Use of WalkerConstellation for constellation propagation

Orekit usage
1

Hi all,

I am starting to use the class WalkerConstellation and its related classes. I want to propagate the orbits of the entire constellation and I was wondering if my understanding of the use of the class was correct.

To propagate an entire constellation, instead of propagating each satellite individually, the most efficient way is to:

  • set an initial state and build a propagator only for this one
  • build the Walker constellation (definition of nb of planes and satellites)
  • propagate the initial state
  • on each step, rebuild the regular slots

I am guessing this is time-efficient but probably less accurate in terms of propagation? The propagation model will make the whole constellation drift as one, whereas if I had propagators for each individual satellite, the drift for each satellite will be slightly different.

Is this correct?

final EphemerisGenerator generator = propagator.getEphemerisGenerator();
SpacecraftState finalState = propagator.propagate(initialDate.shiftedBy(simDuration));
BoundedPropagator ephemeris = generator.getGeneratedEphemeris(); 

// Create Walker Constellation
WalkerConstellation constellation = new WalkerConstellation(1152, 24, 0);
List<List<WalkerConstellationSlot<Orbit>>> constellationSlots;

//Propagate from the initial date one day ahead
for (AbsoluteDate propagationDate = initialDate; 
propagationDate.compareTo(initialDate.shiftedBy(simDuration)) <= 0;
propagationDate = propagationDate.shiftedBy(dt)){

          currentStep++;
          double progress = (double) currentStep / totalSteps * 100;
          if (progress % 10 == 0){
              System.out.printf("Date: %s Progress: %.0f%%\n", propagationDate, progress);
          }

          SpacecraftState spacecraftState = ephemeris.propagate(propagationDate);
          constellationSlots = constellation.buildRegularSlots(spacecraftState.getOrbit());
}
2

I would personally use a different method to handle this.
The idea of WalkerConstellation is to give an initial state for the satellites, and from them to build several independent propagators. These propagators could eventually be grouped later on using a PropagatorParallelizer.
Also note that regardless of the propagator you use, performing the time loop oneself is not the recommended way. What is recommended is to set up a step handler and run the propagator in one sweep from start to finish, letting the propagator call the handler during the propagation. You can find examples of this in the propagation tutorial (look at the intermediate steps section). The design is explained in the propagation documentation. Using step handler is both must faster and in fact simpler as it decouples what is done at each step.

Another point to mention is that WalkerConstellation is based on simple theoretical Keplerian shifts, it is therefore simplified and typically does not include perturbations like zonal harmonics. You can still of course propagate from these initial states using an accurate propagator will all perturbations you want, but you have to be aware you started from something simplified. This means this is more intended to fulfill mission analysis tasks rather than operational tasks (for which you should start with individually estimated orbits).