search

Exawind / hfm-fsi

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

Run smaller case that runs faster, with IEA 15 MW, lower wind speed #101

Open pscrozi opened 4 months ago

pscrozi commented 4 months ago

Nate and Lawrence are working on this.

pscrozi commented 4 months ago

Nate: we need to get the IEA 15 MW running on Frontier. Still waiting for Flight to come back to test on, so that we can test the most recent FSI build. May build on eclipse or attaway instead.

Lawrence is setting up the 7 km x 2 km case. 6.25 m/s case. Lawrence will get a version of this set up on Frontier, ready to go.

psakievich commented 4 months ago

@lawrenceccheung has set up the 7x2 cases and is ready to run them. @ndevelder and @lawrenceccheung to meet and plan runs.

pscrozi commented 4 months ago

Lawrence has AL cases set up and ready to run on Frontier. Nate will set up FSI versions of this at Sandia and then transfer over to Frontier.

pscrozi commented 4 months ago

Nate: been working on this. The build with the exact commit that Jon used on Frontier, with OpenFast on Flight. Has standalone version with IEA15MW. Getting an out-of-bounds error with OpenFast that Nate is looking into, but he might copy in the new line instead. Nate going to replace the files, then try to run the precursor and copy that over. Hoping to have that done soon, probably early Monday morning (5/20).

Lawrence: has 2 cases going, AL with elastodyne (0.005) and beamdyne (0.0001), with different delta_t. We'll be interested to see the differences.

pscrozi commented 4 months ago

Lawrence: elastodyne case works fine. beamdyne OOM error, so need to ensure doing it correctly. Queues on frontier are pretty slow, and getting slower. ALM.

Nate: found the bug in OpenFAST. Now has IEA15MW running with hybrid solver, with precursor, then restarting. That step is done, and able to move to Frontier now. Can use Lawrence's AMR background for reference. Setup and run in the next few days. This is just a single turbine case. Will have to figure out how to do restarts.

pscrozi commented 4 months ago

Lawrence: elastodyn case is running.

pscrozi commented 4 months ago

Lawrence: nothing new on 6/3. 10x10 km case runs quickly. Have a 20x20km case. Need to put things together to get it through the queue, otherwise they take too long.

pscrozi commented 3 months ago

Lawrence: gotten more jobs on queue on Frontier.

pscrozi commented 3 months ago

Lawrence running these jobs while on travel. Precursors are done. Waiting on 2 turbine case, which we figured out was working today.

pscrozi commented 3 months ago

Once we've verified that the 2-turbine case IEA-15MW works, we can set up a larger case. When Lawrence is back early next week, we can set this up.

pscrozi commented 3 months ago

Nate and Lawrence will submit this on Frontier (2 turbine blade resolved case.)

pscrozi commented 3 months ago

2 turbine case, Nate posted in Slack. AMR-Wind is taking 300 seconds per timestep on Frontier, with same recipe as Lawrence used, so probably not the build. Solvers aren't doing a ton of iterations. Might try resubmitting it in hopes that it was a disk issue or something. Was getting 5 seconds per timestep on Flight.

pscrozi commented 3 months ago

Nate: I got past the issue above. Small error by user. AMR-Wind performance is fragile. Following Jon's recipe worked so now getting the same performance on Frontier as on Flight. Now ready for the larger case. Will work with Lawrence to move from 2 turbine case to larger case.

pscrozi commented 2 months ago

Lawrence: we have IEA15MW running with controller on, and comparison vs ALM. If back turbine is waked, this should be useful.

pscrozi commented 2 months ago

Also still running.

pscrozi commented 1 month ago

Lawrence: we're continuing this one. It will be an interesting comparison.