[xisluke]

Hi guys!

Let's suppose to deal with a steady state simulation, using SIMPLE as a Pressure-Based solver. In this case, Fluent allows the user to adopt a "local time step" approach for the pseudo transient term. Fluent, reading on the theory guide, does not use a time-depending term for the steady state calculations, in fact, it allows to set up the implicit URFs only (if the pseudo transient term is deactivated).

At this point, as far as I know, activating the pseudo transient term means somehow modelling the imbalance deriving from the iterative procedure as a transient term, which will disappear once a steady state solution has been reached. Now, implicit URFs and this transient term are linked in someway; fixing a URF, basically means having different time steps across the cells, and Fluent allows to control these time steps imposing a "pseudo time Courant number".

Here it comes my doubt, if we are modelling a steady state equation adopting a pseudo-time step which varies across the mesh, it makes sense that, having cells of different dimensions, it is possible to identify a pseudo time courant number that regulates the changing in the local pseudo time step. But at this point, if we are somehow imposing these time steps, how can it make sense to also impose "(implicit) pseudo time URFs"? What are these parameters? Where are they placed? I tried to control on the Fluent Theory Guide but, as well as mentioning them, it does not give a lot of technical information.

Big thanks!

[LuckyTran]

You are right that it makes no sense to have implicit URFs and a Courant number in the same sentence.

For the particular screenshot in question that is generating your doubt, the URF's are explicit URFs. They are not implicit. The pseudo transient formulation by itself acts via implicit URFs, but the implementation in Fluent allows for also simultaneously specifying explicit URFs on top of it. I admit Fluent doesn't have the most detailed user interface nor documentation but at least it's somewhat intuitively clear that they can't all be the same thing =)

[xisluke]

Originally Posted by LuckyTran

You are right that it makes no sense to have implicit URFs and a Courant number in the same sentence.

For the particular screenshot in question that is generating your doubt, the URF's are explicit URFs. They are not implicit. The pseudo transient formulation by itself acts via implicit URFs, but the implementation in Fluent allows for also simultaneously specifying explicit URFs on top of it. I admit Fluent doesn't have the most detailed user interface nor documentation but at least it's somewhat intuitively clear that they can't all be the same thing =)

That was my first thought too, but when it is necessary to deal with explicit URFs, Fluent specifies it.

For example, if I switch from SIMPLE to Coupled, the pseudo transient term becomes Global (i.e. it is fixed across the cells), therefore it is not necessary to impose a pseudo time Courant number (since it varies across the cells). Then, moving to the "Controls" panel, Fluent it is very clear and allows you to play with "Pseudo Time Explicit URFs", and this makes sense, since the implicit under relaxation is performed by the pseudo transient term.

The most confusing situation then, takes place when a Coupled solver is selected and we are not using the pseudo transient term; in the "Controls" panel I have options for flow Courant number, explicit URFs for pressure and velocity, implicit URFs for every transport equation.

[LuckyTran]

Notice for the steady Coupled solver you don't have an implicit urf for the flow equations or momentum equation. Courant number is just flavor-text.

And although I understand your frustration with the way it is presented, any time you build any implicit solver there will always be an implicit urf implied by your methodology and explicit urf (it might be 1) also being implied. The difference is in whether the particular software you are using has the development experience behind it to implement user options for allowing you to specify both explicit and implicit urf's together (because they've had too many customer cases diverge and it's cost them quite a bit of $$$$$).

If you want to get confused then check out transient cases where you have global and local time-steps, with implicit urf's, with explicit urf's and on top of this you apply preconditioners and multigrid solvers. That's 6 nested levels of relaxation and the multigrid solver uses nested relaxation-prolongation as well. We're talking nested relaxation nested instead nested relaxation! Sounds like something you've seen in Inception but it's very real. I'm not saying you need to understand all these things because really you are asking what is happening at the outer iteration level and not the inner iteration, but it does get a lot more confusing if you want it to be!

[xisluke]

You are right, there are not implicit URFs for momentum and pressure. My main concern was related to the "flow Courant number"; so in your opinion, that input is pretty useless and Ansys guys simply forgot to specify that, when pseudo transient term is activated, URFs are always explicit. Did I get it right?

[LuckyTran]

No, there are always implicit and explicit URFs. Ansys guys are not indeed idiots. Ansys guys simply didn't apply redundant lablels to make it clear that there are explicit and implicit urfs everywhere all the time. They chose (rightfully so in my opinion) to label it Flow Courant number.


The only gripe I have with Ansys is that they don't provide the source code so people can verify these things for themselves. But then again, OpenFOAM does have an open source code and nobody knows how it works.

[xisluke]

Originally Posted by LuckyTran

No, there are always implicit and explicit URFs. Ansys guys are not indeed idiots. Ansys guys simply didn't apply redundant lablels to make it clear that there are explicit and implicit urfs everywhere all the time. They chose (rightfully so in my opinion) to label it Flow Courant number.

Yes, I said a stupid thing; since I have not implicit URFs for pressure and velocity, it is the flow Courant number that implicitly regulates that relaxation within the equations. I'm sorry, I intepreted the "flavour-text" wrong and for some strange reason I didn't notice that there was also another part on your answer. My frustration is driving me crazy ahaha!

Sadly, using a commercial software has its pros and cons (not having access to the code is one of them). In addition, I would say that everything in CFD do get lot more confusing if one wants it to be (but I think also that it is the main reason why some people love it, and some people hate it).