[raunakjung]

Hello, everyone.

I carried out RANS and LES simulation of a fluidic oscillator. No of nodes is 4.5 million. Time step size is 10^-5. Reynolds number is 37500. I have experimental data of streamlines of the internal flow field. When compared to experimental data the streamline by LES is streaky and the smaller vortices aren't captured. Also, the big vortices look messy and not coherent. The results look close to experimental data using SST model where structures are coherent and distinctive.

I am unsure if this is normal using LES. Please help me analyse my results and suggest improvements. Thank You

[juliom]

Are you sure that your grid lies on the inertial subrange. This is my main question. Without that information the les computation is not more but number without sense

[Santiago]

Is the LES field shown an instantaneous field or is it averaged in time?

Sent from my GT-I8190L using CFD Online Forum mobile app

[raunakjung]

Originally Posted by Santiago

Is the LES field shown an instantaneous field or is it averaged in time?

Sent from my GT-I8190L using CFD Online Forum mobile app

This is not instantaneous field. its averaged in time

[Santiago]

1. Over how many circulation times have you averaged your solution?
2. From the plots, no conclusion can be drawn about coherence and vorticity. Regions of non-zero circulation are not, in general, coherent structures.
2. What LES model are you using? The Re is a bit high. Are you resolving the wall directly?
4. Is the grid structured?
5. Have you done any higher order statistics?

Sent from my GT-I8190L using CFD Online Forum mobile app

[raunakjung]

Originally Posted by juliom

Are you sure that your grid lies on the inertial subrange. This is my main question. Without that information, the les computation is not more but number without sense

How to determine if grid is on inertial subrange? I carried out RANS simulation prior to LES simulation to find Turbulent Kinetic Energy and turbulent dissipation energy. This was used to find the length scale and taylor micro scale

Turbulent length scale L=k^(3/2)/ε
Taylor microscale λ=√(10kν/ε)
Kolmogorov scale η=L*Re^((-3)/4)
Turbulence Reynolds number Re*L=k^2/εν


v = kinematic viscosity
I am using the following estimator to calculate the mesh resolution
Δ=max⁡( λ,L/10)

My mesh size is 0.1 mm.

[raunakjung]

Originally Posted by Santiago

1. Over how many circulation times have you averaged your solution?
2. From the plots, no conclusion can be drawn about coherence and vorticity. Regions of non-zero circulation are not, in general, coherent structures.
2. What LES model are you using? The Re is a bit high. Are you resolving the wall directly?
4. Is the grid structured?
5. Have you done any higher order statistics?

Sent from my GT-I8190L using CFD Online Forum mobile app

I am trying to use WALE and LES smagorinsky model. I plan to use wall modeling. No the grid isn't structured. I have used hexahedral mesh. I haven't carried out higher order statistics.

Why is the Reynolds number high? What would be the grid point requirements? It is wall bounded flow.

Thank YOU