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
fix(prt): prevent top exits from partially saturated cells (#2478)
Cyclic particle pathlines were possible in Newton models with DRY_TRACKING_METHOD DROP (the default) due to numerically insignificant upward flows in partially saturated cells. This occurred because PRT did not distinguish the water table from the cell top, so a particle at the water table in a partially saturated cell could obtain an exit solution through the cell's top face, jump to the dry cell above, then drop back down, and so on.
Prevent particles from exiting through the top face of partially saturated cells. If a particle obtains an exit solution upwards through the water table, its behavior now depends on whether the cell top is an assigned boundary face, and the DRY_TRACKING_METHOD setting. If the cell top is an assigned boundary, or if DRY_TRACKING_METHOD is STOP, the particle will terminate. Otherwise it will stay in that location. (Future work may expand on this behavior; this is a minimal bug fix.)
The issue is reproducible on the Keating example, so I added no new tests here. I added a snapshot comparison to Keating to check the new behavior.
This fix allows removing the old ad hoc trap for cycles between vertical faces. Also some other misc cleanup.
Copy file name to clipboardExpand all lines: doc/ReleaseNotes/develop.toml
+5Lines changed: 5 additions & 0 deletions
Original file line number
Diff line number
Diff line change
@@ -97,3 +97,8 @@ description = "Fixes a memory exception that has occurred when running parallel
97
97
section = "fixes"
98
98
subsection = "solution"
99
99
description = "The PRT Model has previously been aware of boundary cell flows but not boundary faces. When an internal boundary face is assigned (an IFLOWFACE adjacent to another active cell), this can lead to cycles (infinite loops) in the particle's pathline, as the two adjacent cells fail to agree on the flow through their shared face. The PRT Model will now terminate particles at assigned boundary faces, whether they are internal or external, if the net boundary flow through the face is out of the cell."
100
+
101
+
[[items]]
102
+
section = "fixes"
103
+
subsection = "solution"
104
+
description = "Cyclic particle pathlines were possible in Newton models with DRY\\_TRACKING\\_METHOD DROP (the default) in the presence of numerically insignificant upward flows in partially saturated cells. This occurred because PRT did not distinguish the water table from the cell top, so a particle at the water table in a partially saturated cell could obtain an exit solution through the cell's top face, jump to the dry cell above, then drop back down, and so on. Particles are now prevented from exiting through the top face of partially saturated cells. If a particle reaches the water table, its behavior now depends on a combination of DRY\\_TRACKING\\_METHOD and whether the cell top is an assigned boundary face. If the cell top is an assigned boundary face, or if DRY\\_TRACKING\\_METHOD is STOP, the particle will terminate. Otherwise, it will stay there until tracking resumes the next time step."
0 commit comments