CAD model for a rotating body

Jonathan Eriksson

Hi, i'm very new to simulations in general. I have done the "ACU-T: 5000 Centrifugal Airblower"-tutorial. I don't undestand how they have created the CAD geometry for the impeller/rotating body. How do you create an appropriate CAD model of a rotating impeller for HW CFD? It appears you need the impeller and some kind of bigger approximating body. Are these the same body? Are one of them surfaces?

Regards,

Jonathan

acupro

There are two typical approaches for rotating bodies, like a blower/fan/pump/etc.

The first would make use of the reference frame approach (or frozen rotor as some solvers call it).  This adds some terms to the equations to approximate centrifugal, coriolis, and angular acceleration body forces in the fluid.  In most cases, this is not applied to the entire domain, but instead to some portion of the fluid volume around the rotating body.  So, yes, this will be a second volume in the domain, for the reference frame - typically created in your CAD package.  This would be some axisymmetric volume around the rotating body itself.  You will generally see differences in solution (sometimes minor, sometimes major) depending on the extent of that volume to which you apply the reference frame.  The benefit to this approach is you typically run this as steady-state, since the mesh is not moving, making the compute runtime quite short.

The second method is to actually rotate that same volume inside the fixed volume.  This time you're actually rotating the mesh - as in ACU-T: 5001.  This will now be a transient calculation, with mesh rotation - thus will take longer to run, as you may need 6 to 10 full revolutions to get a stable solution.

acupro

There are two typical approaches for rotating bodies, like a blower/fan/pump/etc.

The first would make use of the reference frame approach (or frozen rotor as some solvers call it).  This adds some terms to the equations to approximate centrifugal, coriolis, and angular acceleration body forces in the fluid.  In most cases, this is not applied to the entire domain, but instead to some portion of the fluid volume around the rotating body.  So, yes, this will be a second volume in the domain, for the reference frame - typically created in your CAD package.  This would be some axisymmetric volume around the rotating body itself.  You will generally see differences in solution (sometimes minor, sometimes major) depending on the extent of that volume to which you apply the reference frame.  The benefit to this approach is you typically run this as steady-state, since the mesh is not moving, making the compute runtime quite short.

The second method is to actually rotate that same volume inside the fixed volume.  This time you're actually rotating the mesh - as in ACU-T: 5001.  This will now be a transient calculation, with mesh rotation - thus will take longer to run, as you may need 6 to 10 full revolutions to get a stable solution.

Jonathan Eriksson

acupro_21778 said:

There are two typical approaches for rotating bodies, like a blower/fan/pump/etc.

The first would make use of the reference frame approach (or frozen rotor as some solvers call it).  This adds some terms to the equations to approximate centrifugal, coriolis, and angular acceleration body forces in the fluid.  In most cases, this is not applied to the entire domain, but instead to some portion of the fluid volume around the rotating body.  So, yes, this will be a second volume in the domain, for the reference frame - typically created in your CAD package.  This would be some axisymmetric volume around the rotating body itself.  You will generally see differences in solution (sometimes minor, sometimes major) depending on the extent of that volume to which you apply the reference frame.  The benefit to this approach is you typically run this as steady-state, since the mesh is not moving, making the compute runtime quite short.

The second method is to actually rotate that same volume inside the fixed volume.  This time you're actually rotating the mesh - as in ACU-T: 5001.  This will now be a transient calculation, with mesh rotation - thus will take longer to run, as you may need 6 to 10 full revolutions to get a stable solution.

Thank you! Are there any general rules when generating the rotating axisymmetric volume around the impeller to get good results?

acupro

Jonathan Eriksson said:

Thank you! Are there any general rules when generating the rotating axisymmetric volume around the impeller to get good results?

There will be some impact of the size/extents of the reference frame volume on the results - so it's impossible to say what is best for a particular case.  You probably want at least 4 to 6 elements between the reference frame 'boundary' and any other wall.  Sometimes it works to split the difference between the rotor and the fixed wall/housing.

Note that the reference frame approximation can give good comparison in some cases, but if there are geometry features that are not symmetric, as even in the centrifugal fan, squirrel fan, other support structure, etc, the results can be quite far off.  This is due to missing the blade passing effects.