Orbital ideas other than patched conics - merits/issues

[thereaverofdarkness]

I'm aware that Squad does not currently seem interested in implementing any orbital system other than patched conics, due to the complexities involved or for whatever reasons they have for their decision.

Patched Conics is a system in which all orbital functions are based on a sphere of influence, in which only the gravity of the primary body is calculated. Each celestial in orbit around this primary body which has a significant gravitational pull will also have a sphere of influence, though its own orbit is influenced by the gravity of the object it is orbiting.

------------------------------------------------------------------------------------------

Here I would like to discuss other options for orbital mechanics and what advantages or disadvantages they have, what complexities in programming they present, how they will affect gameplay, and in what scenarios they function best or worst.

One idea I had in mind was that all celestials(stars/planets/moons) could be placed on permanent rails but have gravity simulated from all relevant bodies onto your craft(s) and onto the random asteroids. For example, a ship in orbit around Kerbin would be tugged by Kerbin, Mün, Minmus, and the sun, but would not be tugged by other planets. If it got outside of Kerbin's sphere of influence (SOI), it would be tugged by the major planets Moho, Eve, Kerbin, Duna, and Jool, but not by any moons, nor by the minor planets Dres and Eeloo. Now each SOI would have a greater sphere of secondary influence around it, in which you would be affected by its gravity as well as by the other things near you, but it would not alter the other things affecting you. This would cause you to be affected by the gravity of Dres or Eeloo while in interplanetary space if you are close enough to them, or also by Eve or Duna while around Kerbin when they transit close enough.

Interested in hearing all your thoughts, guys!

[Ruedii]

I'm certainly thinking there should be a threshold of gravitational forces below which they shouldn't count.

I don't think the planets pass close enough to each other to create enough force to pass this threshold. In fact, I don't think the pull of Minmus even passes this threshold when you are in orbit of the Mun.

In my opinion, each object should have a range of gravity but those ranges should overlap. I would generally also put a cap on even considering the possibility of 8 objects of influence at a time, since it's very unlikely that you would enter the field of 8 objects, and putting a cap on it would greatly simplify things.

[Nuke]

integrate all the things.

just dont complain when the kracken hurls your homeworld on an escape trajectory into the universe.

[Yoha]

Again, this is not worth the trouble. If you wanted to actually send a spacecraft through the Solar System, you would indeed have to refine the patched conic approximation with a n-body simulation. On the other hand, KSP only needs a simple but seemingly realistic model and patched conics are good enough and relatively simple to use.

The problem of n-body physics is not algorithmic complexity but of analytic solving. Limiting the simulation to three bodies instead of eighteen does not help: even simulating only the attraction of only two bodies on the vessel makes things incredibly more complicated than with only one.

[Kegereneku]

I am no mathematician but I do know for sure one things :

Patched-conic is the reason we can have stable orbit that stay the same for several year.

N-body physics would change this and would require to correct satellite/ship orbit from time to time, by that I mean often enough to be annoying.

However I wish we had Lagrange point, but understand if it's considered "not possible". At worse someone will make a mods (I haven't even searched if there wasn't one already)

[TMS]

Relevant reading: http://forum.kerbalspaceprogram.com/threads/68502-WIP-Principia-N-Body-Gravitation-and-Better-Integrators-for-Kerbal-Space-Program

[RenevB]

Again, this is not worth the trouble. If you wanted to actually send a spacecraft through the Solar System, you would indeed have to refine the patched conic approximation with a n-body simulation. On the other hand, KSP only needs a simple but seemingly realistic model and patched conics are good enough and relatively simple to use.

The problem of n-body physics is not algorithmic complexity but of analytic solving. Limiting the simulation to three bodies instead of eighteen does not help: even simulating only the attraction of only two bodies on the vessel makes things incredibly more complicated than with only one.

Fully agree. Any n-body physics, even limited as in this suggestion, would make it impossible for the average gamer to know if he's in a stable orbit. Patched conics is a simplification, but in my opinion a necessary one for KSP to be considered a game, not a simulation.

[sjwt]

This is one of those things thats on the what not to suggest for a reason, the over all computer power for a cheap hack of N-body physics is higher than KSP can go to start with, and that a hack. It would deal with either say 2D, or mass less particles or weighted particle in a set gravity environment with no change..

It just can not be done. From time to time we get posts linking to some simulators, but at the end of the day if they were a real true solution, it would be Nobel prize winning and rather newsworthy.