Outdoor wind case fails with numerical instability #15563
Description
This case fails in less than 10 min. The fails trace back to
$ git bisect bad
2bf4ee6753e872aa376fd75eccd161f4744832fe is the first bad commit
commit 2bf4ee6753e872aa376fd75eccd161f4744832fe (HEAD)
Author: rmcdermo <[email protected]>
Date: Wed Sep 3 17:13:34 2025 -0400
FDS Source: remove FLUX_LIMITER_SINGLE_COEF parameter
Source/cons.f90 | 1 -
Source/divg.f90 | 158 ++++++++++--------------------------------------------
Source/init.f90 | 29 +++++-----
Source/main.f90 | 12 +----
Source/mass.f90 | 187 +++++-----------------------------------------------------------
Source/read.f90 | 2 +-
Source/wall.f90 | 317 +++++++++++++++++++++++++-----------------------------------------------------------------------------------
7 files changed, 133 insertions(+), 573 deletions(-)
If I change the 'PERIODIC FLOW ONLY' to PERIODIC, the case works. This case is a simple version of the LNG_Dispersion validation cases. I removed the methane gas. The instability occurs at the lateral edge near the ground after 317 time steps. The grid is stretched, which probably complicates things.