[nhhamidsa]

hello friends.

i'm simulating 3D dynamic airfoil using dynamic mesh in ansys - fluent. my problem is that my frequency is small and it's taking long time for just one cycle.

so how can i find the most possible large time step in order to do a correct simulation ??


here is the simulation setup data:
viscosity model = LES
free stream =75.5 (m/s)
frequency of oscillation = 9.26 (1/s)
chord length = 0.2 (m)
RE =10^6


cheers

[FMDenaro]

What about you present time-step?

[nhhamidsa]

for the first attempts i tried with e-5 or 2-e5 . is it true ?
is there any logic or formula for finding the proper time step

[FMDenaro]

From a physical point of view, the time step must be so small to describe the characteristic frequency, therefore

pi/frequency of oscillation > dt

But that could be not sufficient for the numerical stability constraint that, depending on the explicit time marching scheme and spatial discretization, requires a limit on the Courant and diffusivity numbers.

Finally, you have a very high Re therefore for performing an LES you need a refined grid with several decades (or also more) of millions of nodes.

[juliom]

Well, that is a very challenging problem. If you look at the literature you will see that LES is usually performed on lower Re numbers. The computational effort you will do is humongous as professor Denaro suggested. Using a Larger time step will definitely violate your CFL criteria if you are using an explicit scheme.

[nhhamidsa]

what is best viscosity model in terms of computational time ? if i swipe mesh in Z direction using 10 grid my grids will become 1.6 million that is very high. for this number of mesh size which turbulence model(viscosity) will works?

[FMDenaro]

Originally Posted by nhhamidsa

what is best viscosity model in terms of computational time ? if i swipe mesh in Z direction using 10 grid my grids will become 1.6 million that is very high. for this number of mesh size which turbulence model(viscosity) will works?

I do not understand what you are doing... LES is a formulation, not a viscosity model (viscosity is that of the fluid, some SGS model for LES introduces an "eddy" viscosity) and must be run in 3D, it has no physical meaning in 2D.

[nhhamidsa]

i mean which one of this models will be good for 3D modeling of dynamic airfoil?Capture.JPG

[FMDenaro]

Originally Posted by nhhamidsa

i mean which one of this models will be good for 3D modeling of dynamic airfoil?Attachment 57331

The DES formulation could be a compromise while LES requires a 3D geometry but also a very refined grid around the airfoil and some refinement in the wake.

[juliom]

I will suggest to start with a reference, I mean try to reproduce any other case you find in the literature review of your project. Said that, I've seen that this type of flow configuration has been widely studied using RANS, another formulation, very different from LES. In RANS I would use k-w. My guess is that you are using a comercial software. Thus, try to use SST k-w