Dear Timo,
Really Thank you very very much. You clarified everything for me in an
amazing way and finally the results are going in the correct way now and
they make sense despite some numerical error modifying the pressure on
the upper and lower boundaries where the injection well is located.
Maybe it's due to not refining the cells around the well yet. I will
send a separate e-mail later on for the eclipse grid issue for the sake
of consistency with the e-mail subject. Thanks a lot !!
Best Regards,
Mahmoud Atef
---------------------------------------------------------------------------------------------------
Il 2021-04-09 13:13 Timo Koch ha scritto:
Hi Mahmoud,
you also need to extract the correct energy carried by the fluid in
the production well. Otherwise your domain will heat up there.
For solution-dependent point sources you can use the split interface.
First you add your point sources (in addPointSources) only providing
positions as argument to the constructor.
Then you overload the function pointSource (also in the problem) with
something like
template<class ElementVolumeVariables>
void pointSource(PointSource& source,
const Element &element,
const FVElementGeometry& fvGeometry,
const ElementVolumeVariables& elemVolVars,
const SubControlVolume &scv) const
{
const auto& pos = source.position();
const auto& volVars = elemVolVars[scv];
if (pos[0] < 40.0) // injection
{
const Scalar volumeSource = injectionRate; // injectionRate
is positive and m^3/s
const Scalar massSource =
volumeSource*IAPWSH2O::density(injectionTemperature,
injectionPressure);
const Scalar energySource =
massSource*IAPWSH2O::enthalpy(injectionTemperature,
injectionPressure);
source = NumEqVector({ massSource, energySource });
}
else // production
{
const Scalar volumeSource = productionRate; //
productionRate is negative and m^3/s
const Scalar massSource = volumeSource*volVars.density(0);
// using current control volume's water density
const Scalar energySource = massSource*volVars.enthalpy(0);
// using current control volume's water enthalpy
source = NumEqVector({ massSource, energySource });
}
}
This function will be called called for every point source that you
added whenever the point source contribution is added to the equation
residual.
Whether your values are realistic or not I cannot help you. Initial
conditions seem fine.
Hope this helps
Timo
On 9. Apr 2021, at 12:51, Mahmoud Atef Mahmoud Mohamed Aboelseoud
S277151 <[email protected]> wrote:
Dear Timo,
I cannot thank you enough for all your help and guidance. I wrote the
pointsource function as shown below. could you please let me know if
it is OK now or not ? I used 3 pointsources to represent each of the
production well and the injection well because I wanted to represent
fully penetrating wells.
void addPointSources(std::vector<PointSource>& pointSources) const
{
const auto enthalpy = IapwsH2O::liquidEnthalpy(293, 13.0e5);
const Scalar energyFlow = 1.93 * enthalpy; // 0.8 Kg/s *
enthalpy J/Kg
// The injection well (source term)
pointSources.push_back(PointSource({35, 45, 15}, {1.93,
energyFlow}));
pointSources.push_back(PointSource({35, 45, 25}, {1.93,
energyFlow}));
pointSources.push_back(PointSource({35, 45, 35}, {1.93,
energyFlow}));
// The production well (sink term)
pointSources.push_back(PointSource({65, 45, 15}, {-1.93,
0}));
pointSources.push_back(PointSource({65, 45, 25}, {-1.93,
0}));
pointSources.push_back(PointSource({65, 45, 35}, {-1.93,
0}));
}
However, the results I'm getting on paraview are a bit confusing to me
because I imposed Dirichlet boundary conditions to all boundaries
which are basically pressure and temperature gradients and they are
the same as the initial conditions
PrimaryVariables dirichletAtPos(const GlobalPosition &globalPos) const
{
return initialAtPos(globalPos);
}
PrimaryVariables initialAtPos(const GlobalPosition &globalPos) const
{
PrimaryVariables values(0.0);
Scalar densityW = 990; // Kg/m^3
Scalar depth = this->gridGeometry().bBoxMax()[2] -
globalPos[2];
// Hydrostatic Pressure
values[pressureIdx] = 11.0e5 -
densityW*this->spatialParams().gravity(globalPos)[2]*depth; //Pascal
// Geothermal Gradient
values[temperatureIdx] = 358.0 + depth*0.03; //Kelvin
return values;
}
However I could see what looks like numerical errors affecting certain
spots of the pressure boundaries (the low permeability layers on top
and bottom) (picture attached) and I also could see high temperatures
(400+ Kelvin) in the production well while the highest temperature
initially is around 359.5 Kelvin (50 m thick domain) (picture
attached). I would really appreciate it if you could let me know if
there's something wrong here. Another thing causing confusion is that
the simulation proceeds really quickly and when I impose a mass rate
of 0.38 Kg/s for each of the 3 pointsources of a well which is
equivalent to 100 m^3/day for the well, it's like I am not injecting
anything at all so I had to impose higher mass rates !
PS: I used "using IapwsH2O = Components::H2O<Scalar>;" and
"const auto enthalpy = IapwsH2O::liquidEnthalpy(293, 13.0e5);"
because when I tried
"using FluidSystem = GetPropType<TypeTag, Properties::FluidSystem>;"
"const auto enthalpy = FluidSystem::liquidEnthalpy(293, 13.0e5);" it
gave me this error:
error: ‘liquidEnthalpy’ is not a member of
‘Dumux::OnePNIProblem<Dumux::Properties::TTag::OnePNICCMpfa>::FluidSystem’
{aka ‘Dumux::FluidSystems::OnePLiquid<double,
Dumux::Components::H2O<double> >’}
239 | const auto enthalpy =
FluidSystem::liquidEnthalpy(293, 13.0e5);
Thanks again Timo for your help !
Best Regards,
Mahmoud Atef
-------------------------------------------------------------------------------------------------
Il 2021-04-08 15:47 Timo Koch ha scritto:
On 8. Apr 2021, at 13:42, Mahmoud Atef Mahmoud Mohamed Aboelseoud
S277151 <[email protected]> wrote:
Dear Timo,
Thanks a lot for your reply. I didn't use the pointSources because I
didn't know how to assign the temperature inside such function for
the injection well because I want to impose a rate and also a
temperature.
Hi Mahmoud,
for the pointSource for the OnePNI model you need to specify mass
(kg/s) and energy source (J/s) (Indices::conti0EqIdx (=0) and
Indices::energyEqIdx (=1)).
“NumEqVector" has size 2 because you have two equations.
The energy flow can be computed as massFlow*enthalpy(p,T) (neglecting
heat conduction which is usually unimportant here). So the
temperature
enters via the enthalpy.
You can evaluate the correct enthalpy for your fluid system with
using FluidSystem = GetPropType<TypeTag, Properties:: FluidSystem>;
const auto enthalpy = FluidSystem::liquidEnthalpy(T, p);
for a temperature T (in K) and pressure p (in Pa).
In your examples below you build a temporary object of type
FluidState
and set the temperature on this object. However this fluid state
object is used nowhere, so it does nothing.
Consequently you seem to inject mass without accounting for the
energy
the injected fluid carries.
Hope this helps
Timo
I'm using the OnePNI model in my problem. I will try to be more
specific this time as I'm afraid I could be making some mistake that
I'm not aware of. To define both the injection and production wells
inside the sourceAtPos function, I'm writing the following:
NumEqVector sourceAtPos(const GlobalPosition& globalPos) const
{
NumEqVector values;
using FluidState = GetPropType<TypeTag,
Properties::FluidState>;
FluidState fs;
// we define one source term and one sink term
//Injection well (Source Term)
if ((globalPos[2] > 15 - eps_ && globalPos[2] < 35 + eps_) &&
(globalPos[1] > 45 - eps_ && globalPos[1] < 45 + eps_) &&
(globalPos[0] > 35 - eps_ && globalPos[0] < 35 + eps_))
{
fs.setTemperature(293);
values[contiWEqIdx] = 1e-1; // kg/(s*m^3)
}
//Production well (Sink Term)
else if ((globalPos[2] > 15 - eps_ && globalPos[2] < 35 +
eps_) && (globalPos[1] > 45 - eps_ && globalPos[1] < 45 + eps_) &&
(globalPos[0] > 65 - eps_ && globalPos[0] < 65 + eps_))
{
values[contiWEqIdx] = -1e-1; // kg/(s*m^3)
}
else {
values[contiWEqIdx]=0;
}
return values;
}
where contiWEqIdx is defined in an enum as contiWEqIdx =
Indices::conti0EqIdx
and my model dimensions and cells are as follows:
[Grid]
Positions0 = 0 10 30 70 90 100
Positions1 = 0 10 30 70 90 100
Positions2 = 0 10 40 50
Cells0 = 1 2 4 2 1
Cells1 = 1 2 4 2 1
Cells2 = 1 3 1
where there are upper and lower boundary layers with very low
permeability (order of e-25 m^2) each 10 m thick. I'm not using any
refinement for the moment around the wells. Is there anything wrong
I'm making here in the above function and which causes me to have a
very limited range of mass rate below which the simulation proceeds
very quickly and above which I have really high pressures and the
simulator aborts ??!
In fact I want to impose a rate of 100 m^3/day in each well as it is
a common liquid rate and that's why I do the calculations mentioned
in the previous e-mail. but in order to implement what you proposed
Timo in your response which is using
"elemVolVars[scv].density(phaseIdx)" for the density and
"scv.volume()" for the volume, shouldn't I divide the volumetric
rate of the well which is 100 m^3/day by (3 cells x 8 scv) before
dividing by the volume and multiplying by the density ? Anyway, I
tried to follow what you proposed by writing the following for the
injection well for example
NumEqVector sourceAtPos(const GlobalPosition& globalPos) const
{
SubControlVolume scv;
ElementVolumeVariables elemVolVars;
NumEqVector values;
using FluidState = GetPropType<TypeTag,
Properties::FluidState>;
FluidState fs;
// we define one source term and one sink term
//Injection well (Source Term)
if ((globalPos[2] > 15 - eps_ && globalPos[2] < 35 + eps_) &&
(globalPos[1] > 45 - eps_ && globalPos[1] < 45 + eps_) &&
(globalPos[0] > 35 - eps_ && globalPos[0] < 35 + eps_))
{
fs.setTemperature(293);
values[contiWEqIdx] =
(1.157e-3*elemVolVars[scv].density(LiquidIdx))/(24*scv.volume()); //
kg/(s*m^3)
}
where LiquidIdx is defined in an enum as LiquidIdx =
FluidSystem::comp0Idx
but I'm getting the following error:
/home/mahmoud/dumux/dumux/dumux/discretization/cellcentered/mpfa/elementvolumevariables.hh:288:7:
note: candidate expects 1 argument, 0 provided
I would really appreciate it if you let me know how to correctly
accurate for the 100 m^3/day volumetric rate either using the
sourceAtPos function with elaborating the mistakes or using the
pointSource function showing how to impose the 293 kelvin
temperature for the injection well.
I'm extremely sorry for the long e-mail. I just wanted to clarify my
problem a bit more. Thanks in advance !
Best Regards,
Mahmoud Atef
------------------------------------------------------------------------------------------------
Il 2021-04-07 20:27 Timo Koch ha scritto:
Dear Mahmoud,
what you are doing seems correct to me. You can get the relevant
cell
volume with “scv.volume()” so you don’t have to compute that
yourself.
You can also get the density with
"elemVolVars[scv].density(phaseIdx)". But that shouldn’t change
anything if you already used the correct volume in your
computation.
You can also use pointSources (see e.g
https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/blob/master/test/porousmediumflow/1p/pointsources/timeindependent/problem.hh)
which is simpler for the case you describe.
There you just specify the position of the well(s) and the rate(s)
(this time directly in kg/s without the need for considering cell
volumes (that will be done automatically)).
I guess you have to recheck your scenario to see if you would
expect
an influence on pressure for the specified production rate.
Best wishes
Timo
On 6. Apr 2021, at 18:05, Mahmoud Atef Mahmoud Mohamed Aboelseoud
S277151 <[email protected]> wrote:
Hello Dumux team,
I'm defining two wells inside a structured 3D Yasp grid, one is
injection and the other is production and I'm using the MPFA
discretization. The system is a geothermal system for which I'm
defining a certain low temperature for injection. I'm defining
both
wells inside the sourceAtPos method and they are both defined
inside
3 cells in the z direction. each cell is 10x10x10 m^3 Now I
understood from the doxygen documentation that the unit inside
that
source function is Kg/(m^3.s) and I want to impose a rate of 100
m^3/day == 1.157e-3 m^3/s in each well so I'm dividing by (3 x
1000)
and then multiplying by 1000 kg/m^3 for water to get an estimation
of the mass rate to use so I get a value around 4e-4 Kg/(m^3.s)
to
use. However whenever I use that value, the simulation proceeds
really quickly like I'm not injecting or producing anything at all
(static condition) and whenever I use a random higher mass rate
like
0.1 for example, I get very high pressures and warnings during the
simulation and the simulator aborts. Could somebody please let me
know how to correctly account for the 100 m^3/day that I want to
impose in each of the production and injection wells ? tried using
some middle values and the simulation proceeded successfully but I
don't know the basis. Your guidance is much appreciated.
Best Regards,
Mahmoud Atef
_______________________________________________
DuMux mailing list
[email protected]
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux
_______________________________________________
DuMux mailing list
[email protected]
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux
_______________________________________________
DuMux mailing list
[email protected]
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux
_______________________________________________
DuMux mailing list
[email protected]
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux
<high temperature in the production well.png><pressure boundaries
modified.png>_______________________________________________
DuMux mailing list
[email protected]
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux
_______________________________________________
DuMux mailing list
[email protected]
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux
_______________________________________________
DuMux mailing list
[email protected]
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux
