firemodels
Geom Mesh boundary Issue
mesh_test1.fds.txt mesh_test2.fds.txt
mesh_test1.fds This case generates an error in the velocity field at a what appears to be a mesh geom intersection. The second case is a slight modification where the mesh boundary has been moved away from the geometry and runs fine. Suspect there is a bug at mesh interface external boundaries when intersecting geom.
@marcosvanella It looks like we have a bad cutcell definition or something at the boundary. I'll try to work on it, but if you have a clue, please let me know.
@marcosvanella It actually looks like the case PBX slices in mesh_test2.fds also have a problem. In the attached image, you can see a spuriously high value of velocity being output to the slice file that is inside the solid, and this is causing the data bounds to go haywire. But this may be an unused ghost cell value from the CC_STRESS_METHOD. I'm not sure.
We noted similar issue with stress method in one of Kevins terrain cases. Might hit two bugs with one stone. First thing to look is if there is a change in the repo where this started happening, or this was there from the beginning.
Here's a similar case which came back with a different error which might give a clue to what's going on Askervein_TU03A_mesh_test3.err.txt Askervein_TU03A_mesh_test3.fds.txt
I'm bisecting now. @noellecrump2023 if you checkout the code at ca09411e0ef91baca1dd8b70a2a404dca11b2295 and recompile, you should get a version that runs and will give you numbers for your scripts.
e3dbd302d012f34467b3de73b301fded0a93539c is the first bad commit
commit e3dbd302d012f34467b3de73b301fded0a93539c
Author: marcosvanella <[email protected]>
Date: Thu Jul 8 17:08:47 2021 -0400
FDS Source : Set internal TERRAIN_CASE=T when Geom IS_TERRAIN=T, N_ZVALS>0.
This commit #9632 seems very simple, but I double checked and this is indeed the commit where things go bad.
I think I have got a similar issue. Here is a wind flow over a building, and I used GEOM to build the model. However, as you see in the picture, I have got artificial velocity exactly in the cut cell near the building.
@sojoudi This is an artifact of the method that does velocity updates in FDS, we are working on addressing it.