[Sumanth_094]

Hello everyone,


I am doing a steady state simulation for a S-Duct. The duct has a intake lip which resembles the shape of lower half of NACA0006 airfoil and I swept it across my throat of the duct thereby producing an inlet at the leading edge. During meshing, there were some problems with few elements because of the acute angles between the leading edge and inlet surface and I had to pad the inlet by 15mm in order to avoid the meshing problems. This pad gave rise to new surfaces and moved my inlet also by 15mm.



The problem I am facing right now is that the boundary conditions are not gonna be same as what it would have been without the extra pad(extension). I want to use Total pressure at inlet and static pressure at outlet with temperatures being specified at both of them. Since I have these new surfaces because of the new padding, I thought of defining them as openings by defining opening pressures and temperatures but how would i treat my new displaced Inlet now? Is it gonna be having the same pressure BCs as before? I have tried using that and also defining that as an opening and both the simulations failed. The error was "overflow".
I have considered using smaller time steps. I use 0.001 and my model was also SST. Can someone please drop some light onto this topic?


Regards,
Sumanth

[ghorrocks]

Be careful to not ask an XY problem (http://xyproblem.info/). It appears the real problem is you can't mesh the real geometry. You then did some padding and other things and then had convergence problems. So you question should be "how do I mesh this geometry", not convergence problems on your artificial geometry.

Please post an image of what you are trying to mesh and why you cannot mesh it.

[Sumanth_094]

Hello,


Thanks a lot for your reply. I understand what you mean. So, my real problem here was meshing as you said. Without my work around of padding, I have problems as shown in the attachements. The figure 'MODEL' is the model I need to mesh.

The figure 'INLET_mesh' is the mesh problem I have. There are elements missing along the curve and this distorts everything. The image is without prism layers but I also need to create prism layers. When I try doing that the layers are distorted in the place where there are elements missing.


I had looked up how to overcome that problem.
Till now, I tried changing my Triangulation tolerance and also my edge criterion. Have lowered them as much as possible and have tried a series of numbers for those two. Changing edge criterion to 0.02 from 0.2 just decreased the element size but there were still elements missing. Triangulation tolerance is set to 0.0001.
My geometry was repaired to the best tolerance possible.
Can you tell me where I am going wrong?


Regards,
Sumanth

[ghorrocks]

It appears you are using ICEM. In ICEM you have to put curves on edges you want the mesh to conform to. You also have to put points on points you want the mesh to conform to. Have you put curves and points on it?

[Sumanth_094]

Hello,

Yes. I am using ICEM and I do have curves at the point where the mesh is not conforming. Regarding the creation of points, I do not know where to create the points on the Curve because the location where it doesn't conform and the number of such locations varies with seed size.

I have also tried defining thin cuts but since I am new to that aspect of meshing, I wasn't sure if I was doing it right. I had that curve as a seperate part and I defined thin cuts between the surfaces which share that curve. Even that didn't seem to help.

One more problem here was, I was trying to compare the results of different airfoil profiles for the duct Inlet and I was considering using the same seeding for all models so as to have a relatively fair comparision. So, changing seed sizes from model to model wouldnt be apt for my case here.

Thanks a lot in advance. I really appreciate you taking some time to help me out.

[Gert-Jan]

- It should not be a problem to get this meshed in ICEM. But what the real problem is, can be difficult to find out. Did you generate prisms elements (inflation)?
- Using the right mouse button, you can inspect the curves. For example, you can colour them by number or connections and make them thick to make more clear. What do you see? Are the curves continuous. or do they appear dashed?
- The only important curves, are those on your inlet (banana) and outlet (circle). All other curves (and points) are basically obsolete. What if you delete all these obsolete objects?

[ghorrocks]

If you set ICEM up correctly it should not do that, but it has been a while since I used ICEM so cannot tell you what you need to change to fix it. But I can tell you how to fix it if it does occur. ICEM has extensive mesh editting capability. You can either move the nodes of the non-conforming bits to the curve, or you can split the edge into two edges and move the central new node to the curve or many other mesh editting functions. You can also change the boundary face element faces are assigned to if it has goofed that up. And once you have editted the mesh to fix these problems you can smooth it and generate prism layers and so on.

[Sumanth_094]

The curves are smooth and after build topology, the curves are red and there seems to be no problem with that. I have a very acute angle between the inlet surface and the walls and may be that's the reason, its getting distorted but your inputs have been valuable and I will try checking and redoing those steps again. If it still doesn't work, then I would have to manually sit n edit the nodes which isn't the best approach because I have 10 other geometries with a similar problem.

Thanks a lot.

Regards,
Sumanth