[entropie]

Hello dear Foamers,

finally I've implemented in my solver for shape optimization the simpleFoam-equations for primal velocity U and pressure p (they are the same as equations in simpleFoam-solver, OF 2.3.0); and those for adjoint vel. Ua and pressure pa (they are the same as equations in adjointShapeOptimizationFoam, OF 2.3.0 but without alpha and lambda because I use a different gradient method for minimize cost function).


The first equations (U, p) converge after ~1500 iterations, and the adjoint equations (Ua,pa) need too much long:



Can I somehow accelerate this process? I thought, It should take about 5000-6000 iterations till the adjoint equations converge.


The BC's are ok, I've checked it already thousand times.

I thought, that maybe the mesh was too fine. At first, I have had 47000 cells. Now I have 16000 cells, but nevertheless the adjoint equations take too much time.


Have somebody any idea for this?

And thanks a lot in advance for every comment.

[Tushar@cfd]

Please have a note that CFD simulation depends upon many factors.

One of the few factor is:
Try changing the under-relation factors to speed-up OpenFOAM solver.

relaxationFactors
{
fields
{
p 0.3; // try changing these values
}
equations
{
U 0.7; // try changing these values
k 0.7; // try changing these values
epsilon 0.7; // try changing these values
....
}
}

Apart from the under-relaxation factors, there are many other factors that you can play with. Please refer any of the CFD ebook for the same.

Best Luck!

[fpalacios]

Maybe you can run the same simulation with SU2 (su2.stanford.edu)?

Cheers,
Francisco