[Opm] Cell Convergence Debugging
opmuser0 at gmail.com
Tue May 19 06:42:32 UTC 2020
Okay, it is the Volve model that you are trying to get to run. I can get
this this to run on 2020-04 with various modifications due to OPM Flow
not supporting certain keywords. The model is currently running and is
taking the 10 day time steps as per the SCHEDULE file, and it
occasionally chops back down to 3.3 days and then builds back up again
to 10 day time steps, but it does seem to be running okay given the data
input. What I did was:
1. Fixed ROCK keyword error - too many entries.
2. Fixed RSVD keyword error - too many entries.
3. Removed PRIORITY keyword as it results in an Exception.
4. Fixed missing WCONHIST keyword Jan 1,2015
5. Changed grid output option to EGRID.
6. Remove grid file and included fault files and replaced by grid and
files exported from ResInsight.
OPM Flow does not support the PINCH 'NOGAP" option or the ADDZCORN
keyword. So I loaded the original model into ResInsight and then
exported the grid and property data, so I should have the same initial
data as the commercial simulator.
There are many warnings in the commercial simulator's PRT file regarding
connections between different PVT regions and wells connecting to
different PVT regions, this is probably the cause of the numerical
issues. Although the model does use THPRES keyword to force initial
equilibrium; however, as soon as production starts the fluids will mix
and will cause numerical issues as mass conservation is being
invalidated. If you look at the FIPNUM report you can see how much the
fluids are moving around between the different regions, although the
material balance error is surprising good.
I'll send you my files to your email address, but please bear in mind
this is a work in progress, but at least it should get you going.
Spoke too soon.
This model runs okay up to Feb 2014, then just cuts to very small times
steps (E-14) and never recovers from that, even though there are no
convergence messages on the terminal (but there is in the print file :-(
). So the issue is around 26-Feb-2014, so see if you can see anything
happening around that period that may cause problems. It looks like the
water injector I-F-IB is given a zero rate on 29-Jan-2014, instead of
being shut-in, so that might be the problem - that is never a good idea.
_OPM Flow Suggestion to Developers.
1. It would be useful to print a WARNING message every time a well
opens a connection that is in a different PVT region than the
current opened connections.
2. Print a warning message if the target rate is zero and the well is
open, this is a valid field operation, but may be OPM Flow is trying
to solve for this rate - that would be fun.
opmuser0 at gmail.com
More information about the Opm