Von Karman Vorticity

Is it possible with a BGK model to simulate a von karman vortex around a cylinder? Which parameters are the responsables for this phenomena?
Right now i´m ussing a pouiseuille flow with a Re=100 and a cero gradient output in a 400x51 grid, but nothing happens???



Yes it is! But you have to break the symmetry of your simulation. Try putting the center of your cylinder a bit off the middle of the channel (one cell upwards for example). This should do it.

Malaspin thanks for your post,

But still nothing is happening, i broke the symetry, and i fixed Re=100 in the same way it´s done on the lbm code example cylinder, i set the umax of the pouiseuille flow at the inlet, with that i calculate the cinematic viscosity and omega.

Cuould it be a mistake in the unit conversion? I allready read the document posted and i think i´m doing the same as at the example, what i am doing wrong?!!!

Does the relationship between the cylinder´s diameter and the channel´s width special?

This is the only simulation that is missing in order to finish my undergraduate thesis, help me!!!


do you use no-slip or periodicity on top/bottom boundaries (i guess no-slip is more favorable to unstable flow)? what bc do you use on inlet/outlet? zou/he?

Thanks for your post Adam,

At the Top/Bottom BC´s i´m using non-slip boundaries, and at the output i´m using zero gradient on rho and ux, and at the input BC´s with a poiseuille flow. All the BC´s are Zou and He.

Something weird happened. I doenload cylinder.m and fix my model with the exact parameters that Jonas used on his model, to try to understand what is happening with mine, but it reaches inestability. I don´t know why? The only way to keep it stable is to change the omega, but happens the same, the flux stabilizates after some time and the vorticity never appears!!!

Thanks for your help


It is true that the Matlab sample on the LBMethod.org web page did not produce a fully developed vorticity street, as it was mainly intended at illustrating how to program a LB code in Matlab, instead of analyzing the physics of the cylinder flow.

I modified the code cylinder.m by implementing a Zou/He boundary condition, and put it on the web site. When downloading the new version, consider that you may need to reset your browser’s cache to prevent it from providing the old cached version. This version clearly shows a Karman Vortex street, at Re=100, in a 100x400 geometry, with fixed Poiseuille profile at the inlet (Zou/He) and a pressure outlet (Zou/He). Actually, there are several ways of getting a Karman vorticity street, and your von Neumann outlet condition should work as well (try again with a higher resolution).

Note that the Mach number is relatively high (u=0.1) to avoid numerical stability issues with the boundary conditions. These issues can also be circumvented by using a higher resolution, or by implementing a more stable boundary condition.