feat: Add immiscible multiphase flow package #3251

CusiniM · 2024-07-30T00:10:36Z

Implementation of a multiphase immiscible solver for compressible fluids with viscous, gravity and capillary forces:

Included new ImmiscibleMultiphaseFlow solver and associated kernels (currently limited to two phases)
Standard and total mass formulations
Cell-centered Dirichlet and flux boundary conditions
Included new TwoPhaseFluid model for tabular data input
Direct and iterative linear solver compatibility
GPU compatibility
Set of input files with validation cases
Unit tests and documentation provided

src/coreComponents/physicsSolvers/fluidFlow/ImmiscibleMultiphaseFlowFields.hpp

…MultiphaseFlow

CusiniM · 2024-09-17T21:23:08Z

.vscode-codespaces/launch.json

+            "program": "${userHome}/two-phase/GEOS/install-ubuntu22-debug/bin/geosx",
+            "args": ["-i", "/home/rpiazza/two-phase/GEOS/inputFiles/immiscibleMultiphaseFlow/immiscible_2phaseFlow_1d.xml", "-o", "${userHome}/geos-output"],


Remove these changes.

paveltomin · 2024-11-19T23:34:11Z

src/coreComponents/physicsSolvers/fluidFlow/ImmiscibleMultiphaseFlow.cpp

+
+real64 ImmiscibleMultiphaseFlow::setNextDt( const geos::real64 & currentDt, geos::DomainPartition & domain )
+{
+  return PhysicsSolverBase::setNextDt( currentDt, domain );


TODO pressure and saturation change time step selector

Implemented new function:
real64 ImmiscibleMultiphaseFlow::setNextDtBasedOnStateChange(real64 const & currentDt, DomainPartition & domain)

paveltomin · 2024-11-19T23:36:05Z

...onents/physicsSolvers/fluidFlow/kernels/immiscibleMultiphase/ImmiscibleMultiphaseKernels.hpp

+            real64 const dDens_dP = m_dDens_dPres[seri[ke]][sesri[ke]][sei[ke]][0][ip][Deriv::dP];  // dr/dP = dr1/dP1 || dr2/dP
+
+            // average density and derivatives
+            densMean[ip] += 0.5 * density;          // rho = (rho1 + rho2) / 2


this treatment was recently modified, see #3337

New treatment implemented in kernel:
void computeFlux( localIndex const iconn, StackVariables & stack, FUNC && kernelOp = NoOpFunc{} ) const

sorry we had to revert that new treatment and make it an option, see #3467
somehow old treatment works much better for spe11 and it is not a good time to spoil it

We adapted the code to make it optional in the immiscible solver as well, using the parameter m_checkPhasePresenceInGravity

paveltomin · 2024-11-19T23:40:56Z

...onents/physicsSolvers/fluidFlow/kernels/immiscibleMultiphase/ImmiscibleMultiphaseKernels.hpp

+          real64 dPresGrad_dTrans = 0.0;
+          real64 dGravHead_dTrans = 0.0;
+          real64 dCapGrad_dTrans = 0.0;
+          int signPotDiff[2] = {1, -1};


i don't think you need this sign thing

Keeping it for now. We can think of a more elegant implementation in the future.

i think it is just wrong, why no just this:

presGrad[ip] += trans[ke] * pressure; ... dPresGrad_dP[ip][ke] += dTrans_dP[ke] * pressure;

?

If we understand correctly, it's because we need a pressure difference instead of a single pressure value:

Here presGrad = T * (P1-P2)
When we take the derivative with respect to the two pressures we will need: dPresGrad_dP = {dT/dP1 * (P1-P2) + T, dT/dP1 * (P1-P2) - T}. At the last line of code shown dPresGrad_dP = {T, -T} already.
So the second term on the right-hand side needs to be a (P1-P2), instead of just a single pressure. We used the signPotDiff variable to assist taking the difference of the 2 pressures.

i am still confused
the code is more like
presGrad = T1*P1 + T2*P2
right?
the rest should follow the same notation/logic
maybe @CusiniM or @ryar9534 can have a look and check if i am wrong

Someone else can check our understanding, but it's because in
presGrad[ip] += trans[ke] * pressure; // DPv = T (P1 - P2)
we are using the transmissibilities that come from the computeWeights (CellElementStencilTPFA.hpp) function like this:
weight[0][ke] = m_transMultiplier[iconn] * value * (ke == 0 ? 1 : -1);
So T2 = -T1

paveltomin · 2024-11-19T23:43:32Z

...onents/physicsSolvers/fluidFlow/kernels/immiscibleMultiphase/ImmiscibleMultiphaseKernels.hpp

+            potGrad += capGrad[ip]; // DPhi = T (P1 - P2) - T rho g (z1 - z2) + T (-Pc1 + Pc2)
+          }
+
+          // compute upwinding tolerance


compositional is not doing that trick but ok to keep it here as long as it works fine

...onents/physicsSolvers/fluidFlow/kernels/immiscibleMultiphase/ImmiscibleMultiphaseKernels.hpp

paveltomin · 2024-11-19T23:55:57Z

Overall looks pretty good, just some minor things here and there, cleanup and remove commented code

victorapm

I just looked at MGR changes, looks good!

victorapm · 2024-11-20T00:42:18Z

src/coreComponents/linearAlgebra/interfaces/hypre/mgrStrategies/ImmiscibleMultiphaseFVM.hpp

+  explicit ImmiscibleMultiphaseFVM( arrayView1d< int const > const & numComponentsPerField )
+    : MGRStrategyBase( LvArray::integerConversion< HYPRE_Int >( numComponentsPerField[0] ) )
+  {
+    // Level 0: eliminate last density which corresponds to the volume constraint equation


These comments can be removed

…MultiphaseFlow

matteofrigo5 · 2024-12-12T03:21:26Z

I'd remove the file host-configs/Stanford/ubuntu22.cmake before merging this branch into the develop branch.

…MultiphaseFlow

CusiniM added 4 commits July 26, 2024 16:29

Add ImmiscibleMultiphaseFlow physics package.

bf8c1c2

catalog name, register and cmakelist.

22aed17

fix fields.

fd51d91

removed all compositional references.

c90724a

CusiniM commented Jul 30, 2024

View reviewed changes

src/coreComponents/physicsSolvers/fluidFlow/ImmiscibleMultiphaseFlowFields.hpp Outdated Show resolved Hide resolved

CusiniM commented Jul 30, 2024

View reviewed changes

src/coreComponents/physicsSolvers/fluidFlow/ImmiscibleMultiphaseFlowFields.hpp Outdated Show resolved Hide resolved

CusiniM added 2 commits July 29, 2024 17:50

Merge remote-tracking branch 'origin/develop' into feature/immiscible…

e4a9df0

…MultiphaseFlow

Compiles and links.

446e401

Ammara-14 self-assigned this Jul 30, 2024

ryar9534 self-assigned this Aug 1, 2024

Ryan Michael Aronson and others added 6 commits August 1, 2024 11:12

update input file so it runs

0d159ae

adding skeleton for setupDofs

7783a15

testing

5fcd5d8

fluid model

8e132b9

rock-fluid model

c21d047

new kernel file

d59fa4e

rpiazza87 self-assigned this Aug 13, 2024

bd713 marked this pull request as ready for review September 17, 2024 21:05

bd713 requested review from rrsettgast, corbett5, wrtobin, paveltomin, klevzoff, francoishamon, dkachuma, joshua-white, MelReyCG, cssherman and jhuang2601 as code owners September 17, 2024 21:05

CusiniM commented Sep 17, 2024

View reviewed changes

paveltomin reviewed Nov 19, 2024

View reviewed changes

...onents/physicsSolvers/fluidFlow/kernels/immiscibleMultiphase/ImmiscibleMultiphaseKernels.hpp Outdated Show resolved Hide resolved

paveltomin reviewed Nov 19, 2024

View reviewed changes

...onents/physicsSolvers/fluidFlow/kernels/immiscibleMultiphase/ImmiscibleMultiphaseKernels.hpp Outdated Show resolved Hide resolved

paveltomin reviewed Nov 19, 2024

View reviewed changes

...onents/physicsSolvers/fluidFlow/kernels/immiscibleMultiphase/ImmiscibleMultiphaseKernels.hpp Outdated Show resolved Hide resolved

victorapm approved these changes Nov 20, 2024

View reviewed changes

Ammara-14 and others added 14 commits November 21, 2024 12:28

fixed code style and added a dedicated Kernel selector switch

fa42877

Fixed doc references.

83783e6

Used pre-existing Geos readTable function.

84673ee

Addressed Pavel's comments.

b23c33d

code style

557b179

Merge remote-tracking branch 'origin/develop' into feature/immiscible…

cf32759

…MultiphaseFlow

Used the newly introduced initializeState from FlowSolverBase.

3cef9ab

Fixed style.

8bb5ba6

Merge remote-tracking branch 'origin/develop' into feature/immiscible…

e8698c0

…MultiphaseFlow

Changed Total S.A.

3ddd347

Made density averaging optional (need gpu validation).

8d0cd41

Merge remote-tracking branch 'origin/develop' into feature/immiscible…

434c435

…MultiphaseFlow

Added schema.

f39620c

Fixed doc.

16715ce

Added rebaseline infos.

9e32fe5

bd713 added the flag: requires rebaseline Requires rebaseline branch in integratedTests label Dec 12, 2024

paveltomin approved these changes Dec 17, 2024

View reviewed changes

DENEL Bertrand added 3 commits December 17, 2024 16:33

Added back HYPRE_BoomerAMGSetAggNumLevels.

3395073

Merge remote-tracking branch 'origin/develop' into feature/immiscible…

fab919e

…MultiphaseFlow

Removed ubuntu config file.

84613c5

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: Add immiscible multiphase flow package #3251

feat: Add immiscible multiphase flow package #3251

CusiniM commented Jul 30, 2024 •

edited by rpiazza87

Loading

CusiniM Sep 17, 2024

paveltomin Nov 19, 2024

rpiazza87 Dec 7, 2024

paveltomin Nov 19, 2024

rpiazza87 Dec 7, 2024

paveltomin Dec 7, 2024

rpiazza87 Dec 10, 2024

paveltomin Nov 19, 2024

rpiazza87 Dec 7, 2024

paveltomin Dec 7, 2024

rpiazza87 Dec 10, 2024

paveltomin Dec 10, 2024

rpiazza87 Dec 10, 2024

paveltomin Nov 19, 2024

paveltomin commented Nov 19, 2024

victorapm left a comment

victorapm Nov 20, 2024

matteofrigo5 commented Dec 12, 2024

		"program": "${userHome}/two-phase/GEOS/install-ubuntu22-debug/bin/geosx",
		"args": ["-i", "/home/rpiazza/two-phase/GEOS/inputFiles/immiscibleMultiphaseFlow/immiscible_2phaseFlow_1d.xml", "-o", "${userHome}/geos-output"],

feat: Add immiscible multiphase flow package #3251

Are you sure you want to change the base?

feat: Add immiscible multiphase flow package #3251

Conversation

CusiniM commented Jul 30, 2024 • edited by rpiazza87 Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

paveltomin commented Nov 19, 2024

victorapm left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

matteofrigo5 commented Dec 12, 2024

CusiniM commented Jul 30, 2024 •

edited by rpiazza87

Loading