CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM (http://www.cfd-online.com/Forums/openfoam/)
-   -   How hard is it to learn openFOAM? (http://www.cfd-online.com/Forums/openfoam/93062-how-hard-learn-openfoam.html)

YourNameHere October 4, 2011 07:05

How hard is it to learn openFOAM?
 
Honestly, how much effort is required to get up and running with openFOAM? I see a lot of posts on forums suggesting that "you should use openFOAM". So far I have been able to download and install openFOAM on my linux OS. That in itself was a major ordeal. Now, I can rip out page after page of hand-calcs and MS Excel is my b*#@h, but I have no idea when it comes to C++ or any other programming language. As much as I'd like to dedicate the time I don't really want to invest months to learn how to manipulate numerous lines of code. All I want is to mesh my CAD model, enter some parameters and go get a cup of coffee while it solves. FYI the simulation I'm interested in is just a flame from the end of a vertical pipe (flare) and some contours of the resulting radiation. Is it reasonable to expect that I can achieve this with a small investment of time, without the expectation that the results will be in any way accurate? Base it on your first experience with openFOAM and add a factor of safety of 40% because I'm computer illiterate.
Cheers

robertmaier9 October 4, 2011 07:35

It really helps to understand C++, and it is not advisable to go about using OpenFOAM without some understanding of C++ or any programming knowedge for that matter. The first the would be to try do the examples in user guide. This will kind of give you an idea of what OpenFOAM is like. If you're just going to run one case, it is easier to use a commercial software (if available). OpenFOAM simply allows you to control every aspect of the CFD process, unlike the commercial options that act much like a black box.

Depending on how computer savvy you are, you can expect it to take two weeks or several months to truly learn how to use OpenFOAM. I would say it's worth it as it'll give you a nice new set of knowledge that you can use for other things as well! If you're looking for the easy way out, maybe you should look elsewhere. But what do I know, I am still learning OpenFOAM myself (months into it).

elvis October 4, 2011 07:55

Hi,
there is some material online
http://www.tfd.chalmers.se/~hani/kurser/OS_CFD/
http://www.tfd.chalmers.se/~hani/kurser/OS_CFD_2011/
http://www.tfd.chalmers.se/~hani/kurser/OS_CFD_2010/
http://www.tfd.chalmers.se/~hani/kurser/OS_CFD_2009/ etc.

http://www.openfoamworkshop.org/6th_...m/training.htm

=> also take a look at the workshop of the previous years and the http://web.student.chalmers.se/groups/ofw5/Program.htm http://web.student.chalmers.se/group...ingStarted.pdf
http://web.student.chalmers.se/group...sicClasses.pdf
etc.
http://www.cfd-online.com/Forums/ope...thingelse.html

some stuff is at http://openfoamwiki.net/index.php/Main_Page

http://www.caelinux.org/wiki/index.php/Doc:CAETutorials a liitle outdated stuff the tutorials for OF there

elvis October 4, 2011 07:56

If you want something GUI like.

You might give Discretizes a try http://www.discretizer.org/ a free gui tool!

of course there are commercial tools like

http://www.dhcae-tools.com/CastNet.htm
http://www.iconcfd.com/en/services/foampro

but the price tag for these is quite high!

Bernhard October 4, 2011 08:01

If you don't want to spend time on the problem, then OpenFOAM is not the way to go. Otherwise you should hire somebody to do this for you, or you could buy a license for a commercial package, which is usually easier to work at in the beginning. Only entering some parameters is not a really successfull way in CFD in general, also not in commercial tool.

marupio October 4, 2011 10:20

I just checked my notes and was surprised to see it took me 4 months (full time) to learn it... but I learned it top to bottom. Also, I had minimal C++ background.

Another thing to note is that the direction OpenFOAM is taking is towards more user-friendliness. Hrv Jasak demonstrated a working OpenFOAM variation that compiled natively in Windows format. Also, there are several companies that have developed GUI front ends to OpenFOAM... and I suspect there's an effort underway to do so as free software.

akidess October 4, 2011 10:44

I think the first question to answer is if OpenFoam already has the functionality you require built in or not (check http://www.openfoam.com/features/standard-solvers.php as a starting point). If so, then OpenFoam might be what you are looking for. If not you will have to spend considerably more time getting to know some internals of OpenFoam.

djbungee October 5, 2011 05:01

Effort to learn OpenFOAM
 
Hi,

everything that has been said can be confirmed from my side.

What is also important in my eyes is that you get support in learning OpenFOAM. From the starting point that you described I would strongly suggest that you participate in OpenFOAM courses for beginners offered by many institutions.

To solve your specific problem or case you might discuss it in detail on forums or user meetings of local user groups. There you will always get a lot of help from more experienced users.

We have several students here and they always need a lot of briefing in the beginning from our experienced engineers in the CFD team. To start OpenFOAM on your own does not make sense in my opinion. Use the community-based knowledge - your question here was a good start :)

However, if you do neither want to get into CFD, numerical simulation nor programming and just want to simulate a single case - forget it. Use a commercial code (test license), the help of an engineering-service provider (what is your budget?) or ask someone else to do the simulation for you.

Best regards
Ulf.

YourNameHere October 5, 2011 07:15

Thanks everyone for the responses. I was quite interested to learn that OpenFOAM was offered with commercial GUIs. What are people's experiences with dhcae and Icon? When you say they are expensive are we talking ANSYS CFD expensive, FloEFD expensive or lower. Are SGI planning a commercial release. Sorry, I know these are probably annoying questions but I'm really interested.

I am keen to learn CFD but currently I only have access to FEA, and my work outsources our flare analysis to consultants (who use spreadsheeting, not CFD!!). I think we should start to handle some of the analysis in-house, but there is no budget and there won't be until I can show them the value of CFD. Catch22 Right?

Cheers

Bernhard October 5, 2011 08:00

That's not a catch 22. You can just show them the brochure of any CFD license or a random presentation by Hrv Jasak to show examples. If you want to learn CFD, you should attend some courses on CFD to learn all the basics. Just running some codes will disappoint you probably. You should be aware of what you are doing to get reliable results, that is something else then producing nice figures.

elvis October 6, 2011 11:00

Hi,

the Open Source CFD International Conference
http://www.opensourcecfd.com/confere...id=3&Itemid=30 and later conferences
http://www.opensourcecfd.com/confere...itt-Gehrke.pdf
has some presentations so you might get a feeling how capable OF is. Or what it has been used for by some companies.

In one of the oscic-presentations one of the volkswagen employees mentions that making OF "work" @ volkwagen was quite a big investment for vw, but they gained huge productivity for their CFD-Department (=>lots of simulations with high grade of automation).

vkrastev October 7, 2011 09:51

Quote:

Originally Posted by YourNameHere (Post 326607)
Now, I can rip out page after page of hand-calcs and MS Excel is my b*#@h, but I have no idea when it comes to C++ or any other programming language. As much as I'd like to dedicate the time I don't really want to invest months to learn how to manipulate numerous lines of code.

Actually this is not the real problem when dealing with OF...I mean, a basic knowledge of C++ programming is obviously desirable, but a good background in Computational and Theoretical Fluid Dynamics is far more important: rather than developing excellent programming skills, most of the effort in becoming a good OF user (not a developer, which is quite different) should be put in learning the numerics under the Finite Volume method and the behavior of the physical models standing beside the solvers.


Quote:

Originally Posted by YourNameHere (Post 326607)
All I want is to mesh my CAD model, enter some parameters and go get a cup of coffee while it solves.

Well, to be honest, and taking also into account what I've said before, if you are not interested in learning CFD basics a bit deeper than this, I don't really think that OF is the right tool for you...Maybe commercial CFD packages should be a better choice.

V.

alberto October 8, 2011 02:12

Quote:

Originally Posted by YourNameHere (Post 326607)
Honestly, how much effort is required to get up and running with openFOAM? I see a lot of posts on forums suggesting that "you should use openFOAM". So far I have been able to download and install openFOAM on my linux OS. That in itself was a major ordeal. Now, I can rip out page after page of hand-calcs and MS Excel is my b*#@h, but I have no idea when it comes to C++ or any other programming language. As much as I'd like to dedicate the time I don't really want to invest months to learn how to manipulate numerous lines of code. All I want is to mesh my CAD model, enter some parameters and go get a cup of coffee while it solves. FYI the simulation I'm interested in is just a flame from the end of a vertical pipe (flare) and some contours of the resulting radiation. Is it reasonable to expect that I can achieve this with a small investment of time, without the expectation that the results will be in any way accurate? Base it on your first experience with openFOAM and add a factor of safety of 40% because I'm computer illiterate.
Cheers

You do not talk about your background on fluid mechanics. However, based on what you write, I think OpenFOAM is not the right tool for you.

I am not so convinced a commecial code will help you much more. It will be more friendly to use, however you will still need to have an understanding of the physics, of the models, of the numerics, and also have expertise in meshing. Do not trust who tells you CFD is a click & go process, because it's only marketing.

Best,

vkrastev October 8, 2011 10:51

Quote:

Originally Posted by alberto (Post 327126)
Do not trust who tells you CFD is a click & go process, because it's only marketing.

I agree

V.

YourNameHere October 8, 2011 22:56

Quote:

You do not talk about your background on fluid mechanics.
Actually my fluid mechanics is quite good, including advanced study from an engineering and applied mathematics standpoint. Experience-wise I design fluid systems daily using hand calcs and spread-sheeting, and progressively using matlab.

Quote:

Well, to be honest, and taking also into account what I've said before, if you are not interested in learning CFD basics a bit deeper than this, I don't really think that OF is the right tool for you...Maybe commercial CFD packages should be a better choice.
While I see my original post may have been poorly stated and maybe a little antagonistic (I do wonder how many of you set up your model and go and get a coffee while it solves?), responses have been helpful.

Quote:

Actually this is not the real problem when dealing with OF...I mean, a basic knowledge of C++ programming is obviously desirable, but a good background in Computational and Theoretical Fluid Dynamics is far more important: rather than developing excellent programming skills, most of the effort in becoming a good OF user (not a developer, which is quite different) should be put in learning the numerics under the Finite Volume method and the behavior of the physical models standing beside the solvers.
I guess what I wanted to know is: can I use OpenFOAM (with an adequate knowledge of the physics) as a tool, without the depth of programming knowledge required of a developer. The answer seems to be yes.

Quote:

That's not a catch 22. You can just show them the brochure of any CFD license or a random presentation by Hrv Jasak to show examples. If you want to learn CFD, you should attend some courses on CFD to learn all the basics. Just running some codes will disappoint you probably. You should be aware of what you are doing to get reliable results, that is something else then producing nice figures.
Of course its a catch 22. They won't pay for commercial software until its functionality is established, and it can't become established until they pay for it. Its interesting that you feel that just showing them a brochure will suffice to sway their opinion, but state that just running codes will disappoint. The problem is that more proof than colourful brochures is required before they part with 65K+. Of course I care about reliable results, but first one needs to get "a" result. I realize this isn't the place for this discussion, I just don't follow your logic.

Anyway, while working through the initial OpenFOAM tutorials over the last couple of days I'm starting to gain confidence that I might be able to use it. Don't think for a moment I don't appreciate that OF is free.

Regards

alberto October 8, 2011 23:32

Quote:

Originally Posted by YourNameHere (Post 327176)
Actually my fluid mechanics is quite good, including advanced study from an engineering and applied mathematics standpoint. Experience-wise I design fluid systems daily using hand calcs and spread-sheeting, and progressively using matlab.

This makes things much easier for you. If you know fluid mechanics, learning to use (not to develop) OpenFOAM consists in understanding what is the solver that answer your needs, and learn how to set a case up.

Quote:

While I see my original post may have been poorly stated and maybe a little antagonistic (I do wonder how many of you set up your model and go and get a coffee while it solves?), responses have been helpful.
I surely drink coffee... but my cases take longer than drinking a cup of espresso ;-)

Quote:

Of course its a catch 22. They won't pay for commercial software until its functionality is established, and it can't become established until they pay for it. Its interesting that you feel that just showing them a brochure will suffice to sway their opinion, but state that just running codes will disappoint. The problem is that more proof than colourful brochures is required before they part with 65K+. Of course I care about reliable results, but first one needs to get "a" result. I realize this isn't the place for this discussion, I just don't follow your logic.
Just in case, training courses to learn how to use OpenFOAM are nowhere as expensive as the license of a commercial code, and they have a discount if you take both basic and advanced course together. You can see OpenCFD pricing here if you are interested: http://www.openfoam.com/training/schedule.php (Note: I don't work for them, but I tend to recommend them since I know people who was happy of their training, and they re-invest what they earn in extending OF).

Quote:

Anyway, while working through the initial OpenFOAM tutorials over the last couple of days I'm starting to gain confidence that I might be able to use it. Don't think for a moment I don't appreciate that OF is free.
If the company you work for is interested in using CFD more in the future, investing in OpenFOAM might be a good idea, especially if you have large cases, for clear advantages with licenses. Also, OpenFOAM is free, but it has commercial support from OpenCFD/SGI in case you need it.

Best,

vkrastev October 9, 2011 16:05

Ok, things are much clearer now...I can say that Alberto's last post summarizes quite exactly my point of view: for introducing CFD in your company, the choice of OpenFOAM is probably a not so bad idea (and for sure it's better than buying some licenses as "black boxes"...). Of course, and as Alberto said, the main "sponsor" of the code is the free license, which becomes a serious "impact factor" once you start to use largely parallel computing features. And, for what about giving "A" result, don't worry: once you gain some basical experience whit the solvers, settings and mesh quality requirements, you will be able to get a "quick first attempt" result in most of the cases (but of course a carefully prepared OF run CAN give also accurate results ;) ).
So, good luck for your attempt!

V.


All times are GMT -4. The time now is 14:37.