HFSS - Mesh3d error: unable to recover all the surface triangles.

Hi all,

Can anyone tell me what could be causing this error message:

[error] Initial mesh, process mesh3d : Unable to recover all the surface triangles.

The model consists of two cylinders that connect at right angles to each other and are united using boolean unite.

Is this error purely to do with the physical placement / properties of the objects or the mesh,

Which mesh operations should I try to apply to help improve this situation?


Thanks in advance


Cat

Hi,

I would think you should try uniting the structure once again. And check by using "Seperate Bodies" if they are united. OR you can expand teh structure and view.

A s far as meshing is concerned, you can try to assign meshing by length wise and try simulating the structure.

hope this helps

depending on the dimension and relative position of the two cylinders, the "unit" step might not have given you a valid geometry. For example, some critical points might be close to each other, HFSS might have merged some of them and hence not able to recover all the surface triangles.

try different ways of building the geometry. for example, instead of making one solid out of the "unit" operation, build 3 solids by subtracting one cylinder from the other, and intersecting the two cylinders. Clone the "tool" and "target" before performing the subtraction and intersection. try to model only 1 solid at a time by unchecking the "model" option for the other solids.

This kind of problem is not unique to HFSS, it is just that HFSS doesn't give you enough details as to the cause. so you have to try different ways to find out.

thanks for the quick responses :)

I am going to try using length and skin depth based mesh operations, what extreme values could I start off with ?


Cat

hi,

you can start with default values first. then you can assign a maximum length of the mesh as λ0/6 or..λ0/4, where λ0 is the resonant wavelength
hope this helps

well, i'm working below 30MHz, so the wave length is pretty large. I'm starting to think this ain't possible :(

hi,

i guess since the wavelength is more, even the size of the structure increases.....right,
so the structure dimensions would be in meters, well I guess you should first try simulating it....

Do the units effect the result in anyway? Is there disadvantage of having to work in metres?

Thanks


Cat

hi,

I do not think so. Well as the dimension of structure is relative to the range of frequencies you are looking. It is scaled accordingly. As long as you define your frequency range, dimension of the structure for that frequencies and material properties of teh materials you are using. I think you should be able to simulate your structure

Well, I got past the initial error but now I have a strange second error.

I have added an extra tubing inside one tube to form a coaxial inner pin conductor.

I have tried all sorts of mesh operations, sometimes the mesh making will take ages but still produces an error:


[error] Initial mesh, process mesh3d : Unable to recover all the surface triangles

I am getting real dumb struck now so I have attached the HFSS project file. Could anyone please look at it and explain what is causing the error?


See Below...


Thanks


Cat

Any ideas folks?

could you please zip it . I mean .hfss extension

I have zipped it with .zip extension now. I tried to just include the hfss file but the .hfss extension is not allowed.

Let me know if there are any more problems with opening the project.


Thanks

Cat

Try to see the error message in the html file. This kind of error usually occur when you have two faces with ratio >1e6.

I looked at your model quickly although I could not debug completely.
You might want to spend sometime simplifying your problem though.

(1) I was a little vague about the feeding structure, but do you really need to compute the field INSIDE the long narrow coax? Probably not. You could use an internal port at somewhere near the opening. If you still need the scattering effect from the feed, you could just leave the solid metal tube (or long narrow box will be better) in place.

(2) You could also use a virtual object around the antenna since the aspect ratio between the smallest and largest feature is pretty large.

(3) Before running simulation, you could select all abjects and set the meshing option of 'maximum normal deviation' to 45 degrees or larger, so that the mesher will not create really fine mesh around the curved surface. Since the tube diameter is really small comparing to the wavelength, it does not need the fine mesh.

good luck,

why i can't open the "test.hfss"

and to Rito . could i make a friend with you ?
my email is :fanqikai@163.com
:)

with rito's advice,have you solved it?i wonder

Maybe there is some so lillte object that HFSS cannot mesh.

does your cylinder have a thickness? and are you using any other material apart from pec? check on the intersections and the edges ..zooom in and check.

I am not able to open the project at all