I am trying to simulate a 2D fluid flow past a cylinder at a reynolds number of 100. The cylinder is of about 166 lattice unit diameter. I know that the stability is dependent on relaxation time and umax. The tau is set to 1 and umax is 0.1. All these values were calculated to simulate the given problem. The total lattice is about 500*1000. But after about 200 iterations, all the values are going to NaN. How is this happening even though my tau and umax is in the safe zone? And how can i fix this? Thnx in advanceā¦

You tau is fine, the Umax is high. Try to reduce it to something like 1e-04 and see if it works.

Hi Githin,

Before running your program:

1- be sure that the unknown values in the corners are solved correctly.

2- the diameter of the cylinder (500/3) is quite big for the speed that you set. The LBM has compressibility errors of higher-order expansions. Therefore your program would crash since pressure behind the obstacle drops so sharply. You should solve Poisson equation separately, decrease the velocity speed or increase the inlet velocity slowly (until you reach the Umax) before the main loop of your program.

3- set parabolic velocity in the inlet flow if you want to model developed fluid flows (it helps to decline the errors which come from unknown values of the corners and reduce computing time).

4- set parabolic velocity for all nodes before the main loop (in order to get the final result sooner).

The best,