[oozcan]

Hi all,

I am working on 3d heat exchanger.

I have just completed mesh and I have a warning like inflation created some stairtep mesh at some locations

What does it mean and how can i resolve it?

Kind Regards,

[scipy]

It means that you tried to grow prisms on two walls that are facing each other or are merging into a tight space/corner and prism layers from each wall started running into each other. ANSYS Meshing will first try to "shrink" these prism layers (up to a specified maximum aspect ratio, default might be 50 I think, I usually go to 200) by compressing them down, once it can't compress any more and they're still running into each other - it's going to start removing the top layer of prism elements, first 1, then 2, then 3 etc. resulting in something that looks like stair-steps. Since picture is worth a thousand words:



Green elements are actually new tetraheadrals which transition from the side of the prism to the rest of the volume. But as you can see, they are all of quite terrible quality (high aspect ratio, high skewness). It's probably best to set up a "gap factor" so prisms stop growing in places where collisions are unavoidable (there you'd have a non-conformal interface between the sides of the prisms and tetras, nothing to worry about - it's normal). However, I don't know if you have any control over this in ANSYS Meshing. Therefor, you would probably have better luck if you just used AM to generate a decent surface mesh of pure triangles and then move on to Fluent Meshing (ex TGrid) where you could play with gap factors, max. aspect ratio and other fine things.

[oozcan]

Originally Posted by scipy

It means that you tried to grow prisms on two walls that are facing each other or are merging into a tight space/corner and prism layers from each wall started running into each other. ANSYS Meshing will first try to "shrink" these prism layers (up to a specified maximum aspect ratio, default might be 50 I think, I usually go to 200) by compressing them down, once it can't compress any more and they're still running into each other - it's going to start removing the top layer of prism elements, first 1, then 2, then 3 etc. resulting in something that looks like stair-steps. Since picture is worth a thousand words:



Green elements are actually new tetraheadrals which transition from the side of the prism to the rest of the volume. But as you can see, they are all of quite terrible quality (high aspect ratio, high skewness). It's probably best to set up a "gap factor" so prisms stop growing in places where collisions are unavoidable (there you'd have a non-conformal interface between the sides of the prisms and tetras, nothing to worry about - it's normal). However, I don't know if you have any control over this in ANSYS Meshing. Therefor, you would probably have better luck if you just used AM to generate a decent surface mesh of pure triangles and then move on to Fluent Meshing (ex TGrid) where you could play with gap factors, max. aspect ratio and other fine things.

First of all thank you for all,

First I have carried out inflation and mapped face meshing to provide two contacted surface will be proper and proper interface.( is it enough?)

Second one I have put it attached photo maximum aspect ratio.As you can see max 210.23 and in the other photo I have uploaded I have changed manually 200 in x axis ? is it right?

By the way max skewness is about 0.96 and I haven't decreased it much.

Last one is I dont know what T-Grid is and I dont have enough time to learn about it because I need to give my dissertation as soon as possible.

Kind Regards,

[scipy]

Since you did not post any screenshots of your geometry and I don't know if you're dealing with both a solid and a fluid part of the heat exchanger - I can't really understand or answer your first question regarding mapped face meshing and inflations.

Second part you're misunderstanding what's going on. You are looking at the mesh metrics (quality report) and the things that you're changing are just min/max values that a graph will show, not the actual elements. You don't really have control over the things that I listed (that's why I recommended going with Fluent Meshing).

[oozcan]

Originally Posted by scipy

Since you did not post any screenshots of your geometry and I don't know if you're dealing with both a solid and a fluid part of the heat exchanger - I can't really understand or answer your first question regarding mapped face meshing and inflations.

Second part you're misunderstanding what's going on. You are looking at the mesh metrics (quality report) and the things that you're changing are just min/max values that a graph will show, not the actual elements. You don't really have control over the things that I listed (that's why I recommended going with Fluent Meshing).

actually I am dealing with fluid-fluid .Coupled wall has been defined by two fluid domain and I havent given any wall thickness for heat conduction.

you can see the mesh in attached.and air flow through in semi-tubes and the other part,that is, outside of tubes are external fluid flow domain.Therefore I havent used sweep and multizone as this project has many tubes.Instead I have just used inflation and all circular wall parts are done mapped face meshing to have common faces ( that is interface, and not to have problem named face-handedness)

I have decreased max aspect ratio about 198 , 200 but this time skewness has gotten a little bit more value.I couldnt adjust both values.

Anyway,that is last situation attached. you can see the max skewness and max aspect ratio?

how can I fix this problem? you can see max aspect ratio 198 bla bla bla.

Kind Regards,

[scipy]

If this is your geometry then I don't understand why you're not using sweep.. It seems ideal for this purpose, it will save you a lot of elements and increase mesh quality drastically.

However, even with your current mesh you could try doing a simulation run. Fluent can handle max. skewness even up to 0.98-0.99 if it's only a small number of elements in the domain. However, your average skew of 0.5 says that your mesh is of rather poor quality in general. If I were you I'd avoid all this mapped face meshing when using tetra. As far as interfaces go, they can be non-conformal.. solver will not care much.

[oozcan]

Originally Posted by scipy

If this is your geometry then I don't understand why you're not using sweep.. It seems ideal for this purpose, it will save you a lot of elements and increase mesh quality drastically.

However, even with your current mesh you could try doing a simulation run. Fluent can handle max. skewness even up to 0.98-0.99 if it's only a small number of elements in the domain. However, your average skew of 0.5 says that your mesh is of rather poor quality in general. If I were you I'd avoid all this mapped face meshing when using tetra. As far as interfaces go, they can be non-conformal.. solver will not care much.

Hi,

if I havent done mapped face meshing in circular wall faces , those faces are tetra and inflation meshes would have been gaps or overlapping that's why I used mapped face.

As for sweep meshing,


I have selected only just one face in Scr/Trg ,however, I need to select all wall faces Which one is proper to selection ( automatic,manual source, manual source and target, automatic thin, manual thin)? and how can I use it ?

Actually I couldnt know how I use it maybe I couldnt make it right ,

As for Inflation mesh, I do not know where I define inflation whether or not inflation in sizing box or inflation in mesh control.

Moreover two body parts have different form each other boundary layer thickness and layer .if I use sweep method how can I enter two different boundary layer thickness?

Kind Regards,

[scipy]

I'm going to stop replying now because it's being counterproductive. All of this information is contained within the: ANSYS pdf documentation, several ANSYS channels on YouTube (official ones contain both multizone and sweep meshing) and this forum. It seems that me being willing to help you here and there is being interpreted as "teach me everything so that I don't have to do anything on my own".

For interfaces to work there does not have to be node-to-node conformality between the two meshes, but again.. if you went to the Fluent user's guide or tutorial guide and just searched for non-conformal interfaces, you would've found all of this already.

Good luck with your project.

[oozcan]

Originally Posted by scipy

I'm going to stop replying now because it's being counterproductive. All of this information is contained within the: ANSYS pdf documentation, several ANSYS channels on YouTube (official ones contain both multizone and sweep meshing) and this forum. It seems that me being willing to help you here and there is being interpreted as "teach me everything so that I don't have to do anything on my own".

For interfaces to work there does not have to be node-to-node conformality between the two meshes, but again.. if you went to the Fluent user's guide or tutorial guide and just searched for non-conformal interfaces, you would've found all of this already.

Good luck with your project.

Thank you for your replying,

Before you wrote this post , I have already tried to do multizone mesh.On the other hand first mesh I have shown you is still being analyzed in FLUENT in different URFs to avoid some vortex.anyway

Thank you for your help thus far,

Kind Regards,