User:Al Da Best/gsoc2012project
Non-vacuum gravity simulator
The current simulation available in BRL-CAD has not been developed in depth, and is quite limited in its implementation. It can currently only simulate very basic situations in vacuums. For this project I aim to vastly improve this area by adding additional features such as variable fluid for the object to fall through, a variable ground surface which would allow simulating reactions from it through use of the coefficient of restitution and other such formulae. Another feature would be to allow for variable gravity, since gravity is weaker at greater distances from the Earth, and indeed other large masses. After some thought and looking at the available physics engines I have opted to go with the Bullet engine. The reasons for this include its superior capability to constantly check for collisions, combined with its relatively fast simulations compared to, say, ODE and it is already integrated to some extent into BRL-CAD. All my work would follow on from what has been done by Abhijit, who created the current simulation systems during his ESA-Summer of Code project. During improving this project I would make sure to keep in mind compatibility between both MGED and Archer, to make sure it is ready for the future of BRL-CAD.
So there are 3 clear features I want to implement into the simulation system, fluid dynamics, ‘rebound’ mechanics and improved gravitational system.
This is an area with hugely far reaching possibilities, with CFD potential in the long run, and open source libraries such as OpenFVM being available, however to begin with I intend to keep this relatively simple. To start with, this would consist of having a user-defined fluid, or more specifically if desired, the density of the fluid. From this the software would be able to calculate an estimate of the drag coefficient and thus drag which will result in a more correct representation in the simulation if a user wished to see the position of an object at certain time intervals. I would also want to add the functionality for a variable fluid as this would allow the density to increase say if an aircraft was simulated to drop from 30,000 feet. This would involve the creation of several functions that would include calculating the surface area of the object facing perpendicular to the firm plane, a function to either calculate or retrieve (if it is a standard shape) the value of its coefficient of drag, which would then be passed to a larger set of functions and it would more or less directly counter gravity. This does open up another potential path of functions, for moments. Now, whilst calculating the drag on the surface of the object is a good start, there is much more potential that would allow for the calculation of the rotation of the object as it falls through space, by resolving the forces caused by drag. This would of course mean that the centre of gravity of the object will need to be found, and that is relatively easy to do, but will depend on how complex the shape is and may require some estimation. An issue with objects rotating is that they will create variable drag, and whilst the software will be able to calculate the drag at every point in the simulation, there will be a loss in accuracy since the fluid will be assumed to be moving at a user pre-defined velocity, and this won’t be affected by the motion of the object as it realistically should be.
There is already some implementation of this within the simulation, however it is in a very early stage and I feel has much more potential than it currently uses. As I briefly mentioned I would like to allow for variable materials. Expanding further upon this, and taking into account the makeup of the Earths’ surface, I would very much like to implement some water physics. By that I mean for the software to be able to simulate the effect of dropping an object from high atmosphere into water, and allowing for forces such as upthrust as this would allow for simulating say, satellites dropping from orbit. The collision physics would need a lot of testing to ensure it could hope with very high velocity impacts (e.g. a comet, just in case).
Improved Gravitational System
The simulation currently can certainly handle gravity, however only at a basic level of things falling to the ground. As far as I can tell, it is very lightly implemented, just passing a value of -10 for gravity to the bullet class, this would be replaced by a separate function that would take into account the gradual increase of the acceleration due to gravity since at 500km away from Earth gravity is a fair amount smaller, and simulating that distance would make a noticeable difference between a standard value of 9.81 and the true values. I would want to improve this to take into account other user specified masses. For example, if something were to drop from orbit, there would be a sizable gravitational force upon it from the Earths’ moon. Furthermore, if we are simulating several large objects, to increase accuracy it would be ideal for the simulation to take into account the gravitational pull each object will have on the others, however minor it is. This function would need to be called repeatedly, to calculate the gravity for each step of the simulation to allow for it to constantly change. There will be some overhead will this of course, however the calculate is very simple. It would vary however, depending on the Gravitational Constant, which is different for different masses. This would be user input with presets (Planets in the Solar System, and some others) available.
A system that can calculate the drag resulting from a falling object, as well as calculate rotation due to that drag.
A system that can consistently cope with any collisions with a surface that takes into account the force exerted upon the object by the surface and also allows for said surface to be water, in which the fluid system will take over again.
A system that can calculate gravity at any distance from the surface, which will also take note of the gravitational effect from other large masses.
A method, either command line or GUI, that will allow the user to specify as many options as is possible to allow for a simulation that is suited to the users’ needs.
BRL-CAD interests me because I have some background in CAD and modelling work and I expect to be doing a lot more of it as the years progress on my course and potentially in a working environment in the future. The simulation project caught my eye specifically, as whilst I am interested in several of the ideas, simulations are an area I like to use and would be very interested in improving.
Whilst my knowledge of C/C++ are not as strong as I would like, I am confident that I can learn a lot from this project and build my skills up before the coding begins properly to a more than adequate level. I believe that I have all the mathematical ability that I will need to complete this project, including knowing how and where to use which formula's. Naturally, this would help reinforce what I have been taught which will be useful during exams and the coming years.
I would undoubtedly continue to work on BRL-CAD after this project is complete, most likely on the simulation software as I still have several ideas for improving that, ranging from full computational fluid dynamics modelling to a fully featured fracture system.