Bug: Orbit constantly changing!

[NJC2]

Is anyone else having problems with orbits changing and ships not having electricity when switching to them? I started this career save in 1.0.2, but I didn't have this problem until a few days ago after 1.0.4. I'm playing career mode on a custom hard mode setting, which means things are going quite slow. I am running mods, but I have re-produced this in stock KSP without mods.

I finally managed to orbit the Mun for the first time in this career. After the burn to intercept the Mun, I set myself up to have a 25k Pe when I arrived. After crossing the Mun's SOI my Pe jumped to 50k. I made a correction burn to bring it back down to 25k, then circularized at 25k x 25k. I EVA'd my scientist to grab the experiments I collected in high space and reset them to collect low orbit and gathering low orbit biome science.

This mission completed 3 contracts, so before I left the Mun I switched back to the Space Center to check and see if any new contracts were available that I could complete with this ship. I was in the Sun and the batteries were full when I left, but when I switched back it was out of power and was all of the sudden in a 48k x 30k orbit and both the Ap and Pe were dropping at a pretty fast rate. I was on the opposite side of the Mun for the exit burn, and only had just enough fuel to make the return burn. The orbit dropped down to 32k x 26k while waiting to be in position for the return burn.

Today accepted some new contracts, one of which was to gather science from space around Kerbin. I already had satellites in orbit from previous contracts, so I could easily knock that one out. I switched to an existing satellite to transmit science and it was out of electricity, but it was in the sun and immediately began charging up. It was in a stable 500K x 400K polar orbit and fully charged when I left it, but its orbit began dropping just like the ship in orbit around the Mun.

I decided to install a fresh KSP and start a new sandbox game to see if it was mod related or my save, but it is still happening. I installed one mod, HyperEdit, so I could quickly throw a ship into an orbit. I loaded the stock vessel Orbiter 1A and placed it in a 30K circular orbit around the Mun. After almost an hour, its orbit has dropped to 29.1K x 28.7K, and is still dropping.

[kiwi1960]

Only thing I can suggest is that you install RCS units and then activate them to help maintain your orbit. (Mechjeb does that for you, I believe.)

I once had a space station that i tried to dock with, using mechjeb, and i watched do multiple attempts, failing, backing off and trying again...

It was then that I noticed the station was dropping out of orbit, it started at 100K ... and dropped to below 98K before I noticed. I got my kerbals into a pod and escaped but that took time, by the time I was able to try and save the station, it was already too late. That was in a version before version 1.0 ... to this day, I don't know why it did that. It started off in a stable enough orbit.

I suppose these things just happen, its never happened since, so maybe your "bug" isn't really a bug.

EDIT: Have you considered that while the Mun has no atmosphere, it still has gravity? Maybe that is your problem. 30K might be a tad low.

Edited July 4, 2015 by kiwi1960

[Stargate525]

I know there is voodoo that you can do with phantom forces and part clipping... To rule that up, throw a probe core up there in hyperedit and see what happens to it.

[NJC2]

I know part clipping can cause phantom forces, which is why I don't clip parts in my own designs. The stock craft does have clipped parts, so I threw a Stayputnik core in the same 30k x 30k orbit and left it running for an hour. To rule out any torque forces I disabled the reaction wheels and drained the battery in the VAB before HyperEditing it in orbit. This time the Ap began rising and the Pe lowering, but after an hour the Ap and Pe were ~100m lower than where it started out.

I just remembered while typing this that I may have 1.0.2 on a backup drive. I'll check that out later and see if it happened in 1.0.2, but I'm pretty sure it didn't because I would have noticed it when fulfilling satellite contracts.

[NJC2]

I did still have 1.0.2 on my backup drive and repeated the experiment. I put a Stayputnik core in a 30k orbit and left it running for an hour. It still happens, but it's not as bad. After an hour the orbit had only dropped by 10m.

[sal_vager]

Hi NJC2, floating point errors will accumulate over time, and while Squad has tried many things to minimize the issue it's not something that is likely to ever be eliminated, sorry.

Severe changes in orbit are due to the claw bugging out though, or part clipping as have been mentioned, and a reproducible case would be welcome, without using any addons by the way, as even having plugins loaded can affect KSP.

[quyxkh]

Can't you just run off the orbital parameters when there's no active thrust? If there's no atmosphere and you're only counting one gravity source, there's no need to sim, you can just calculate directly, no?

[ibanix]

This is a really, really annoying issue for anyone trying to make SatCom systems with RemoteTech.

Is there any way around this issue, even at the loss of other accuracy? Can the orbit simulation calculations be turned OFF in some cases? Once I get a craft into a stable orbit with no SoI change and no atmospheric effects, I don't want to have the orbit being recalculated unless a dV change happens.

[Alshain]

I don't think it is as simple as turning it off when there is no thrust. Your craft may not be under thrust but that doesn't account for the force of that craft that.is.about.to.crash.into.you... LOOK OUT!

You either calculate physics, or you don't.

[ibanix]

You either calculate physics, or you don't.

That can't be true, because we have the 'on rails' mode where these things don't happen.

[steve_v]

I don't think it is as simple as turning it off when there is no thrust. Your craft may not be under thrust but that doesn't account for the force of that craft that.is.about.to.crash.into.you... LOOK OUT!

There is absolutely no reason to do collision physics, or anything beyond relatively basic orbital calculations if the orbital path will not intersect any other objects. Why would there be?

No thrust & no external forces besides gravity == no strange orbital perturbations. If this is happening on rails, it's pain mathematical error. If it's happening off-rails, it's still mathematical error, with more variables involved. Or we could just blame Unity some more.

[sal_vager]

HarvesteR mentions the inaccuracies with orbits here if you're interested.

[ibanix]

HarvesteR mentions the inaccuracies with orbits here if you're interested.

That appears to be about the SoI transition, not the issues with 'regular' orbits.

[sal_vager]

Well if you read it...

This is caused not by an imprecision with the estimation itself, nor is this a floating-point inaccuracy issue for once;

He does say that imprecision of orbits are often caused by "floating-point inaccuracy".

[ibanix]

That's not much help, I already knew that =)

[sal_vager]

This post references a post explaining the issue but it was lost during the forum wipe, might be worth looking for an archived copy via Google.