Mini-CFD

[CGN]

CFD application, this can be useful for those looking for a CFD Software that is easy to use.

http://www.raczynski.com/pn/fluids.htm


http://www.sinmec.ufsc.br/cfd/doc/uk/index.html

[yipster]

hmm, maybe, only 2 ( tube) demo's again and seen better...
thanks anyway, like to keep up with these!

[CGN]

Do you know of any CFD sofwtare that uses windows enviroment that can be on the 100 - 200 dls range without being crippled or very limited on the node number?, by all means is this not the greatest for sure but for introduction to CFD i think is a good alternative.

[yipster]

i said thanks but knew you were gona bite me. i fully agree that a working CFD program that works for boat drawings would be just great but havent really found what i'm looking for (yet). over 2000 results on "CFD program" there is even a CFD forum, guess i'm not trying hard enough
some more: http://capella.colorado.edu/~laney/software.htm
good one: http://www.acricfd.com/applications/answer/default.htm
and "uflow" attached

[CGN]

I have to do it, I have no ties to this sowtare or any relation, the only purpose of the link is to share that there is a good IMO application that users like me that we don't now how to use a sofwtare fo CFD or does not have endless pockets or coding or understanding fortran is not an option (besides reading books) is just an alternative.

Yipster, thanks for the acricfd link, answer 16K looks good

Cheers!

[Tim B]

For the first peice of code on this thread...

Treat with extreme caution... consider the wing example... The far-field is WAY too close to the wing (and the section geometry's crap to say the least).
I'll post a few NICE CFD piccys when I have the time (perhaps later this evening).

Two things are key though...

1) CFD is not simple... The quality of the result depends largely on the quality of the grid (and the question, of course). It takes a long time to learn, and there is no point trying to give a simple introduction, because, unusually, any simplifying assumptions will be completely wrong.

2) Cheap CFD is to be handled like nitroglycerin (with extreme care). Cheap codes are rarely robust, and often have not been extensively verified against origenal test data.

CFD is a term that is often bounded about as a solution for everything. In 10 years time, it may be a good answer, but for the moment, it is a fledgling science. I'm afraid you have to understand the problem totally, before you start using CFD. It is very easy to make mistakes (yes, I have run a case for a hull at 6kts backwards).

Furthermore, most good 2D Euler cases will take around an hour or two to solve on a good P4. Navier Stokes will require lots of memory (at least 750MB), and Navier Stokes 3D... well, I'm speccing a new machine to 1.25GB of Ram, and running it in parralell with another 1GB and .75GB machine. This cluster may JUST do a sensible hull.

Incidentally, many programmers write code for Linux because of it's superior memory management.

Tim B.

PS. Will post piccys later.

[Tim B]

I've known about capella.colorado.edu for some time, if you can download any software do tell me! The problem is that I tend to look for certain types of code (Structured solvers maily).

ACRI look like they are more interested in pipe-flows and lubrication in bearings, which is not really our area of interest... If I can just reboot my machine I'll post some nice pictures...

Tim B.

[yipster]

i'm not underestimating anything, phew tell me about it
2 gig however is here today and woulndnt it be nice...
"for the acricfd link, answer 16K looks good" you mean their application form right? works tho!

[CGN]

no, no, i'm talking about the cost of the version of 16K nodes and i was being honest, the cost looks right to me, I now is not a Boat design application but it can be used for other stuff, ventilation or exhaust systems for example.

Thanks

[Tim B]

Hummm, 16 thousand nodes eh? check out the screenshots for the 2D grid below. 7860per plane (15720 in total). The Hull below was done with 350,000 cells as I remember. that works out at a 5-6 hour run time for each case.

I'm well aware of the size of RAM available, some 64-bit machines even boast maximums of 16GB of RAM (practically only 8 at present). There is, however, a limiting factor, commonly known as cost. 2GB of RAM is nice, but then consider the flexibility of the system. There is no need for a laptop to have 2GB of RAM, even to do CFD. The sort of jobs that need 2GB of ram or more are far more suited to desktops or clusters that can run at 80-100% cpu usage for a few days at a time.

Incidentally, I wouldn't go anywhere near a CFD code with the grids I've pictured, they're WAY too coarse even for Euler calculations. I wouldn't be surprised to see 25000 or 30000 cells (or similar nodes if you want) and N-S grids I'd expect to see 50000 to 75000 cells. There may be some special cases.

Food for thought,

Tim B.

PS. If this spec. will do useful CFD I'm a frenchman! "PC min 64KRAM, 256K and a fast machine recommended. Windows 98 or later, NT or XP."

PPS. There are the two grids, I'll upload the others from Linux

[Tim B]

right... back to linux, now to post the other piccys...

[yipster]

a golden oldy in computer flow dynamics

[Leo Lazauskas]

Nice pics, Tim.

Can your solver do infinitely thin wings in inviscid, incompressible fluids? I.e. can your software work as a Laplace equation solver?

I'm asking because I have had a long interest in wings of circular planform. Calculating the lift and induced drag of these wings in ideal fluids is a very tough problem, but at least an exact solution exists. I'd be interested to see what some CFD codes can do.

Regards,
Leo.

[Tim B]

It can do sails 2D Euler, should be able to do sails in 3D Euler and Navier Stokes too.

Circular wings need not be 0 thickness, but it should be able to do it. Sadly I havn't got a 3D grid generator at the moment, nor do I really have enough time to write one. Eventually, I'd like to do an automatic hull grid generator. That gives about 4 general different hull forms to write a grid generator for.

There are several projects that must be finished first.

Tim B.

[jam007]

Does anyone have som info on a program called ADFC http://adfc.sourceforge.net/
I just downloaded it and GiD that seems to be nessesary. Will try to work through the example tomorrow but there is just one small problem. The manual is in spanish and I do not even know how to say hello in that language

Anders M