# tau is close to .5

Dear all.
this is my first quastion in openlb forum. i think iti svery usefull for lBM reserchers.
Im investigating turbulence flow with lattice boltzmann method, but i have sereious problem. first of them is when i choose Re number near 10^4 and N=30 and U=.1 then :

Re=(U*N)/nu then nu=.0003
so ta=3nu+.5 then ta=.5009
this ta is near .5 and we know that this cannot bo converg
so what can i do for solving it.

one that i can do is increase N. but i think it take memory and maybe cause unstablity.
whats your idea for my problem.

Hello,

i will make two remarks on you simulation. The first one is that i guess that you wnat to simulate an incompressible fluid. The LBM is a weakly compressible solver. And in order to simulate “incompressible” flows you have to be in the small Mach number regime. The maximum U you take should therefore be around 0.02.

The second remark which is probably more important is that you are trying to solve a problem with a very unresolved grid. So your simulation should not converge, since you are not resolving the dissipative scale of your system. In order to resolve it you should have N ~ Re^(3/4) grid points per physical direction.

Hope it helps a bit.
Orestis

dear malaspin
thanks for your answer, you are right,if i dont use turbulent modeling my grid cant resolve small eddy, so i use LES for solve small eddy with smagorisky model, but i have a serios problem, i choose d2q9 and my domail consist 300*60 lattice, asn a obstacle with 30 lattic is mounted in chanel,
i use uinform flow aa inlet and zerogradian in oulet,and no slip condition for surfaces, but my problem is after some iteration the error function come in order 10^-4 and it will come constant and dosnt reduce. so my hidrodynamic parameter always change and they never converg. i want toknow is it tru for nature of lES. if yes sowhat can i do for exit my data. i want to say how can i compare my result to existing one (LARGE-EDDY SIMULATIONS WITH A
MULTIPLE-RELAXATION-TIME LBE MODEL by li shi lou)

dear malaspin
thanks for your answer, you are right,if i dont use turbulent modeling my grid cant resolve small eddy, so i use LES for solve small eddy with smagorisky model, but i have a serios problem, i choose d2q9 and my domail consist 300*60 lattice, asn a obstacle with 30 lattic is mounted in chanel,
i use uinform flow aa inlet and zerogradian in oulet,and no slip condition for surfaces, but my problem is after some iteration the error function come in order 10^-4 and it will come constant and dosnt reduce. so my hidrodynamic parameter always change and they never converg. i want toknow is it tru for nature of lES. if yes sowhat can i do for exit my data. i want to say how can i compare my result to existing one (LARGE-EDDY SIMULATIONS WITH A
MULTIPLE-RELAXATION-TIME LBE MODEL by li shi lou)

How is your error function defined?

hi
my error function is difference of velocity magniutude of two timestep as:
sigma(u(x,t+1)^2–u(x,t)^2/u(x,t)^2)