I have a code constructed to simulate the flow around the cylinder with LBM, using a body force to generate the pressure difference (using something similar to what was proposed by Gou and Zheng). The system is quantized and it works well for flow test as Poiseuille flow and Driven cavity.
But in the case of flow around a cylinder I have an error. By increasing the length x of my domain, the resulting velocities are modified (increase slightly), as if the length of the domain in x affect the results, which is incorrect.
I consider for this reason that the boundary conditions are inadequate. I am currently using the boundary conditions of Zou and He for pressure, both inlet and outlet of the flow. I noticed the mistake, since initially i had a domain of 0.1m on the y axis and 0.175m on the x axis. By increasing the length x to 0.2m, it is that the results are changed (about 5% compared with the original result).
I wonder then if the employment boundaries are correct for this type of simulations, in order to know if I should modify them or if I should look for the error elsewhere.