CFL condition violation--large velocity on water/land boundary

0 votes
asked Jun 29, 2017 in AEM3D by [email protected] (220 points)
I am attempting to run some simulations on a very shallow lake, but I am getting this error:
 * in subroutine sl_momentum_hi_cfl                               
 * ELM method stopped before complete
 *  Cell:       333326
 *  ijk:          114          18          29
 *  uvw:   0.0000000E+00   1571.319      6.9824216E-04

The CFL condition is being violated, so I reduced the timestep drastically but I still can't get the simulation to converge.  The interesting thing is that the (114,18) is a land cell, but (113, 18) is a wetted cell, so it's on the boundary.  The cell (113,17) is the dam outlet.  It looks like the v-direction velocity is becoming non-physically large, and setting off the CFL condition.  Any ideas on how to correct this?

1 Answer

0 votes
answered Jul 11, 2017 by dallimor (5,460 points)
Hi Deena, sorry I just saw this.  It is probably worthwhile making some movies of your output before the crash to see if you can see what's going on.  Sometimes high velocities are caused by instabilities in the free surface so it may be worth looking at the HEIGHT output.

commented Jul 13, 2017 by [email protected] (220 points)
Thank you, Chris.  I figured out that this occurs at an elevation of about 272.001m, and the bottom of the outflow cell is located at 272m, so an instability in velocity arises due to the very shallow volume in the outflow cell.  The workaround I determined was to restart the simulation slightly below 272m and close that outlet in the restarted simulation.  Not the most elegant, but it works!
Welcome to Hydronumerics Q&A, where you can ask questions and receive answers about the software we develop and distribute


  • Suite 103 757 Bourke St Docklands Victoria 3008
  • Suite 103 757 Bourke St Docklands Victoria 3008
  • +61 3 9037 0892
  • [email protected]
  • Twitter