Conversation
| sigma11 = applied_loads[0] / area | ||
|
|
||
| tol = 1e-4 | ||
| tol = 1e-1 |
There was a problem hiding this comment.
Flagging this so I remember. I need to go check if this test is still sensitive enough to catch errors in the code with the new test tolerance (i.e., verify the test still fails when I break the code).
There was a problem hiding this comment.
This isn't the test that I was most concerned about, so this change is okay in that regard.
Looks like the failed CI run is this one: https://github.com/NLRWindSystems/SONATA/actions/runs/21037290795/job/60488705197
That fails for sigma33, so I am going to push a commit where the tolerance is still loosened on that component, but remains at 1e-4*sigma11 for the sigma11 stresses (right now 10% error is accepted for sigma11 on this simple beam in tension, which seems a bit loose).
…n beam in tension.
This PR: