[olwi]

Hi,

I'm solving an electric problem, with generation of negative and positive charge carriers. The negative charges (electrons) have a mobility (~velocity) 1e5 times larger than the positive charges (ions). It means that when a molecule is ionized, the positive ion stays where it is, but the electron is leaving the scene very quickly...

I obviously need to solve on the time-scale of the fast electrons. I would like to solve the transport equations for both electrons and ions. Problem is that the physics for the ions is almost compleletly local on that time scale, and the matrix is nearly diagonal.

I could in principle live without the convection and diffusion terms in the slow equations and solve the resulting ODE:s, but that means I need to learn how to do that in OF, and it would be more neat to treat the equations the same.

After this preamble, my questions:

1) It seems an fvMatrix refuses to solve if there are no spatial operators. Is there any way or any solver that could be forced to update the resulting ODE, without rewriting the code?

2) If I do include my very slow spatial operators, the solver finds so little work to do in each timestep, that it often refuses to iterate. This means that the ions produced in that timestep just dissappears, as no update is done. (i) Is lowering the aboslute tolernace to exetremly small numbers the only way to make it crunch? (ii) Is there any solver that can be set to do at least ONE iteration, always?

Apologize for the long post, but still hope someone read all the way...

Thanks in advance for any ideas,

Ola