# Convergence of Eulerian-Lagrangian part. tracking

 Register Blogs Members List Search Today's Posts Mark Forums Read

 July 29, 2004, 09:50 Convergence of Eulerian-Lagrangian part. tracking #1 Sven Guest   Posts: n/a Calculating a discrete phase particle tracking in Fluent 6, I am trying to determine a reasonable time step size for the solving of the tracking equations. By trying a single particle injection, and stepwise decreasing the lenght scale or increasing the step lenght factor I expect the solution to converge to some value of residence time for the particle. Instead I find that the residence time do not stabilize to a certain converged value, but tend to fluctuate in a random pattern when decreasing the time step size (even to a very small time step). Could it be that my grid is too coarse or something..? I know that the minimum time step that is accepted by Fluent is the time step that corresponds to the "length" of a control volume. Thanks for any tips or explanations!

 July 29, 2004, 11:21 Re: Convergence of Eulerian-Lagrangian part. track #2 George Guest   Posts: n/a If your flow is turbulent and the particles have stochastic properties, then you will not get the same residence times. Regards George

 July 29, 2004, 15:06 Re: Convergence of Eulerian-Lagrangian part. track #3 wxl Guest   Posts: n/a I found even if for laminar flow, particle trajectory will not converge as I reduce the time step. Something is not right in Fluent. And I found the particle reflect boundary condition at the centerline of an 2D axisymetric tube is doing funny.

 July 30, 2004, 06:21 Re: Convergence of Eulerian-Lagrangian part. track #4 Sven Guest   Posts: n/a Thanks for your sharing your experience! That is what I also suspect - that there is something strange with the software. Perhaps somebody have tried the same thing in some ofter software environment? And in that case, any recommendations for choise of software? (laminar duct flow with free surface)

 July 30, 2004, 06:34 Re: Convergence of Eulerian-Lagrangian part. track #5 Sven Guest   Posts: n/a Thanks for your comment, George! I am modeling laminar flow in a duct, which means that there is no stochastic handling of the particle equations. I am at this moment calculating the laminar flow using a finer mesh. I am thinking that this could yield in the possibility of smaller time steps in the particle tracking calculations.