Learning Forum

[zels496]

I am running Iincompressible LES simulation for NACA0012 aerofoil. My simulation domain is 3 c upstream, 7 downstream and 3 c to upper and lower boundaries and 0.1c in spanwise direction. The LES simulation is initialized using RANS steady state simulation.

RANS steady state velocity field

i used two boundaries as inlet (velocity inlet) and i define the velocity components ( ucos(10.,usin(10.,0) and other two as pressure outlet, while the front and back are periodic.

The solver used for solving momentum equation is PISO solver ,fixed time step of 1e-6 is usedand max number of iteration/time step is 4

The schemes used in the simulation second order upwind for momentum,second order for pressure and least square cell based for the gradient

The relaxation factor is 0.3 for pressure and monetum

The problem is that the continuity residual does not reach the set residual convergence 1e-6 after 4 iteration/time step

  iter        continuity      x-velocity        y-velocity      z-velocity      time/iter

 

  61163   1.9651e-05   3.8508e-07   3.3605e-07    3.6221e-07   0:001  20

  61164   2.4330e-05   7.6474e-07   6.7809e-07    7.3229e-07   0:001  19

  61165   2.0330e-05   5.4125e-07   4.7720e-07    5.1500e-07   0:00:27  18

  61166   1.9703e-05   3.8512e-07   3.3598e-07    3.6219e-07   0:00:28  17

  step flow-time lift flow-time drag

  9845 9.8450e-03 1.0801e+00 9.8450e-03 3.8341e-02

  Flow time = 0.009844999760389328s, time step = 9845

 

is there something wrong in my case setting?

 

[Amine Ben Hadj Ali]

Have you run precursor RANS simulation where you quantified the gird size and time step size required to resolve a certain amount of turbulence kinetic energy?

[zels496]

Thanks for you reply Abenhadj

after running the RANS simulation, I used the result as initial solution for LES using (init-turb-vel-fluctuations)

I use this time step (10e-6) as it was quiet enough to keep convective courant less than 1 over the simulation domain

The y+ value over the aerofoil is between (1-3) and the first cell height (1e-5).

and x+ and z+ values are 200 and 150 respectively  

no of nodes on aerofoil is 312 and 21 nodes in the spanwise direction

chord length=0.3048m

free stream velocity=71.3 m/s and Rynold number is 1.5 million

best regards

Zein

[Amine Ben Hadj Ali]

Still coarse for real LES. Please re-check the grid in precursor run or use SBES.

[zels496]

I changed simulation to SBES but still have the same problem with continuity residuals

 

[Amine Ben Hadj Ali]

As I said still coarse your run: you need to quantify in precursor run if you are having enough resolution.

Moreover you can start from scratch now and not from steady-state result.

[Amine Ben Hadj Ali]

Add more information about your model (even the LES one) + pciture of let say 100 time steps where one can easily check the behavior of the residuals.

[zels496]

I have check y+ value and it is between 1-3, while x+=200 and Z+=150,so this means that the grid resolution is not enough for LES but i think it is reasonable for SBES

[zels496]

 

This is the residuals for LES simulation

The LES subgrid model is WALE.

The mesh is around 1.7 million control volume

 

[Amine Ben Hadj Ali]

SRS methods are not only for wall regions: if you are not paying the appropriate resolution in the bulk you are just doing an expensive URANS.

Residuals looks okay: better to plot small number of them to see whether you have a monotonic decrease or not.

[zels496]

This is the residual plot (no of iteration/time step=5) for 40 time steps

is there guidlines for the mesh away from the walls?

[Amine Ben Hadj Ali]

You might allow more iteration per time step: Say max 10 and set Report Interval to 3.

LES or SRS Grid cannot just be judged by just looking into it. But what you are sharing is fine probably very fine in the regions of no interest. For that reason a precursor run. Seems that the wake region is not really well resolved if the wake does deflect to top or bottom.

 

 

[zels496]

i will run the simulation for zero angle of attack on the same mesh and check if the continuity residuals are affected or not

[Amine Ben Hadj Ali]

Have a fun!