Pressure-Velocity Coupling in CFX
Hi,
is Pressure-Velocity Coupling in CFX enabled as default? From Fluent, there is an option to chose SIMPLE or SIMPLEC schemes for the Pressure-Velocity Coupling. So I am wondering does CFX has this option? pls shed some lights. Many thanks |
so i know, it isn´t enabled by default. but you can fix it in the expert parameters.
|
Pressure-velocity in CFX is implemented using Rhie-Chow algorithm and is enabled by default.
|
Hi,
may I know where the setting is? The help files only briefly mentioned about the theory. Pls advice |
What do you really want to do? Turn it off at all or adjust some settings? Keep in mind that any changes in Rhie-Chow algorithm can change solution and/or convergence and therefore are not recommended.
|
Hi Zef,
I am actually doing a study on CFD of airfoils and one of my objective is to see how the results would varies with different settings and parameters. So I am wondering is there any settings for Pressure-Velocity Couplings that I can play around with. In Fluent, there are SIMPLE and SIMPLEC for the P-V couplings so I am wondering CFX would have something like this too. Could you shed some light? |
There are two ways to adjust Rhie-Chow parameters in CFX. If you are using version 11 with SP1 then you can turn beta features on. With this option checkbox "Velocity Pressure Coupling" in Solver Control->Advanced Control will appear. There are several parameters in it, but be aware that it is a beta feature and some of these parameters will not actually work. In CFX version 12 "Velocity Pressure Coupling" is not hidden, but contains much less parameters to adjust.
The second was is to use expert parameters. In order to know which of the expert parameters are dealt with Rhie-Chow algorithm, you should ask technical support. In my opinion none of these settings actually affects on the converged solution. It may affects only on speed of convergence. |
Hi Zef,
currently I only have access to version 11. Anyway, many thanks for sharing with me. |
Hi,
Sorry to be brutally honest, but all the previous replies don't know what they are talking about. CFX is a fully coupled solver and so the pressure-velocity coupling is inherent in the solution procedure. SIMPLE, SIMPLEC and PISO are required when you have uncoupled solvers like Fluent. Adam S - Turning PV coupling off does not make sense. You cannot turn it off. Zef - Rhie Chow is not PV coupling. It is how the pressure is interolated from the nodes to cell centres. The main thing here is to avoid decoupling of adajcent cells. PV coupling is totally different. LSC - I strongly recommend you not fiddle with the Rhie-Chow settings unless you know exactly what you are doing. 99% of the time there is something else fundamentally wrong with the simulation and this sort of thing will not help. Glenn Horrocks |
Hi Glenn,
now I understand how it works. Thanks for sharing! |
Hi,
Just to expand my previous answer a bit: Uncoupled (or segregated) solvers like Fluent solve the 3 momentum equations (ie velocity) sequentially, then using the updated velocity field they calculate the pressure correction equation for continuity. This is repeated until convergence. Coupled solvers like CFX have all 3 momentum equations AND the pressure equation in the same matrix so they are solved together. It does not need PV coupling as that is taken care of in the matrix solution. Coupled solvers take more time per iteration and use more memory as the matrix is bigger, but coupled solvers usually converge much faster as you are only converging on the non-linear terms. Segregated solvers need to converge both the non-linear terms and the PV coupling. Regards, Glenn Horrocks |
Hi Glenn,
the add on was very helpful. Now I understand why each iteration in CFX takes a longer time when compared to Fluent. Again, thanks a lot for sharing |
Quote:
|
Hi,
As you correctly explain, Rhie Chow is how the PV coupling interpolates the pressure. Therefore it is an interpolation scheme, not a PV coupling mechanism. It is, however, an integral part of a PV coupling mechanism so that is why the options are grouped together. While Rhie-Chow does act on the mass-carrying terms, it does so by introducing a pressure redistribution term. So my original reply was not totally wrong, but not quite right either :) Glenn Horrocks |
Hi Glenn,
I understood. Thank you for explanation. Best regards, Zef |
Hello everyone ,im really new in CFD techniques ,and im actually doin my thesis on a problem of combustion chamber in CFX 10.(its a combustion tubular chamber from a HITACHI turbine).
and if CFX doesnt use algorithm like SIMPLE,SIMPLEC,SIMPLER,where can i find the information of the way this software works with velocity-pressure? sorry for my english thanks very much Mauricio Caamaño. |
Hi,
Read the documentation. That is the nest place to start and I can't count the number of times I end up saying it. The documentation also has references to more detailed articles about specific bits of the software so if you want to delve deeper you should chase those up. But as an overview the discussion on this thread should get you started. Also you should try to get V12, the current version. V10 is quite old and there has been lots of enhancements since then. Glenn Horrocks |
student in problem
i have all the documents of ansys cfx ,theory,modelling,tutorials etc... but..i really didint see this subject.:(
with Rhie-Chow algorithm can i explain how its cfx solves velocity-pressure coupling?. but ansys cfx 12 really has a difference with cfx 11 or 10?. thanks very much for your help. im full studyng all with a book called "Introduction to computatioanl fluids,Malalasekera" i think its a very good book for starter in this problems. once again thanks very much for your advise.:D Mauricio Caamaño Flores |
Quote:
Quote:
Quote:
Quote:
Also note the CFX coupled solver technology came originally from TASCFlow. You may be able to find some old TASCFlow documentation to help describe the basics. Glenn Horrocks |
thanks very very much for your help Glenn, it was very helpful your comments ,i really apreciate your help.:D
Mauricio Caamaño Flores. |
Dear Glenn Horrocks and thread participants, I think the sentences below are true. But could you find if there is something false or incomplete?
1. In pressure-based continuity solvers, the pressure field is indirectly specified via the continuity equation by means of a PV coupling algorithm (SIMPLE, SIMPLEC, PISO, ...), and the density field is then extracted from pressure via an equation of state. (True or false?) 2. In density-based continuity solvers, the density is directly specified via the continuity equation, and the pressure is then extracted from density via an equation of state. (True or false?) 3. The checkerboard effect (velocity-pressure decoupling and oscillating solutions), which is avoided by the Rhie-Chow algorithm, may occur only on pressure-based continuity solvers, and not on density-based. (True or false?) 4. The checkerboard effect may occur only on segregated solution sequence, and not on coupled. (True or false?) 5. The Rhie-Chow algorithm is an improvement for the discretization/interpolation schemes (Central Difference, Upwind, Power-Law, ...). (True or false?) |
1. True, keeping in mind that your list of PV coupling algorithms is not complete. CFX uses a coupled approach which does not use SIMPLE, SIMPLEC, PISO but solves all the equations simultaneously. But you can call this just another PV coupling algorithm if you like.
2. I have not done much with density based solvers but I think you are correct. 3. Rhie-Chow is designed to avoid checker-boarding is correct. As I have not done much with density based solvers I cannot say if checkerboarding is an issue for them. 4. False. Checkerboarding is caused by the adjacent pressure and momentum control volumes not being coupled. This can occur regardless of whether the approach is segregated or coupled. For instance CFX uses a coupled PV algorithm but still requires Rhie-Chow to stop checkboarding. 5. False. Rhie Chow is how the pressure field is interpolated to the momentum control volumes to avoid checkerboarding. Central difference, upwind etc are how the advection term is discretised. These are totally different things. |
Glenn Horrocks, thanks for your thorough replies.
As far as I understood from [1], Rhie and Chow (1983) suggest that a third-order pressure gradient term should be included to the interpolated expression of the cell face velocity, according to the equation below: https://drive.google.com/file/d/1SuA...ew?usp=sharing This links the cell face velocity directly to the driving pressure difference across that face. Even though [1] later says that "Rhie and Chow’s pressure interpolation practice involves the addition of a third-order pressure gradient term", this term is added to the expression of the cell face velocity, not pressure. Thus, shouldn't Rhie-Chow algorithm be called a velocity interpolation method instead? In addition, in the equation above, isn't that third-order pressure gradient added to the standard Central Difference velocity interpolation accross the face? This is why I called the Rhie-Chow algorithm as improvement to the interpolation schemes. Please let me know where are my flaws. [1] Versteeg and Malalasekera, 2007. An Introduction to Computational Fluid Dynamics. 2nd Edition. |
These questions are starting to get a bit academic.
The point of Rhie-Chow is to improve the coupling of the P and V equations as a naive coupling leads to checkerboarding. OK, so it does that by changing the interpolation on the velocity; so you could call it a velocity interpolation scheme. But just remember the point of why you are doing it. Upwinding, central difference and the others are for the advection terms in the momentum equation (and the other conservation equations). Rhie Chow replaces the naive central difference approach to handle the fact that the velocity equations require the pressure gradient and when implemented on a co-located grid this leads to odd-even decoupling. So Rhie Chow is not an advection scheme. This presentation explains it very well: https://personalpages.manchester.ac....omphydr/p5.pdf |
To reaffirm Glenn's point, the Rhie-Chow algorithm does produce a velocity at the faces; however, it does not interpolate from neighbor's velocities but it interpolates the "momentum equation" from the neighbors and it produces a new equation for pressure when substituted into the continuity equation (a two lines summary of course).
If it were a "classic" interpolation scheme of a variable, there would be no contribution from the advection, diffusion and pressure gradient term in the equation. So, it is NOT a "traditional" velocity interpolation scheme. |
All times are GMT -4. The time now is 00:22. |