Weird stuff happening at Eve

[EdFred]

So, I was heading to Eve in sandbox mode to test some Eve landers/ascent vehicles.

I established a 107x107 orbit, then dipped the PE to about 95km, decoupled from my transfer/orbiter stage, and then re-established the 107km orbit with the orbiter. I then switched back to my lander when I noticed something odd: While my craft was in Eve's atmosphere, and above 90km, my apoapsis would increase. Everything was off. In fact my engines hadn't even been activated, so there wasn't any inadvertent burning. Since I was barely scraping the atmosphere, my apoapsis climbed every time I did an aerobrake. It actually climbed from 107km to 112km (might have been 116km). The periapsis would decrease each time, however.

Finally once my craft dipped below 90km the AP started to come down, but only for the time the craft was below 90. So as my craft would aerobrake from 97 to 90 the AP would climb, below 90km the AP would decrease, and then the AP would climb again as the craft went from 90 to 97. Eventually the craft stayed below 90km long enough that the AP would finally have a net descent for the orbit. I though it had something to do with the construction of the craft, but after making the descent and ascent it happened again. After rendezvousing with the orbiter stage at an orbit of 115km, I detached the lander can. The lander can consisted of 4 massless batteries, 4 massless solar panels, 2 RCS thrusters, a MechJeb module, and a clamp-o-tron. I burned the rest of the RCS on the unmanned lander can which left the PE above 90km but below 97km, and sure enough the AP would rise while the craft was in that same range.

Anyone else experience this? Does it happen at Kerbin, Duna, Jool, or Laythe?

Mods installed:

Enhanced Navball

MechJeb

Kerbal Engineer

Kerbal Alarm clock

Two other parts mods which had NO parts on this craft.

[magnemoe]

Was it under physical warp? My guess is some rounding error as you was in the absolute edge of atmosphere, if you aerobrake your Ap should drop far more than your Pe

[NFUN]

Did you make sure your ship isn't haunted?

[EdFred]

Was it under physical warp? My guess is some rounding error as you was in the absolute edge of atmosphere, if you aerobrake your Ap should drop far more than your Pe

I thought that might have been the case, but it happened at 1x 2x 3x and 4x warp. I had A LOT of orbits to watch it happen.

[TrooperCooper]

You enter the atmosphere, you got to deal with drag. It slows you down, which is why your Periapsis decreased. At the same time, depending on aerodynamics of your craft, the atmospheric stream generates lift, which carries you upwards and raises your apoapsis. Happens to me regularly when aerobraking, especially in spaceplanes for obvious reason. Though, I use FAR and I m not sure how much this experience can be transfered to stock aerodynamics...

It can even be observed during the ascend into LKO with a plane, AFTER switching off all engines and just gliding upwards... apoapsis rises, periapsis decreases...

[EdFred]

Currently experiencing the same thing with a 1 man lander can, 1 Z-1K battery, 2 of the 2x3 solar panels, a FL-T200, and a 48-7S. No matter the orientation, it happens.

Edit: Now happening even outside the atmosphere. PE down, AP up. Slower than when in the atmo, but AP still rising.

Edited August 17, 2014 by EdFred

[cantab]

I've experienced an unexplained orbit shift like this before, when my Mun periapsis was gradually changing with no engine thrust and the same change regardless of orientation. It seems to be some sort of bug. You can work around it by (non-physical) timewarping whenever possible, though that's no help if you're trying to aerobrake.

[Renegrade]

I've experienced an unexplained orbit shift like this before, when my Mun periapsis was gradually changing with no engine thrust and the same change regardless of orientation. It seems to be some sort of bug. You can work around it by (non-physical) timewarping whenever possible, though that's no help if you're trying to aerobrake.

Yeah, I've had instances where space stations would slowly accelerate out of orbit and/or rotate spontaneously even without engines, RCS, or trim. Sometimes even small vessels do this, but it seems to impact larger, more complex ones more.

My research suggests that the forces on the parts may be out of balance due to a floating point issue, so there's a net motion and/or torque of the entire vessel.

Specifically the forces of the vessel's components acting on each other, which would normally be in equilibrium, but thanks to the not-real approximation of floats, are slightly off, so force A doesn't exactly cancel out force B, resulting in random acceleration vector C and torque D which shouldn't be there.

There may very well even be a bias in the way that Squad doing math (optimized math is not necessarily correct math in the world of FPUs or SSE), wherein a careful study may allow them to minimize the error by re-ordering operations such that errors tend to cancel each other out. Possibly. Or the people at PhysX. Or Unity. Or someone.

[problemecium]

I've observed that in 0.24, orbits will decrease in eccentricity while physics is enabled and the ship is near and just past its apoapsis (and the opposite before the apoapsis). If I had to guess, I'd say that the orbit's eccentricity is increasing because you are near, but just before, your periapsis. In terms of precision it's a slight improvement over what we had up through 0.23 where the eccentricity would change by a significantly larger amount every time one exited timewarp.