[FLASH-USERS] ppm_modifystates: Nonconvergence in subroutine rieman

[John ZuHone]

Sean,

If you're purposefully setting up a cloud in NON-hydrostatic equilibrium, setting ppm_modifystates will not be really useful for you... it's there for setting up initial atmospheres in hydrostatic equilibrium. The parameter itself modifies the way the Riemann states are calculated, so something (not sure what exactly) is going wrong with that. However, I don't think it should be set for your purposes.

I'm not sure what you mean by "It seems that the timestep calculation isn't adequately taking gravity into account." As it is, the FLASH code will compute the timestep of the hydrodynamics based on the velocity and sound speed of the gas, so your timestep will be set by whatever is the hottest and/or fastest cell:

dt = min(C*dx/(|v|+c_s)) (over all directions)

where C is a constant. You can tune this constant in your flash.par by setting "cfl" lower, if you haven't already.

Best,

John ZuHone

On Apr 18, 2011, at 1:03 PM, Sean Lynch wrote:

> Hello,
> 
> I am running a simulation with a self gravitating spherically symmetric cloud in a non-equilibrium state. When allowed to run, it contracts and then rebounds and oscillates until finding the equilibrium solution.
> 
> I am using the multigrid module for gravity and gamma=5/3 equation of state. When I run this I get huge total energy errors (60%). After some testing it looks like the timestep is too large during the initial collapse and maximum compression. It seems that the timestep calculation isn't adequately taking gravity into account.
> 
> Looking through the documentation I found the runtime parameter ppm_modifystates which looks like it might fix the problem I'm having. However when I run the same exact simulation with ppm_modifystates = .TRUE. the simulation fails with "Nonconvergence in subroutine rieman"
> 
> Can someone shed some light on why setting this option would cause the simulation to fail like this?
> 
> Thanks,
> Sean