I did not see anywhere in CFX-Pre to specify incompressible flow. I guess it can be realized by setting the fluid as constant property air (or water,etc.). Does someone know exactly? If the density is included into the simulation equation, will it consume more CPU time? That is another point I want to set the simulation as incompressible flow. Thanks

Dear Dr. Bian,

The ANSYS CFX solver checks if the density is a constant value. If it is constant, it simplifies the equations a bit. There is no mechanism to force an incompressible simulation directly..

constant density => incompressible?

Good luck, Opaque..

Using a constant density fluid will do it. To run incompressible you need variable fluid properties and you require the Total Energy equation, rather than Thermal Energy.

Difference in CPU time will be minimal.

-Robin

Although the CPU time will not increase a lot to solve the RANS including the density, I suspect that this will affect the convergence speed because the density and static pressure are coupled and need to be gradually adjusted.

Dear Dr. Bian,

Now I am confused... if the density is constant, there is no coupling between pressure and density.. The number of equations are still the same, and density is never updated..

ANSYS CFX uses a pressure-velocity based algorithm for all speed flows.. There is no difference if the density is constant, or not..

Would you mind explaining what your concern is about?

Thanks, Opaque..

Hi Dr. Bian,

The variation in density is accounted for in the linearization of the equations. It is updated at the start of each iteration, but will have little effect on convergence. Especially if the flow is low speed. The details of the discretization are included in the documentation.

Compare two run's to see.

Regards, Robin

Let me explain my concern in a little more detail:

If the density is a variable in the equation, it will be updated after each iteration by static pressure and/or temperature. When the updated density is used, it will affect the solution of velocity and pressure. So at least the pressure and density is linked to increase the difficulty for convergence. If the compressibility is big (like air), definitely the solution will be more complicated (the worst is shock).

For my case, the compressiblity is small, so I am thinking the calculation will be faster if I set the case as incompressible flow.

Thanks for your input. Actually, the flow speed of my cases (turbomachinery) sometimes is not small. So it is moderate compressibility (around 10%). Because the case is big, I just want to simplify it and make the simulation run faster.

Robin, I am afraid that Dr. Bian is absolutely correct, both in his reasoning, and in the fact that cases such as his will run much faster incompressibly. I have spent most of the last 4 years running turbomachinery cases in CFX, and can tell you from experience that while the CPU time per iteration may not be hugely less (although it is less), the speed of convergence is typically much faster.