search

Exawind / hfm-fsi

Input files and data for HFM FSI Work
0 stars 0 forks source link

Run NREL5MW with old mesh #70

Open psakievich opened 5 months ago

psakievich commented 5 months ago

Turn on theta and span ramping, update bc's, shorten the precursor, and overset definitions.

kevmoor commented 5 months ago

Neil did this, and it gave negative volumes. This does not have tilt built in, so not the same issue. After changing the tilt on the split mesh in the other 5MW issue, getting same error, so now consistent. Will check nalu output and check against openfast output files, sanity checking. Feels surprising to get negative volumes on split mesh, unless openfast and nalu don't agree on where the mesh is and torquing it to match. Best next step is to take small time increments of VTK data. While doing that, may kick off case without tilt as well.

Had stuff running prior to openfast changes, but Ganesh has had things running since then, but he is the only one. Question of if Ganesh found something and forgot to mention, or just found it in passing. A git diff on repos might be useful.

pscrozi commented 5 months ago

Ganesh figured out the negative volumes issue with the NREL 5 MW old mesh. Still working on a fix and root cause determination. Trying the same rounding approach that he used for the IEA-15MW.

Ganesh will see if it runs and will submit the batch of runs. He won't need help unless it doesn't run through.

psakievich commented 5 months ago

@ndevelder got a case that ran out to 27 seconds at 14m/s (4-5 revolutions). Looks like it may be starting to have issues with convergence of linear systems. Starting issues with SDR. To get this working he removed a the root beamdyn node and running without any mesh transformation tilt in the nalu-wind input deck (mesh was pre-tilted)