[Fabian_W]

Hello,

this is my first post, so I introduce myself quickly. I am Fabian, a Postdoc in Germany. I do topology optimization and now work on an OF based project. Someone else wrote most of the solver and other stuff, my task is to change the code in such a way, that a second problem is solved afterwards. I do it in the same code.

I have good C++ knowledge but am totally new to OpenFOAM.

The forward problem is solved for h with dynamic boundary conditions defined based on the time in 0/h as inlet code. The forward problem is nonlinear with implicit time stepping.

I need to solve a so called adjoint problem which is linear, needs to be solved with explicit time stepping and goes backwards in time. I do the backwards in time by going forward in time (resetting my time object) but have for the solution variable the boundary conditions with other times in the dynamic 0/lambda code.

I have

fvScalarMatrix lambda_eqn(
phi * C * fvm::ddt(lambda) == -fvc::laplacian(k, lambda)
);
lambda_eqn.solve();

and hope, this solves my problem with an explicit time stepping.

in fvSolution I have

solvers
{
h
{
solver PCG;
preconditioner DIC;
// tolerance 1e-06;
tolerance 1e-14;
relTol 0.1;
}

lambda
{
solver PCG;
preconditioner DIC;
// tolerance 1e-06;
tolerance 1e-14;
relTol 0.1;
}


hFinal
{
$h;
relTol 0.0;
}
}

In fvSchemes I have

laplacianSchemes
{
default none;
laplacian(k,lambda) Gauss linear uncorrected;
laplacian(k,h) Gauss linear uncorrected;
}

For the forward problem the inlet code of 0/h is executed (I do a command line output) and then it is written "DICPCG: Solving for h, Initial residual ..."

For the adjoint problem again the code from 0/h is executed and then it is written " diagonal: Solving for lambda, Initial residual = 0, Final residual = 0, No Iterations 0"

My problem is, that the inlet code from h is executed but not from lambda ?!

Any help or hint is very much appreciated!

Thank you very much, Fabian