging filter
and more closely analyse the causes of the divergence. After fixing
the error the tracking data can be reprocessed and the latest
state is determined.
This classical
tool accepts sets of tracking observations in form of antenna
ponting parameters (azimuth, elevation, range, range rate) or
a smoothed subset of them and it determines the average value
over the data arc for: the state vector, a (optional) correction
to the solar radiation force, and up to four sets of antenna pointing
bias. The iterative WLS algorithm can be augmented by an optional
a priori state, error covariance matrix, and a set of "considered"
parameters. A much used WLS feature is the possibility for the
user to preview the preprocessed tracking data and cull, modify
or delete individual or grouped observations according to user-defined
patterns, entered into a special dialog. Data culling can be graphically
accomplished in drag-and-drop mode. This quickly and effectively
solves problems like occasional bad tracking data. The number
of iterations needed by the application to converge is usually
four or less. Assuming a WLS over a period of 4 days, from end
of interactive data setup to state vector result reception up
to 60 seconds are needed. The WLS application is mainly used to
cross-check the Kalman Filter behaviour. The effects of possible
small solar array misalignments or of undetected in-plane components
of the automatic attitude control system (the latter one is only
valid for the DFS bus) are much better estimated by examining
the long-term orbit determination via WLS than by seeing the instantaneous
picture as given by the Kalman Filter.
This application
( see Figure 4 ) is used for
near-GEO orbit maneuvers. The application supports nine main maneuver
types (from arrival on-station to insertion into graveyard orbit),
but only the following ones have been operationally used by Telekom:
Templates are designed and stored by the user for repetitive maneuver scenarios, but the application allows non-standard maneuvers to be designed and validated in less than 30 minutes. The application uses the Kamel's mean elements in order to simplify the equations of motion and thrust equations. The planning process is template assisted and database driven. Repetitive maneuvers require the user to select fields, rather than enter parameters. Any time a maneuver is planned, its inclination and longitude effects on the spacecraft orbit are calculated and graphically displayed against the satellite deadband limits. Should deadband violations be detected as consequence of an incorrectly planned maneuver, the application requires the analyst's explicit ackowledgment, and it logs this reaction. All satellite maneuvering constraints are automatically checked for and the FDS shall accomodate for the related strategy changes. The user is then asked for acknowledgement. The planning run can be continued by the user, say for analysis studies, even if fundamental contraints are violated, but in this case the application shall deny the output of the maneuver message, thus preventing the maneuver execution under such circumstances.
Some 120 parameters
(see Figure 5) can be dynamically
projected in the future or in the past, at any desired step. They
deal with ephemeris for a single satellite, Sun, Moon and some
selected stars, and, if depending on tracking facilities, for
single or multiple antennas. The parameters are grouped in Stationkeeping
Checks, Tracking Data, Sun&Moon-related Ephemeris and so on
for operational convenience, but can be picked up at will, if
so needed. The position and velocity elements can be calculated
either as via numerical integration or by means of an upgraded
Kamel analytical integration method.
The scheduler (shown in Figure 6 ) is the deskleft application available to all analysts. From this calendar-like interface the user can:
The interesting feauture of the scheduler is that it works networkwide, i.e. if the appropriate privileges are granted, the instantaneous overview of all activities for the entire satellite fleet is available to everybody involved in the flight dynamics activities. An appropriate access management system coordinates the access to the databases so that only one user can work with a satellite at a time. Whenever a master file is changed, the event status is changed and violations are recomputed and redisplayed onto all workstations.
The Telekom DFS-Kopernikus satellites are maneuvered in longitude (EW) on a weekly and in inclination (NS) on a two week basis. TVSAT-2 is operated in EW and NS on a two week cycle. It results that the Flight Dynamics System has to cope with 11 maneuvers on a fortnight basis.
The filter is a powerful and elegant application which reaches a stable solution after two tracking pases (6 hours). The Filter eliminates the daily need for the analyst to retrieve tracking data and process them with a Weighted Least Squares Estimator. The following table summarizes the orbit determination precision operationally attained by the filter in GEO:
Satellite DFS-2 DFS-3 TVSAT-2 One-Sigma a (Km) 2.9E-2 2.8E-2 1.5E-3 e (-) 7.8E-7 4.8E-7 2.9E-7 i (deg) 3.8E-4 4.6E-4 2.1E-4 M (deg) 1.4E-1 1.2E-1 4.8E-2 AP (deg) 3.1E-1 2.1E-1 1.2E-1 RAAN (deg) 2.1E-1 9.5E-2 9.2E-2 Long. (deg) 9.1E-4 1.4E-3 3.5E-5 Drift (deg/day) 4.9E-4 3.6E-4 1.9E-5
The application allows repetive maneuvers to be planned in 20 sec, assuming that only the maneuver time is changed from the previous similar maneuver. If the only indication given to the application is the maneuver type and maneuver day, the planning time increases due to the optimization algorithms usage, but stays well below the two-minutes limit.
The three maneuvers per cycle (DFS bus) and their impact on the mid-term orbital behaviour can be outlined within an hour. This includes the actual planning time needed by the software and the integrated check of all the maneuvers together by the planning analysts and the cross-check by the second analyst.
Once that all maneuver in a cycle are planned as a whole, there is still a need to refine each of them, in order to take into account the performance of each single maneuver on the following one within the cycle.
The state vector used for maneuver optimization is first checked for plausibility against a set of well-defined criteria including, but not limited to, the three sigma error on the best knowledge orbit determination and the mid-term noise of the Kalman Filter solution. If the Kalman Filter noise increased in the past few days up to the 7 m level, the WLS is used to optimize the filter solution, until the Kalman Filter noise reverts to the 3 to 4 m level. This step may be needed approximately once every 2-months and requires some fifteen minutes to be completed.
If enough tracking data were collected since the last maneuver, a calibration is performed. In case the calibration shows a significant deviation from the expected value, this fact is further analyzed. The calibrated maneuver is then passed on to the Kalman Filter again, which provide the FDS engineer with the best knowledge orbit determination. Such a procedure can be carried out in some ten minutes or less.
The term calibration
is used for the analysis phase aimed to determine whether the
maneuver was actually executed as it was planned. The FDS carries
out this task by means of two methods.
The Station Acquisition/Relocation
application stacq is used to produce a sequence of maneuvers which
represents a fuel-efficient transfer from an initial orbit to
a user-specified target orbit. The application may be used to
plan a maneuver sequence to acquire station from an initial drift
orbit following Apogee Motor Firing, or to plan a relocation in
which a geostationary satellite is moved from one longitude to
another.
The maneuver sequence consists of any or all of the following maneuvers:
The user identifies which maneuvers are to be included in the sequence, and specifies targets for each of the selected maneuvers. The spacecraft Reaction Control System model is defined using standard FDS dialogs, or completely specified via user inputs. A feature of the application is its ability to construct a sequence around a set of pre-defined maneuvers. This allows the user to develop a large sequence of maneuvers which satisfies many constraints by successively running the application to achieve user-defined intermediate target orbits. In the given example ( Figure 7.1+7.2 ) a relocation simulation of TVSAT-2 from 0.60 W to 28.50 E is shown.
German Telekom Flight Dynamics Group can support all aspects of geostationary missions. The operational FDS has successfully been used to support the stationkeeping of over 22 satellites. Due to the flexibility of the FDS, various satellite bus types from different spacecraft manufacturers can be readily accomodated and supported.
The FDS was designed around the following guidelining principles:
In addition to this German Telekom and its customers rely on:
besides that, German Telekom provides the following workpackages for external customers: