Dear Marco,

Why do you not think before you post something like that? OpenFOAM has been developed by Nabla Ltd.! Sorry, but you are a bad advocate for a good cause -- do your research and maybe learn a bit more about CFD...

For industrial users, switching to an open-source code like OpenFOAM, might not be as straightforward as you like to think. It is, however, not as impossible as Opaque would like us to believe. Another misconception that many people seem to have is that open-source is charity -- companies like OpenCFD need to make profit, because there are people living on it. This means they need to buy groceries and pay off mortgages or pay their rent. My personal opinion is that open-source offers excellent business opportunities for small companies which do not have the resources to develop a well tested and well documented code. Still, those companies do create profit by selling consultancy or customised developments. Yes, it is possible that OpenFOAM, or another code, will become serious competition for the big commercial players in the CFD field. Other people have different opinions and they are allowed to. Some people are aware of this and they try to defend their ground -- they are also allowed to do that. Personally I don't think that we are going to see an abrupt change between closed-source or open-source CFD software, but open-source is an interesting and working approach. Anybody denying this does go on the same thin ice as somebody who predicts Fluent's death within the next couple of years.

-- O.

Hi,

I got some questions...

What do you think will happen, when all those academic users grow up with openfoam and start some kind of business or go into industry? What do you think will happen, when some supplier start offering cfd calculations with better quality, faster results and/or cheaper prices due the use of opensource and e.g. the chance to invest in hardware? What do you think will an OEM do, when there is need for automatic optimizations with some hundreds of 'quick and dirty' calculations expecting the results on the next day?

Regards!

Dear Opaque I mentioned the SINDA story (www.sinda.com)(C&R are not the only ones in this field, and not to mention the clones) just to remind the readers that the absence of a Preprocessor did not stop or delay the humans from space exploration, of course, that argument is not valid today, as I personally, believe that an intelligent User Interface is a vital piece in today's competitive industrial environments.

Dear O

OpenFOAM was developed by open cfd and not nabla. Why would Opaque compare OpenFOAM with Nabla otherwise. I do not hesitate to admit that i have stepped in CFD not long ago and i am relatively inexperienced in this field. However, i have to say that your arguments for OpenFOAM were not over kill either. How long have you been working for on CFD? It sounds like you dont know much either!

marco

*sigh*, you are still posting very confident responses without knowing the background. OpenFOAM is a GPL release of FOAM. The background to FOAM is that it was originally started by Henry Weller with help from Charlie Hill when they were at Imperial College. Later on people like Mattijs Janssen and Hrvoje Jasak also made significant contributions. The main architecht of OpenFOAM was and still is Henry Weller. The team behind FOAM started the company Nabla Ltd sometime around 2000. For a few years they developed FOAM as a commercial product. Rumour says that Nabla's original funding came from a cooperation with Fluent, but I don't really know if that is true or just a false rumour. In 2004 Nabla was dicountinued and they stopped selling FOAM. Instead FOAM was released under GPL under the name OpenFOAM. At the same time they also started the company OpenCFD in order to support OpenFOAM. Most of the original FOAM developers are now connected with OpenCFD.

GPL licence permits using in commersial package too, so commercial vendors do not miss any things and could costomize and use opensources and include them in his package (of course under some definite condition)

"How long have you been working for on CFD? It sounds like you dont know much either!"

12 years

Dear O

In that case you should had done better for the cause of OpenFOAM. Thanks Pete that was genuinely informative. I infact have learnt quite a lot in the course of this discussion. I also would love to contribute something to OpenFOAM one day!

marco

As far as parallel CFD costs being prohibitive... ANSYS could decide to change it's pricing at any point (to be less prohibitive) and Openfoam would become irrelevant at least from that standpoint.

As far as open source CFD goes... well, if open source for engineering tools was a reality, and was going to work well, then it would have taken over the CAD market a long time ago. It has not, so I'm not so worried about open source CFD. The same logic applies to operating systems by the way. Microsoft is still around, even though it kind of sucks.

OpenFOAM is nothing more than a toy, bottom line. It's a total piece of crap, especially from an architectural standpoint. It is at least 100-200 man years behind the commercial vendors in becoming a general purpose CFD tool.

As far as support goes, fair enough, commercial CFD support may suck. However, that is a factor of 'infinity' better than non-existent support with OpenFOAM.... if something is wrong, fix it yourself. This makes no sense.

In then end, I think there is a place for both commercial and open source CFD. Open source CFD is just not going to 'take over the world'. Linux has certainly not taken over the world either and they had a much better chance.

Open FOAM can have their little piece of the pie, plenty of money for everyone!

Got to agree with you about the programming disaster part. I had downloaded it once and tried to look into it. It is so difficult to understand. People keep on saying that we could add things to it. Off course you could add things to it, but it is not so easy as one might want to believe. OP is okey, but still no where near commercial solvers.

It is very well known, how ANSYS decides to change prices. Doubling the price from one year to the next one is not a reliable business for a customer. This means that the code becomes unreliable.

Fair enough. I know nothing about their pricing policies. It was just a thought.

I can understand that OpenFOAM looks very attractive from the parallelization for LES standpoint, especially for flows around boxes and cylinders etc....

However, unlike RANS, real LES is still not a practical design tool. Computing power is probably still decades away from this being realized.

Ah, but what if OpenFOAM could run as a native application under Windows? Find out how at:

http://www.symscape.com/openfoam-on-windows

Using a cross compiler - that's how. The process of cross compiling may be deemed advanced, but the result is a native application that isn't slowed by Cygwin Linux emulation.

One last thought, there are 'commercial codes' that use Nutcracker or Exceed (expensive cousins of Cygwin) to emulate Unix on Windows anyway, so why single out OpenFOAM as deficient for using Cygwin?

[ruedavallejo]

Why not to take advantage of both sides of the coin? Using the commercial as a supporting tool of the open source and viceversa.

[sbaffini]

This thread is quite old but...anyway...

I'm a medium experienced user of Fluent with no experience with unix/linux systems (just like user) and a Fortran 95/2003 background (i'm a student).

In my opinion the whole point about the accuracy of the solvers is meaningless. Up to now, LES is still not a viable way, mostly because of the available computing power but also because (and this is the other side of the coin) until we'll be in the 2nd order age everything about LES will be far from accurate or, at least, from accuracy control.
Actually, i think that LES will take decades to be part of the industrial practice...Spalart from Boeing being the happy exception (actually DES, but the biggest part of computations is still Euler with BL).

So, if the whole point is about RANS/URANS i hardly believe that there can be differences between second-order/interface-capturing schemes implementation in the different codes (open or not) which are not dominated by the user experience and which cannot be made small enough by a proper mesh refining (when correct boundary conditions are used). The same is true for the different models employed in the solvers, from turbulence to acoustic, to MHD etc. I hardly believe that in the whole world there aren't people able to get this models work in OpenFoam much better than the Fluent people does...come on...it's just a matter of time.

The big deal about numerics is, in my opinion, about robustness, speed, flexibility and all the grid stuff. While robustness actually means "CFD for dummies" (i will not be surprised, in these days of crisis, to find out that now some of the economy guys are doing CFD...) all the other points require a very big effort, which means a lot of money invested. While Ansys can do this, the OpenFoam community would need a combined effort of a lot of people, but i think it will never happens...probably a medium/big sized company, strongly involved with CFD, could sometime decide to invest in such a big project but, at this point, why don't start from scratch?

As stated by someone else, and i somehow agree, OpenFoam is OpenSource but this doesn't means that it's easy to get in to it. If i had to make a choice i would probably pick a commercial pre/post processing package (which i think, now, are almost invaluable) and put all my effort in a highly efficient, highly scalable, easily readable (which could also mean without a GUI...in fact Fluent works the same way without it) flow solver. In this case, starting from scratch would not probably be that hard and i had the choice "to go exactly where i want". Maybe the lack of experience would be paid somehow in flexibility, but easily recovered with the time.

This brings us to the main question...Who uses what? I think that medium/big sized companies and research center actually have their own codes and probably Fluent also (or CFX or whatever). This don't even excludes that there could also be some OpenFoam users in those companies. The point is, do you actually think that OpenFoam/Fluent are the best ever CFD products in the history whose results are unachievable by anyone else? They're simply not. For several reasons.

So, we are talking about medium/small sized companies not completely involved with CFD. In this case a CFD investment could not be a proper choice for several reasons. As seems the case, these companies rely, almost completely, on commercial CFD packages and yes, they're looking for robustness, speed and flexibility (how could that be different?).

I'm not saying that OpenFoam hasn't this capability but it has a steeper learning curve and an almost advanced knowledge of C/C++ and linux is required to just start learning how to use it. Also, it is a matter of fact, windows based systems are the norm and the unavailability of OpenFoam for these systems makes it's diffusion more harder than should normally be. Let's face it, don't you think that a simple windows installation procedure with the capability of easily importing commercial software's grid files would highly help the OpenFoam widespread use? Up to now, in my opinion, the OpenFoam approach seems pretty classy. That is, yes it's open source but only if you want to work it out, fix its bugs and, of course, if you have linux....i mean, it's not that open.

However, for the next future something will change. First, as pointed out by someone else, little OpenFoamers are growing up, and will bring with them the necessary knowledge for the start up in the companies they will create or in which they will work.
And second, which hasn't been stressed enough (not at all actually), available computing power is growing up; if it's not yet the case, in the next future every single PC will be an highly advanced multicore machine. There will be a moment in the next future, let's call it a "break even point", when the single license per node approach will not be a feasible one any more (consider an office with 10 8-core machines = 80 licenses ).

Something will change, i don't know exactly what, probably a single license will be scaled to an 8-node basis or something similar. It will probably depends on the policies of the software vendors and how fast will be to adapt to the new scenario. However, and this is a fact, the whole bandwagon of software vendors is much more a resellers company than a CFD one (paradoxically, the Boeing was such a company in the same years in which much of these companies born); they simply show up 5-10 years later with well established results without any significant contribution if not deliberately scre... the provision because of the robustness (you don't know how cra... is the LES in Fluent). With the increasing use of OpenFoam, is likely to happen that someone will show up with front end CFD results making them available via OpenFoam. This will probably change something. Maybe not. But the OpenFoam community will surely give a contribution to the field which will continue to grow.

[gocarts]

SymLab Running OpenFOAM simpleFoam Solver

Here at Symscape I'm working on a simulation environment that might satisfy some of the points you raised concerning OpenFOAM:

• GUI driven pre/post environment for OpenFOAM - yes
• Native Windows version - yes
• Standard Windows installation and operation - yes
• Import Fluent/CFX - yes
• Export Fluent - yes
• Release date - coming soon

[aerogt3]

Opaque,

You are making some *very* unfair comparisons here.

Fluent is not a pre-processor. You are comparing meshing in open source stuff to ICEM and using that in your argument against open foam vs fluent. I don't know if you know this, but ICEM is not inside of fluent

Fluent is a solver. CFX is a solver. Openfoam is a solver. Let's compare these solvers. On your original points:

a - Geometry enviroment: I do not think so.

Fluent doesn't really have this either.

b - CAD import: what CAD formats does it support? None.

Fluent can't open CAD files.

c - Meshing in OpenFoam: it does not qualify to be compared to any decent commercial product in the CFD, FEA or even plain meshing market. You read the messages in the OF forum and cfd-online, and even OF supporters are asking for decent meshing of complex geometries.

Neither fluent nor OF can mesh.

d - Physics pre-processing: can OF be used in a modern design environment along those like CFX, FLUENT, CCM, CFD-Design, etc? I do not think so.

Neither can fluent.

e - Post-processing: it does not even has its own, but relies on Paraview. Can it do quantitative post-processing for common modeling applications, simulation reports?

Paraview does what fluent does, and more, and better.

- Radiation: surface to surface modeling, discrete ordinates, or Monte Carlo for starters. What about non-gray media modeling? None to be seen that compares to either FLUENT, CFX, or STAR-CD

Fair enough. But those aren't really fluid dynamics problems . . . most fluent users don't solve that stuff.

- Combustion: Yes, it has some simple mixture fraction based models, but where are flamelets, ISAT, EDC, of ECFM?

More features that most fluent users like me pay for, but don't use.

- Turbulence: that is probably the one with several implementations; however, not even close to the pletora of models you will find on commercial codes. LES is not state of the art as Marco called. It is a fashionable model just because the computers are faster these days. The concept is more than 40 years old, and it is just receiving more attention lately.

Very few people use S-A or RSM. Most use KE and KW, KWSST. OF has these.

- Stress modeling: well, I leave that for the FEA guys to evaluate.

What kind of FEA features does fluent have?

Should I continue? Was I way off the mark? No, I was exactly on the mark. CFX, FLUENT, STAR-CD are not just a solver

Yes, they are.


I am a 5 year fluent user. It's good stuff. But let's face it - its only advantages over OP re:
-a "GUI" - cortex, which sucks and is a 1980's relic.
-tools for the really weird areas of CFD - radiation, etc. I design aircraft and racecar external aero. if I want radiation i'll get thermal desktop or a proper thermal toolset.
-post processing built in, though its inferior to paraview IMO
-support and documention
-ease of use/learn

I am going to openfoam training this year and I will start using it to consult instead of fluent. Yes, I still need ICEM and perhaps tgrid, but fluent is a solver. Openfoam can easily replace it for 90% of users.

http://www.totalsimulation.co.uk/services.htm They use openfoam, IIRC. So do quite a few F1 teams I believe. Openfoam is a viable alternative. Don't let the text base scare you off. Pretty gui's are useless and fluent doesn't even have one anyways.

[mr_fluent]

Originally Posted by aerogt3

I am a 5 year fluent user. It's good stuff. But let's face it - its only advantages over OP re:
-a "GUI" - cortex, which sucks and is a 1980's relic.
-tools for the really weird areas of CFD - radiation, etc. I design aircraft and racecar external aero. if I want radiation i'll get thermal desktop or a proper thermal toolset.
-post processing built in, though its inferior to paraview IMO
-support and documention
-ease of use/learn

if you think openfoam (as it stands today) can replace 90% of fluent (even only as solver) you know nothing about Fluent.


What you call relic as GUI is best thing about Fluent. It only gives you basic operations there. If you follow its GUI in logical order (mesh->defination->solution->post processing) you could set up and do simulation without even touching its helps.

But of course to understand what Fluent provides you got to understand CFD.
Fluent's GUI hides advanced stuff from user, its very helpful because one can not get confused by it. For advanced user it provide a lot of things you could even not know just by simply looking at it.

You are trying to boast openfoam as solver. So i will only point out few solver options

1. Fluent provides you, pressure based coupled and segreggated solvers.
It also provide you density based compressible solver.
2. It provides you with Full multigrid options, it also provides you with AMG solver. (try full multigrid with openfoam and let me know).
3. Fluent provides you two sets of fractional step solvers.
(plus openFOAM is PISO based , fluent not only gives you option of PISO, but you could chose SIMPLE and SIMPLEC also).

4. There are number of descritisation schemes available with fluent. How many options openfoam provides. (actually gamma is i think what openfoam provides and is very useful).

there are many differences. But main thing is one has to pay a lot for Fluent. (but i think thats not what you are arguing).

[aerogt3]

Originally Posted by mr_fluent

if you think openfoam (as it stands today) can replace 90% of fluent (even only as solver) you know nothing about Fluent.

That's a pretty bold statement. But your post is full of assumptions, most of them false . . .

What you call relic as GUI is best thing about Fluent. It only gives you basic operations there. If you follow its GUI in logical order (mesh->defination->solution->post processing) you could set up and do simulation without even touching its helps.

You cannot mesh in fluent. The GUI is terrible. It crashes, it fails all the time, the mouse interface is rubbish, rotating and displaying things is a huge pain. ICEM has a proper GUI. Fluent's is terrible. The menus going from left to right does not mitigate this.

Fluent's GUI hides advanced stuff from user, its very helpful because one can not get confused by it.

Another terrible feature. The TUI features are impossible to ever stumble upon. It's like building a car with a navigation system that is underneath the car buried behind the exhaust system where you can't see or find it.


For advanced user it provide a lot of things you could even not know just by simply looking at it.

You are trying to boast openfoam as solver. So i will only point out few solver options

3. Fluent provides you two sets of fractional step solvers.
(plus openFOAM is PISO based , fluent not only gives you option of PISO, but you could chose SIMPLE and SIMPLEC also).

You can also use coupled, but is foam not a good solver because it doesn't have that option? The core of a solver is bigger than p-v coupling. You can get into the semantics of all these small features.

My bottom line point is that I can obtain equally accurate results with foam given the same input mesh. Maybe it will take 10% more iterations, or I'll have to use a text interface. But as a solver it can do the same job.

[ruedavallejo]

OpenFOAM® CoAuthor to be Visiting Researcher at Pointwise This Autumn:

http://www.pointwise.com/pr/pr-08g-jasak.shtml