[Julian121]

I am trying to simulate a multistage axial compressor.
A periodic sector of IGV, Rotor and Stator in addition to an intake has been considered.
I have used bladegen to design the blade rows.
After importing them from Turbogrid to CFX-Pre, the blade rows are placed in different angular positions, although the extent of angular sectors (theta) are the same at least for two domains.
My questions are:
Why does CFX-Pre place them in different angular positions?
Does it affect the result?

[Julian121]

Let me explain more:
I need two parts (upstream of rotor and rotor) be united into one part.
I used Bladegen to design the rotor.
The upstream part of rotor has been created by Solidworks.
The rotor has been meshed by Turbogrid, while I used ANSYS Meshing to mesh the upstream part.
The upstream part of the rotor has less height than the rotor domain, because a type of casing treatment is placed on the top of this part.
How can I merge these two domains?

Originally Posted by Julian121

I am trying to simulate a multistage axial compressor.
A periodic sector of IGV, Rotor and Stator in addition to an intake has been considered.
I have used bladegen to design the blade rows.
After importing them from Turbogrid to CFX-Pre, the blade rows are placed in different angular positions, although the extent of angular sectors (theta) are the same at least for two domains.
My questions are:
Why does CFX-Pre place them in different angular positions?
Does it affect the result?

[AtoHM]

Hi, if I understand correctly, you asked two questions: Whether the displacement will make a difference and how to merge the parts.

The first one, I think does not matter. Is it a steady state or transient simulation? What kind of interface are you using between stationary and rotating domains? To the left in CFX you can also apply a transformation (see Turbo Rotation). Use RMB on the mesh part you want to transform, hit "transform mesh" and choose Turbo Rotation. You can now rotate it to your desired pitch angle.

The second one: Take the meshes as you have, apply the same rotation speed and direction to the upstream part of the rotor as you used in the rotor domain. Because of the different mesh types, you will need a GGI interface between the rotor and the additional part. But, and thats the point, you won't have a frame change and therefore no mixing plane or something. This should effectively "merge" the parts for the simulation. Dont forget to adjust the wall options for counter-rotating wall and rotating.

[Julian121]

Thank you for your answer. I really appreciate it.
I have used a mixing plane interface between the IGV and the Rotor and the Rotor and the Stator.
I used frozen rotor between the casing treatment and the rotor and the upstream of the rotor and the casing treatment.
For the interface between the casing treatment and the rotor, should the shroud be divided into two parts to define the interface between the casting treatment and the rotor and another counter rotating shroud boundary? (Image 1)
I also need to define another interface between the casing treatment and the rotor. Similarly, should the inlet of the rotor be divided into two parts to define two boundaries? (Image 2). Should Design modeler be used to split the boundary for this purpose?
Is it clear how much pitch angle is used in CFX? Does bladegen show this angle? In my case it seems that the rotor domain has rotated -2.1 degrees, so I rotated the upstream part this angle but I do not know where this angle comes from!

Originally Posted by AtoHM

Hi, if I understand correctly, you asked two questions: Whether the displacement will make a difference and how to merge the parts.

The first one, I think does not matter. Is it a steady state or transient simulation? What kind of interface are you using between stationary and rotating domains? To the left in CFX you can also apply a transformation (see Turbo Rotation). Use RMB on the mesh part you want to transform, hit "transform mesh" and choose Turbo Rotation. You can now rotate it to your desired pitch angle.

The second one: Take the meshes as you have, apply the same rotation speed and direction to the upstream part of the rotor as you used in the rotor domain. Because of the different mesh types, you will need a GGI interface between the rotor and the additional part. But, and thats the point, you won't have a frame change and therefore no mixing plane or something. This should effectively "merge" the parts for the simulation. Dont forget to adjust the wall options for counter-rotating wall and rotating.

[Julian121]

I mean as the casing treatment does not fully cover the rotor shroud, does CFX need an interface between the casing treatment and the rotor and a counter rotating shroud boundary condition for the part that has no contact with the casing treatment or just an interface is enough?

[AtoHM]

No I think you got it right. Physically correct should be an interface where treatment and rotor are in contact. The other part however is the shroud and hence a rigid wall so needs an appropriate bc. The interface is only there to enable change of reference systems or different mesh resolutions on both sides.

[Julian121]

I am faced with convergence difficulty where the momentum residuals oscillate after 20 time steps.
I have followed the convergence suggestion in order to find the location where the maximum residual exist. I used CFD-post and the location of maximum residual lies close to the suction side of the stator near to the hub. Also, in this location Mach number reaches zero.
The simulation is done in the design mass flow where no flow separation is expected.
I have checked the mesh in this area and it seems that mesh is fine.
My question is: does the flow separation in this area is responsible for the increased momentum residuals which prevents the convergence criteria to be met? And how can this be fixed?

[Julian121]

Originally Posted by AtoHM

No I think you got it right. Physically correct should be an interface where treatment and rotor are in contact. The other part however is the shroud and hence a rigid wall so needs an appropriate bc. The interface is only there to enable change of reference systems or different mesh resolutions on both sides.

Thank you for the answer.
I do not know how the shroud surface should be divided in order to apply two different BCs.
I tried to export the rotor and the casing treatment mesh to ICEM to find the intersection and apply different parts but it seems ICEM cannot be used for this purpose.
Should the geometry be exported to Solidworks to divide the shroud to two parts?