I working on 3-d transient analysis of flow in an ic engine using moving mesh.when i executed the program,it went upto 97 time steps and then a message came that "negative densities found in more than 100 dell so run stopped ".I have already used relax factors 0.1 which is quite low so what can be one plz tell me

negative densities can arise due to many reasons like use of large timestep, high pressure relaxation factor, high residual tolerances and many more.

try reducing pressure relaxtion to 0.05 to begin with and see what happens.

also u can reduce the time step also.

i dont think this is any problem due to moving mesh but still it is better to check for proper movement of the cells as desired by you by doing a mesh preview run.

explaining the problem a detail more might help u get some proper advices from the forum.

Arnab

It may depend on the physical process as well. For instance, I got this problem in a Diesel engine simulation. That was (partly) due to a bad low of evaporation of the spray (there is no grid consistence with Lagrangian spray).

The discretisation scheme is important too. In this kind of configuration, an UW scheme is more stable.

an UW scheme is more stable?

What MARS?~~~~~

I can confirm all the given advice, but in some special cases I suppose that simple the mesh quality invokes negativ density, because some cases run with the special solver / numeric setup and other very comparable cases don't run. Even with more restricted settings (and very small time steps) refuses the cases to run. So I think the mesh quality is the only left explanation for this. Please don't use the MARS scheme for starting the case, and never use this for density. I prefere starting with LUD and especially UD for density. When the case is stable I make a restart with MARS (when I really need sharp pressure fronts)

concerning UD for density. Just to be sure to think of the same thing:

I think every one is using it per default. My understanding is that for density CD 0.01 means 99% UD scheme. So don't switch density to UD in "Primary Variables" -> "Differencing Schemes".