[ssbear]

Problem solved. Thanks.

[ghorrocks]

Before you look at anything like turbulence - make sure your model is physically consistent. How have you set the pressure? How can the simulation match the inlet and outlet flows?

Is the fluid incompressible or compressible?

[ssbear]

Originally Posted by ghorrocks

Before you look at anything like turbulence - make sure your model is physically consistent. How have you set the pressure? How can the simulation match the inlet and outlet flows?

Is the fluid incompressible or compressible?

Hi Ghorrocks,

Thanks for the reply. The inlet and out size is the same therefore with the same velocity, the flow is the same. I did not have any settings for the pressure in this model (just default settings), will that be a problem? The fluid is set to be air at 25 C. Should I change it to Air ideal gas to make it compressible?

The transient case run good for 100 timesteps and then the residuals jump up again....

[ghorrocks]

It appears your boundaries are not physically consistent. You cannot define a velocity inlet and outlet. Even if you have matched them up so the flow rate in and out should be the same, tiny numerical approximations (eg mesh discretisation, numerical accuracy) mean that they are not equal. This means the flow in does not equal the flow out and therefore this will never converge steady state. Also a transient flow will have a hard time converging as well.

Read the CFX documentation regarding choice of boundary conditions. In short, either the inlet or outlet should be a velocity condition (which sets the flow rate) and the other should be a zero pressure boundary (which sets the pressure). This will make your boundaries physically consistent. The term for this is "well posed".

Until you fix this fundamental problem with your simulation your results will be rubbish and convergence all over the place.

[ssbear]

Originally Posted by ghorrocks

It appears your boundaries are not physically consistent. You cannot define a velocity inlet and outlet. Even if you have matched them up so the flow rate in and out should be the same, tiny numerical approximations (eg mesh discretisation, numerical accuracy) mean that they are not equal. This means the flow in does not equal the flow out and therefore this will never converge steady state. Also a transient flow will have a hard time converging as well.

Read the CFX documentation regarding choice of boundary conditions. In short, either the inlet or outlet should be a velocity condition (which sets the flow rate) and the other should be a zero pressure boundary (which sets the pressure). This will make your boundaries physically consistent. The term for this is "well posed".

Until you fix this fundamental problem with your simulation your results will be rubbish and convergence all over the place.

Thanks again for your comment. My problem is that the inlet and outlet is actually in one mechanical unit. The are air from the outlet will be treated and then send back to the warehouse through inlet. Therefore the flow through each outlet should be same as each inlet. If I use velocity inlet and pressure outlet, the flow through each outlet will be different based on the pressure, and that is not a valid assumption in my case.

In this case do you have any suggestions for for me the set up the boundary conditions?

Thanks

[ghorrocks]

If I use velocity inlet and pressure outlet, the flow through each outlet will be different based on the pressure, and that is not a valid assumption in my case.

If the inlet is a flow rate boundary and the outlet is a pressure boundary then the inlet flow rate will equal the outlet flow rate by continuity. You do not need to set them to be the same. In fact if you set them both to the same flow your simulation is mathematically impossible to solve, as it is badly posed.

Please read the section in the documentation on choosing boundary conditions. If you do not choose well posed boundary conditions nothing is ever going to work.