Long simulation time using LES turbulence model?
 

Hi all,

I have a simple geometry similar to a cylindrical engine shape, with the left piston-like boundary moving toward right direction. N2 is the only species. However, when I run the simulation, the time step size is limited by turbulence iteration? My geometry size is about 0.39*0.05*0.05, and I tried both coarse and refined mesh. For coarse mesh, dx*dy*dz=0.04*0.0125*0.0125, with total cell number around 250; For refined mesh, dx*dy*dz=0.015*0.01*0.01, with cell number around 1000; The time step size for coarse case is dt=3e-9s, and for refined case it's 5e-11s. Attached is the "turbulence.in" I used and "a.log" CONVERGE generated. I am wondering if there are any problems in the turbulence model I set up.

Thanks for any help or suggestions from you!

Hi all,

The current LES turbulence model is Viscous one equation (default), and I just tried another LES model -- Dynamic Structure, and the time step size is highly improved? Are there any details showing the difference of these LES models, or which one is better for my simulation case?

Best,
tmy

tmy,

In order to know more about the different types of LES models available in Converge and their differences, you should take a look at the theory manual section 15.3 LES models. In our experience, the different LES models will have not a drastic impact on the results for an engine type of simulation. In addition, I would take a look at the file time.out to see what is the timestep limiter (cfl_u, cfl_d, cfl_mach) which might give you a hint on the reason why the simulation is so slow. I would also pay attention to the initial condition for turbulence in the domain.