I’m a new user of SPIS, and recently I’ve been doing simulations of spacecraft surface charging. I started with an aluminum spherical spacecraft and a spherical boundary in the NASA worst GEO environment. However, I found that the size of the boundary sphere affects the charging potential. A bigger boundary results in a larger potential. So what is the correct answer? I think maybe it’s because my global parameters aren’t set accurately, such as boundary condition. I don’t know.
Thanks in advance for answers!
That would imply that your simulation is unstable, as only instabilities depend on the size of the domain. You need to make sure you are resolving the relevant effects, it can be an issue with having a too small a volume (a large potential on the boundary of the volume is one indication of this), or also likely, poor mesh resolution or too large timesteps. The list of possible error sources is large, so take time to go through the documentation to guide you.
If you had a very complex simulation, or are very limited in what you can simulate, you could look into the boundary conditions and the relevant documentation.
Also, I’m confused about one other thing. In the localparameters of the external boundary group, IncomPart controls whether there are injected particles or not. But I’m not sure how properties like the number of injected particles are controlled. This may not help the simulation substantially, but I really want to know it.
For PIC particles, the avPartNbinCell (or a parameter named like that) dictates how many PIC particles SPIS starts with, per cell. But the real control you have is in manipulating the mesh resolution, increasing the number of cells (and as such, the number of particles) whenever the situation calls for it, (e.g. small debye length, if applicable, or fine details on the mesh)
GEO Worst case correspond to case where the Debye length is about a few hundreds of meters, thus, if your boundary is not a few times the Debye length (~ kilometric), you will have an effect. SPIS proposes a models that allow to have a simulation domain smaller than the Debye length (Robin conditions for E, Fluid or backtracking model for parts…).