[tareqkh]

Hello All,

I am just working on a case where I have to perform simulations in symmetry, axiymmetric and full 3D grids. When I performed the first simulations using wedge boundary condition i.e. axisymmetric and then compared with the symmetry, the results were not the same. I am expecting to get the results with a tiny deviation but am not getting what it supposed to be. I am using the same boundary conditions the grid was the same. Everything was exactly the same except that I am extruding 5 degree in the axismmetric where as 90 degree for the symmetry grid. I don't know what could be the reason of this big deviation. Has anyone experienced such an issue before?

Any suggestion is appreciated.

Best,

[tareqkh]

Hello All,

I have been trying to understand the math and assumptions behind the wedge boundary condition that is implemented in OpenFoam to simulate axisymmetric flow usually expressed in cylindrical coordinates. I understand that they are employed since the NS equations implemented in OpenFoam are in Cartesian coordinates. Is there good documentation why this trick works?

I have verified results obtained using the wedge BC versus a full 3D simulation of an axisymmetric CD nozzle and as hoped, I got pretty decent agreement at a much lower computational cost. Now I am interested in implementing volume forces (for an actuator disk model) but I cannot wrap my head around how I need to implement them to play nicely with the wedge boundary condition and an axisymmetric assumption. I know how to compute the Cartesian components of the forces so am I just going to implement those? I would appreciate any insight anyone may have.

Thanks in advance,

[m_ridzon]

I'm probably less experienced than many others here, but I might be able to add some small comment to this statement/question. The 3D Navier-Stokes equations are derived in many college textbooks in Cartesian coordinates. Usually in that same area of the book, they will go further to derive them in cylindrical coordinates. A classic, well known book that I know for sure has them, is Schlichting's "Boundary Layer Theory." I cannot say exactly how OpenFOAM uses them in its source code to complete the computations, since I'm not that well versed on the deeper inner workings of the code. But in terms of understanding the math that defines cylindrical NS equations, check out a fluids book; many have the cylindrical system derived therein.

[BoatsNJos]

Hello!

I am relatively new to OpenFOAM and hoped to find out if anyone knows how the wedge boundary condition works or can point me toward any reading that could help me understand.

I've searched the forums but I haven't yet seen an answer.

I am using OpenFOAM to hopefully simulate axisymmetric electromagnetic cases so understanding what happens at the boundary is very important to me.

Any information anyone could provide would be deeply appreciated.

Joanne

[tareqkh]

I have used wedge BC's in my simulation a lot so setting up a case wouldn't be so hard if that is your question https://cfd.direct/openfoam/user-guide/boundaries/. However, I don't really know how the trick works in terms of math since everything is written in Cartesian coordinates.

Let me know if you need more help,

[BoatsNJos]

HiKhamlaj,

Thanks for your speedy reply! Building the geometry is fine, it's how the BC works is the problem I'm having.

I really appreciate you taking the time to help all the same

[tareqkh]

I have asked the same question a long time ago. However, nobody seems to know how it works!

[wyldckat]

Greetings to all!

I merged several threads into this one, since they were all on the same topic.

I don't have any documented reference about how wedges work in OpenFOAM, but as far as I know, the concept is that the wedge boundary is a special type of cyclic boundary condition.
The difference is that:

• Cyclic boundaries assume that the adjacent cell to the boundary will be part of a domain and not part of the other corresponding cyclic boundary (that's one of the reasons why "empty" exists).
• Wedges act as a one-to-one cyclic boundary that only has to care about a single cell for each wedge-face-pair.

This also means that a wedge can allow rotational flow around the axis, which would explain why a 90 degree cylinder with symmetry boundaries would not give the same results as with the wedge.


However, there are considerable limitations when using wedges: the cells near the axis can have extremely small volumes (OpenFOAM usually only works in 3D, so cell volumes always count), which can result in considerable numerical problems, given that the data is not handled as a 2D case that only accounts for face area.


Furthermore, there have been a few cases that I've seen that the wedge face pairs are too close to each other, that it's not possible to numerically have a significance difference between the centers of those faces. For example, a worst case scenario would be to have a distance smaller than the diameter of a single molecule of water... which is reaaaally small and something that only specialized CFD solvers can handle.

Best regards,
Bruno

[m_ridzon]

Originally Posted by wyldckat

Greetings to all!

I merged several threads into this one, since they were all on the same topic.

I don't have any documented reference about how wedges work in OpenFOAM, but as far as I know, the concept is that the wedge boundary is a special type of cyclic boundary condition.
The difference is that:

• Cyclic boundaries assume that the adjacent cell to the boundary will be part of a domain and not part of the other corresponding cyclic boundary (that's one of the reasons why "empty" exists).
• Wedges act as a one-to-one cyclic boundary that only has to care about a single cell for each wedge-face-pair.

This also means that a wedge can allow rotational flow around the axis, which would explain why a 90 degree cylinder with symmetry boundaries would not give the same results as with the wedge.


However, there are considerable limitations when using wedges: the cells near the axis can have extremely small volumes (OpenFOAM usually only works in 3D, so cell volumes always count), which can result in considerable numerical problems, given that the data is not handled as a 2D case that only accounts for face area.


Furthermore, there have been a few cases that I've seen that the wedge face pairs are too close to each other, that it's not possible to numerically have a significance difference between the centers of those faces. For example, a worst case scenario would be to have a distance smaller than the diameter of a single molecule of water... which is reaaaally small and something that only specialized CFD solvers can handle.

Just FYI...An interesting discussion is here (2D Axisymmetric Model in OpenFOAM 4.0). Therein, it's concluded that the OpenFoam solver requires a 5 degree wedge, but the documentation is sketchy in terms of conveying that detail.

M Ridzon