[charan3007]

Hi everyone,
i simulated a small aircraft geometry in su2 and compared it with fluent. it is simulated for 1million Reynolds number, Mach number=0.1. i have use compressible SST, with FGMRES & LU_SGS, ROE scheme. the solution is getting converged in 600 iteration but i have noticed that SU2 is predicting Stall angle AoA=18, while fluent is predicting stall at AoA=14. and i have post processed the data where it is noticed that su2 is predicting more skin friction coefficient than fluent but interestingly the Cd is predicted similar to fluent. the following are my question:
1) is this prediction of more cl prediction is common in all flow regimes of su2.
2) if so, what could be the possible reason for it?
3) how y_plus, Cl and skin friction coefficient influence each other?

i will be very thankfully if anyone answer there. i have attached photos of cl
forum.jpg

[pdp.aero]

Originally Posted by charan3007

Hi everyone,
i simulated a small aircraft geometry in su2 and compared it with fluent. it is simulated for 1million Reynolds number, Mach number=0.1. i have use compressible SST, with FGMRES & LU_SGS, ROE scheme. the solution is getting converged in 600 iteration but i have noticed that SU2 is predicting Stall angle AoA=18, while fluent is predicting stall at AoA=14. and i have post processed the data where it is noticed that su2 is predicting more skin friction coefficient than fluent but interestingly the Cd is predicted similar to fluent. the following are my question:
1) is this prediction of more cl prediction is common in all flow regimes of su2.
2) if so, what could be the possible reason for it?
3) how y_plus, Cl and skin friction coefficient influence each other?

i will be very thankfully if anyone answer there. i have attached photos of cl
Attachment 86593

I am not going into details but I can say this is not proper way of comparing two solvers. You need to get into details and focus on specificity when you wanna compare accuracy, consistency and validity of two solvers. Based on 600 iterations, I assume you have been using a very course mesh. The difference in your solutions you see it is very normal for a very course mesh.

If you really want to compare fluent vs SU2 (which I think they will give you a very very similar accurate RANS result at the end), pick a standard test case and dive deep into details starting from a coarse, medium and fine meshes. Showing, your solutions get closer and closer to the exact let's say value, and solutions show asymptotic behavior. They don't depend on the way you did discretize domains. You can even start from a 2D or a 3D standard test case and then do the similar study for your aircraft configuration.

Over estimation of C_l max for a full aircraft configuration in RANS solution doesn't depend on specificity of solvers. It is a physical phenomena that needs to be really studied in a proper way.

Good luck,

[charan3007]

Originally Posted by pdp.aero

I am not going into details but I can say this is not proper way of comparing two solvers. You need to get into details and focus on specificity when you wanna compare accuracy, consistency and validity of two solvers. Based on 600 iterations, I assume you have been using a very course mesh. The difference in your solutions you see it is very normal for a very course mesh.

If you really want to compare fluent vs SU2 (which I think they will give you a very very similar accurate RANS result at the end), pick a standard test case and dive deep into details starting from a coarse, medium and fine meshes. Showing, your solutions get closer and closer to the exact let's say value, and solutions show asymptotic behavior. They don't depend on the way you did discretize domains. You can even start from a 2D or a 3D standard test case and then do the similar study for your aircraft configuration.

Over estimation of C_l max for a full aircraft configuration in RANS solution doesn't depend on specificity of solvers. It is a physical phenomena that needs to be really studied in a proper way.

Good luck,

hi thank you for a quick reply, actually i compared naca 0012 wing results in which su2 is on higher side compared to fluent. Based on Naca 0012 wing exercise i moved to this geometry. Naca 0012 wing i used transition sa in su2 and transition sst in fluent the results are varying, i do understand difference of turbulence model influence the results( iam attaching a picture of the naca 0012 wing ). in this aircraft geometry i used compressible sst and the mesh is same for su2 and fluent.
so as you said if it is a courser mesh refining the mesh also influence fluent results?? ( i will refine and run the cases)forum_1.png and does mesh requirement for SU2 is different for fluent?? and i observed a different in y plus predicted by SU2 & fluent.