CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   enGrid (https://www.cfd-online.com/Forums/engrid/)
-   -   Prismatic boundary layer (https://www.cfd-online.com/Forums/engrid/152610-prismatic-boundary-layer.html)

KateEisenhower May 5, 2015 05:16

Prismatic boundary layer
 
Hello everybody,

I am struggling with getting a proper prismatic layer in enGrid. First of all, not every option in the create prismatic boundary layer window is clear to me:

Improve existing layer: Clear. I check it every time I want to improve an existing layer. If I didn't create one before or if I want to create one from scratch, I uncheck it.

Improve points: This option relocates the points on adjacent boundary codes which do not get a prismatic layer in order to achieve the correct prismatic layer thickness, right?

Use safe mode: What does this mean?

Strict height limiter: This option limits the height of the prismatic layer strictly. My studies showed that this doesn't work, especially where the layer meets an adjacent boundary code without a prismatic layer.

Use absolute heigth: Here you can switch between defining the prismatic layer height absolute or relative to the edge length of the underlying surface cells (see below). In which use cases would it make sense to have a prismatic layer thickness proportional to the edge length of the underlying cells? -> I always have this checked.

Number of iterations: I ran 99 iterations (maximum) and it didn't show a significant improvement.

Relative height of first cell: As stated above, I can't see a use case for this. This doesn't mean there isn't any, it is just not relevant to me.

Absolute height of first cell: I enter here the expected height of my boundary layer (in my case around one inch). I am planning to divide this layer lateron.

Stretching factor (hint): Is this the ratio between two consecutive cells in the prismatic layer normal to the surface? I don't understand this since you can only create one layer here. What does hint mean?

Ratio last layer/far-field: What is the meaning of this? I thought you were building first the prismatic layer and then the rest of the volume mesh. This means I don't have any far-field right now, hence no ratio.

Number of height relaxations: I don't get big oscillations in height, so I guess the default value for this is okay.

Number of normal vector relaxations As with the height, I don't seem to get unstable solutions, so I guess the default value is fine. Any comments on the two relaxation factors are welcome. Is it worth playing around with these?

After several days of trying I only managed to get a prismatic boundary layer which adjusts to the edge length of the underlying net. It seems to me, that the absolute height option just doesn't work.

Has anyone managed to get a decent prismatic layer with the same height all over the surface with enGrid?

Best regards,

Kate

edit:
I am adding some information (and my follow-up questions) here, which I found on another topic (http://www.cfd-online.com/Forums/eng...er-height.html).

Taxalian complains, that changing the boundary layer parameters doesn't seem to have any effect. The interesting part of this thread is the respond of Oliver Gloth, who seems to be the head developer of enGrid. He states
- "that one cannot specify the total height of the boundary layer mesh, because it is a feature which is not required." What does the absolute height of first cell option specify then?
- "the farfield ratio leads to a decent transition to the far field mesh" What does this have to do with the creation of the prismatic layer? Shouldn't this layer be specified when creating the volume mesh?
- "Once you have an isotropic mesh (appr. ratio of 0.3-0.8) you don't need to use prisms." Is this supposed to be a general statement?
- "An important parameter, however, is the weighting between relative and absolute height of the first cell. Relative size means the height is a fraction of the edge length on the surface and absolute is absolute." My research showed so far, that the absolute height option gets you a rather bad quality prismatic layer. The entered values don't seem to have an impact at all. Now I am playing around with the relative height option. This gets me a layer with beautiful constant height. Downsides are that I can't specify how high it is and that it collapses at the boundary with another patch (which also gets a boundary layer).
- "The next release will possibly have the option to prescribe the initial height in the same manner that the surface resolution is prescribed at the moment." Is this release out already? What is the initial height?

Again, any comments are welcome.

Best regards,

Kate

KateEisenhower May 18, 2015 07:51

1 Attachment(s)
Hello again,

I attatched a very easy case of two cubes which act as boundary for a fluid domain. The prismatic layer is already created. It does not matter if I choose absolute or relative height for the boundary layer, under no circumstances am I able to adjust the height of the prismatic layer.

Feel free to try it yourself. Has anyone an idea what is going wrong here?

Best regards,

Kate

Chati14 September 15, 2015 05:44

Hi KateEisenhower :)
Have you done any progress controling the boundary layer creation?

I'm facinng the same problems as you, I want to control better the boundary layer formation, to not have problem dividing the layer afterwards.
I don't know neither what the options do or if they even do anything (for me, fx, using absolute height create the same boundary layer as with relative height :confused: )

Have you tried the new version? EnGrid 1.5

I hope we can help each other

Best Regards,
Mario

KateEisenhower September 15, 2015 05:57

Hello Mario,

unfortunately not, I've given up on this. I experienced that enGrid is not capable of creating a decent prismatic layer with not-so-low height on a curved surface because it adjusts the prismatic layer height to the curvature or cell size on the surface or both. In my opinion prismatic layer height should depend on the size of the boundary layer, which is not necessarily linked to the curvature and not at all to the cell size at the surface.
Anyone, please correct me if I'm wrong. I think enGrid is a great tool which is kind of easy to use. But for applications like this I think it is not suitable.

I just noticed that my example only contained plane surfaces. If it is the same in your application you should try to adjust the height by changing the cell size on the surface.

I don't know of any version 1.5:
http://sourceforge.net/projects/engrid/files/engrid/

Best regards,

Kate

Chati14 September 15, 2015 07:41

Hi KateEisenhower,
Thank you for your coments and hints.

I don't know if it's called 1.5, but there a "new" master branch that has changes in the boundary layer operations.
http://www.cfd-online.com/Forums/eng...w-release.html
https://github.com/enGits/engrid/issues/52

My case is a cylindrical pipe with a disturber inside. Like you said, it seems that the prismatic layer height is adapted to the curvature or cell size on the surface mesh :(

Best Regards,
Mario

KateEisenhower September 15, 2015 07:48

Yes I even posted there. I remember now. Finally I didn't try this new enGrid version because I switched to gmsh for meshing.

Have a nice day,

Kate


All times are GMT -4. The time now is 07:06.