Talk:Computational fluid dynamics

From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Physics / Fluid Dynamics  (Rated C-class, High-importance)
WikiProject icon This article is within the scope of WikiProject Physics, a collaborative effort to improve the coverage of Physics on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
C-Class article C  This article has been rated as C-Class on the project's quality scale.
 High  This article has been rated as High-importance on the project's importance scale.
This article is supported by Fluid Dynamics Taskforce.
WikiProject Computer science (Rated B-class, High-importance)
WikiProject icon This article is within the scope of WikiProject Computer science, a collaborative effort to improve the coverage of Computer science related articles on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
B-Class article B  This article has been rated as B-Class on the project's quality scale.
 High  This article has been rated as High-importance on the project's importance scale.
edit·history·watch·refresh Stock post message.svg To-do list for Computational fluid dynamics:

Top Priorities

Low Priorities

  • Fix broken Wikilinks
  • Add additional images (search Wikimedia Commons or create/upload your own images) to help illustrate basic CFD concepts and make the whole article a little more accessible
  • Add an "Applications" section addressing more prominent uses of CFD (e.g. climate change, vehicle design)

Topics to include:

  • Structured/unstructured (& main articles, & significant issues, e.g. kinetic energy conservation)
  • Parallel CFD
    • Birds eye view of the problems associated with it
    • Where else to look
    • Applications
  • conservative equations, etc
  • truncation error, stability analysis
  • verification
  • validation and uncertainty quantification

Last to-do list update: (replace with your signature when updating the to-do list)

Charlesreid1 (talk) 18:26, 5 November 2010 (UTC)

Public CFD codes[edit]

Why only make reference to commercial codes? Why not include publicly-available codes? A list is already available in .

Because we didn't know about them. Feel free to add anything you know. DJ Clayworth 13:08, 29 Jul 2004 (UTC)
My point was actually that it's not necessary to mention specific commercial codes, unless you plan to be exhaustive. The idea is to not impose a commercial bias to this entry. CFD codes started in the public domain. I would just leave out any mention of commercial codes altogether until there are representative lists for public and commercial codes.
Agreed. If someone wants to make a comprehensive list, they should make a Wikipedia list article. Otherwise, codes should only be discussed if they are significant in the history or use of CFD. --Charlesreid1 (talk) 00:05, 5 November 2010 (UTC)

Why not mention commercial codes as well? It's just letting people know what's out there. (Incidentally, I'd be interested to know what you mean by "CFD codes started in the public domain". What codes do you mean? My understanding is that some CFD started in academia, but some in research labs too. Even academics might not think their work was 'public domain'.) DJ Clayworth 21:46, 30 Jul 2004 (UTC)

My point is that there are public and commercial codes. Why mention specifics about only one
of these types without mentioning the other? This is not supposed to be an advertisement page for
industrial interests. You're right that many academics are using non-publicly
available codes, though. On the otherhand, publicly-available codes tend to come mostly from
academic research.
My point about CFD starting in the public domain is that CFD started as publicly-funded research.
The work by Richardson in the ... what, 20's ... was well publicised and the human algorithms
could easily be used by anyone who wanted to use them. Later, computer work at Los Alamos was
publicly, not privately, funded and published soon after. --Andy Froncioni 15:59, 7 Aug 2004 (UTC)

I have no objection whatsoever to publicly available codes being mentioned in the article. The only reason I didn't do it was that I didn't know any of them. DJ Clayworth 16:52, 9 Aug 2004 (UTC)

It's normal in Wikipedia to use lowercase for article titles except proper nouns. A lot of people write "Computational Fluid Dynamics" but we also have fluid dynamics, finite element analysis and so on. DJ Clayworth 14:01, 25 Aug 2004 (UTC)

To add to this discussion, the list of codes is getting pretty big and I'm wondering if it would help to create subcategories such as Open Source vs. Commercial Codes. These are distinct software types, unlike other subcategories such as incompressible vs. compressible codes(a lot of CFD software handles both) or solution method (e.g. Finite Element vs. Finite Difference vs. Finite Volume.)

To my opinion, the list has become ridiculously big and contain many things that should not be there. It seems to me that only the most relevant and generalist CFD softwares should be cited here. What is the point of a link to a "commercial CAE software for injection molding", the mesh maker should be in a mesh page etc...
I propose to keep only CFX, FLuent and STAR-CD for commercial codes and OpenFOAM and maybe Openflower for free codes. I am far from having an exhaustive sight of all the CFD codes, so some more cultivated CFDers of you should probably correct this list, but I really think we should drastically reduce the list. ZondeR, 28 Nov 2006
As I mentioned above, if someone is making a comprehensive list it should be a Wikipedia list article. The codes mentioned in the CFD article should only be those significant in the history of CFD (for example, the first commerical CFD code). --Charlesreid1 (talk) 00:07, 5 November 2010 (UTC)

Problems with the article[edit]

Why is there not any mention of LES(large eddy simulation) or DNS(direct numerical simulation) (fixed)

More importantly, what about Langrangian Techniques? Don't you think that Lagrangian and Euler based methods deserve a say in this article?

I think that there should be an introduction to the difference between the Lagrangian and Eulerian approach and how Lagrangian ones are better suited to handle interactions with rigid bodies (that are usually modelled with a Lagrangian approach). Lagrangian-Eulerian methods are also used widely in the field of CFD for computer graphics, where the main problem is the fluid-solid (and vice-versa) interaction and the free surface between the fluid and the air (that's the visible part of fluid flow, so it's important for CG). Maybe a reference to the level-set method is also needed as it is a link to this book

Added 5/1/2005 I've cleaned the article up a bit, and pointed out the key issues with regard to discretization at the very beginning. User:Rtfisher

In the article it says, "While it is possible to directly solve the Navier-Stokes equations for laminar flow cases, turbulent flows require the introduction of a turbulence model."

This is not true since DNS does not use any turbulence model, although it solves the Navier-Stokes equations directly.

"Boundary conditions are defined. This involves specifying the fluid behaviour and properties at the boundaries of the problem. For transient problems, the initial conditions are also defined"

This is not true since initial conditions are specified for steady-state problems as well. In steady state problems, the solution evolves from the intial guess to the final steady state solution. (unsigned)

In that case it isn't an initial condition, it's an initial guess. By definition, initial conditions are the conditions for when the time coordinate is equal to zero. If there is no time coordinate, there cannot be initial conditions. So the statement is correct. --Charlesreid1 (talk) 00:10, 5 November 2010 (UTC)

There is probably no mention of LES because because nobody felt competent to write it. It's not exactly a well-known subject. I worked in the field, and I'm not competent. Feel free to add what you know. Be bold in editing. DJ Clayworth 04:00, 18 Nov 2004 (UTC)

The "Boussinesq hypothesis" link that is included under the RANS subsection should refer to Boussinesq's assumption regarding a scalar eddy viscosity and not to his approximations regarding buoyancy-driven flows. ( 20:38, 18 April 2007 (UTC))


Does Suhas Patankar warrant a mention in this article? If so, could someone more knowledgable than I whip up a sentence to do so. Thanks. -- John Fader 21:27, 17 Feb 2005 (UTC)

We need, I think, a paragraph or two at the beginning that explains CFD for the layman. Someone with no mathematical background coming to find out what CFD is all about needs some information before we start getting into discretization method. It should go first because if they don't get something they understand in the first paragraph they will stop reading. DJ Clayworth 11:20, 3 May 2005 (UTC)

I disagree, I think some mention of the most common discretization methods (a sort of "where to go" if you're interested in discretization methods) is definitely called for in the article (although I agree it doesn't belong in the introduction). For example, Patankar would be an excellent reference for the control volume subsection of the discretization section. Finite volume, finite element, and particle methods would also be good to mention, since most discretization methods in use fall into one of these four categories. If someone is qualified to add brief descriptions of another discretization method, that would be great. In any case, each would be a short description linking to a main article. --Charlesreid1 (talk) 00:13, 5 November 2010 (UTC)

HYSYS is not a CFD app[edit]

HYSYS is not a CFD application. It is a steady state & dynamic process flowsheet simulator. It has a built in dynamic pipe unit (simple compressible gas model - 1 dimensional) and links to other commercial dynamic pipeline simulators (also 1 dimensional models).

External links[edit]

The external links are getting out of hand. Wikipedia is not a link farm. I suggest we cut this down to the bare minimum, and if we can't agree on that then we cut out all the external links. Computer-aided design doesn't have any external links to codes and is much better for it. DJ Clayworth 21:06, 18 July 2006 (UTC)

Well, I don't know which version you mean, but if you mean current external links under the headline "External links", I think around 10 links looks fine. Links under "Software", on the other hand, are way too many and could easily be tenfold if you link to all software that has to do with CFD in one way or another, solvers, mesh generators, post-processing, et cetera. Most of those links could be found if you go to the links under "External links" anyway, so they are not necessary for finding CFD software. Anders Ytterström 01:20, 31 August 2006 (UTC)

Yep, it's the Software ones I meant. DJ Clayworth 04:51, 31 August 2006 (UTC)

I have commented out all the software links. This is an encyclopedia, not a directory listing of all the software firms. I will delete them if not one provides any good suggestion as to why those links should be kept here. Please refer to following pages, before making any suggestions,
What Wikipedia is not
Wikipedia is not a directory
External links
Links normally to be avoided
Other links that I have removed,
CFDnet The page seems to be last updated in 2001. Not sure if it is of any use.
Cavity flow Would be better to have this link on a separate article related to cavity flows. It is too specific to be included here.
Usenet discussion group on CFD One can find many online discussion groups. No point including all of them here. One can easily find them.

myth 03:14, 17 January 2007 (UTC)


I fixed the contents' levels: DNS, LES etc. all under "Turbulence Methods"..

Separate articles about the turbulence models[edit]

Perhaps the sections about turbulence models (DNS, RANS, LES) should be integrated more with their respective separate articles, to avoid having the same information in more than one place? -- Ehdr 15:33, 16 March 2007 (UTC)

The general approach should be to give a brief description of each turbulence model, as it pertains to computational fluid dynamics, and use the {{Main}} template to point readers toward the full article if they're interested. --Charlesreid1 (talk) 00:16, 5 November 2010 (UTC)

What does a typical cfd software toolchain look like?[edit]

Can anybody help me figure out the file formats and programs involved in completing cfd calculations etc.? Starting with blender, the open source 3D modeling program, I can get a mesh file (gmsh can read this), but then the majority of the cfd simulators do not seem to have a lot of documentation on how to set up a simulation (like the specifications, parameters, the types of materials, the density etc. etc.), but maybe I am missing something? And after that, how can I relate that to lift, thrust, and other properties involved in making sure, say, an airplane could fly? -- kanzure (talk) 03:04, 4 March 2008 (UTC)

I posted that question over at cfd-online noon earlier this day. Since then I have found that FreeFEM++ works nicely, as well as OpenFVM, and salome/OpenFOAM still questionable (though salome at least can run -- having some java environment config problems with OpenFOAM, as awesome as the whole package looks). I am stil having troubles envisioning what the entire toolchain should look like when I get it all set up. I have posted to the ff++ mailing list, the gmsh mailing list, and I am about to go write to the arocket group as well. Maybe they will have some ideas. I am surprised that this is not more thoroughly discussed on the internet yet, I suspect it is because of the perceived "math is hard" barrier. -- kanzure (talk) 03:07, 4 March 2008 (UTC)

Cleanup templates[edit]

I've gone through and added multiple cleanup templates, particularly {{Refimprove}}, {{Unreferenced section}}, {{Weasel}}, {{Who}}/{{Whom?}}, {{Citation needed}}. The style of writing in the article is terrible, it sounds more like an editorial than an encyclopedic treatment of CFD. The sections could be greatly improved by simply adding some additional references and getting rid of weasel words and editorial phrases, but more work is sorely needed. It's frustrating to see such an important topic get such a poor treatment. --Charlesreid1 (talk) 18:04, 5 November 2010 (UTC)


This article states: "With high-speed supercomputers, better solutions can be achieved." This would imply that merely switching to a faster computer is going to help, which is of course not true. Higher mesh resolution will generally help get a "better" solution as will improving physical models - both of these require more computational resources, so if you want to get your answer in the same time, you'll need a faster computer, but a faster computer will not in itself do much for you in terms of accuracy, assuming that's what "better" is supposed to mean here.

Lowering Article Quality Rating[edit]

I've lowered the article quality from C to Start due to the lack of citations and overall plethora of unverified claims. I'm also including the Wikipedia editorial team's assessment table/criteria below. I believe this article very clearly fits in the Start class, as it lacks citations, has had a {{Refimprove}} template since 2008, and leaves a majority of readers looking for a different source of information about CFD.

Accordingly, the number one task for improving this article should be to include additional references for unverified claims.

WikiProject article quality grading scheme

--Charlesreid1 (talk) 18:13, 5 November 2010 (UTC)

Combustion models for CFD[edit]

Can someone review this newly created article? Not sure if it should be a stand alone. Hell In A Bucket (talk) 13:56, 4 December 2012 (UTC)

CFD in Aerospace Accident Investigations[edit]

Does someone know if CFD is regulary used in Aerospace accident investigations. I read about CFD investigations related to space missions, but is it typically used for investigations within the aerspace domain? — Preceding unsigned comment added by (talk) 15:24, 12 August 2013 (UTC)

Merge with Fluid simulation[edit]

I see no clear distinction between this article and the article Fluid simulation, so I propose that we merge them together. What do you say? —Kri (talk) 19:10, 18 September 2014 (UTC)

  • Oppose I think they are about two different topics and have two different goals although solve similar problems. XFEM Skier (talk) 15:40, 25 September 2014 (UTC)
  • Oppose These are different topics. CFD is about numerically solving the equations of fluid dynamics as well as possible. Fluid simulation in computer graphics is about simulating the appearance of fluid elements in a graphics scene or animation. If a fluid simulation looks good, it doesn't matter if the physics is correct. An example of this is Fourier synthesis of ocean waves. This technique can produce a good looking ocean, but isn't numerically solving the underlying equations of fluid dynamics. --Mark viking (talk) 17:55, 25 September 2014 (UTC)
Well, I would say that the Fourier synthesis is a form of CFD and that it does solve the underlying wave equations approximately after making some assumptions about the flow, since it is based on theory, which is really what all other CFD methods do as well. :) Although it is not a very powerful method and can only be applied to a very limited set of fluid dynamics problems, like those that are interesting in computer games and computer graphics. —Kri (talk) 20:51, 25 September 2014 (UTC)
Respectfully, I disagree. As the Mastin, et. al., 1987 paper on Fourier synthesis of ocean waves describes, this particular fluid simulation is based on Fourier synthesis using a weighting based upon a Pierson–Moskowitz spectrum. The spectrum itself is an empirical, statistical one derived from ship wave data. While there is simulation of the appearance an ocean surface, there is no underlying numerical simulation of any equations of fluid dynamics. Mastin, et. al., were faking it and doing a good job of it--it was a big advance in computer graphics. But it wasn't CFD. --Mark viking (talk) 23:56, 25 September 2014 (UTC)
The Fourier synthesis method uses a differential equation to update the phase angle in the Fourier transform of the free surface elevation. This equation can be derived from the boundary element method applied to incompressible, irrotational ocean water, by linearizing the boundary integral equation and transforming it to the frequency domain. This results in that different eigenmodes map against different wavevectors. If this is not CFD, where along the transformation from BEM does it stop being CFD? When does the differential equation stop being an equation of fluid dynamics? —Kri (talk) 10:53, 27 September 2014 (UTC)
About the usage of a spectrum, that just affects the initialization of the system, so I don't see how that would affect the method itself. —Kri (talk) 11:10, 27 September 2014 (UTC)
  • For: The only difference that might exist between the terms "CFD" and "fluid simulation" is in the way they are normally used, and would in that case be (as XFEM Skier pointed out) their goals. Although essentially, they are two different terms for the same concept—they both simulate fluids by solving fluid dynamics equations numerically and can both utilize the same methods (even though the usability of the different methods may vary depending on the goal). Sure, CFD is a branch of fluid mechanics, but fluid simulation may also be seen as such. The two terms also seems to be used interchangeably by Jerry Tessendorf in a TED talk I recently bumped into. —Kri (talk) 17:49, 19 November 2014 (UTC)
  • Oppose The two subjects have fundamentally different objectives. — Preceding unsigned comment added by (talk) 22:20, 6 December 2014 (UTC)
  • Oppose Arguably fluid simulation (in the sense used here) is a subtopic of CFD, but even taking that to be correct, if they were merged into one article then I would suggest splitting the computer graphics part of the merged article off into an article of its own. The amount of content in the current fluid simulation article is over the top for its significance in CFD, it's sufficiently notable for an article of its own, so it should continue to have one. Djr32 (talk) 19:37, 26 December 2014 (UTC)
  • Oppose Fluid simulation and CFD are essentially synonyms. The current 'fluid simulation' article should be re-titled to something more appropriate and descriptive, e.g. 'fluid animation' or 'fluid simulation in computer graphics' or something along those lines. Christopherbatty (talk) 01:42, 23 February 2016 (UTC)
  • Oppose Fluid simulation and CFD are basically the same thing. However, I agree that the current fluid simulation article has the wrong name and should be called fluid animation (or something similar) instead. I'll probably do the rename in a few days unless someone disagrees. ArguMentor (talk) 11:46, 27 February 2016 (UTC)
You and Christopherbatty make a good point. I'd support a better fitting rename. --Mark viking (talk) 21:12, 27 February 2016 (UTC)

Parallel computing[edit]

This is missing from the solution methodology. I think it is an important and interesting item and is also one in the to-do list topics to include of this article. I propose a section with a few bullet points regarding

  • The need for parallel computing
  • Methodology
    • Shared vs. distributed parallel computing
    • Partitioning (principles and methods)
  • Various protocols (MPI, PVM, etc)
  • Best practices
    • Nodes/CPU core
    • Overlapping
  • Parallel performance definition
  • Massive parallel (thousands of CPUs)

I will soon drop a draft. Sankgeo 10:47, 14 March 2016 (UTC) — Preceding unsigned comment added by Sankgeo (talkcontribs)