Error relating to developments on wells

[vqueulain]

When I run my simple test case, I got a warning related to region pressure average :

***** WARNING
***** LOCATION: /path/to/geos/develop_d5fccb5/src/coreComponents/physicsSolvers/fluidFlow/wells/CompositionalMultiphaseWell.cpp:437
***** Controlling expression (should be false): useSeg
WellControls referenceReservoirRegion not set and well constraint fluid property calculations will use top segement pressure and temp 

So I added the referenceReservoirRegion="Reservoir" and now I got the following error message about wells :

***** ERROR
***** LOCATION: /path/to/geos/develop_d5fccb5/src/coreComponents/physicsSolvers/fluidFlow/wells/CompositionalMultiphaseWell.cpp:459
***** Controlling expression (should be false): !region.hasWrapper( CompositionalMultiphaseStatistics::catalogName())
***** Rank 3: compositionalMultiphaseWell (simulationA_prop.xml, l.36): No region average quantities computed.  WellControl wellControls1 referenceReservoirRegion field requires CompositionalMultiphaseStatistics to be configured for region Reservoir 

** StackTrace of 10 frames **
Frame 0: /path/to/geos/develop_d5fccb5/install-spack-generated-Release-2025-09-17-13-22/lib/libfluidFlowSolvers.so 
Frame 1: geos::WellSolverBase::initializePostSubGroups() 
Frame 2: geos::CompositionalMultiphaseWell::initializePostSubGroups() 
Frame 3: geos::dataRepository::Group::initialize() 
Frame 4: geos::dataRepository::Group::initialize() 
Frame 5: geos::dataRepository::Group::initialize() 
Frame 6: geos::ProblemManager::problemSetup() 
Frame 7: geos::GeosxState::initializeDataRepository() 
Frame 8: main 
Frame 9: __libc_start_main 
Frame 10: _start 
=====

This case runs without warning and errors before this commit: 09dcac5cb815b7e02c07956c91372b82c6a35108

Here is my case:
case.zip

For the tutorial case Multiphase Flow with Wells, there are warnings with the XML smoke file: 'deadOilEgg_smoke_3d.xml'

[vqueulain]

Can I ask your help @tjb-ltk ?

[tjb-ltk]

@vqueulain I'll have a look.. thanks for uploading the model

[tjb-ltk]

The warnings in deadOilEgg case are letting the user know that conditions used to compute fluid properties associated with the constraint are at the top segment. This was the default behavior before that commit.

When you added referenceReservoirRegion="Reservoir" that tells GEOS to use region average P&T compute the constraint fluid properties... This is common method in voidage replacement modeling or history matching... The region information is pulled from CompositionalMultiphaseStatistics. The naming could be better, basically the option needs ave quantities over a region and this region is defined via the Statistics object.
The error
"referenceReservoirRegion field requires CompositionalMultiphaseStatistics to be configured for region Reservoir "
is stating that wasn't setup...

I removed the keyword referenceReservoirRegion="Reservoir" and your model is running. Did you see a difference in results using current code and code before res vol constraint commit ? There should not be, at least due to that commit... Since you have both builds please let me know if they match or not... Thanks

[vqueulain]

Hello Tom, thanks for your help.
I ran 2 tests yesterday.
My original test didn't run to the end but did in previous versions.
Here is its log:

job_GEOS_P4_1ertest_git.txt

Is it ok to still use top segment contraints? If yes, shouldn't this message be an info rather than a warning ?

For my second run, I added referenceReservoirRegion="Reservoir" in WellControls and cycleFrequency="1" in the PeriodicEvent compflowStatistics. This example didn't run to the end.
Here is the log:

job_GEOS_P4_2ndtest_git.txt

What is missing to configure it correctly?

[tjb-ltk]

Take a look at well control docs it mentions cycleFrequency ....
Try setting

its still fine to use top segment conditions..... WARNING vs INFO question is debatable
in reservoir engineering workflows not using region ave pressures for a reservoir volume constraint , would be an ERROR.. so I picked the middle ground . and if trying to match industry codes, ref regions are used for res vol constraints..

run finished with above setting

[tjb-ltk]

not sure why the setting info didn't show up in msg
in PeriodicEvent for statistics set.
beginTime="0"
cycleFrequency="1"

[MelReyCG]

Hello everyone.

Seems like Valérie still does not achieve to run her case with these setting. Also, I noticed the value you suggested are the default ones.

I now understand the warning, maybe we can add a parameter to silent it when desired (especially if there are cases in the integrated tests).

On another hand, I was wondering if we could keep this averaging component internal. The user should only define which is the reference region, and GEOS should create the infra in background so the solver can have the necessary stats. I can do that it later if you want.

[tjb-ltk]

agree on need for infrastructure code to facilitate option....

I was testing with debug mode, where the model did run... but the model doesn't run in release mode... added a few write statements and dbg version doesn't run.. seems like something initialized, or ordering.... should have checked both versions... sorry for the confusion...

Any ideas on this error. @MelReyCG is it the wrong check, or to early flow sequence to do this check ?

  // Check if regions statistics are being computed
  GEOS_ERROR_IF( !region.hasWrapper( CompositionalMultiphaseStatistics::catalogName()),
                 GEOS_FMT( "{}: No region average quantities computed.  WellControl {} referenceReservoirRegion field requires CompositionalMultiphaseStatistics to be configured for region {} ",
                           getDataContext(), wellControls.getName(), regionName ));

[tjb-ltk]

A fix is in following PR to check if the ave quantities are computed at the time they are used. This could be improved but its not clear how to validate if region stats are configured for the region when the constraint is defined...

#3829

[tjb-ltk]

bsver.xml

bsver.txt

@vqueulain attached are the results from a modified version of your case. I just reduced nx and ny to speed up run

[MelReyCG]

@tjb-ltk I left a review on your PR. My proposal should solve any issue related to user event setting interfering with reference region averrage pressure computation.

[MelReyCG]

The key thing is that user should only enter the reference region name for the wells. User should not get requested to add any component that serve to internal solver computation, it should be kept hidden.