Opened 4 years ago

Closed 4 years ago

Last modified 3 years ago

#1879 closed help (answered)

run failure with Error Message: Convergence failure in BiCGstab

Reported by: dpolson Owned by: um_support
Component: UM Model Keywords:
Cc: Platform: ARCHER
UM Version: 8.5



I'm running an N216 version of HadGEM3-A for 01/09/2000 to 01/09/2001 and it's crashing after a couple of months with the error:

? Error in routine: eg_sl_helmholtzApplication 21617299 is crashing. ATP analysis proceeding…

? Error Code: 1
? Error Message: Convergence failure in BiCGstab

job is xmqwa,
output is in ~dpolson/output/xmqwa003.xmqwa.d16133.t014453.leave
and /work/n02/n02/dpolson/xmqwa/

I'm not familiar with why the model would be crashing so I'm hoping you can point me in the right direction.


Change History (5)

comment:1 Changed 4 years ago by grenville


This is typically the result of some numerical instability in the model. It's hard to say why this has happened. You need some more diagnostics — try dumping on a few time steps prior to the failure and examine examine the fields.

How does this model differ from the standard HadGEM3-A if at all.


comment:2 Changed 4 years ago by dpolson

Is should be a pretty standard run, the only thing I really changed was to add some aerosol related diagnostics.

IS there a way to perturb the data from the previous month slightly to avoid encountering the same instability?


comment:3 Changed 4 years ago by grenville


Changing dump frequency may be enough to do that. Alternatively if you write a dump just prior to the failure, then reconfigure that dump and carry on the run with it, that may work.

You could also try reducing the time step - maybe just to get past the instability, then return it to it's normal value.


comment:4 Changed 4 years ago by grenville

  • Resolution set to answered
  • Status changed from new to closed

Closed for lack of activity

comment:5 Changed 3 years ago by willie

  • Platform set to ARCHER
Note: See TracTickets for help on using tickets.