SU2 LES, discrete adjoint
 

Dear SU2 community,

based on the papers I have seen SU2 seems to be a nice code for performing URANS simulations. In combination with the Ffowcs Williams and Hawkings analogy and the algorithmic differentiation for receiving the discrete adjoints an acoustic optimization framework was already presented (An Efficient Unsteady Aerodynamic and Aeroacoustic Design Framework Using Discrete Adjoint, Zhou et al.).

For broadband noise predictions a LES model would be required. Receiving design sensitivities for a geometry for influencing this far field broadband noise (or at least the hydrodynamic sources around the geometry) would be interesting.
So e.g. OpenFoam offers LES but no adjoints. SU2 in turn offers algorithmic differentiation of the entire solver for receiving adjoints but "only" URANS.

Finally, is it a) possible with SU2 and if yes b) would it be an own dissertation to perform the following:
1. Implement a simple LES model (e.g. Smagorinsky)
2. Use the algorithmic differentiation of the entire solver to receive discrete adjoints

Any answer or literature recommendation to educate myself and other possible readers would be appreciated :).
Almost nothing relevant found regarding LES in the forum. Just the following:
Hybrid RANS/LES models, where someone mentioned it is very simple to include DES into SU2.
https://www.cfd-online.com/Forums/su...es-models.html

Thanks ahead

Hi,

we are already working on DA with (D)DES. Have a look at this branch in case you want to know more about the implementation of the latter.

Tim