Learning Forum

[helen.durand]

Good morning We have been working on a system where there are several T-junctions. However, the simulations show that we are not capturing the mixing behavior with the current case, where we have an "O-Grid" type of mesh. We ran some tests to see if either the geometry or mesh could be the issue. In particular, we tested only one T-junction using the same mesh size and boundary layers as the one presenting the problematic behavior. The simulation shows that the mixing behavior is captured in the T-junction. We also ran a test in which we do not have an "O-Grid" method at each pipe using the ANSYS Fluent default mesh generation (without any boundary layers) and the results indicate that we are capturing the mixing behavior. The geometry and mesh do not seem to be the cause of the issue based on the tests above. Do you have any thoughts on why this may be happening? Would the connections/contact regions play an important role if we use an "O-Grid" method at each pipe? We used shared topology for conformal geometry and removed the connections under "connections" in ANSYS Fluent meshing to define boundary conditions at different surfaces of the geometry.
In advance, thank you very much for your support Research Group Discovery

[DrAmine]

Do both meshes have same mesh quality metrics like for aspect ration, orthogonal quality, skewnes and cell size ratio? How are you assessing the mixing?

[DrAmine]

And how are you creating the Ogrid mesh?

[helen.durand]

Thank you, DrAmine. They have quality metrics in the same range from acceptable to excellent in terms of orthogonal quality and skewness. The aspect ratio ranges from 1.16 to 9.38 (most of the elements are closer to 1.16).
We are not using an exact "O-Grid" mesh. We created a "square block" in the center of each pipe to have more control of the mesh (see attached figure "CurrentMesh.png").
We are assessing the mixing by using the CFD-post results and looking at whether the temperature towards the end of T-junctions is more uniform (the difference between the maximum and minimum temperature in the radial direction should be very close to each other). The expected mixing behavior is shown in figure "ANSYSExpectedMixingBehavior.png". However, for the problematic case, we have what is shown in figure "ANSYSNotExpectedMixingBehavior.png". We are not sure why we cannot capture the mixing behavior in the last case.
.

[DrAmine]

Please insert instead of attaching the pictures.

[helen.durand]

Please, find below the pictures for reference.
CurrentMesh.png:
ANSYSExpectedMixingBehavior.png:

ANSYSNotExpectedMixingBehavior.png:

[helen.durand]

For some reason, the other two figures are not being displayed. From the figure with the mesh and the description above, would you be able to provide any help?

[DrAmine]

Okay for this simple geometry the mesh does not look as one expects for an "o-grid" type of mesh (other would say a bad looking mesh). You said you removed all contacts etc. can you please verify in Fluent if you have any interfaces.

[helen.durand]

Thank you, DrAmine. In ANSYS Fluent setup, we do not have any interfaces. The reason why we removed the connections/contacts in ANSYS meshing is that we want to add boundary conditions on all inlet and outlet surfaces of the pipes. The mesh is conformal as we used share topology.

[DrAmine]

Again the mesh should be enhanced. Stick to uniform mesh if you cannot create a decent ogrid. Xou have large cell size change from BL to core and thatcis baddctoo.