[Jokert]

Hi,

I simulated an unsteady flow by injecting two oscillating flows in anti-phase in a rectangular 2D domain. I extracted the velocity field for each time step.

In Matlab i created a massless particle tracking script using a Taylor-Galerkin integration scheme. In my loop i calculate each particle position but i noticed the positions are very dependent on the time step in the particle tracking script. If i make this time step smaller the particles still move with the flow field but they travel less far.

Does this time step needs to be the same as the time step i used in the unsteady CFD solver?

Thanks in advance!

[FMDenaro]

Originally Posted by Jokert

Hi,

I simulated an unsteady flow by injecting two oscillating flows in anti-phase in a rectangular 2D domain. I extracted the velocity field for each time step.

In Matlab i created a massless particle tracking script using a Taylor-Galerkin integration scheme. In my loop i calculate each particle position but i noticed the positions are very dependent on the time step in the particle tracking script. If i make this time step smaller the particles still move with the flow field but they travel less far.

Does this time step needs to be the same as the time step i used in the unsteady CFD solver?

Thanks in advance!

are you solving the particle tracking using a velocity database ? What about the time interval of the sampling of the velocity field?

[Jokert]

Originally Posted by FMDenaro

are you solving the particle tracking using a velocity database ? What about the time interval of the sampling of the velocity field?

Yes i do, i use the velocity database(velocity field) calculated from the unsteady solver for every timestep. The sampling interval is the same as the time step for solver.

[FMDenaro]

Originally Posted by Jokert

Yes i do, i use the velocity database(velocity field) calculated from the unsteady solver for every timestep. The sampling interval is the same as the time step for solver.

ok, I suppose you have a second order time integration but what about your interpolation on the spatial grid?
However, if the flow contains relevant oscillations within the frequency pi/dt you must integrate the Lagrangian problem using the velocity field computed at each time step dt ...