Opened 3 months ago

Closed 3 months ago

#2271 closed help (fixed)

Extended NAE domain with boundary conditions from global model

Reported by: LSaffin Owned by: willie
Priority: normal Component: UM Model
Keywords: convergence Cc: L.Saffin@…
Platform: Monsoon2 UM Version: 7.3

Description

I'm trying to run a limited area run using an NAE run but with an extended domain so that it is 500x700 gridpoints. The job is xjjhl and uses the boundary conditions generated by the job xjjhm.

In the .leave file
(/home/d03/lsaffi/output/xjjhl000.xjjhl.d17256.t085359.leave). I can see the model crashes about 8-hours into the run during the stash output but the problems seem to start earlier than that. Past 3 hours into the run the pressure solver occasionally doesn't converge and eventually keeps producing NaN values. This issue can also be seen in the output (/projects/diamet/lsaffi/xjjhl/) where, after the first few hours, all the values outside the lateral boundaries are NaN.

Any idea what I have done wrong?

Thanks,

Leo

Change History (4)

comment:1 Changed 3 months ago by LSaffin

  • Cc L.Saffin@… added

comment:2 Changed 3 months ago by willie

  • Keywords convergence added
  • Owner changed from um_support to willie
  • Status changed from new to accepted

Hi Leo,

It fails to converge at time step 40 (3.3 hrs) , recovers for a bit and then fails to converge until it fails. I don't think you've done anything wrong, it's just that the model is a bit unstable with this data (I checked the start dump and LBC files and they're fine). So I think you should try halving the time step and see if that improves things.

Regards
Willie

comment:3 Changed 3 months ago by LSaffin

Hi Willie,

That's worked perfectly, thanks.

comment:4 Changed 3 months ago by willie

  • Resolution set to fixed
  • Status changed from accepted to closed
Note: See TracTickets for help on using tickets.