[Getdp] CFD

[Kubicek Bernhard]

This effects the Solver, so the answer would be the C Code.
CFD solvers are usually finite volume methods, with sophisticated methods to find the matrix elements that couple neighbour cells (e.g. upwind differencing scheme, and follow ups, see Patankar's book for starting). Additionally, a lot of secrets are in the way the different equations sets are coupled. E.g. Fluent is famous for its segregated solver, meaning that the transport and energy equation are individually solved and only after solving recombined again. Pressure is corrected in a completely different way. (Please don't cite me on this, I am not on my terrain here). 
Also coupled solvers exists (all in one big matrix for each time-step). 
However, much of the nice numerical tricks is quite secret, as they are one of the major factors to the price of Fluent.

Thomas Jung/Fraunhofer (I guess) and us are however able to couple Fluent to GetDP, by heavy use of UDFs and external converters.
Honestly, I don't know if there are free CFD programs, where on can implement this kind of functionality.

Commercially, there is MPCCI as Glue-Code, _however_ it is _very_ expensive, and at least half a year ago it was not legally allowed to implement the necessary APIs within GPLed software (e.g. GetDP).
My glue-tools are available (although buggy), but only for money due to some high-level management decision (same price-range as MPCCI non-academic). Personally, I prefer GPL.
However, in case you need calculations for a customer-project, please don't hesitate to contact us. Our research institute has lots of expertise with fluid-dynamics ( http://www.arsenal.ac.at/products/products_mob_cfd_cs_en.html ) as well as high current/voltage systems (e.g. field simulations of power-stations, legal certificates, ...) in industrial contexts.

nice greetings,
 bernhard

-----Ursprüngliche Nachricht-----
Von: getdp-bounces at geuz.org [mailto:getdp-bounces at geuz.org] Im Auftrag von Matt Koch
Gesendet: Mittwoch, 24. Januar 2007 05:52
An: Christophe Geuzaine
Cc: getdp at geuz.org
Betreff: Re: [Getdp] CFD


Hi All,

so, is this something that would require a modification or extension of 
the source code in C, or could this be done directly in a GetDP .pro 
file? It seems to me that if there is a way to access quantities of the 
current and the previous time step, then an iteration loop inside a time 
loop within the Resolution section could be one possible approach?

Thanks,

Matt Koch


Christophe Geuzaine wrote:
> Alan Bromborsky wrote:
>> What, if anything, needs to be done to getdp to make it suitable for
>> solving  computational fluid dynamics problems?
>
> We would need to implement numerical schemes adapted to whichever 
> problem you want to solve (upwinding, etc.)
>
>> _______________________________________________
>> getdp mailing list
>> getdp at geuz.org
>> http://www.geuz.org/mailman/listinfo/getdp
>>
>
>
_______________________________________________
getdp mailing list
getdp at geuz.org
http://www.geuz.org/mailman/listinfo/getdp