Thanks for pointing this out @EnnaDelfen! I can reproduce and see the same commit as the culprit. The velocity update/nodal projection seems to be the first thing to go haywire: the initial iterations are ~the same, the plt00000 file has small diffs in reaction rate but nothing too concerning, then the plt00001 file shows diffs in velocity and gradp, then everything starts showing significant diffs beyond that.
Investigating further...
Discussed in #632
Originally posted by EnnaDelfen March 2, 2026
Hi all,
In the latest version (Feb 26 41e51c1), the triple flame seems to be broken. I am investigating but in case anyone from the current development team has a clue as to what the issue may be it may be quicker ?
To be more precise, the dtConv is stuck to 7.731042256e-08, and the solution exhibits weird kinks on the velocity profile at coarse fine boundaries
EDIT: it seems the regtest stopped working after this massive commit c8b3030
Thanks for pointing this out @EnnaDelfen! I can reproduce and see the same commit as the culprit. The velocity update/nodal projection seems to be the first thing to go haywire: the initial iterations are ~the same, the plt00000 file has small diffs in reaction rate but nothing too concerning, then the plt00001 file shows diffs in velocity and gradp, then everything starts showing significant diffs beyond that.
Investigating further...
Discussed in #632
Originally posted by EnnaDelfen March 2, 2026
Hi all,
In the latest version (Feb 26 41e51c1), the triple flame seems to be broken. I am investigating but in case anyone from the current development team has a clue as to what the issue may be it may be quicker ?
To be more precise, the dtConv is stuck to 7.731042256e-08, and the solution exhibits weird kinks on the velocity profile at coarse fine boundaries
EDIT: it seems the regtest stopped working after this massive commit c8b3030