[CFD-fellow]

Hi
In my simulation I know maximum pressure and velocity in my domain and Im looking for a way to force the solver to clip these values (not clipping properties).
For example in Fluent one can set min and max values of pressure velocity and some other variables, and Im sure CFX has too but cant find it!
Any help would be appreciated.

[ghorrocks]

This is possible but is a really bad idea. It is a bad idea because it means the gradients of pressure and velocity go bezerk where it clips and that makes the numerics horribly unstable. You should allow the solver to undershoot and overshoot as it progresses to convergence. You will only make things harder by clipping.

[CFD-fellow]

Hi Glenn
I know that this is not a good task but I want to try it. Where can I adjust these parameters?

[ghorrocks]

I can't guarantee this will work, but try this:

Open {CFX_ROOT}/etc/VARIABLES

Find the pressure section. It has a keyword "Component Bounds Flag = No", switch that to Yes. Then the "Component Lower/Upper Bounds = ..." keywords should (might?) become active.

And try your luck in the velocity section. Even less guarantees that it works there.

And don't forget to backup the VARIABLES file before you do this otherwise you might kill your CFX installation.

[singer1812]

Something that might work without involving changing Variables section is to add a source term turns on or off with your criteria. It will not strictly limit the variable, but you can influence the magnitude with a source or sink.

For example if you want to clip temperature at the lower limits you can do the following. The below example will pump energy into a node when its temperature falls below 20K. The trick on the source terms is to get the equation on the source to behave in a way you want it to (dont want to be pumping in too much energy as the temperature will over compensate):

SOURCES:
EQUATION SOURCE: energy
Option = Source
Source = Density*1004[J kg^-1 K^-1]/tstep*(20 \
[K]-T)*step(20-T*1[K^-1])
Source Coefficient = Density*1004[J kg^-1 K^-1]/tstep
END
END

[Opaque]

Clipping solution variables such as velocity, pressure, enthalpy, turbulence kinetic energy, etc is a bad practice since you are forcing the solver to satisfy two conditions: the solution the equation is trying to converge to, and the maximum/minimum artificially imposed value. You will observe erratic convergence if any at all.

If your variable is physically limited, i.e. bounded, the transport equation must include such constrain such the equations fed to the solution algorithm can converge to both constrains: conservation of something, and bounded.

Singer1812's suggestion is a good idea, though I rather see it with the Source Coefficient only since the addition of the source could let it converge to an unexpected solution, and you will need to verify the source added is negligible to rely on the converged results. This is approach is basically an additional under-relaxation of the equations greater than the one imposed by the software.

My 2 cents.