fix: use mass and energy consistently for single phase solvers #3485
Conversation
paveltomin
changed the title
| @@ -241,46 +241,38 @@ DECLARE_FIELD( temperatureScalingFactor, | |||
| NO_WRITE, | |||
| "Scaling factors for temperature" ); | |||
|
|
|||
| DECLARE_FIELD( mass, | |||
There was a problem hiding this comment.
mass went to single phase file since it is used there, not for compositional
| @@ -594,8 +634,6 @@ void SinglePhaseBase::initializeFluidState( MeshLevel & mesh, arrayView1d< strin | |||
|
|
|||
| // 2. save the initial density (for use in the single-phase poromechanics solver to compute the deltaBodyForce) | |||
| fluid.initializeState(); | |||
|
|
|||
| updateMass( subRegion ); | |||
There was a problem hiding this comment.
this call is redundant, updateMass already called in updateFluidState few lines above
| @@ -291,13 +260,11 @@ class SurfaceElementAccumulationKernel : public AccumulationKernel< SurfaceEleme | |||
| void computeAccumulation( localIndex const ei, | |||
| Base::StackVariables & stack ) const | |||
| { | |||
| Base::computeAccumulation( ei, stack, [&] () | |||
| Base::computeAccumulation( ei, stack ); | |||
There was a problem hiding this comment.
nice... introducing mass cleaned up accum and flux kernels a fair bit and probably reduced computation at the same time ... same for introducing energy variable..
| stack.dSolidEnergy_dPres = dSolidVolume_dPres * m_rockInternalEnergy[ei][0]; | ||
| stack.dSolidEnergy_dTemp = solidVolume * m_dRockInternalEnergy_dTemp[ei][0] + dSolidVolume_dTemp * m_rockInternalEnergy[ei][0]; | ||
| } | ||
| {}; |
There was a problem hiding this comment.
i could not remove this, somehow compiler gets always confused
paveltomin
added
type: cleanup / refactor
paveltomin
requested review from
rrsettgast,
CusiniM,
castelletto1,
frankfeifan,
ryar9534 and
jhuang2601
as code owners
|
you may want to keep an eye on this: #3460 |
Yes thanks |
paveltomin
changed the title
paveltomin
requested review from
corbett5,
wrtobin and
dkachuma
as code owners
paveltomin
changed the base branch from
develop
to
feature/byer3/single_phase_prop_refactor
paveltomin
requested review from
tjb-ltk
and removed request for
wrtobin,
corbett5 and
dkachuma
| @@ -257,53 +263,93 @@ void SinglePhaseBase::updateMass( ElementSubRegionBase & subRegion ) const | |||
| { | |||
| GEOS_MARK_FUNCTION; | |||
|
|
|||
| using DerivOffset = constitutive::singlefluid::DerivativeOffsetC< 1 >; // TODO check if this is correct | |||
There was a problem hiding this comment.
The assumption with this is that someone doesn't make pressure second index in isothermal run... to be safe I would use
geos::internal::kernelLaunchSelectorThermalSwitch( isThermal, [&] ( auto ISTHERMAL ) {
integer constexpr IS_THERMAL = ISTHERMAL();
using DerivOffset = constitutive::singlefluid::DerivativeOffsetC< IS_THERMAL >;
if constexpr(IS_THERMAL) // for the thermal bits
{
}
}
| } | ||
|
|
||
| void SinglePhaseBase::updateEnergy( ElementSubRegionBase & subRegion ) const | ||
| { | ||
| GEOS_MARK_FUNCTION; | ||
|
|
||
| using DerivOffset = constitutive::singlefluid::DerivativeOffsetC< 1 >; | ||
|
|
There was a problem hiding this comment.
by definition this is thermal routine so this is fine
paveltomin
added
the
flag: requires rebaseline
paveltomin
removed
the
ci: run integrated tests
should go after #3460
unified where possible
for poromechanics - hard to do because porosity is being updated inside the assembly kernel, so left it as is
fixed glitch with double porosity update in poromechanics when it was updated both in flow solver and in poromechanics kernel using different equations leading to potential inconsistencies