1

I'm using STK 11.2 to find a minimum-time finite transfer between two coplanar orbits. The first orbit is circular and has a radius 7000km, the second orbit has a semimajor axis = 10000km and eccentricity = 0.1.

For the optimization I use Design Explorer Optimizer (in this version there is no SNOPT or IPOPT). Unfortunately, the solution doesn't converge.

What is wrong in my configuration?

enter image description here

enter image description here

enter image description here

enter image description here

enter image description here enter image description here

enter image description here

enter image description here

enter image description here

enter image description here

Leeloo
  • 833
  • 11
  • 29
  • different question but somewhat related and of interest to future readers: Software for low-thrust trajectory optimization – uhoh Sep 23 '20 at 04:17
  • First you should move the first propagate out of the TargetSequence. There is nothing to optimize. Second: why to you have a duration as a target? I think you can remove this to. – CallMeTom Sep 23 '20 at 10:29
  • @CallMeTom The first propagate is in the Target sequence because I want to find the optimal True anomaly to start the maneuver. Duration is a target because I want to minimize it. – Leeloo Sep 23 '20 at 10:46
  • @Leeloo: since you have not specified the target AoP and you start on a circular orbit, there is nothing like an optimal true anomaly? (execpt the numerical error produces a pseudo one) – CallMeTom Sep 23 '20 at 10:51
  • @CallMeTom Well, I've changed this. However, this doesn't affect the convergence. – Leeloo Sep 23 '20 at 11:18
  • @Leeloo, i quite sure, there is not one probem, but several. The AoP one was for sure. The next I would recommend to try changing the Maneuver Type to "impulsive". Look if it converge then. My experiance is, that non-impulsive manuevers always struggle to converge. First when the sequence is stable enough to converge with an impulsive maneuver try to make it more realistic – CallMeTom Sep 24 '20 at 05:35
  • if/when it converge, try a engine with more thrust at first, and try to reduce the thrust step by step till you reach your wanted thrust ... the sequence will also not converg, if the thrust is to low to reach your goal with one burn – CallMeTom Sep 24 '20 at 05:43
  • @CallMeTom If you have an STK, may I ask you to publish an answer? – Leeloo Sep 24 '20 at 05:52
  • I am ready to help you with experience I have and to point errors you made and I found. But I fear you will have to work out the answer by yourself, at least I am not ready to spent time on working out the whole scenario for you. Another topic, why this sequence may not converge: if you apply thurst and reach the target SMA, it is not given, that the ecc will be correct. So after raising your apo you would need a second burn to make the peri fit. – CallMeTom Sep 24 '20 at 06:21
  • @CallMeTom I changed the maneuver type to impulsive, the thrust increased to 1000 N, as the control used the cartesian components of velocity. It hasn't converge either (in several iterations it stops). – Leeloo Sep 24 '20 at 06:23
  • @CallMeTom May be I should use Differential Corrector first? – Leeloo Sep 24 '20 at 06:24
  • @Leeloo: yes! try the DC first. In general start with a very easy sequence and if/when it converge make it step by step more complicated. – CallMeTom Sep 25 '20 at 05:34
  • @CallMeTom Tried, the same thing, it doesn't converge. I've added a bounty for this question. – Leeloo Sep 25 '20 at 12:41
  • are you still trying to reach an orbit AND eccentricity? – CallMeTom Sep 26 '20 at 13:50
  • @CallMeTom I checked them separately, in my case even semi-major axes doesn't converge. – Leeloo Sep 26 '20 at 14:25
  • 1
    okey ... so to sum up, you have a circular orbit ,one impulsive maneuver and a new SMA.. and this does not converge?? strange... ehh, do you still have the propagate element before the sequence? I said you need to move it out, but does it has an abort criteria? – CallMeTom Sep 28 '20 at 11:00
  • @CallMeTom Yes, I checked your suggestions and it doesn't still converge. If you have an STK, it would be great if you check and publish a solution. – Leeloo Sep 28 '20 at 12:10
  • since i do not have an Astrogator license on my private comupter and since this is nothing I can do just quick during the lunch break, I am afraid I cannot help you with a ready to use solution. – CallMeTom Oct 01 '20 at 05:05

1 Answers1

2

From what I can tell from the parameters in your screenshots, the program has performed a perfectly valid optimisation.

What's more concerning is that the text in you question doesn't match the parameters clearly shown in the screenshots. If there's a configuration page where does values are actually used, why is this not included in the question?

The optimiser has been tasked to achieve these constraints:

  • Obtain a semi-major axis between 8000km and 8010km. The value it reached, 7999.57km, is within the tolerances you have specified. The values here are not the 10000km you claim in the question.

  • Achieve an eccentricity between 0 and 0.01. The value it reached is 0.00068, safely within. This is not your claimed input of 0.1

Given those two constraints, which has been met, it has minimised the time to a journey under 24 days.

The second concerning thing is the notice about "stop after specified duration". Since the value there is the exact same as what the optimiser output, is it perhaps the case that the provided time is not enough to reach the target orbit? 24 days is not a very long time as far as low-thrust transfers are concerned.

parameters

SE - stop firing the good guys
  • 43,384
  • 3
  • 143
  • 244