Smoothed Particle Hydrodynamics - for astronomical model demonstration

Update: Some time after working on this project, I was approached by Universe Sandbox and hired to work on the upcoming version Universe Sandbox². Here I have worked on various tasks primarily dealing with computational physics and mathematics. A new YouTube channel with fresh content has been created to showcase the current state of development, as well as other projects, I may be working on.

The theories explaining many astronomical observations come from an iterative process of forming a general theory, designing a model, testing and refining. This project aims to retrace the steps leading to some of the astronomical theories which are now considered fact. 



The focus will not be on state of the art and very complex astronomical models and the maximal degree of realism. Instead it will be on creating the simplest possible models which still realistically demonstrate the mechanisms at work at a level of accuracy that would make them relevant for introductionary astronomy courses. 
The SPH method will be used to create models for some astronomical problems. The models will be described, implemented and tested to provide a comparison between the observations and the models. 
This project aims to demonstrate those models and do it in such a way that it can be considered a dynamic lab experiment where the effect of changing the parameters can be observed live. Motivation 
In the astronomical field various very complex SPH models are being put to everyday use forming new theories, but the focus is naturally not on old previously solved problems. This means that there is a lack of illustrative physical simulations of those problems.

The paper can be viewed here as a pdf.

A couple of videos can be found on youtube. Unfortunately they turned out a little dark during compression so they should probably be viewed in full screen to be interesting.


Proto planet collision


A protoplanet is impacted by a slightly smaller protoplanet. The collision blasts material outwards, but not far enough to form as moon. This is a "failed" experiment trying to form an orbiting moon. This problem, along with a successfull experiement is described in the paper.
 

Fluid planet/moon stretched by gravity


A fluid planet with zero initial momentum is accelerated towards a gravity source. As the fluid planet gets closer to the gravity source, it is deformed by the difference in gravitational pull on the near side and the far side.
 

Moon breaking up inside the Roche limit


A fluid moon is given an initial motion vector which puts it in orbit arround a gravity source. The orbit is however within the Roche radius and the difference in gravitational pull (the tidal forces) breaks the moon into smaller pieces which eventually form an orbiting disk around the gravity source. Transferance of momentum due to viscosity in the orbiting material makes some material fall into the gravity source while other material is thrown out in a higher orbit.
 

Conservation of angular momentum for particle cloud collapsing under its own gravity


A cloud of particles with small random motion callapses under its own gravity. As the cloud shrinks in size, it begins to spin faster in order to conserve angular momentum.
 

Barnes Hutt spatial subdivision for 1000 particles


This video shows the dynamic spetial subdivision tree used for the Barnes Hutt N-body approximation of gravity. This is explained in detail in the paper
 


Earlier simulator

The videos below are of an earlier 2D version of the simulator which actually had walls surrounding the scene. This causes the particles to sometime bounce off the walls. The simulations are otherwise interesting enough to watch.


Rotating particle system with gravity forming planets

A rotating disk of particles clump together and form many orbiting "planets". These planets collide and combine into a few larger planets.

 

Rotating system with gravity forming planets

Another rotating disk forms objects and during this we see a matter bridge connecting the central body with the orbiting ones. Eventually the tidal forces "breaks" the moon and the low orbiting parts are caught by the planet and this speeds up all the moon material. This is both due to the tidal bulge on the planet accelerating the moon which is in a prograde orbit and the transferance of momentum from the low part of the "broken" moon to the higher orbiting parts. This throws the last of the moon out into a higher safe orbit outside the Roche radius. Here it stays. 


Formation of solar system, not physically correct

Yet another "solar system creation" with the old version of the code. Here the "planets" are held together both by gravity and by surface tension. Obviously, on an astronomical scale, surface tension has no relevant effect, but still this shows some interesting things such as sling shots, planets breaking up when close to a gravity well and so on.

Ċ
SPH.pdf
(1667k)
Thomas Grønneløv,
Apr 12, 2011, 1:16 AM