Burn Time and Special Relativity?

[maddog59]

I've noticed something odd and am not sure if it's something I'm doing or if anyone else has seen this: it seems as though during particularly long burns, whether w/ chemical or nuclear engines, that the burn time indicator seems to slow down significantly. For example, last night I set up an asteroid intercept that should've gotten me to within about 5km of my target. It used a single Mainsail engine, and ran for nearly 4 minutes. Once I started the burn, I set a timer for 3 minutes at the 3 min/4 sec mark . When my timer went off there was still more than 20 seconds of burn time to go, rather than the expected 4 seconds. When the burn completed, I was more than 11,000 km off from my target. 

I tried it a few more times, using both MJ2 maneuver utility and manually, with the same result. I was able to see that the seconds on the burn time indicator at some point began ticking down much slower than the other timers. 

I saw this in earlier attempts to make planetary xfers using nuclear engines and thought it was my imagination, or something I'd set incorrectly, but now I'm not so sure.

Thoughts from the hive mind? 

[steuben]

Special relativity doesn't really kick in till you are around 1/2c and even then the clocks run at about 85% of normal to the outside viewer. To the inside view they still run at 100% of normal. 

What you are most likely seeing the fact that rocket science is hard. Well one aspect of it anyways. Rocket science is mathematically hard.  You thought high school calculus was hard, this stuff makes HS Calc look like a cake walk. There are very few closed form equations in rocket science. unless you are dealing with spherical cows in a vacuum on a frictionless plane. Things get really ugly when dealing with varying masses and curving space-time. Pretty much one of the reasons that the Good Lord Math! asked the Good Lord Electronics! for the miracle of electronic analog computers.

So to keep things simple some approximations are made. The majority of what you are seeing the accumulated errors from the approximations. They are much more visible with longer burns than short ones.

The errors are kind of ignored since usually the burns are live monitored rather, than in a set and forget chain, so they can be extended as required.

[maddog59]

9 minutes ago, steuben said:

Special relativity doesn't really kick in till you are around 1/2c and even then the clocks run at about 85% of normal to the outside viewer. To the inside view they still run at 100% of normal. 

Thanks; I was actually being facetious - I like to try to inject some humor or whatever in my titles to generate interest/curiosity. I'm somewhat familiar with Einstein's work, but that's a story for another space/time. 

From a gameplay perspective, if what you say is correct than it seems like that amount of error makes it much more difficult to make those planetary transfers than I'd anticipate. I don't have much experience w/ nuke engines, but since they have such low TWR and require exceptionally long burns, I'm not sure  how to account for this - shorter burns that result in multiple orbits until one gets to a point where a short burn will finally let the ship escape Kerbin's (or another planet's) SOI and be on it's way?

[AHHans]

While what @steuben writes is correct as such, the actual differences that you see are not due to rounding errors but due to the difference between an instantaneous maneuver and a long burn.

The prediction from the maneuver node is done with the assumption that the whole maneuver is applied to the trajectory as a single, instantaneous "kick". But in reality the burn takes some nonzero time, which results in a different orbit than the prediction. If the burn is short (a small fraction of the orbit) then the difference is also short, so that you may not notice it. But on long burns the difference becomes large: you not only wind up on a different trajectory than predicted but you also need more dV.

To deal with this you can split your burn into multiple burns: to save dV you can split your burn over multiple orbits. Do one or more short burns at the PE until you are on an elliptical orbit close to the edge of the SOI (but still inside) and then do the final escape in one (probably longer) burn. Ta actually arrive at the target do what they also do in RL: make one or more correction burns somewhere during the transfer.

[maddog59]

1 hour ago, AHHans said:

While what @steuben writes is correct as such, the actual differences that you see are not due to rounding errors but due to the difference between an instantaneous maneuver and a long burn.

The prediction from the maneuver node is done with the assumption that the whole maneuver is applied to the trajectory as a single, instantaneous "kick". But in reality the burn takes some nonzero time, which results in a different orbit than the prediction. If the burn is short (a small fraction of the orbit) then the difference is also short, so that you may not notice it. But on long burns the difference becomes large: you not only wind up on a different trajectory than predicted but you also need more dV.

To deal with this you can split your burn into multiple burns: to save dV you can split your burn over multiple orbits. Do one or more short burns at the PE until you are on an elliptical orbit close to the edge of the SOI (but still inside) and then do the final escape in one (probably longer) burn. Ta actually arrive at the target do what they also do in RL: make one or more correction burns somewhere during the transfer.

Doing course corrections is how I've been doing it ... I hadn't thought of first establishing such a high orbit, then committing to a maneuver that would be so much smaller a fraction of the orbital period. Thanks, I'll have to give that a try.  

[AHHans]

1 minute ago, maddog59 said:

I hadn't thought of first establishing such a high orbit, then committing to a maneuver that would be so much smaller a fraction of the orbital period. Thanks, I'll have to give that a try.  

The tricky part is figuring out the correct place for the PE: you want it to be at the place where you need it for the final escape burn.

I don't have a good procedure to find that. Up to now I've essentially just "winged it".

[king of nowhere]

three minutes is still not all that large. but if done in kerbin orbit, you still get some inefficiency for cosine losses. 16 extra seconds out of a 3 minutes burn? 8% more deltaV required, could be a reasonable result

[maddog59]

On 10/14/2021 at 5:08 PM, king of nowhere said:

three minutes is still not all that large. but if done in kerbin orbit, you still get some inefficiency for cosine losses. 16 extra seconds out of a 3 minutes burn? 8% more deltaV required, could be a reasonable result

Well, I usually plan my missions with a ~10% margin, and in that particular instance it was a major annoyance: it had taken me about an hour of fiddling with the various widgets to set up a single node that would get me to within 5 km of the target, and after all of that effort I ended up with a separation of more than 11,000 km. Plus, using more fuel for the course correction really made for a skin-of-the-teeth completion. 

[Kryxal]

On 10/13/2021 at 3:18 PM, AHHans said:

The tricky part is figuring out the correct place for the PE: you want it to be at the place where you need it for the final escape burn.

I don't have a good procedure to find that. Up to now I've essentially just "winged it".

Starting from an orbit with a period of 30 minutes or so, a "kick" of about 430 m/s will result in an orbit of about 90 minutes.  Plan that "kick" 3 orbits before your target departure time.  Similar methods can be used so long as you're short of the Mun's orbit, so under about 800 m/s in any combination, past that planning gets more complicated.