Almost captured with no injection burn!

[X-SR71]

So I thought of sharing these with you guys, since I've always thought it wouldn't be possible to get captured with gravity alone.

So I was hanging around Duna in a orbit inside Ike's orbit when the game predicted an encounter and collision with Ike (solid orange path, no periapsis).

I thought 'well, I'll decrease the orbit a bit'. Set up a maneuver node, and the new orbit was represented by the dotted orange path.

Except at the end of the new predicted path it also predicted a straight encounter and capture with Ike!

I thought 'no way this is gonna work', but surely enough, after the burn, I had the encounter and capture predicted with no injection burn:

I was pretty amazed since I've never had this happen before!

Too bad it didn't actually work. When close to the descending node of the second picture Ike ended up capturing and putting me into a collision course

[Xyphos]

It's not possible, and you should NEVER trust the prediction system, EVER.

[cantab]

I wouldn't go quite so far, but yes, there are cases where the game fails to predict an SOI transition.

KSP uses the patched conic approximation, which means that if you enter an SOI then you must leave it unless something alters your orbit. Floating-point rounding errors might allow a capture like this though.

[SkyRender]

It's actually theoretically possible in real life to be captured solely by gravity. It's very unlikely, of course, but due to the fact that gravity is not uniform across a body, it's entirely possible. That said, the encounter would have to be very close to already being captured in the sphere of influence of the body that it was trying to capture around, and the gravitational variation would have to be both significant and properly distributed relative to the incoming object's trajectory.

It's very possible to have this happen in KSP because the SOI transition maths are less than perfect. A simple rounding error is all it takes.

[Kobymaru]

Also, for OP: capture without a third object is mathematically impossible, both in KSP and real life.

However, if your closing velocity (the speed at which you approach Ike) is very small, and you just "strafe" the SOI, then you make one single almost-orbit and then exit it again.

It's not possible, and you should NEVER trust the prediction system, EVER.

I understand why you would say that, because the prediction system has failed me numerous times already. But what else should we use? Use 4D vector math with pen on paper, to figure out our space-time trajectories?

Edited August 2, 2015 by Kobymaru

[Korbit]

It's not possible, and you should NEVER trust the prediction system, EVER.

I'm very new to the game, but what I've tried to do when possible is to "insure" the prediction by making tiny adjustments to the burn and/or the time of the burn and see if the prediction still holds true. If only a tiny tweak will negate the prediction, I consider it precarious and try when possible to find an alternate route.

[Hannu]

KSP have some errors in SOI change detections. Therefore orbit predictions are often inaccurate. It is very annoying especially around Jool.

Such a capture needs 3-body interaction in real world. Third body takes energy and leaves system with higher velocity. However, KSP's simplified gravity model loses always energy when ship enters into SOI. It is not very difficult to get such an orbit with bodies which have small SOI compared to their mass. Jool's large moons and Ike are good candidates.

[Skorpychan]

Doesn't that first image show your encounter with Ike passing THROUGH Ike?

[Sharpy]

A gravity assist of a moon can certainly let the planet capture given body. So can aerobraking.

[Kuu Lightwing]

Huh, funny. that shouldn't work. What does work, though is using some moon for capturing. It saves so much dV at Jool...

[Sandworm]

Also, for OP: capture without a third object is mathematically impossible, both in KSP and real life.

However, if your closing velocity (the speed at which you approach Ike) is very small, and you just "strafe" the SOI, then you make one single almost-orbit and then exit it again.

I understand why you would say that, because the prediction system has failed me numerous times already. But what else should we use? Use 4D vector math with pen on paper, to figure out our space-time trajectories?

Gravity waves. It's possible in a 2-body system if the loss due to gravity waves or like effects is enough to reduce the velocity of the inbound object. It would probably require something silly like a cloud of gas with a periaps a few feet from the horizon of a supermassive black hole, but there is almost certainly some way to do it even with just two spheres.

[haltux]

A gravity assist of a moon can certainly let the planet capture given body.

Not in KSP where n-body gravity is not simulated. If it does in KSP it is a bug / floating point error.

So can aerobraking.

No, not without a third body. Without thrust, a body that touch the atmosphere is doomed to go back to the atmosphere again and again and crash at some point.

I am no expert but you can easily notice that there Earth does not have plenty of satellites.

[Red Iron Crown]

KSP uses the patched conic approximation

It's not even an approximation in KSP, it's how KSP's universe really works.

A gravity assist of a moon can certainly let the planet capture given body.

Not in KSP where n-body gravity is not simulated. If it does in KSP it is a bug / floating point error.

I believe Sharpy was referring to the use of a gravity assist from a moon to capture around the parent planet, which is definitely possible without n-body. I use a Tylo assist to capture around Jool almost every time I visit there now that aerobraking is more hazardous.

[User Unrelated]

I had this happen with Gilly before. I thought the same as you "Well, let's just see what this does"

I can't find the screenshot of the actual encounter, but it just wanted to swing way around and exit gilly's SOI about 10* before it would complete a whole orbit.

[BudgetHedgehog]

I understand why you would say that, because the prediction system has failed me numerous times already. But what else should we use? Use 4D vector math with pen on paper, to figure out our space-time trajectories?

He said don't trust it, not don't use it. Use it all you want, but know that it can be, and sometimes is, wrong.