[mechy]

Hi all
I need an adjoint shape optimization solver in openfoam which can moves the geometry. in other words it uses grid points as the design parameters( instead of porosity in adjointShapeOptimizationFoam )

does this solver exist ?

If not I need some references which explain its algorithm with more details specially about calculating the gradient of object function

any help will be appreciated.

[grgkarpou]

Hi,
Basically, you need the surface sensitivities.
The ones at the adjointShapeOptimizationFoam are volumetric sensitivities (design variables = porosity). You can calculate the surface sensitivities very easily.
Just google surface sensitivities and you will definitely find it. This field will be calculated at walls and then you can move the points with a mesh motion tool or something similar (in-house).

Note: The adjointShapeOptimizationFoam implementation is not actually right from the mathematical point of view. So before you calculate the sensitivities you have to correct the solver first.

G

[samiam1000]

Hi,

I am gonna use the adjointShapeOptimizationFoam solver, too.

What's wrong with it?

[brugiere_olivier]

Hi,

I'm using a solver which comes from Chalmers : http://www.tfd.chalmers.se/~hani/kur...seFiles.tar.gz

I've no problem with this new version. But I don't find how I can calculate surface sensitivities. I read Othmer's paper but I don't understand how I can implement this equation. So If somebody can give some informations thanks a lot.

[mechy]

Originally Posted by grgkarpou

Hi,
Basically, you need the surface sensitivities.
The ones at the adjointShapeOptimizationFoam are volumetric sensitivities (design variables = porosity). You can calculate the surface sensitivities very easily.
Just google surface sensitivities and you will definitely find it. This field will be calculated at walls and then you can move the points with a mesh motion tool or something similar (in-house).

Note: The adjointShapeOptimizationFoam implementation is not actually right from the mathematical point of view. So before you calculate the sensitivities you have to correct the solver first.

G

Hi George
can you tell me what is wrong in adjointShapeOptimizationFoam ?

[Nigò]

Hi I'm also interested in adjoint and I'm trying to implement the surface sensitivity, can you post again the Chalmers link to the solver because seem than now doesn't work.

[elvis]

I tried the links and they seem to be okay
http://www.tfd.chalmers.se/~hani/kurser/OS_CFD_2013/

http://www.tfd.chalmers.se/~hani/kur...Adjoint_UN.pdf
http://www.tfd.chalmers.se/~hani/kur...ortAdjoint.pdf
http://www.tfd.chalmers.se/~hani/kur...seFiles.tar.gz

[crixman]

Hi all,
wanted to reopen this topic to inform you about the Sig Numerical Optimization.

http://openfoamwiki.net/index.php/Si...l_Optimization
https://www.linkedin.com/groups?gid=8137157

We are willing to improve the adjointShapeOptimizationFoam solver, so if anyone is working on that and would like to give its contribution feel free to join the group!

[fpalacios]

Originally Posted by crixman

Hi all,
wanted to reopen this topic to inform you about the Sig Numerical Optimization.

http://openfoamwiki.net/index.php/Si...l_Optimization
https://www.linkedin.com/groups?gid=8137157

We are willing to improve the adjointShapeOptimizationFoam solver, so if anyone is working on that and would like to give its contribution feel free to join the group!

Have you tried SU2 (su2.stanford.edu)?

Cheers,
Francisco

[crixman]

Not yet but will start soon! Thanks for the update, I wasn't aware that SU2 could perform adjoint optimization!

[fsifsi]

Hello All,
Anyone have boxWindCase file ? I want to test adjointSensitivityFoam solver in OF 2.3.0.
Thanks

[brugiere_olivier]

Hi Le,

you can find Simon Lindberg's work on :

http://www.tfd.chalmers.se/~hani/kurser/OS_CFD_2014/

On this site you will find solvers and report.

Regards