| Author | Message | ||
     Jens Baumann (max) Member Username: max Post Number: 4 Registered: 07-2004 |
Hello, at the moment I run a quiet large 3D simulation in flexpde 3.10a. An interesting message appeared: "Exceeded regrid limit. 32 cells exceed errlim. Max err is 2.14e-4. Continue gridding?" I clicked no and got the result. So everything is OK, but I would like to know: What is the (default) critical number of cells to exceed the given limit in order to make this message appear? Can one set this number of cells? Is it possible to locate the cells (by viewing in a figure or by coordinates)? Thanks. | ||
| Robert G. Nelson (rgnelson) Moderator Username: rgnelson Post Number: 324 Registered: 06-2003 |
FlexPDE applies a limit to the number of regrids, in an attempt to avoid excessive regridding activity in pathological cases. If the error estimate in any cell is above the ERRLIM limit, all cells exceeding the limit will be split and the problem re-run. If the total number of solution grids exceeds the limit, the warning is issued. Obviously, you have asked for a smaller ERRLIM than the default 1e-3, or you would not have seen this warning at a maximum error of 2e-4; the problem would have terminated. You can continue from this warning, and the problem will continue regridding until all cells are below threshold, or the number of additional regrids exceeds the limit. If the limit is reached again, you will get another warning. In this case, you can decide whether a worst-cell error of 2e-4 warrants continued computing or not, and continue or stop. The default is eight grids before warning. The limit can be set by SELECT GRIDLIMIT=<number> In BATCH problems, the problem will merely terminate normally when the limit is reached. See GRIDLIMIT and BATCH in the Help Index. | ||
| Robert G. Nelson (rgnelson) Moderator Username: rgnelson Post Number: 325 Registered: 06-2003 |
PS At present there is no way to locate the offending cells. |