Kerbol System Planets Simulated in Universe Sandbox - any help with Kerbin system?

[Drew Kerman]

Javascript is disabled. View full album

I searched but didn't find any simulations for the Kerbol system built in Universe Sandbox. No doubt because anyone who tried to recreate the entire system found that it's completely unstable under n-body physics. Well, I only wanted a high-level simulation of the planets so by setting all their masses/densities to 0 they had no gravitational influence upon each other. As you can see in the images above, using RK4 and letting the sim run forwards at automated steps for accuracy, the results are very approximate even 10 Earth years from the start of the game. My attempts to recreate the Kerbin system, however, have not yet been as successful, no doubt due to the smaller size. I've included both the Kerbol and Kerbin systems for download here. Use as you wish. I'm personally going to start filling up my Kerbol system with asteroids I'm tracking to get some nice-looking visualizations.

Additionally, all planet diameters are accurate and if you are using Texture Replacer, copy the planet textures into your My Documents/Universe Sandbox/Media folder and they will be used to texture the planets.

Edited October 4, 2014 by Gaiiden

[Dkmdlb]

Well, I only wanted a high-level simulation of the planets so by setting all their masses/densities to 0 they had no gravitational influence upon each other.

Well, uh, obviously it was stable then, if there is only one source of gravity. It's not really N-body is it, if only one thing has any mass?

[Drew Kerman]

Well, uh, obviously it was stable then, if there is only one source of gravity. It's not really N-body is it, if only one thing has any mass?

yea.... that was the whole point. The system is unstable under n-body simulation so I had to remove that factor to create a proper simulation of the planets

[metaphor]

So.. what happens if you do include mass for every body in the system? Does Vall crash into Jool? Does Minmus get ejected into interstellar space?

[Drew Kerman]

search it up on YouTube, someone did it already

[Drew Kerman]

unfortunately I continue to be unable to simulate the Kerbin system - even approximately. I suspect it's mainly due to the fact that Mun and Minmus have orbits of 0 eccentricity. Universe Sandbox doesn't like that very much, best I can get is something like 0.000001 for the eccentricity value and at this scale that seems enough to throw things out of sync within just a few days. The Kerbol system is large enough that Kerbin's 0 eccentricity value takes longer to become a problem.

However I have been able to achieve my goal with the Kerbol system, importing asteroids into their proper orbits:

It's too bad that Universe Sandbox doesn't have better tools for visualization. For example I can't set up camera paths or hide certain bodies.

Wait WTF Moho??? Aww crap...

Edited September 30, 2014 by Gaiiden

[andrewas]

So.. what happens if you do include mass for every body in the system? Does Vall crash into Jool? Does Minmus get ejected into interstellar space?

IIRC, the system as a whole and most planetary systems are stable, at least in the short term. But the Jool system falls apart quickly, with the outermost moon ejected very quickly and the next one ejected after a year or so.

[Drew Kerman]

Wait WTF Moho??? Aww crap...

Nevermind, turns out I was just working too quickly to test out things after learning the SFS file stores Mean Anomaly at Epoch in radians rather than degrees for some reason. I forgot to enable RK4 simulation and that threw off Moho.

Tested out 28 asteroids. So far everything is showing up where it should be. The process isn't very fast - these 28 asteroids took me the better part of 2.5hrs to get into the simulation. But now that I have the method down it doesn't take as long. A short video of what it looks like:

Pink are short-period asteroids whose orbit remains within Dres, yellow are sun grazers that come well within Moho at periapsis, and orange are long-period asteroids whose orbits extend beyond Dres. I'll have plenty of Near Kerbin Objects to add in blue as well.

Good news too on the camera tools - the devs have told me Universe Sandbox^2 will have some eventually

Edited October 1, 2014 by Gaiiden

[Dominatus]

This makes me wonder what would have to be tweaked with planetary orbits to stop the system from falling apart... Eccentricity, orbital period, distance orbiting, etc.

[Drew Kerman]

still can't get the Kerbin system to remain approximate. Even after a few days at RK4 on high accuracy time dilation it's out of whack. I guess the scale is just too small to properly simulate a single-body system? Anyone have any ideas? Kerbin mass was used from the Wiki like I did for Kerbol, and Mun and Minmus were placed exactly like I did for the planets in the Kerbol simulation, which stays approximate for at least 10 years (maybe more, didn't test further)

[metaphor]

The Mun is much bigger and closer compared to Kerbin than the real Moon is to the Earth. Minmus is also pretty far out in Kerbin's SOI. In an n-body simulation I would expect Minmus's orbit to be unstable, so maybe that's what's happening there.

[Drew Kerman]

there's also the 0 values for ECC and INC I forgot to mention. Ubox doesn't like those values

[Armchair Rocket Scientist]

This makes me wonder what would have to be tweaked with planetary orbits to stop the system from falling apart... Eccentricity, orbital period, distance orbiting, etc.

As others have pointed out, Minmus is really far out in Kerbin's SOI. Perturbations from Kerbol and Mun would eventually cause its orbit to become so eccentric it passed outside Kerbin's Hill sphere and escaped into interplanetary space. Moving both it and the Mun inward might help, but there might need to be a stabilizing resonance or something to prevent the Mun from throwing it out of the system. Gilly I'm not sure about. Due to its enormous mass and close orbit Ike and Duna will be pretty much impossible to mess with. However, tidal forces will quickly reduce Ike's eccentricity to a near-perfect circle.

Jool's moons are unstable because they're WAY too big relative to Jool. Jool is barely over a hundred times more massive than Tylo - a ratio similar to Earth's moon. In reality all the gas giants in our solar system are around 5,000-10,000X more massive than all their moons put together. Oversized moons produce far larger perturbations on the orbits of the other moons, making the system unstable.

If you want a "realistic" simulation, a good bet is to start by plugging in the values from the 10X stock RSS config, then reducing the masses of Jool's moons by 100.

To add one thing: the reason you can't accurately simulate the movements of all the planets in US2 is that as an N-body simulator it will by nature eventually develop rounding errors. To my knowledge it cannot recognize that it's simulating a two-body system (or an N-body system with every mass but one set to zero), and will not put objects "on rails" in Keplerian orbits. In addition, the "auto timestep" feature will usually set the timestep high enough that there are still significant rounding errors, it just sets it low enough that the system doesn't fly apart.

[Camacha]

there's also the 0 values for ECC and INC I forgot to mention. Ubox doesn't like those values

I believe the standard all moons simulation has zero eccentric orbits. That's US², not the first version.