Time-step in particle tracking

[Cabaset Elise]

Hello,

I am currently using GeoDict (FilterDict module) and I have a question regarding the temporal resolution used to integrate the particle equation of motion in a Lagrangian framework.

I could not find clear information in the documentation about how the time step is defined or controlled when solving the particle equation of motion. Is the time step fixed or adaptive? And where can this information be accessed?

In addition, when visualizing particle trajectories, I noticed that a spatial resolution of 0.5 voxel is used by default. Is this parameter related to the temporal resolution (time step), or is it only used for visualization purposes?

More generally, how can one assess the accuracy or convergence of particle tracking simulations with respect to the time discretization in GeoDict?

Thank you in advance for your help.

[Jürgen Becker]

Dear Elise,

we use a semi-implicit scheme to compute the particle motion. The time step is adaptive. The time step size is computed in each step such that the distance traveled by the particle in one step is bound.
We have not published the details of the implementation. 

It is possible to increase or decrease the step size by a user-defined factor. For this, you can use the TimeStepScaling expert setting:


But there are some caveats: if you decrease the time steps too much, some particles might erroneously be detected as "not moving" or "filtered". If you increase the time step size with a factor larger than 1.0, particle-wall collisions might go undetected which can lead to various errors.

Not all the computed steps are stored for particle trajectory visualization - the typical distance traveled in a time step is much smaller than 0.5 voxel. You can change this distance on the Output options tab - be aware that smaller numbers will lead to much larger *.gpt files. 

Regards,
Jürgen