[heng03313]

I am trying to model a real world situation. It is a 140+ meters long pipe with the inlet having a fan blowing air at 1.8bars gauge and the outlet to atmospheric. The resultant velocity, from what I know, is of (single digit) m/s around the inlet, to about 20 m/s at the outlet. These values suggest that the Mach number is less than 0.3. However, the pressure difference at the inlet and outlet is so big, that the density varies from about 3.2 kg/m^3 at the inlet to 1.225 at the outlet.

So the question is, do I ignore compressibility effects (since M<0.3) and use constant density, or do i use an ideal gas formulation to account for the density? I've tried using ideal gas, but the solution for a steady state calculation can never converge. And a constant density formulation just doesn't make sense.

Could anyone help me bounce off ideas on why an ideal gas formulation doesn't converge? Might it be due to the Mach number being too low? Resulting in diverging oscillations? Or could it be due to mesh size? Due to the length of the pipe, I am using 2cm cell size (compared to 35cm pipe diameter), before my computer blows up having too many cells.

I have been using constant density for now, using the average value of the inlet and outlet, with no issues. However, using constant density means getting velocity results which are (fairly) constant too? I am getting about 18-20 m/s throughout the pipe, from inlet to outlet.

[Bruno Machado]

can you solve it for an incompressible condition and use it as an initial condition for a compressible one? than you can compare and see if the difference is significant for your purpose or not.

[heng03313]

I've tried that. it diverges the moment I use ideal-gas.

[Antanas]

Originally Posted by heng03313

I am trying to model a real world situation. It is a 140+ meters long pipe with the inlet having a fan blowing air at 1.8bars gauge and the outlet to atmospheric. The resultant velocity, from what I know, is of (single digit) m/s around the inlet, to about 20 m/s at the outlet. These values suggest that the Mach number is less than 0.3. However, the pressure difference at the inlet and outlet is so big, that the density varies from about 3.2 kg/m^3 at the inlet to 1.225 at the outlet.

So the question is, do I ignore compressibility effects (since M<0.3) and use constant density, or do i use an ideal gas formulation to account for the density? I've tried using ideal gas, but the solution for a steady state calculation can never converge. And a constant density formulation just doesn't make sense.

Could anyone help me bounce off ideas on why an ideal gas formulation doesn't converge? Might it be due to the Mach number being too low? Resulting in diverging oscillations? Or could it be due to mesh size? Due to the length of the pipe, I am using 2cm cell size (compared to 35cm pipe diameter), before my computer blows up having too many cells.

I have been using constant density for now, using the average value of the inlet and outlet, with no issues. However, using constant density means getting velocity results which are (fairly) constant too? I am getting about 18-20 m/s throughout the pipe, from inlet to outlet.

So how you specify boundary conditions?

[LuckyTran]

I've had similar convergence problems on very long domain when using ideal gas / real gas equation of state. For large domains you need a very good initial guess or the solution can diverge rather easily. For small domains, the initial conditions get washed out pretty fast but for large domains they can persist for many iterations and cause problems.

The fix for me was to play with the under-relaxation factors and reduce them just enough to get a stable simulation and slowly increase them to default values. I usually started with the energy equation, as properties tend to depend more on temperature than pressure. But, for large domains the pressure can also have a significant influence. Is your density change dominated by temperature or pressure changes? If pressure effects are also significant, I had better experience using the coupled p-v solver rather than SIMPLE or PISO, but still with reduced urf's for initial iterations.

[Antanas]

Btw you can set incompressible ideal gas low for density in material prop.