You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently the solver will crash reliably if there are stuck parcels and the mesh is changed (e.g. in a parallel run after load balancing where each processor will have a freshly assigned mesh). In such a case the parcels stuck properties (patch and patch face that a parcel is stuck to) are not valid anymore.
Edit: This happens only if a stuck parcel is desorbed after the mesh has changed. Setting a finite residenceTime is therefore necessary to reproduce the issue.
As there is no simple / elegant way to get a global patch / face identifier in parallel runs and this would also introduce additional overhead, the most elegant solution seems to be to relocate the parcels when the solver is restarted.
See related pull request for a fix.
The text was updated successfully, but these errors were encountered:
Currently the solver will crash reliably if there are stuck parcels and the mesh is changed (e.g. in a parallel run after load balancing where each processor will have a freshly assigned mesh). In such a case the parcels stuck properties (patch and patch face that a parcel is stuck to) are not valid anymore.
Edit: This happens only if a stuck parcel is desorbed after the mesh has changed. Setting a finite residenceTime is therefore necessary to reproduce the issue.
As there is no simple / elegant way to get a global patch / face identifier in parallel runs and this would also introduce additional overhead, the most elegant solution seems to be to relocate the parcels when the solver is restarted.
See related pull request for a fix.
The text was updated successfully, but these errors were encountered: