CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Dynamic mesh questions plus what the frack is up with named selections (http://www.cfd-online.com/Forums/fluent/115048-dynamic-mesh-questions-plus-what-frack-up-named-selections.html)

Maurosso March 22, 2013 09:33

Dynamic mesh questions plus what the frack is up with named selections
 
Hello,

First I'd like to get it off my chest that fluent in cooperation with workbench and system coupling is driving me NUTS!

Now...to specify the problem(s).

My main issue is that I just can't figure out (and I've ran through the help numerous times) how declaring dynamic mesh surfaces and zones works.

I'm working on a 3d, transient, FSI problem using system coupling. At this stage, the flow is laminar, one-phase, non-compressing. Instead of a inlet boundary condition, I'm using a moving bar defined in Mechanical. With the use of FSI surface, the displacement of the bar is translated onto fluid, squeezing it out (see pic1).

The multi-mesh is due to my learning of creating interfaces between different cell zones. Plus, further down the road I want to use different smoothing and remeshing options in different cell zones. But to do that, I really need to understand how the frack do i choose zones for dynamic meshing.

I've completed the system coupling FSI tutorial with the oscilating plate, and done my own simulation with diffusion based smoothing (to be viewed here: http://goo.gl/S8gX6 <-youtube link)

But I still don't grasp the concept, and Ansys Help...isn't helping.
Here is the onslaught of questions:

1. Should you ever pick the interior of the fluid as a deforming zone? Why not? Shouldn't it just work that once I choose my zone as a whole...it simply applies my chosen method of deformation (in this case, only smoothing).

2. If you choose a surface for deformation, that is planar, do you have to specify that it is a plane, with the normal and point on plane coordinates? I just don't understand this...the surface is part of a named selection, it is defined, it's should be obvious for the solver that that is the case. Isn't faceted enough in this case?

3. Pics 2, and 3, are the result of what happens most of the times in my simulation. Only the cell layer touching the FSI interface is deforming. This result is the effect of spring based smoothing. Diffusion based method simply won't start, and will crash the simulation on the first fluid iteration (WHY?! It worked earlier...).

4. To clarify no. 3-> Then defining the deforming zones, I did the following. I have two named selections: "wall", and "symmetry". "Wall" named selection contains the two narrow faces, the "symmetry" named selection contains the two large front and back faces. Do I need to create 4 separate surfaces, with different names, and declare that they are planes and additionally define that their normals and lay-on point? What if my surface is irregular? That is usage of the "faceted" option?

5. There are two dialog boxes in dynamic meshing. The "Mesh method settings" where one can pick the methods etc., and the "Dynamic mesh zones" box. In both you can specify the parameters of the cell zone. It is specified in the help, that the options from Dynamic mesh zones, will override the ones from mesh method settings dialog...So why the frack are they even there? Or is it that if the default values in the Dynamic mesh zones are left unchanged, that the Mesh method setting are the one in command?

That is one thing...the other is the ridiculous crashing of fluent when I change the name or structure of my "Named selections". It's like...that the crap, I set up half of my simulation, every thing is almost ready, when I realize I need one more named surface. When I go back to Ansys Meshing, and create a new Named selection, or rename an old one (because of let's say a typo that just bothers me), and try to read the mesh again...the whole thing crashes because the id numbers don't match...how stupid is that? I could even live with it clearing all the settings, and allowing me to re do them...But NO! It just crashes and I need to get out of fluent, clear all the settings, from workbench, and then start from scratch. It just infuriates me...

tl;dr-> What is the secret behind picking walls, surfaces, zones for dynamic mesh smoothing and remeshing?

Pic1:
http://i.imgur.com/DHnphlF.jpg

Pic2:
http://i.imgur.com/nEK6MkH.jpg

Pic3:
http://i.imgur.com/zNA9YcK.jpg

Help?

Maurosso March 24, 2013 07:17

anyone? pls....

vasava March 25, 2013 05:59

I have tried to understand and answer your questions. Let’s see what we can do here.

First of all for conjugate heat transfer and for one way Fluid-Stricture interaction (where motion of wall affects the fluid motion) does not necessarily need the use of structural module. If you specify the mesh motion via dynamic mesh or ‘mesh motion’ in Cell Zone Condition that should suffice.
1. Should you ever pick the interior of the fluid as a deforming zone? (No, the deformation of a domain is governed by deformation of its boundaries.). Shouldn't it just work that once I choose my zone as a whole...it simply applies my chosen method of deformation (As far as mesh motion or deformation is concerned fluent does not do anything automatically. You have to be very specific about what you input. And if your inputs are not viable fluent will not progress.)

2. If you choose a surface for deformation, that is planar, do you have to specify that it is a plane, with the normal and point on plane coordinates? I just don't understand this...the surface is part of a named selection, it is defined, it's should be obvious for the solver that that is the case. Isn't faceted enough in this case? (By default every surface unless declared is wall for fluent. Of course if you forget to preprocess your boundaries you can always change it in fluent.)

3. Pics 2, and 3, are the result of what happens most of the times in my simulation. Only the cell layer touching the FSI interface is deforming. This result is the effect of spring based smoothing. Diffusion based method simply won't start, and will crash the simulation on the first fluid iteration (WHY?! It worked earlier...). Implementing mesh deformation or dynamic mesh sometimes requires specifying conditions for neighboring walls/faces/boundaries also especially the ones which are directly affected by motion of the mesh. This shows how specific you have to be with fluent.

4. To clarify no. 3-> Then defining the deforming zones, I did the following. I have two named selections: "wall", and "symmetry". "Wall" named selection contains the two narrow faces, the "symmetry" named selection contains the two large front and back faces. Do I need to create 4 separate surfaces, with different names, and declare that they are planes and additionally define that their normals and lay-on point? (If you have only one fluid you need not make separate zones and thus you need not make separate walls.) What if my surface is irregular? (whatever shape it may be, I would not make more surface and confuse myself. Especially in the scenario when there is so much going on at the same time).

5. The dynamic mesh options are useful when mesh within the domain is modifying with each time step or iteration. Here you could specify the exact mesh motion and control other parameters in order to obtain proper mesh motion and avoid negative mesh volumes. The mesh motion/ frame motion specified in ‘cell zone condition’ is different than the dynamic mesh. If you specify both of them one has to be overridden by the other.

The feature of fluent where you could read different mesh by making minimum modification to your current case is very useful. However (in my opinion) this feature is useful only in cases where you make minimal changes to your mesh. For example you want to test flow inside a pipe and want to test different diameter cases. Here you only modify the size of the domain and not the name of the faces/boundaries. Some may like and some may not. Despite this feature it is strongly advisable to check and double check what you want from your mesh and case.


All times are GMT -4. The time now is 17:45.