Hi all,
I want to use the solution of a turbulent flow from PyFR as an input for a Lagrangian simulation. The problem is computing the trajectories of the particles advected by the flow are sensitive to local errors in the fluid velocity field and can diverge. The trajectories can be computed one at a time since they do not interact and they are not coupled to the solver (flow not affected by the presence of particles). Because of that, I dump my flow solution from PyFR and convert it to .VTU so I can after read it with the Lagrangian software and use VTK libraries to extract velocities at different positions. The problem is in the conversion from .pyfrs to .vtu information is lost through the -d N linearisation stage and the computational advantage of having a coarser mesh it is lost. Furthermore, making N high, when resolving particle trajectories, makes the inverse search from particle positions to fluid cells really expensive in an unstructured mesh. Would it be possible to dump the internal PyFR polynomial information local to each cell so it can be used in a post-processing stage? This would be useful for any kind of post-processing doing integration over the fields.
Regards,
Eduardo