THERM does not mesh

[Vladimir Pezel]

I have a lot of trouble with the THERM mesher. It is very unstable. It does not want to mesh this file. I simplified it to almost an orthogonal polygon assembly. I am afraid that I will loose the actual model geometry accuracy just so it can mesh. It keeps pointing to a single point repeatedly, even though there is nothing there that could be problematic.

Also, when it meshes one model successfully, it does no mesh it anymore if I rotate the same geometry 90 deg. How is that possible?

You guys need to seriously improve the mesher!

[Robin Mitchell]

There seem to be two problem areas in this file -- I am wondering how you created it, given that it seems that there are points very close to each other. 

The first error the mesher encounters initially looks like there isn't anything wrong. 

But if you zoom in several times to that area, you will see weird "double points" including the point that the mesher identified with a red circle.

Using the false die mold method, I deleted the gray polygon and recreated it, and that particular mesh error went away

Here is a link to what the false die mold method is

https://windows.lbl.gov/sites/default/files/software/THERM/Therm7-05-14-DrawingTips.pdf -- go to page 5-107 

You basically surround the polygon you want to delete and recreate with other polygons, so that when you delete it, you can then just use the Paint tool to automatically fill it, and then delete the extraneous surrounding polygons. In general it is better to let the program match points, using the fill tool, rather than trying to do it yourself. 

I then generated BCs again, tried to run the simulation, and got another mesh error. Again, on zooming in to the point, there are a few points really close together

Again, I deleted the gray polygon, and refilled the resulting void (this time it was completely surrounded by other polygons) with the Paint tool. 

That seemed to solve the problem because after generating the BCs again and running the simulation, the file meshed without issue. 

I have attached the fixed file. I used THERM 7.7.10

Robin 

--
You can view the THERM forum online at: http://windows.lbl.gov/software/THERM/forum
---
You received this message because you are subscribed to the Google Groups "THERM" group.
To unsubscribe from this group and stop receiving emails from it, send an email to LBNL-THERM+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/LBNL-THERM/bce75a96-79b4-41f4-8a06-b78cf6658ce1o%40googlegroups.com.

--

Robin Mitchell

Building Technology and Urban Systems Division
Energy Technologies Area
Lawrence Berkeley National Laboratory

[Vladimir Pezel]

Thanks Robin.

I generally got to the same result but manually. Where you see 2 points is in fact what the actual geometry is and mesher was complaining about one of the points. To make it work, I often remove these geometry features and heavily simplify it. 

Btw, that area was created with a "Paint Fill" tool, to create the 'Frame Cavity NFRC 100" volume (green area). So the system also did not match the points.

On the second problem area, I actually removed that sliver and created an orthogonal corner. That meshed. But it is these kind of areas that are frustrating. I loose so much time trying to finagle the points and in the process the actual model is loosing fidelity.

Why the two points that snap to each other are actually not on top of each other, I am confused. The tolerance of the geometry snapping is not aligned with the tolerance for the mesher. That should probably not be the case.

I will try the reverse die method. Also using 7.7.10.0

Thanks.

Vlad

[Robin Mitchell]

In general, it's worth zooming in as far as you can to see if there are points that are too close together that the mesher might be complaining about, rather than to just delete the point circled. 

I am wondering if some of the issues you are having might be because of a bug in 7.7.10, that the program actually allows overlapping points / polygons, which is very bad. 

You might want to revert back to 7.7.07 (alas the 7.7.10 files are not readable by 7.7.07, so it would be for the next project you are working on) until we fix that problem. We are in the process of putting together a new version of THERM which should be out within the next month (?) that fixes that problem plus others (but nothing probably that would affect your modeling, mostly to do with shading systems). 

--

You can view the THERM forum online at: http://windows.lbl.gov/software/THERM/forum
---
You received this message because you are subscribed to the Google Groups "THERM" group.
To unsubscribe from this group and stop receiving emails from it, send an email to LBNL-THERM+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/LBNL-THERM/a26faf6a-b2bb-446e-9f36-c16504f31eeao%40googlegroups.com.