Hi there
I want to investigate how relaxation parameter influences on the solution.
I read Timm Krüger famous paper: Shear stress in lattice Boltzmann simulations, based on eq. 35 of the paper, if Re and Ma are fixed, we can adjust Tau with the adoption of resolution.
For this, I use Poiseuiile2d case and change resolution of the geometry from 120 to 1500, and calculate relaxation time using getTau(). As I know, best choice for tau is between .8 and .9, But when I increase resolution, simulation time is increased.
Furthermore, in the base setup of the program, Tau is about .5, and results sound good! why?
Is it reasonable?
How can I change the relaxation time in the program?
If I change Ma number, I think that i cannot compare my results.
as i mentioned, i have expected that Tau in the program must be between .8 to 1.1.
I want to find best parameters for the program. my case is Poiseuiile problem. Re = 100 and Ma = .02 (Ulb)/.58 = .0344
thus, I can adjust Tau and Resolution (based on Timm Paper’s eq . ).
for fixed Ma and Re, the spatial resolution and the relaxation parameter are coupled via $\tau - 0.5 = const * N$ where N is the number of grid points along a given axis, cf. Eq. (35) in the paper.
By increasing N, you have to increase \tau in such a way that the equation is still satisfied.
Example:
If \tau = 0.6 and N = 20, we find that the constant must be 0.005. From this, it follows that for N = 40 you have to take \tau = 0.7 and for N = 80 one takes \tau = 0.9.