Hi Max,

you should first check if the problem is the linear solver by switching to a 
direct solver. If you use Dumux 2.x, that's setting a
SET_TYPE_PROP(YourTypeTag, LinearSolver, UMFPackBackend<TypeTag>);
in the problem file, for 3.0 you have to change the corresponding line in the 
main file.

Let us know how this changes things and we can decide what to do next.

Concerning the other issue about only one affected element, I'd proceed as you 
and use a graded mesh to check if it spreads over more elements once it's fine 
enough.

Kind regards
Bernd



On Thu, Mar 7, 2019 at 7:53 PM +0100, "Maximilian Johannes Lueftner" 
<[email protected]<mailto:[email protected]>> wrote:



Dear DuMuX experts,

i have adapted the simulation from the test folder
„dumux/test/porousmediumflow/mpnc/implicit/thermalnonequilibrium" .

I used the YASPGrid for 1D with grading:
struct Grid { using type =
Dune::YaspGrid<1, Dune::TensorProductCoordinates >; };

With 800 Elements (400 in the porous domain and 400 in the outflow)
and with 1600 Elements it works fin, but with more than 2000 Elements
I get the following error message:

Solve: M deltax^k = rNewton: Caught exception: "FMatrixError
[luDecomposition:/.../dumux/dune-common/dune/common/densematrix.hh:909]:
matrix is singular"

If I try it with grading I receive the following error message:

Solve: M deltax^k = rNewton: Caught exception: "NumericalProblem
[solveLinearSystem:/.../dumux/nonlinear/newtonsolver.hh:354]: Linear
solver did not converge"


Could you please help me to solve this problem? I tried it myself
several times but was not successful until now.

I suspect that an even finer grid is needed, since only in the last
element of the porous area does the temperature of the vapor increase
dramatically. Or does this behavior have a different reason? When the
heat flow is doubled, however, superheated steam is still calculated
only in the last element.


Thank you in advance and best regards,

Max

_______________________________________________
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

Reply via email to