LBM does it stand for lattice Boltzman Model? you flow is compressible?

Hi Ahmad

Yes, It is Lattice Boltzmann and No, Flow is incompressible.

Uniform density is trivial in incompressible flows and add nothing valuable to the simulations. The question is whether you can initialize the system with the velocity field resembling that in the steady state? This might speed up thermalization, but not in a critical matter. If thermalization, i.e. reaching the steady state, is really too time-consuming then you probably are using a huge mesh. If so, try to solve your problem on a coarser lattice and then use the results to initialize the actual simulations, using interpolation where necessary.

Dear Zbigniew

Hi

Can you help me to understand how can i use a coarser mesh data for a finer one?

I am coding LBM and i can extract data on each node. But i dont know how to do what you suggested 

Thanks in Advance

Hi Mr. Alimoradi,

Although LBM is a simple and powerful  tool in modern cfd, it has some limitations. Intrinsically LBE has transient  nature and it is better not to use it for steady state problems. However it seems that convergence of your  code is slow and breaking. Boundary condition settings is a crucial subject  in LBM, so I suggest  reconsidering your BCs more carefully, because proper BC setting can make diverging code to a converging one. 

Dear Amir,

Don't code LBM unless you really have to. Use Palabos or other freely available implementation that was developed by many people for many, many years and is well tested and documented.  

As far as interpolation is concerned, try and read something about multigrid methods. While the idea behind it is rather simple, it is not easy to explain it in few words here.

Amir,

Can you tell us what is the size of your problem? What is the computational time? How you determine that you've reached the steady state? What is u_max at the steady state? What is the initial condition for the velocity?

Hi

I am newbie in LBM and i was trying to simulate Lid Driven Cavity.

Grid is 100 by 100 and i didn't even reach steady state in 50000 lattice time( about 15 min run time ).

I use difference between successive iterations for checking for steady state and

difference in velocity field was about  6e-7 at that lattice time.

For that i tried to ask for a solution for this.

Oh i almost forgot i use u=0.1 for the speed of the moving lid.

 Hi,

I think that you need a LB scheme with an accelerated convergence to a steady state solution. A seminal work has been proposed by Guo 

https://journals.aps.org/pre/abstract/10.1103/PhysRevE.70.066706

In the very short, you can see that the acceleration is promoted simply by a factor \gamma that is adopted in the equilibrium populations.

Best,

Alessandro

https://journals.aps.org/pre/abstract/10.1103/PhysRevE.70.066706