[SOLVED] Complex output from StessSolve
-
- Posts: 2237
- Joined: 25 Jan 2019, 01:28
- Antispam: Yes
Re: Complex output from StessSolve
Apparently it is truly imaginary!
Re: Complex output from StessSolve
Yea, physics still works , if only I could see and pass it to another solver
-
- Site Admin
- Posts: 4812
- Joined: 22 Aug 2009, 11:57
- Antispam: Yes
- Location: Espoo, Finland
- Contact:
Re: Complex output from StessSolve
Hi
Maybe you could replace "Harmonic Analysis" with "Harmonic Mode" in the sif file. This could help in the postprocessing in this case.
So what is the problem generally: most solvers, also including StressSolver, were initially written as a real valued solvers. Then at a later stage a possibility was added to use the initial matrix equation Mx''+Kx=F also to solve eigenmodes or harmonic solution. In these cases the resulting field is complex so the original matrix equation is made complex on-the-fly. The results for the harmonic case are then copied back to the real valued solution vector, Var % Values. This allows for the visualization of the real part, but not the harmonic part. You still have the full complex solution in Var % EigenVectors, but it is not easily available.
When we were solving some rather cumbersome cases with harmonic FSI coupling it was not enough to do this change to harmonic system on-the-fly since we needed to have full access to the complex displacement field and complex flow field. Hence a new strategy for treating the harmonic case was introduced. There the system is two times bigger and the component 2::2 is associated to "Diplacement Im" and the component 1::2 to "Displacement Im". The changes were quite significant so the old procedure is still maintained as "Harmonic Analysis" and the new one is called with "Harmonic Mode".
For eigenvectors we have similar issue. I guess the primary output method is VTU files. I just added a keyword for ResultOutputSolver that can choose between the type of field the keyword is
and will commit this after running tests. This is hardly perfect since it will only allow to save either component. One could perhaps have two instances of the solver with different names.
It would be ideal if VTU files would include complex valued datatype. But I guess this is not the case.
-Peter
Maybe you could replace "Harmonic Analysis" with "Harmonic Mode" in the sif file. This could help in the postprocessing in this case.
So what is the problem generally: most solvers, also including StressSolver, were initially written as a real valued solvers. Then at a later stage a possibility was added to use the initial matrix equation Mx''+Kx=F also to solve eigenmodes or harmonic solution. In these cases the resulting field is complex so the original matrix equation is made complex on-the-fly. The results for the harmonic case are then copied back to the real valued solution vector, Var % Values. This allows for the visualization of the real part, but not the harmonic part. You still have the full complex solution in Var % EigenVectors, but it is not easily available.
When we were solving some rather cumbersome cases with harmonic FSI coupling it was not enough to do this change to harmonic system on-the-fly since we needed to have full access to the complex displacement field and complex flow field. Hence a new strategy for treating the harmonic case was introduced. There the system is two times bigger and the component 2::2 is associated to "Diplacement Im" and the component 1::2 to "Displacement Im". The changes were quite significant so the old procedure is still maintained as "Harmonic Analysis" and the new one is called with "Harmonic Mode".
For eigenvectors we have similar issue. I guess the primary output method is VTU files. I just added a keyword for ResultOutputSolver that can choose between the type of field the keyword is
Code: Select all
Eigen Vector Component = String "re" ! or "im" or "abs"
and will commit this after running tests. This is hardly perfect since it will only allow to save either component. One could perhaps have two instances of the solver with different names.
It would be ideal if VTU files would include complex valued datatype. But I guess this is not the case.
-Peter
Re: Complex output from StessSolve
Hi Peter,
thank you very much for the fast help, "Harmonic Mode" works fine for me.
Cheers
Raphael
thank you very much for the fast help, "Harmonic Mode" works fine for me.
Cheers
Raphael