Can a mesh be too refined for SU2?
 

Hi!


First of all I'll introduce miself, I'm an aeronautics student and I'm doing a research about actuation in the boundary layer. I'm using SU2 to implement the actuation model. I'm validating some results against Fluent for some cases.


The problem is the following: I'm simulating a low speed 2D NS incompressible simulation with turbulence (SA model). I have a very refined mesh with around 6000 elements covering the airfoild (and around 200k in total). A partner of mine has done the simulation with ansys and obtains reasonable results. On the other hand I obtain a stepped Cp profile after converging to a residual of ^-8 with Su2.


Have any of you experienced a similar problem?


Thank you in advanced,


Javi


[img]https://imgur.com/2kmWM5i[\img]


https://imgur.com/2kmWM5i https://imgur.com/a/HKFCdGxhttps://imgur.com/a/HKFCdGx
https://imgur.com/a/qVaIYPZ
This image is my result for 0 angle of attack and 10 meters per second.

Hi,

I wonder how did you generate such fine mesh? Was it by refining previous meshes? This is important since unless the mesh refinement algorithm incorporates some kind of correction to account for the real geometry (which, I think, SU2 doesn't have), the new nodes are aligned with old ones and do not quite follow the real geometry (i enclose two examples, one without geometry correction, another with geometry correction; adapted meshes are depicted in red in both cases). This can cause oscillations in the cp, especially near suction peak past the leading edge.

Hope this helps

Hi and thanks for the answer!


The mesh was generated with ANSYS by my partner (who is working with FLUENT kind in paralel to my work in SU2) and exported to .cgns format. I don't know what procedure did she follow to generate it but I guess it was just a single run with very fine sizing at the surface of the airfoil and an inflation basically.


Javi