Learning Forum

[Zwernjayden]

I am running a VOF-to-DPM simulation similar to this video but with evaporating liquid and droplet DPM instead of inert. Its a case of two impinging liquid jets, where one is ethanol/water mixture and the other is liquid oxygen. The simulation runs fine, except when I enable enable evaporation/condensation in the Multiphase menu (interface heat flux method), even if the temperature in the entire domain stays below the saturation temperture of the liquid. I get "floating point exception" consistantly around the time the liquid jets enter the main impingment zone, where the mesh adaption takes place, so I think its likely related. I have tried the Lee evaporation model and get the same result.

Here is how the adaptive mesh is defined

REFINE

Type: Cells More Than

Derivative Option: Curvature

Curvature of: Phases, volume fraction

Phase: ethanol/water or liquid oxygen

Cells having value more than: 1e-10

COURSE

Type: Cells Less Than

Derivative Option: Curvature

Curvature of: Phases, volume fraction

Phase: ethanol/water or liquid oxygen

Cells having value less than: 1e-14

CONTROLS

Maximum refinement level: 3

Min orthogonal quality: 0.2

An expression is used to only refine cells in the main domain, not the injector elements

Adaptive time stepping is used, with a global courant number of 0.9.

 

The solution always errors out just after the iso surface enters the main domain and the adaptive meshing kicks in. The adaptive meshing works fine when evaporation is off though

 

The mesh is initially very course due the the use of adaptive mesh refinement

Here is what the residuals usually look like. The one of the gas species always seems to be having issues

 

[Rob]

As cells are refined you may find the solution time scale changes very rapidly, and especially as you evaporate which adds a lot of volume to a cell relative to the amount of liquid. 

Have a look at the Lee model limitations (specifically regarding multispecies and partial pressure) and then look at the options to export a DPM injection from the VOF to DPM simulation. Trying to do secondary break up to DPM and evaporation is going to be difficult to converge. 

[Zwernjayden]

 

Is there something I can do to try to resolve the issue of large volume being added to the cells, previously containing liquid? Do I need to define the mesh refinement/coursening in a different way? It seems to be running if I disable adaptive mesh refinement. I have the issue even when not using the Lee model, and I have tried with both constant courant number and constant time step so it seems less likely thats the problem

 

[Rob]

If it's failing without Lee model check the time step. As you refine you decrease the cell size (as you'd expect) which also then requires a smaller time step.