separating physics and numerics?
Folks,
I am thinking whether one can come up with a general CFD code that would use a variety of options for discretization for a variety of options for the physics equations. Something like a parser that would automatically interpret the physics equations according to the chosen discretization scheme, rather than using a hard-wired discretization scheme. Anyone heard of somebody trying anything like that? Thanks for any info or thoughts. -Maxim- |
Re: separating physics and numerics?
Well, have a look at the FOAM web site (http://www.nabla.co.uk): the code is written (in C++) as a library of operators talking the language of partial differential equations. Each operator creates a matrix and the matrices are then summed up to create the final linear system on which a solver is called. There are examples on the web site and all top-level codes are shipped in full source. Typically, they will be a couple of hundred lines long, so there's no need to be scared :)
This approach allows you to pick-and-mix your physics any way you like. The discretisation is then controlled "per-operator" and "per-equation"; with some work you can even go from FVM (that's the work-horse) to the FEM (used for automatic mesh motion and such like). If you need to know more, there are a couple of papers you can read: Weller, H.G.; Tabor G.; Jasak, H. and Fureby, C.: A Tensorial Approach to CFD using Object Orientated Techniques, Computers in Physics, 1998 v 12 n 6, pp 620 - 631 Jasak, H.; Weller, H.G and Nordin, N.: In-Cylinder CFD Simulation Using a C++ Object-Oriented Toolkit, SAE World Congress, Detroit, 2004. The paper describes some parts of the FOAM design and includes example internal combustion engine simulations using FOAM. Enjoy, Hrv |
Re: separating physics and numerics?
|
Re: separating physics and numerics?
or just check out their website at http://cfdlab.engr.utk.edu
|
All times are GMT -4. The time now is 06:29. |