Re: [DuMux] Calculate solution at a given position

[Etienne Ahusborde]

Dear Timo,

Thanks for your reply.

I will try your proposal.

Regards

Etienne

Le 18/02/2024 à 00:36, Timo Koch a écrit :

Dear Etienne,

yes there is. But it’s not direct. You can use evalSolution (from 
dumux/discretization/evalsolution.hh) but for that you need an element.
To find the element that your designated point is in you can use the 
boundingBoxTree of the gridView for an efficient query.
In dumux/geometry/intersectingentities.hh you find 
intersectingEntities, which you can use similar to this


|auto entities = intersectingEntities(point, 
|gridGeometry.boundingBoxTree())

auto element = gridGeometry.element(entities[0]);

to get the element that contains your point.

Best wishes
Timo


On 13 Feb 2024, at 12:10, Michel Kern  wrote:

Bonne question,  je ne trouve pas non plus a+
Michel

Envoyé depuis mon appareil Galaxy


 Message d'origine 
De : Etienne Ahusborde 
Date : 13/02/2024 10:34 (GMT+01:00)
À : dumux@listserv.uni-stuttgart.de
Objet : [DuMux] Calculate solution at a given position

Dear Users,

I'd like to inquire whether there's a feature in DuMuX that can
calculate a specific quantity, such as pressure, at a designated point.

Thanks in advance

Regards

Etienne

___
DuMux mailing list
DuMux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux
___
DuMux mailing list
DuMux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux
___
DuMux mailing list
DuMux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

[DuMux] Calculate solution at a given position

[Etienne Ahusborde]

Dear Users,

I'd like to inquire whether there's a feature in DuMuX that can 
calculate a specific quantity, such as pressure, at a designated point.


Thanks in advance

Regards

Etienne

___
DuMux mailing list
DuMux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

[DuMux] Parallel computations with poromechanics

[Etienne Ahusborde]

Dear Dumux community,

I would like to know if it is possible to run parallel calculations when 
coupling two-phase flow and elasticity as in the test present in 
dumux-3.7.0/test/multidomain/poromechanics/el2p


I had understood, perhaps wrongly, that is was not possible but I found 
recent slides presenting DUMUX according to which it seems to be possible.


Could you please confirm me this point, and if it is really possible 
which linearsolver should/could be used?


Thanks in advance

Etienne



___
DuMux mailing list
DuMux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

Re: [DuMux] Coupling DUMUX + Phreeqc

[Etienne Ahusborde]

Hi Nikolai,

Thanks a lot. It will help us a lot.

Best regards

Etienne

Le 16/03/2023 à 10:37, Nikolai Andrianov a écrit :


Dear Etienne

Please have a look at the coupling example in 
https://git.iws.uni-stuttgart.de/andrian/dumux/-/blob/master/dfm-ownc.


Best regards,

Nikolai

*From:*DuMux  *On Behalf Of 
*Etienne Ahusborde

*Sent:* Thursday, 16 March 2023 10:28
*To:* DuMuX User Forum 
*Cc:* sara.tabrizineja...@univ-pau.fr
*Subject:* [DuMux] Coupling DUMUX + Phreeqc

Dear Dumux community,

With my colleagues, we are trying to couple DUMUX with Phreeqc. We are 
not specialist with cmake and we have some difficulties in the 
compilation process. We tried to use the  Find Packages option.


We started to build a project phreeqc with duneproject command. Then 
we modified the PhreeqcMacros.cmake file in


phreeqc/cmake/modules and created the FindPhreeqcRM.cmake file (both 
files are enclosed).


We can mention that the library libphreeqcrm is present in /usr/lib 
and the file PhreeqcRM.h is present in /usr/include


Unfortunately, when we use the command ./dune-common/bin/dunecontrol 
--opts=cmake.opts all, we obtain the following information


-- The following REQUIRED packages have not been found:

 * PhreeqcRM

Can you please help us by explaining how we could do?

Thanks in advance.

Regards

Etienne


___
DuMux mailing list
DuMux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux___
DuMux mailing list
DuMux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

[DuMux] Coupling DUMUX + Phreeqc

[Etienne Ahusborde]

Dear Dumux community,

With my colleagues, we are trying to couple DUMUX with Phreeqc. We are 
not specialist with cmake and we have some difficulties in the 
compilation process. We tried to use the  Find Packages option.


We started to build a project phreeqc with duneproject command. Then we 
modified the PhreeqcMacros.cmake file in
phreeqc/cmake/modules and created the FindPhreeqcRM.cmake file (both 
files are enclosed).
We can mention that the library libphreeqcrm is present in /usr/lib and 
the file PhreeqcRM.h is present in /usr/include


Unfortunately, when we use the command ./dune-common/bin/dunecontrol 
--opts=cmake.opts all, we obtain the following information


-- The following REQUIRED packages have not been found:

 * PhreeqcRM

Can you please help us by explaining how we could do?

Thanks in advance.

Regards

Etienne
# File for module specific CMake tests.
find_package(PhreeqcRM REQUIRED)
include_directories(${PhreeqcRM_INCLUDE_DIRS})
set(LIBS ${LIBS} ${PhreeqcRM_LIBRARIES}) find_path(PhreeqcRM_INCLUDE_DIR PhreeqcRM.h)
find_library(PhreeqcRM_LIBRARY libphreeqcrm)

set(PhreeqcRM_LIBRARIES ${PhreeqcRM_LIBRARY})
set(PhreeqcRM_INCLUDE_DIRS ${PhreeqcRM_INCLUDE_DIR})___
DuMux mailing list
DuMux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

Re: [DuMux] Poroelasticity : Mandel's problem

[Etienne Ahusborde]

Dear Dumux community,

First, we wish you an happy new year.

Since our last questions about flow and poromechanical coupling on 
Mandel problem, thanks to the help of Yue Wang, we have been able to 
perform simulations by using the coupling manager for fully implicit 
approach and we greatly thank you for that.


Then we tried to reproduce in version 3.5 the test case « 4.2 CO2 
injection heterogeneous » proposed by Beck in Accuracy of fully coupled 
and sequential approaches for modeling hydro- and geomechanical 
processes, Computational Geosciences (2020) 24:1707–1723.


We used the implicit model el2p from multidomain example. We looked in 
details the source of the code in version 2.12 but when we tried to 
consider the same boundary conditions in version 3.5, we obtain some 
convergence issues. The only solution to reach solver convergence is to 
impose Dirichlet conditions on all boundaries for all displacements but 
results are obviously different.


Has somebody already tried to perform this case in a recent version of 
DUMUX?


Moreover, we would like to compare the results using implicit and 
sequential approach by using a fixed stress scheme where the two 
sub-problems (flow and poromechanics) are decoupled and solved 
sequentially. Unfortunately we are facing some difficulties on how to 
set up communications between the two sub-problems.


Like the flow problem is solved with a cell-centered scheme, we need to 
extrapolate the pressure in order to express it from cell to vertex. 
This is necessary to pass it as argument to the poromechanics problem 
which is formulated with the box method (vertex).


The same problem occur in the other direction, from poromechanical 
problem to flow problem where an interpolation is necessary to pass from 
vertex data to cell data.


Are there some existing tools in DUMUX to make some conversions in a 
simple way ?


Thank you for your answer

Best regards

___
DuMux mailing list
DuMux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

[DuMux] Poroelasticity : Mandel's problem

[Etienne Ahusborde]

Dear Dumux community,

With my colleagues, we are trying to perform an hydro-mechanical 
coupling model with Dumux.


We have first investigated the option to use the model "el2P" supplied 
in the test section of the multidomain coupling in version 3.4. In order 
to validate the coupling, we have been interesting in reproducing Mandel 
's problem benchmark by taking into account of what has been carried out 
in Martin Beck 's publications and source code from 
https://git.iws.uni-stuttgart.de/dumux-pub/beck2019a .


Unfortunately, we are not capable to obtain good results on the pressure 
field where the Mandel-Cryer effect should be captured. As depicted on 
the images that we join to this mail, we are obtaining good results for 
horizontal and vertical displacements but not for the pressure which 
decreases directly at early times.


We have no idea of where comes from the issue, this is why we are kindly 
asking if someone has already been facing this situation.


Thank you for your help




comp_pressure.pdf
Description: Adobe PDF document


Hdisplacement.pdf
Description: Adobe PDF document


Vdisplacement.pdf
Description: Adobe PDF document
___
DuMux mailing list
DuMux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

Re: [DuMux] Singular matrix and MPI

[Etienne Ahusborde]

Hi Dmitry ,

We already had the same problem.

We understood that it is due to some "DUNE_THROW" commands.

Our solution (probably not the best) was to remove several "DUNE_THROW" 
commands and to replace them with a simple std::cout to keep the 
information that something is wrong.


Regards

Etienne

Le 29/06/2022 à 19:36, Dmitry Pavlov a écrit :


Hello,

In one porous flow simulation, I get this message:

Newton iteration  6 done, residual = 3.1654e+13
Solve: M deltax^k = rNewton: Caught exception: "FMatrixError 
[luDecomposition:/home/dpavlov/DUMUX/dune-common/dune/common/densematrix.hh:939]: 
matrix is singular"


... and it is fine. Matrix can be singular, no objection here. After 
this, the Newton solver reports that it did not converge and restarts 
with a smaller time step, and eventually it proceeds.


However, when I run the same simulation with MPI, I get this

Newton iteration  8 done, residual = 7.2437e+13
Solve: M deltax^k =

and the program hangs.

So it seems that it misses the singularity in the MPI mode and tries 
to proceed with the current time step, to no avail.


How this can be cured?

Best regards,

Dmitry


___
DuMux mailing list
DuMux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

___
DuMux mailing list
DuMux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

Re: [DuMux] Model coupling in DUMUX

[Etienne Ahusborde]

Dear Timo,

Thanks a lot.

Indeed, it seems that the strategy for 1ptracer is exactly what we want 
to do so it will be a very good starting point even if the two phase 
case is more complex.


I will have probably other questions soon.

Regards

Etienne

Le 08/03/2022 à 18:34, Timo Koch a écrit :

Dear Etienne,

unfortunately there is currently no example of such a simulation setup with 
two-phase flow.
But there is a simple example for one-phase flow + transport example: 
https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/tree/master/examples/1ptracer

Consider two things:

(1) The "multi-domain” framework with coupling managers is only really needed 
for assembling coupled multi-domain Jacobians.
In the sequential case you describe this should not be necessary. Basically the 
first solver dumps everything the second solver needs to know into some 
containers (std::vectors for example)
and the second problem/spatialparams take these values. In the main fail you 
just solve two problems after each other and exchange data in-between.
This is for example done in the 1ptracer example (only that there the tracer 
quantities don’t couple back but that’s the same process).

(2) There is unfortunately no very consistent way of accessing the time 
discretisation information. This means if for example for your reactive 
transport problem you need both the saturation at the old time step and at the 
new time step,
you need to manually overload the storage term evaluation in the base local 
residual class and make sure that the correct quantities for the respective 
time level are obtained from e.g. the spatial params (if you decide to store 
saturation from the other problem there).
The spatial params would store both old and new saturations from the  two phase 
flow model in this example.

Hope this is good as a starting point
Timo




On 8. Mar 2022, at 18:13, Etienne Ahusborde  
wrote:

Hello DuMuX users,

With my colleague we work on the simulation  of thermo-hydro-chemical processes 
in porous media.

We developed a fully implicit scheme and now we would like to compare it with a 
sequential scheme.

For this we would like to solve a non isothermal compositionnal two phase flow 
and then a reactive transport problem.

For instance, the first subproblem would compute saturations, pressure, 
temperature  while the chemistry will be explicited. Then, these quantities 
would be given to the second subproblem that would calculate mole fractions of 
all chemical species and concentrations of minerals. The 
dissolution/precipitation of minerals could modify the new porosity that would 
be given to the first subproblem

We looked how DUMUX could do this. It seems that the different coupling modes 
currently available (boundary, embedded mixed dimension, conforming mixed 
dimension facet) are not suitable but maybe we are wrong.

We think that a strategy similar to the one used to coupled flow and 
geomechanics (/geomechanics/poroelastic/couplingmanager.hh) could be a good 
point of departure?

Could you please confirm us what in your opinion could be the best way to 
proceed using the current DuMuX capacities?

Thanks in advance

Regards

Etienne


___
DuMux mailing list
DuMux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

___
DuMux mailing list
DuMux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

[DuMux] Model coupling in DUMUX

[Etienne Ahusborde]

Hello DuMuX users,

With my colleague we work on the simulation  of thermo-hydro-chemical 
processes in porous media.


We developed a fully implicit scheme and now we would like to compare it 
with a sequential scheme.


For this we would like to solve a non isothermal compositionnal two 
phase flow and then a reactive transport problem.


For instance, the first subproblem would compute saturations, pressure, 
temperature  while the chemistry will be explicited. Then, these 
quantities would be given to the second subproblem that would calculate 
mole fractions of all chemical species and concentrations of minerals. 
The dissolution/precipitation of minerals could modify the new porosity 
that would be given to the first subproblem


We looked how DUMUX could do this. It seems that the different coupling 
modes currently available (boundary, embedded mixed dimension, 
conforming mixed dimension facet) are not suitable but maybe we are wrong.


We think that a strategy similar to the one used to coupled flow and 
geomechanics (/geomechanics/poroelastic/couplingmanager.hh) could be a 
good point of departure?


Could you please confirm us what in your opinion could be the best way 
to proceed using the current DuMuX capacities?


Thanks in advance

Regards

Etienne


___
DuMux mailing list
DuMux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

Re: [DuMux] How to prescribe an uniform background hydraulic head gradients in DuMuX?

[Etienne Ahusborde]

Hi Timo,

First, I would like to wish a happy new year to everyone.

Thanks a lot for your answer.  We will try your idea.

The problem when imposing only a flux on the left side is that the 
groundwater flow is unfortunately not uniform. The test case that we 
want to consider comes from this reference 
https://www.sciencedirect.com/science/article/abs/pii/S1750583615001899


Regards

Etienne

Le 23/12/2021 à 10:24, Timo Koch a écrit :
On 15. Dec 2021, at 18:16, Etienne Ahusborde 
<mailto:etienne.ahusbo...@univ-pau.fr>> wrote:


Dear DuMuX,

With my colleagues, we are interested to perform a test case of 
injection of CO2 into a reservoir with uniform background hydraulic 
head gradients.




The configuration is represented in the figure above and the 
groundwater flow is 5.5m / year.


We are wondering how it is possible to impose this groundwater flow 
in DuMuX.


We inject CO2 as a source term and we tried to impose a flux of water 
on the upstream boundary with a velocity of 5.5m / year. 
Unfortunately, it does not seem to be the good way to do.





Dear Etienne,

what was the problem with the boundary condition approach? This seems 
like the solution with the smallest perturbation of the solution 
within the domain.
I don’t fully understand the meaning of the conditions you want to 
set. Since the groundwater flow pressure field will be influenced by 
the CO2 plume and in a non-linear way fixing a hydraulic heads or even 
superimposing a gradient within the domain additively would influence 
the physics.


However, maybe in your setup this wouldn’t be important.
In that case you, could modify the flux implementation, e.g. 
Dumux::DarcysLaw (or more specifically the specialisation for your 
discretisation method) and add a position-dependent background 
pressure field (e.g. via a spatial parameter interface) to the 
pressures appearing there.
Your pressure solution would then be the pressure difference to that 
background pressure field, so to minimise the influence on the 
solution the background pressure field should probably be centred 
around 0.
You can set your modified Darcy’s law class via the property 
“AdvectionType”.
As I haven’t tried something like that myself I can’t guarantee this 
works as expected.


Best wishes
Timo


Has someone an idea to how we could impose that?

Thanks in advance for any help

Regards

Etienne

___
DuMux mailing list
DuMux@listserv.uni-stuttgart.de <mailto:DuMux@listserv.uni-stuttgart.de>
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux


___
DuMux mailing list
DuMux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

[DuMux] How to prescribe an uniform background hydraulic head gradients in DuMuX?

[Etienne Ahusborde]

Dear DuMuX,

With my colleagues, we are interested to perform a test case of 
injection of CO2 into a reservoir with uniform background hydraulic head 
gradients.


picture

The configuration is represented in the figure above and the groundwater 
flow is 5.5m / year.


We are wondering how it is possible to impose this groundwater flow in 
DuMuX.


We inject CO2 as a source term and we tried to impose a flux of water on 
the upstream boundary with a velocity of 5.5m / year. Unfortunately, it 
does not seem to be the good way to do.


Has someone an idea to how we could impose that?

Thanks in advance for any help

Regards

Etienne

___
DuMux mailing list
DuMux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

Re: [DuMux] Oil + gas simulation

[Etienne Ahusborde]

Hello everyone,

From my understanding, the main difference between the CO2 model and 
2p2cmodel is the management of phase appearance/disappearance and the 
criterion on which is based the switch of variables.


In a old version of the handbook (2.6), we could read

*The CO2 model: CO2Model*

See TwoPTwoCModel for reference to the equations used. The CO2 model is 
derived from the 2p2c
model. In the CO2 model the phase switch criterion is different from the 
2p2c model. The phase
switch occurs when the equilibrium concentration of a component in a 
phase is exceeded, instead
of the sum of the components in the virtual phase (the phase which is 
not present) being greater
that unity as done in the 2p2c model. The CO2VolumeVariables do not use 
a constraint solver for
calculating the mole fractions as is the case in the 2p2c model. Instead 
mole fractions are calculated
in the FluidSystem with a given temperature, pressure and salinity. The 
model is able to use either
mole or mass fractions. The property useMoles can be set to either true 
or false in the problem file.
Make sure that the according units are used in the problem setup. 
useMoles is set to false by default.


I hope it can help

Regards

Etienne

Le 26/03/2021 à 14:59, Beatrix Becker a écrit :

Dear Dmitry,

I am not sure about the reasons for the separate implementation in the 
CO2 model. Maybe someone else can shed light on this? The phase 
equilibrium at least seems to me equivalent to the fugacity method, 
provided BrineCO2 is used.


In any case, you want to use the 2p2c or 2pnc model, am I right? As 
long as you make sure that useConstraintSolver is set to true in the 
2p2c model, the MiscibleMultiPhaseComposition::solve() will call the 
FluidSystem::fugacityCoefficient(), as you correctly remarked. As a 
first step, I would recommend trying to use BrineCO2 as it is, with 
the 2p2c model (use constant salinity) on a test case of your choice. 
Only remove the call to Brine_CO2::calculateMoleFractions() in 
fugacityCoefficient() and instead set xlCO2 and xgH2O to constant 
values. In addition, you may have to set isIdealMixture() to true. 
Check if the results are as expected. Then copy the content of 
fugacityCoefficient() over into your custom fluidsystem.


Let me know if this helps or not!

Best regards,

Beatrix


On 26.03.21 1:38 PM, Dmitry Pavlov wrote:

Dear Beatrix,

Sorry, I got a bit confused and here I am, again asking for advice.

I took a look on BrineCO2. I learned that it is used in 
tests/porousmediumflow/co2/implicit/problem.hh. That problem also 
uses TwoPTwoCCO2 model.


The TwoPTwoCCO2 model has its own implementation of VolumeVariables: 
TwoPTwoCCO2VolumeVariables.


The completeFluidState() method in that model calculates the mole 
fractions and sets them directly. I do not see calls to 
fugacityCoefficient() there, either direct or indirect, am I missing 
something?


I do see that a similar (but more general) TwoPNCVolumeVariables 
class calls MiscibleMultiPhaseComposition::solve(), which in turn 
calls FluidSystem::fugacityCoefficient(), but, I guess, there was a 
reason to override that behavior in TwoPTwoCCO2VolumeVariables?



Best regards,

Dmitry


On 17.03.2021 15:30, Dmitry Pavlov wrote:

Dear Beatrix,

Thank you, your answer is very useful, I will take a look at brineco2.

Best regards,

Dmitry

On 17.03.2021 15:28, Beatrix Becker wrote:

Dear Dmitry,

2p2c can handle gas, yes. As for the equilibrium (I suppose you 
want to fix mass fractions), you can have a look at the fluidsystem 
brineco2 and how the fugacityCoefficient() method can be 
manipulated to fix mass fractions to a certain value:


https://git.iws.uni-stuttgart.de/dumux-repositories/dumux/-/blob/master/dumux/material/fluidsystems/brineco2.hh 



Dumux calculates equilibrium mole fractions by setting the fugacity 
of a component in one phase equal to the fugacity of the same 
component in the other phase. For two components that results in 
two equations with four unknowns (i.e., the mole fractions). To 
close the system of equation the sum of the mole fractions of a 
phase has to be one.


I hope that answers your question.

Best wishes,

Beatrix


On 17.03.21 11:50 AM, Dmitry Pavlov wrote:

Hello,

After a series of useful simulations of oil + water + some soluble 
components, I am now facing a task of oil + gas simulation. It is 
a simplification, but, as far as I understand, a more-or-less 
common one: water is considered immobile (hence nonexistent), gas 
in injected into oil; some gas gets immediately dissolved in oil, 
and the ratio of "gas in gas phase" / "gas in oil phase" is 
constant everywhere at all times. I call this an equilibrium, 
hoping this is the correct term. No temperature is considered.


I would be very grateful for pointers as to what DuMux models and 
classes I can use for this task.


Can 2p2c model handle gas? Is there an easy way to naturally write 
the aforementioned equilibrium into the equations? Should I deal

[DuMux] Second order scheme for time derivative

[Etienne Ahusborde]

Dear Dumux Users,

With my phD student we are interested in the numerical simulation of 
convective dissolution of CO2.


We would like to implement a second order scheme for the time derivative :

$$\frac{\partial Storage}{\partial t}\approx \frac{3 
Storage^{n+1}-4Storage^n + Storage^{n-1}}{2\Delta t}$$


We identified that we have to modify (at least) the files 
fvlocalassemblerbase.hh and fvlocalresidual.hh.


We have to create a variable "prevprevElemVolVars_" to store 
$Storage^{n-1}$ in fvlocalassemblerbase.hh


When we just try ro declare  ElementVolumeVariables prevElemVolVars_ in 
fvlocalassemblerbase.hh as a private variable (similarly to   
ElementVolumeVariables curElemVolVars_ and ElementVolumeVariables 
prevElemVolVars_) without any use of this variable, we obtain the 
following error during the compilation :


/home/eahusbor/DUMUX_PYRENE/DUMUX_3.1/dumux/dumux/assembly/fvlocalassemblerbase.hh:107:37: 
error: no matching function for call to 
‘Dumux::CCTpfaElementVolumeVariables, 
Dumux::TwoPTwoCVolumeVariablesImplementation5>, int>, Dumux::FluidSystems::BrineCO2Dumux::HeterogeneousCO2Tables::CO2Tables, 
Dumux::Components::TabulatedComponent, 
true>, Dumux::FluidSystems::BrineCO2DefaultPolicy >, 
Dumux::CompositionalFluidStateDumux::FluidSystems::BrineCO2Dumux::HeterogeneousCO2Tables::CO2Tables, 
Dumux::Components::TabulatedComponent, 
true>, Dumux::FluidSystems::BrineCO2DefaultPolicy > >, 
Dumux::SolidSystems::InertSolidPhaseDumux::Components::Constant<1, double> >, Dumux::InertSolidStateDumux::SolidSystems::InertSolidPhaseDumux::Components::Constant<1, double> > >, double, 
Dumux::TwoPNCModelTraits<4, true, true, Dumux::TwoPFormulation::p0s1, 4> 
>, false, true> >, false>, false>::CCTpfaElementVolumeVariables()’

  107 | , elementIsGhost_(elementIsGhost)

We don't understand this issue.

Can somebody help us?

Thanks in advance

Regards

Etienne

___
DuMux mailing list
DuMux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

[DuMuX] Cornerpoint in DuMuX: problem to install opm-grid

[Etienne Ahusborde]

Dear DuMuX users,

I would like to use a mesh in the grdecl format using the version 3.1.

So I tried to execute the test 
dumux/test/porousmediumflow/2p/implicit/cornerpoint


I loaded the modules opm-grid and opm-common but when I compiled 
everything, I had the following error during the compilation of opm


Boost version: 1.65.1
-- Found the following Boost libraries:
--   date_time
--   system
--   unit_test_framework
-- A library with BLAS API found.
-- A library with BLAS API found.
*CMake Error at /usr/share/cmake-3.10/Modules/FindLAPACK.cmake:361 
(message):*
  A required library with LAPACK API not found.  Please specify library 
location


I don't understand because I think that  I have installed all the 
package required by lapack using the command:


sudo apt install liblapack*

Using the command whereis lapack, I obtain:

lapack: /usr/lib/x86_64-linux-gnu/lapack /usr/share/man/man3/lapack.3.gz

Maybe I have to specify in the option file the exact path to find the 
lapack library. I had the same problem with blas but after the 
installation, the blas library was automatically found.


Can you help me please?

Best regards

Etienne


___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

Re: [DuMuX] CO2 storage simulation : range of temperature in co2values.inc

[Etienne Ahusborde]

Hi Christopher,

Thanks a lot for your reply.

Best regards

Etienne


Le 26/04/2020 à 12:13, Christoph Grüninger a écrit :

Salut Etienne,

use the file
   dumux/material/components/co2tables.inc
instead. This provides data up to 400°K. The files from the examples are
smaller to reduce the runtime of the tests.

You cannot run extractproperties yourself, as it is not open source.
Someone run the software and DuMuX includes only its output.

Bye
Christoph


Am 25.04.20 um 11:16 schrieb Etienne Ahusborde:

Dear DuMuX users,

I would like to perform a simulation for CO2 geological storage with a
temperature equal to 358K.

Unfortunately, in the file "co2values.inc", the temperature range is
only between 290K and 340K.

So I would like to know if it is possible for the users to generate an
other file.

It is specified that the file was generated using ./extractproperties
290 340 50 10 1 495 but I can not find this function.

Thanks in advance for your recommendations.

Best regards

Etienne

___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

[DuMuX] CO2 storage simulation : range of temperature in co2values.inc

[Etienne Ahusborde]

Dear DuMuX users,

I would like to perform a simulation for CO2 geological storage with a 
temperature equal to 358K.


Unfortunately, in the file "co2values.inc", the temperature range is 
only between 290K and 340K.


So I would like to know if it is possible for the users to generate an 
other file.


It is specified that the file was generated using ./extractproperties 
290 340 50 10 1 495 but I can not find this function.


Thanks in advance for your recommendations.

Best regards

Etienne

___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

Re: [DuMuX] Initialization of the 2 P problem

[Etienne Ahusborde]

Dear DuMuX users,

As you have probably guessed, I sent the previous email to the wrong recipient. 
Sorry.

Best regards

Etienne 

Etienne Ahusborde  a écrit :

> Bonjour Khadija,
>
> Je ne sais pas quel est trop problème mais à mon avis, il faut que tu 
> demandes à Mustapha ou à moi si on peut t'aider.
> Si  ce n'est pas le cas, il faut décrire en détail le problème que tu 
> rencontres et ce que tu as essayé de faire avant d'envoyer un mail 
> sur la liste de DUMUX.
> S'ils voient qu'on pose des questions sans avoir cherché la solution 
> ou bien qu'on ne leur explique pas bien le problème, ils ne vont pas 
> répondre.
>
> Cordialement
> Etienne
>
> khadija talali  a écrit :
>
>> Dear Dumuxers;
>>
>> I'd like to ask how to initialize the 2p problem.  My new primary variables
>> are P_n and P_w. I have imposed an initial condition on P_n and P_w but the
>> program didn't take the good initial conditions.
>>
>> I have implemented as follow in the model.hh file:
>>
>> void init(problem &problem)
>>
>> {typeparent:: init(problem);}
>>
>> but that's didn't work.
>> Best regards.
>> --
>> *Talali Khadija*
>> Doctorante en mathématiques appliquées
>> Laboratoire de modélisation mathématique et informatique
>> Ecole Nationale supérieure d'arts et de métiers (ensam) Meknès
>> Marjane 2,B.P. 15250 Al-Mansor, Meknès
>> Tel: +212 6 20 85 25 95
>> e-mail: *talali.khad...@gmail.com *
>>___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

Re: [DuMuX] Initialization of the 2 P problem

[Etienne Ahusborde]

Bonjour Khadija,

Je ne sais pas quel est trop problème mais à mon avis, il faut que tu demandes 
à Mustapha ou à moi si on peut t'aider.
Si  ce n'est pas le cas, il faut décrire en détail le problème que tu 
rencontres et ce que tu as essayé de faire avant d'envoyer un mail sur la liste 
de DUMUX.
S'ils voient qu'on pose des questions sans avoir cherché la solution ou bien 
qu'on ne leur explique pas bien le problème, ils ne vont pas répondre.

Cordialement
Etienne

khadija talali  a écrit :

> Dear Dumuxers;
>
> I'd like to ask how to initialize the 2p problem.  My new primary variables
> are P_n and P_w. I have imposed an initial condition on P_n and P_w but the
> program didn't take the good initial conditions.
>
> I have implemented as follow in the model.hh file:
>
> void init(problem &problem)
>
> {typeparent:: init(problem);}
>
> but that's didn't work.
> Best regards.
> --
> *Talali Khadija*
> Doctorante en mathématiques appliquées
> Laboratoire de modélisation mathématique et informatique
> Ecole Nationale supérieure d'arts et de métiers (ensam) Meknès
> Marjane 2,B.P. 15250 Al-Mansor, Meknès
> Tel: +212 6 20 85 25 95
> e-mail: *talali.khad...@gmail.com *
>___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

Re: [DuMuX] Problems with test_richards_box_parallel

[Etienne Ahusborde]

Hello,

I don't know if my remark will be relevant but when we started with 
parallel computing with version 2.10 of DUMUX , we had a similar problem 
("each process simply solves the full problem sequentially").


The source of problem was the choice of the linear solver. AMGBackend 
was mandatory.


Best regards

Etienne

Le 28/02/2018 à 15:47, Bernd Flemisch a écrit :
Strange. It seems that everything is compiled with MPI support. It is 
executed on two processes that don't recognized that they are invoked 
in a parallel setting. So each process simply solves the full problem 
sequentially.


I try to investigate further.

Kind regards
Bernd

On 02/28/2018 09:42 AM, Daniel Leitner wrote:

Hello,

please find attached the outputs,

Thanks,

Daniel


Am 2018-02-28 um 09:16 schrieb Bernd Flemisch:

Hi Daniel,

can you also provide the output of
make clean; make VERBOSE=1 test_richards_box
and of
ctest --verbose -R test_richards_box_parallel
?

Thank you. Kind regards
Bernd

On 02/28/2018 08:47 AM, Daniel Leitner wrote:

make test_richards_box








___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

Re: [DuMuX] Outflow boundary conditions

[Etienne Ahusborde]

Dear Georg,

In fact, we had the same idea and I can confirm that it works (see 
figure in the enclosed file).


Thanks a lot

Etienne


Le 28/11/2017 à 09:01, georg.fut...@dlr.de a écrit :

Hi Etienne,

I think an outflow boundary condition will only work correctly if the phase 
that is leaving the domain is perfectly mixed, i.e. there is only transport 
perpendicular to the outflow boundary face. You might be able to solve your 
problem by appending a kind of pipe to your outflow. This way you are able to 
use the outflow condition at the end of the pipe. I am not 100% sure whether 
this will work.

Best regards
Georg

-Ursprüngliche Nachricht-
Von: Dumux [mailto:dumux-boun...@listserv.uni-stuttgart.de] Im Auftrag von 
Etienne Ahusborde
Gesendet: Dienstag, 21. November 2017 13:50
An: DuMuX User Forum; Mohamed Id Moulay
Betreff: [DuMuX] Outflow boundary conditions

Dear DuMuX users,

With my colleagues we are working on an example of reactive flow in porous 
medium with an outflow boundary condition.

To emphasize our issue, we have considered a simpler problem using the 1p2c 
model where a component X1 is injected in pure water (see the enclosed figures 
geometry.png and permeability.png).

There are two zones of injections named Inflow1 and Inflow2 (where a flux is 
imposed and a dirichlet boundary conditions is imposed for the concentration of 
X1) and an outflow zone. The remaining boundaries are impermeable.

The pictures X1cell.png and X1box.png depict the concentration of the component 
X1 after 125s of injection respectively with the cell centered and box schemes. 
We can see at the outflow the presence of two pikes at the border between the 
outflow and the impermeable zones.

These pikes are unphysical since the maximum value of the concentration should 
be equal to the value of the dirichlet condition (that is 0.3).

We would like to know is somebody ever had and fixed this problem?

Thanks in advance.

Best regards

Etienne

___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux


___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

Re: [DuMuX] Memory Problem with AMG in parallel computation in version 2.10

[Etienne Ahusborde]

Hi Bernd,

I reply instead of my colleague Mustapha.

1- Yes it is the result of a sequential run.

2- Yes it is the result of the unmodified co2 test 
(test/porousmedia/co2/implicit/test_ccco2.cc). We have only increased 
the duration of the simulation (Tend=1e6 to 1e7s) to have a better sight 
of the problem.


The increase of memory consumption is easily visible with the command top.

Best regards

Etienne and Mustapha

Le 08/02/2017 à 13:06, Bernd Flemisch a écrit :

Hi Mustapha,

as Christoph says, we should try to make the example as minimal as 
possible.


On 02/08/2017 09:56 AM, Mustapha El Ossmani wrote:


We have run a simple test (co2 model in DUMUX)in some iterations with 
valgrind and we obtained the messages that can be found in the 
enclosed filewhich confirms that there is a loss of memory.



Is this the result of a sequential run? If no, can you also post that?
Is it the result of an unmodified test from 
test/porousmediumflow/co2/implicit? If no, do you see the same problem 
for such an unmodified test? If again no, can you share the 
modifications which you did that trigger the behavior?


Kind regards
Bernd


Have you ever met this issue ?

Best regards

M. El Ossmani


Le 07/02/2017 à 06:22, Christoph Grüninger a écrit :

Hi Mustapha,
I have never seen such problems but I am no solver expert. What you 
can do:


* Do see this problem only in parallel or also in sequential code? 
Only with AMG or also with GMRES and ILU?
* Reduce the problem as much as you can. The best would be a minimal 
piece of software that only depends on Dune-istl that reads in your 
matrix and repeatedly solves your linear system with the seen 
undesired memory consumption. Not sure whether this is possible and 
whether such minimal setup would still have the problem at all. When 
you reduce your current problem, the problem might vanish, too. This 
can help to find the cause of your issue.
* Analyze the problem with Valgrind or AddressSanitizer. Having it 
reduced might be beneficial.
* Turning on all compiler warnings and carefully evaluating them 
might help. But there are false positives but it can help.
* Maybe its worth repeating your question at the Dune mailing list 
d...@dune-project.org  as there are 
more Users and developers of istl.


Bye,
Christoph

Am 03.02.2017 um 11:14 schrieb Mustapha El Ossmani 
mailto:m.elossm...@ensam.umi.ac.ma>>:




Dear DuMu^X developers,

We are performing parallel computation with AMG solver. Due to 
problems of convergence in the newton's method, in amgproperties.hh 
we  set  the Preconditioner from Dune::SeqSSOR to Dune::SeqILU0 :


typedef 
Dune::BlockPreconditionerVType> > Smoother;
 // typedef 
Dune::BlockPreconditionerVType> > Smoother;


It seems that there is some problems of memory loss with ILU0 
preconditionner. Indeed we can see that the memory of computation 
is continually


increasing, until the computation stops with the following error 
message :


Solve: M deltax^k = rslurmstepd: Job 936902 exceeded memory limit 
(41146808 > 41058304), being killed

slurmstepd: Exceeded job memory limit

We can notice that this problem does not occur with SSOR as 
Preconditionner.


Have you ever met this issue ?

Best regards

M. El Ossmani

University of Pau

___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de 


https://listserv.uni-stuttgart.de/mailman/listinfo/dumux



___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux




___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux



--
___

Bernd Flemisch phone: +49 711 685 69162
IWS, Universität Stuttgart fax:   +49 711 685 60430
Pfaffenwaldring 61email:be...@iws.uni-stuttgart.de
D-70569 Stuttgarturl:www.hydrosys.uni-stuttgart.de
___


___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux


___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

Re: [DuMuX] Error in Parallel computation on version 2.10 of DuMuX

[Etienne Ahusborde]

Hi Bernd and Christoph,

Thank you for your reply.

In fact, the same problem occurred (computation stopped in a Newton 
iteration without crashing) in a sequential run.


For the solver, we have this line in our problem.hh file 
(SET_TYPE_PROP(HeterogeneousProblem, LinearSolver, 
Dumux::AMGBackend );) and  #include 
 without any other modification.


So by default, the solver is the BiCGSTABSolver.

I have installed and compiled SuperLU but I don't know how to link with 
DUMUX. Can you give me please, the options to put in the  file 
optim.opts in the section "CMAKE_FLAGS" to link with SuperLU or UMFPack?


Best regards

Etienne


Le 13/12/2016 à 22:50, Bernd Flemisch a écrit :


Hi Etienne,

as Christoph points out, the effect you see is that a problem occurs on one 
process and other processes are waiting. I don't expect the linear solver to 
cause the deadlock, since quite some care is taken there to avoid exactly that. 
I recommend that you recompile with debug options. This might give you a more 
meaningful error message. If not, you have to enter the joy of parallel 
debugging.

It is very unusual that the AMG is so much slower. As Christoph says, solving 
directly on the coarsest level can be important. You can try to make a minimal 
example on top of a standard Dumux model so that we can take a look.

Kind regards
Bernd


  Tue, 13 Dec 2016 16:09:25 +0100
  Christoph Grüningerwrote:

Hi Etienne,
as I did never parallel runs with DuMuX myself, I can only guess.

My guess is, that your linear solver does not converge and you don't see
the error because some process is waiting for the result from a process
that terminated due to the failed convergence. So it will never
show the error or terminate.


For the AMG solver, are there some additional parameters to give?

Do you use SuperLU or UMFPack or an iterative method after you
preconditioned with AMG? Markus Blatt often advises to use the direct
methods as the iterative ones are prone to breakdown. Might be worth a try…

Bye
Christoph

--
AddressSanitizer is not expected to produce false positives.
If you see one, look again; most likely it is a true positive!
[http://clang.llvm.org/docs/AddressSanitizer.html]


___

Bernd Flemisch phone: +49 711 685 69162
IWS, Universitaet Stuttgart  fax: +49 711 685 67020
Pfaffenwaldring 61email: be...@iws.uni-stuttgart.de
D-70569 Stuttgarturl: www.hydrosys.uni-stuttgart.de
___
___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux


___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

Re: [DuMuX] Error in Parallel computation on version 2.10 of DuMuX

[Etienne Ahusborde]

Dear DuMuX users,

After Bernd's email,  with my colleague Mustapha, we used the AMG solver 
for our parallel computations.
As Mustapha told, it solved the problem but only partially since the 
computations stopped without explanation as shown below


Newton iteration 2 done, maximum relative shift = 20.7159
Solve: M deltax^k =

The computation does not really crash but it seems to be frozen at this 
step.


So to isolate the problem, we performed the same computation in a 
sequential run with the AMG solver.
The computation was OK but 10 times longer than the same one performed 
with the default solver (ILU0BiCGSTABBackend).


Have you ever met this issue? Is it normal?

For the AMG solver, are there some additional parameters to give?

I want to precise that we use a sequential approach to solve two-phase 
flow with geochemical transport. We solve firstly a two phase flow with 
the co2 model and then a transport reactive problem.
The problem occurs in the co2 flow step and  not in the transport 
reactive step.


Thank you in advance for your reply.

Best regards

Etienne

Le 29/11/2016 à 19:17, Bernd Flemisch a écrit :

Hi Mustapha,

in Dumux 2.10, you have to change the linear solver to the AMGBackend 
to perform parallel runs. Add in the problem file:

#include 
...
SET_TYPE_PROP(LensProblem, LinearSolver, AMGBackend);

We are setting the default to AMG soon so that you don't have this 
inconvenience anymore.


Let us know if this solves your problem.

Kind regards
Bernd


___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

Re: [DuMuX] Discontinuous flux function

[Etienne Ahusborde]

Hi Bernd,

You will find in the enclosed file two questions concerning this point.

Best regards

Etienne

Le 21/03/2016 16:57, Bernd Flemisch a écrit :

Hi Etienne,

yes, this is what we do. Can you give more details on why/when this is 
problematic?


Kind regards
Bernd

On 03/21/2016 10:48 AM, Etienne Ahusborde wrote:

Dear DuMuX Community,

With my colleagues,  we would like to know how Dumux treats the 
computation of the convective flux at the interface of two heterogeneous

media for the implicit scheme.
It seems that for the absolute permeability, an harmonic mean is used 
while for the mobility, a weighted upwinding is used (with by default
mobilityUpwindWeight_=1 in 
dumux/imlicit/common/implicitdarcyfluxvariables.hh).
We wonder if an other specific treatment is performed since upwinding 
can produce non-physical solutions.


Thanks in advance for precisions.

Best regards

Etienne

___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux







remark.pdf
Description: application/download
___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

[DuMuX] Discontinuous flux function

[Etienne Ahusborde]

Dear DuMuX Community,

With my colleagues,  we would like to know how Dumux treats the 
computation of the convective flux at the interface of two heterogeneous

media for the implicit scheme.
It seems that for the absolute permeability, an harmonic mean is used 
while for the mobility, a weighted upwinding is used (with by default
mobilityUpwindWeight_=1 in 
dumux/imlicit/common/implicitdarcyfluxvariables.hh).
We wonder if an other specific treatment is performed since upwinding 
can produce non-physical solutions.


Thanks in advance for precisions.

Best regards

Etienne

___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

Re: [DuMuX] Solve an ode in DuMuX

[Etienne Ahusborde]

Dear Johannes,

Thanks for your reply.
The problem with this approach is that we need a boundary condition for 
the equation corresponding  to the ODE while in principle, the ODE only 
needs an initial condition.


Best regards

Etienne

Le 11/12/2014 10:50, Johannes Hommel a écrit :

Dear Etienne,

the approach you describe is exactly the approach I am currently 
using, with Q in (DStorage/Dt + div F - Q =0) being a function of the 
composition of the water phase concentrations.

Thus, I can tell you that it works.

Best regards,
Johannes Hommel


On 12/11/2014 10:38 AM, Etienne Ahusborde wrote:

Dear DuMuX users,

I try to explain quickly my issue. I have a problem with several 
aqueous concentrations Ci and an other solid concentration S.
All the concentrations Ci are solution of coupled partial 
differential equations (PDE) that depend on S while S is solution of 
an ordinary differential equation (ODE) depending on S and Ci.

So I would like to know if in DuMuX it is possible to solve an ODE?

Currently, we have tried to use the general  framework (DStorage/Dt + 
div F - Q =0) and to take an flux F equal to 0 in the equation 
corresponding to S.

I think that it could work but it's not really accurate.

Thanks in advance for your reply.

Best regards

Etienne Ahusborde
___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux




___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

[DuMuX] Solve an ode in DuMuX

[Etienne Ahusborde]

Dear DuMuX users,

I try to explain quickly my issue. I have a problem with several aqueous 
concentrations Ci and an other solid concentration S.
All the concentrations Ci are solution of coupled partial differential 
equations (PDE) that depend on S while S is solution of an ordinary 
differential equation (ODE) depending on S and Ci.

So I would like to know if in DuMuX it is possible to solve an ODE?

Currently, we have tried to use the general  framework (DStorage/Dt + 
div F - Q =0) and to take an flux F equal to 0 in the equation 
corresponding to S.

I think that it could work but it's not really accurate.

Thanks in advance for your reply.

Best regards

Etienne Ahusborde
___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux

Re: [DuMuX] Choice of solver for parallel computations

[Etienne Ahusborde]

Hi Dumux,

After many test, I can confirm that the best choice for parallels runs 
seems to be the amg solver. We use SUPERLU as solver for the coarse grid 
(we have not tried yet UMFPACK).
For our implicit problem, we used the box models for spatial 
discretization and we found that the cell centered approach seems to be 
faster or at least more stable.

All the tests have been done with the co2 model

Then, we have put our developments in the version 2.6 to test our whole 
problem. For sequential run, everything is OK so we have tried to 
compile in parallel.
Unfortunately, we have the following message error (the full message of 
compilation is enclosed):


/home/eahusbor/DUMUX_2.6/dune-common-2.3.0/dune/common/parallel/indicessyncer.hh: 
In member function ‘void Dune::IndicesSyncer::calculateMessageSizes()’:
/home/eahusbor/DUMUX_2.6/dune-common-2.3.0/dune/common/parallel/indicessyncer.hh:664:72: 
error: wrong number of template arguments (1, should be 5)
 CollectiveIterator collIter = remoteIndices_.template 
iterator();

^
In file included from /usr/include/c++/4.8/bits/stl_algobase.h:65:0,
 from /usr/include/c++/4.8/bits/char_traits.h:39,
 from /usr/include/c++/4.8/ios:40,
 from /usr/include/c++/4.8/ostream:38,
 from /usr/include/c++/4.8/iostream:39,
 from 
/home/eahusbor/DUMUX_2.6/dune-common-2.3.0/dune/common/parametertree.hh:11,
 from 
/home/eahusbor/DUMUX_2.6/dumux-2.6.0/dumux/common/basicproperties.hh:29,
 from 
/home/eahusbor/DUMUX_2.6/dumux-2.6.0/dumux/modelcoupling/common/simplecoupledproblem.hh:26,

 from simple1p1ptestproblem.hh:28,
 from Durlofsky.cc:25:
/usr/include/c++/4.8/bits/stl_iterator_base_types.h:118:12: error: 
provided for ‘templateclass _Pointer, class _Reference> struct std::iterator’

 struct iterator

We don't understand the problem so we wonder if you have already met the 
same issue? I can precise that we use an additional library call GNU 
scientific library (gsl) and may be it could be the reason of this 
conflict


Thanks in advance

Best regards

Etienne

Le 21/10/2014 17:47, Bernd Flemisch a écrit :

Hi Etienne,

for me, it is very difficult to make sense of speedup times on today's 
multi-core CPUs. In addition, for the AMG solver, I am not so 
surprised, because there is no parallel overhead at all for one core. 
The speedup from two to four cores seems very reasonable. In order to 
go into more detail, you should do the following:


- Set the verbosity level of the linear solver to 2 by, e.g., providing
[LinearSolver]
Verbosity = 2
in your .input file. Send the complete output of the runs to this list.

- Report/compare the times that are needed by the linear solvers only, 
as given in the lines "Assemble/solve/update time:" of the output. 
Comparison is only fair for the same number of Newton steps there.


- Do you use SuperLU for the coarse solves? If not, you might want to 
apply istl-2.3.0.patch in the patches/ directory to dune-istl in order 
to reduce the tolerance of the iterative coarse grid solver.


I would recommend the AMG for all parallel runs. As you experienced, 
the other Dumux parallel solvers get into trouble already for very 
moderate process numbers. Since we are not able to maintain and 
definitely not able to improve these solvers, we most likely will get 
rid of them in the not-to-distant future.


Kind regards
Bernd

On 10/21/2014 03:07 PM, Etienne Ahusborde wrote:

Dear DuMux developers,

With my colleagues we are working on two-phase multicomponent flow 
with reactive transport and application to CO2 storage.
The problem is decoupled into two subproblems. The first one is 
devoted to a two-phase two-component flow (H2O-CO2) while the second 
one is devoted to a one-phase multicomponent reactive transport 
problem for the other components.


We have successfully performed sequential simulations. No we would 
like to perform parallel computations but we have some difficulties 
to chose . In a previous mail, Bernd told to a DUMUX user that for 
your parallel computations, you used AMG solver. Consequently we 
performed some tests to compare the performances of the AMG solver in 
comparison with the solver by default and we found a strange behaviour.


To explain our problem, we have installed the last version 2.6 of 
DUMUX. We use the CO2 model on a 3D mesh (8 elements). The 
problem deals with injection of CO2. We use ALUGRID for the mesh. We 
want to do only one iteration (DtInitial = 86400, TEnd = 86400) and 
compare the time of simulation on 1, 2, 4, 8,... processors. The 
times of simulation are displayed below:


#procs  bicgamg
1 212.089315.145
2 118.446401.046
4 292.873244.24
6 not doneStopped coarsening because of rate breakdown

For the bicg, with 4 processors, Newton solver

[DuMuX] Choice of solver for parallel computations

[Etienne Ahusborde]

Dear DuMux developers,

With my colleagues we are working on two-phase multicomponent flow with 
reactive transport and application to CO2 storage.
The problem is decoupled into two subproblems. The first one is devoted 
to a two-phase two-component flow (H2O-CO2) while the second one is 
devoted to a one-phase multicomponent reactive transport problem for the 
other components.


We have successfully performed sequential simulations. No we would like 
to perform parallel computations but we have some difficulties to chose 
. In a previous mail, Bernd told to a DUMUX user that for your parallel 
computations, you used AMG solver. Consequently we performed some tests 
to compare the performances of the AMG solver in comparison with the 
solver by default and we found a strange behaviour.


To explain our problem, we have installed the last version 2.6 of DUMUX. 
We use the CO2 model on a 3D mesh (8 elements). The problem deals 
with injection of CO2. We use ALUGRID for the mesh. We want to do only 
one iteration (DtInitial = 86400, TEnd = 86400) and compare the time of 
simulation on 1, 2, 4, 8,... processors. The times of simulation are 
displayed below:


#procs  bicgamg
1 212.089315.145
2 118.446401.046
4 292.873244.24
6 not doneStopped coarsening because of rate breakdown

For the bicg, with 4 processors, Newton solver does not converge with 
Dt=86400 and the simulation needs 4 iterations to reach Tend so the 
computation is slower than for one processor.
For the amg solver, the behaviour is even more strange since the 
computations is faster with one processor than with two processors. With 
6 processors, the computation stopped but I think that it is due to the 
small size of our mesh.
We expected that the amg solver will be faster that the bicg but it is 
not the case.


Have you already found the same behaviour with your simulation? Is it 
normal in your opinion? Which solver do you advise us?


We have also seen in the mail announcing the DUMUX 2.6 release that a 
new direct solver UMFPACK is available. We would like to know how it is 
possible to test it.


Thanks in advance.

Best regards

Etienne
___
Dumux mailing list
Dumux@listserv.uni-stuttgart.de
https://listserv.uni-stuttgart.de/mailman/listinfo/dumux