Transmissibility output

[MelReyCG]

This PR aims to add a new component: the StencilOutput, that allow to add a CellToCellOutput task node in the xml to output ConnectionData (global id pairs, transmissibilities).

For now, by using the writeCSV="1" attribute, the CellToCellOutput will output this data in a CSV. Here is an exemple:

• XML excerpt

  <Solvers gravityVector="{0.0, 0.0, 0.0}">
    <SinglePhaseFVM
      name="singlePhaseFlowSolver"
      […]
    </SinglePhaseFVM>
  </Solvers>

  <Tasks>
    <CellToCellOutput
      name="stencilOutput"
      flowSolverName="singlePhaseFlowSolver"
      writeCSV="1"
      logLevel="1" />
  </Tasks>

  <Events maxTime="10.0">
    <SoloEvent
      name="stencilOutputEvent"
      targetCycle="1"
      target="/Tasks/stencilOutput" />
  </Events>

• First 10 lines of the generated CSV

time [s] , A global id , B global id , A transmissibility [(Pa*s*rm3/s)/Pa] , B transmissibility [(Pa*s*rm3/s)/Pa]
       4 ,           0 ,        2715 , 7.55463e-18                          , -7.55463e-18                        
       4 ,           0 ,       47546 , 8.06167e-18                          , -8.06167e-18                        
       4 ,           0 ,      134459 , 6.63826e-18                          , -6.63826e-18                        
       4 ,           0 ,      165202 , 7.78139e-18                          , -7.78139e-18                        
       4 ,           1 ,        4642 , 1.83325e-17                          , -1.83325e-17                        
       4 ,           1 ,       53907 , 1.76737e-17                          , -1.76737e-17                        
       4 ,           1 ,      184341 , 1.81705e-17                          , -1.81705e-17                        
       4 ,           2 ,        5459 , 1.7414e-17                           , -1.7414e-17                         
       4 ,           2 ,       63907 , 1.86145e-17                          , -1.86145e-17                        
       4 ,           2 ,      177257 , 1.86145e-17                          , -1.86145e-17                        

(the CSV has been aligned here for readability)

[paveltomin]

not sure it makes much sense to print the same transmissibility value twice but it's ok, not a big deal

[karimifard]

Maybe it will be more useful to output two half-transmissibilities.

[paveltomin]

could you please check if all that will work for https://github.com/GEOS-DEV/GEOS/blob/develop/inputFiles/compositionalMultiphaseFlow/co2_hybrid_1d.xml ?
not sure hybrid discretization has the same things as TPFA

[dkachuma]

I suppose the key here is that fvManager.getFluxApproximation( m_solver->getDiscretizationName() ); might return null if the solver doesn't have a flux approximation. So I would add a validator (postProcessInput()) to the solverName field of this to ensure that the solver provided has the appropriate flux approximation.

[dkachuma]

I suppose the key here is that fvManager.getFluxApproximation( m_solver->getDiscretizationName() ); might return null if the solver doesn't have a flux approximation. So I would add a validator (postProcessInput()) to the solverName field of this to ensure that the solver provided has the appropriate flux approximation.

[CusiniM]

It seems that we are trying to only write a file for rank 0 but I don't see when you gather all stencil data there? This would only work for small cases....
In any case, why not using the timehistory and writing in a parallel format instead? @wrtobin opinions?

Besides debugging I don't really see why someone would want to write out transmissibilities.

[codecov]

Codecov Report

Attention: Patch coverage is 11.53846% with 138 lines in your changes missing coverage. Please review.

Project coverage is 55.62%. Comparing base (f91800b) to head (c5df35b).

Files	Patch %	Lines
...physicsSolvers/fluidFlow/StencilDataCollection.cpp	0.00%	87 Missing ⚠️
src/coreComponents/common/Units.hpp	14.58%	41 Missing ⚠️
...physicsSolvers/fluidFlow/StencilDataCollection.hpp	10.00%	9 Missing ⚠️
...vers/multiphysics/CoupledReservoirAndWellsBase.hpp	0.00%	1 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #3091      +/-   ##
===========================================
- Coverage    55.69%   55.62%   -0.07%     
===========================================
  Files         1032     1034       +2     
  Lines        87739    87841     +102     
===========================================
+ Hits         48863    48864       +1     
- Misses       38876    38977     +101     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[MelReyCG]

@CusiniM

It seems that we are trying to only write a file for rank 0 but I don't see when you gather all stencil data there? This would only work for small cases.... In any case, why not using the timehistory and writing in a parallel format instead? @wrtobin opinions?

Besides debugging I don't really see why someone would want to write out transmissibilities.

This proposal is not completely finished, but it now takes into account all connections within all ranks. The CellToCellDataCollection now can be used for bigger cases.

To explain why we want to output this data, the goal is to compare the transmissibility data from other simulator(s), in order to understand the differences between the simulations.

[rrsettgast]

@MelReyCG Can you resolve conflicts?

[CusiniM]

Can you remove these? Coz they are also written in French.

[CusiniM]

Maybe supporting the output for all the other stencils could be useful too. I m thinking of debugging of fractured problems.