Simplifying the geometry

[Guido Antoniol]

Good evening,I apologize for my bad english. I write this post because I'm calculating a thermal bridge and at the time of the calculation, I am required to simplify the geometry or Increasing Quad Tree Mesh Parameter File Options in Therm. I tried both solutions, but I can not solve the problem. I attach the file and I thank everyone who can help me.

thank you very much

[Jaanus Hallik]

Hello Guido,

The problematic areas seems to have very branched PVC structures around the frame cavity. Simplifying the geometry through subdividing those PVC parts in few areas will be sufficent. Improved example THM file is attached to the post.

All the best,

Jaanus Hallik

---

University of Tartu, Estonia | Energy efficient building core lab

http://www.tuit.ut.ee/en/core-facilities/energy-efficient-building-core-facility

[Robin Mitchell/LBNL]

Guido

One of the problems with this file is that scale of the model -- it is quite a large model, because the walls are very "long" (1789 mm in height), with very intricate detail in the window frame. THERM has difficulty meshing a model like this. 

I was able to get the file to run by doing the following:

•  made each of the wall sections shorter -- I basically cut them in half, and I think there is still enough there for the model; in particular the isotherms on the vertical wall section are parallel, so there is no need to model more height on that wall. 
  
• For the PVC frames, you had introduced lots of extra points in the middle of the frame elements when making new polygons. This adds a lot of extra points into the model, which forces the mesher to do a finer mesh there when it is not needed. So I simplified the number of points in the frame polygon. The attaches images show some of the issues.

I have also attached the file that I was able to run, called Attacco alto PT-RM.THM

Let me know if you have any questions.

Robin

[Robin Mitchell/LBNL]

Attached is a screen shot of the final frame polygons

Robin

On Monday, January 5, 2015 11:52:40 AM UTC-8, Guido Antoniol wrote:

[Jaanus Hallik]

All points correct however the "sufficient" length of the flanking elements depends on the purpose of the calculation and underlying norms. So if this junction would be in European country (Italy, I would assume?) where iso 10211 norm describes the calculation methodology then the cut-off plane should be at distance equal to 3 times the thickness of the flanking element or 1m whichever is larger of the two (so even little bit more than in the original THM).

All the best,

Jaanus

---

University of Tartu | Energy Efficient Building core lab

[Guido Antoniol]

Good morning, I wanted to thank everyone for the help given!

[Robin Mitchell/LBNL]

It could be that a European standard requires modeling a longer wall segment, but the THERM mesher may not be able to model that, depending on the level of detail in other parts of the model. And it may not be possible to simplify the detailed areas enough to model the longer wall segments. Probably incrementally increasing the wall segments until the mesher fails again would be one way to approach it, starting from the simplifications of the frame cross section already offered in this post.

Robin

On Monday, January 5, 2015 11:52:40 AM UTC-8, Guido Antoniol wrote:

[Jaanus Hallik]

I agree that there are some limitations with the mesher and for those cases you will need actual geometrical simplifications. But from our own experience I just wanted to share that our team has done hundreds of Therm calculations and we haven't really had situation where we needed to make considerable simplifications in geometry or shorten the flanking elements while doing ISO compliance calculations. Even for slab-on-ground-to-window-sill junctions where the floor layers adjacent to fully detailed window frame are extended up to 4 meters and soil volume up to 20 meters down and outwards. Typically we have utilized 8 / 2% / 10 combination for "QT mesh parameter" / "Maximum Error" / "Maximum iterations", but for very difficult junction we have found a solution with 7 / 5% / 15 easily within ISO accuracy limits.

I have faced a few impossible meshing situations while trying to model the full cross section of entire building just for fun, but hey it is really outside "normal" usage scenario :) .

Anyway, happy new year and happy modeling with Therm. All the best,

Jaanus

---

University of Tartu | Energy Efficient Building core lab

[Robin Mitchell/LBNL]

It's great to know that you have been able to use THERM for these models !

Thanks for the feedback.

Robin

On Monday, January 5, 2015 11:52:40 AM UTC-8, Guido Antoniol wrote: