Hello

I'm trying to compute the flow in a duct at 20 bar with a bulk inlet velocity of 30 m/s. At 1 bar and 6 bar, all went fine but at 20 bar it doesn't want to converge. It says things like : Mach number is 5.05 and speaks about expert parameters. I'm not an expert and as it worked fine at 6 bar I'm wondering what is wrong. The gas is air as an ideal gas, I solve with the Thermal model. Does someone has an idea of what is going on ?

Thank you

Regards

Olivier

Exactly what pressure are you changing? From the sounds of it you are running a velocity inlet and possibly a static pressure outlet. So, are you changing the outlet static pressure?

Neale

Hello,

Thanks for the reply. I'm running a velocity inlet with experimental inlet velocity profile data. It's a semi-developed velocity inlet. At the exit I specify a relative pressure of 0. It's the reference pressure that I switched from 1 bar to 6 bar, and then to 20 bar.

The thing is that with a coarse mesh it works fine at 20 bar. I mean that with a y+ larger than 100 it converges. But if I try to get a y+ between 20 and 100, I get this message concerning a Mach number larger than 5 and everything goes wrong. I run a Reynolds stress model, because validation has shown that the other models do not give good results in my geometry.

Thanks for the help.

Hi,

From thermodynamics point of view, the assumption of ideal gas is not valid at high pressures. So you might want to change the "air as ideal gas" assumption. I don't know whether this will help you in getting better convergence, but it will help you in getting more realistic solution.

Regards, Swapnil

Thanks a lot.

Cheers

You didn't specify if you created a new case/def file or used CCL and edited the old res file.

I have experienced similar difficulties with restarts from an existing res file with a new reference pressure. I suspect a bug in the code for restarts with different reference pressures, because I can get reasonable results when I change my initial guess for pressure from automatic with value to value and specify an appropriate value. Perhaps it's not so much a bug as missing communication between what the user thinks he's setting: reference pressure for the domain and relative pressure for bc / initial guess and what the solver might actually be doing with absolute / relative pressures. Of course if CFX would implement absolute pressure BCs , we wouldn't think of starting up a case with a BC of 20 bar with an initial guess of 1 bar would we?

In my case I was changing Re and pressure level so my velocity was not changing much. If you're just changing pressure level but not Re, your velocity will probably change a lot so you might have to do the same with your velocity initial guess. Too bad you can't just easily scale your velocities in the Post GUI for an initial guess like you could in TASCflow.

Martin,

You don't have to specify a reference pressure. If you want your boundary conditions to be absolute pressures, just make your reference pressure zero!

Robin

Dear Robin, Could you please explain how CFX deal with pressure in more detail? I do not understand exactly about pressure in CFX. For ideal and general fluid, -when there is the hydrostatic pressure, we don't need to specify it in the boundary condition, right? -When you said that zero reference pressure mean absolute pressure, what does this absolute pressure mean? It mean the pressure term that compose of the atmospheric pressure and the guage pressure, right? Then how can CFX know the atmospheric pressure? Thank you very much.

Atit

I thought the reference pressure was there to prevent round-off errors in single precision?