[Archive] poissonBCType 2

Message by Guillaume Tcherniatinsky:
Hello to the SPIS community,
I have an issue with boundary conditions on SPIS which I can’t solve alone… so your support could be very helpful, I post it here in case someone has a idea.
I am currently running a simulation with electron temperature much higher than ion temperature and ion kinetic energy (10eV for electron temperature using Boltzmann model, 2eV for PIC phoelectron temperature versus 0.026 eV for PIC ion temperature and 0.035eV for ion kinetic drift energy). This means that ion trajectories are disturbed by the sheath at a much larger distance than electrons. Thus:

  • I need a box much larger than 10 debye lengths-distant (debye length = 0.74m, box distance from spacecraft = 25 m)
  • I need to use a boundary condition where potential is not null and ions are already accelerated when injected.
This is the reason why I am using the type 2 boundary condition (poissonBCType = 2), which I was told suits better this case. However, in the simulation I am running, I get potential on the boundary that are very negative (up to -14V when spacecraft potentiel is -63V) which does not look like a physical phenomenon to me. I did not find documentation on that option so I can't understand why it occurs. Do you know for that specific model of boundary conditions:
  • How the number of PIC particles injected at each time step is computed taking into account the external boundary potential? (I was told it uses Liounville's theorem, but how does SPIS implement that exactly?)
  • How does SPIS compute the potential on external boundary? What does "mimicking a 1/r² decay" means?
  • How does SPIS compute the velocity distribution of PIC ions at injection? Is it normal to the surface with its energy corresponding given by external surface potential?
Thank you very much for your help. Guillaume

Message by Aldo:
Hello Guillaume
I’m facing a similar problem in a simulation of my own and I’ve been trying to solve it.
I would like to know if you have time working on the topic, because I think I could provide some advice but I’m not precisely an expert, so I may say something that is obvious to you
Please let me know.

Message by Guillaume Tcherniatinsky:
Hello Aldo,
I’m trying to run that simulation for quite a long time now and not having much success, so I have some time working on this and of course any inputs would help me improve my reflexion :slight_smile: . Thank you!

Message by Aldo:
Hello Guillaume
I’m facing a similar problem in a simulation of my own and I’ve been trying to solve it.
As far as I understand from “How to control NUM from UI in SPIS 5”, if you use the parameter “poissonBCType=2” it will turn ALL your boundaries ElecFieldBC into that kind (r-2 decay). I’m not sure you want that, maybe you’re injecting particles only from one boundary and the others need a constant null electric field or other ElecBC.
You can set the ElectricFieldBC and ParticleBC at Group Edition level for each boundary in your box, not sure if you already knew that.
Maybe you need an ElecFieldBC made for you, so you may try a Fourier Boundary condition that combines both V and E, or at least I’ve been thinking in that possibility at my simulations but haven’t tried it yet.
Also, as far I understand from the “How to Control…” document, Ions and electron (and any other desired species) are injected at a Vx, Vy and Vz velocities at the “Plasma” tab in the Global Parameters sheet, that is the “drift velocity” of the species. If any of the “Vi” component needed isn’t there, you shout add it with de “add global parameter” button, then you will be controlling the particle injection velocity direction. But again, not sure if your asking that in “… Is it (the velocity) normal to the surface…”
And regarding your results, you didn’t mention photo emission so I’m assuming you’re not implementing it, is “reasonable” that your spacecraft is absorbing a lot of ambient electrons and has no photo emission to compesate them, thus your potential is “very negative”. But it doesn’t mean that those results are correct, specially because your ElecBC.
And most importantly, I understand from your statement “…(the result) does not look like a physical phenomenon” that you need some documentation on the spacecraft charging topic. If that’s the case, I recomend you “Potentials of surfaces in space” (Whipple, 1981). I really, really, really gives you light on what is happening in your simulations.

Message by Guillaume Tcherniatinsky:
Thank you for your time and help!
Yes, I understand the documentation just the same as you, i.e. poissonBCType 2 sets all boudaries to r-2 decay. In my case I actually didn’t want external potential to be Dirichlet, because ions are perturbated at very large distance because of their very low energy. But since I get absurd results on boundaries anyway, I guess your idea of switching back to Dirichlet is better anyway. I will tell you if it works.
However I am not sure to understand how you simulate r-2 decay using Robin local boundary conditions? If you know about that it could help me a lot!
Concerning injection velocities, yes, the method used by SPIS to compute injection velocities of the ions in poissonBCType 2 isn’t described anywhere in documentation, I know about this special computation from people at ONERA who advised me to use that method. The key is that as potentiel can be non-zero on boundary, injecting particles without taking that starting energy due to potential into account (and using strictly drift velocity) wouldn’t be consistent. And even in Dirichlet BC, I understood that PIC particles cannot strictly be injected at drift velocity because it wouldn’t take their particle temperature into account. So I guess a thermic velocity is chosen at random following Maxwell-Boltzmann isometric distribution and that thermal velocity is added to drift average velocity to get particle injection velocity.
You’re right about photoemission, it wasn’t explicit but I actually use photoemission with sun at 4UA, AL2K body and ITOC coating on panels.
Thank you very much for the documentation tip, I will have an eye on that!

Message by Aldo:
Hello again Guillaume
I suggested Robin boundary conditions for modelling non equilibrium sheaths as an hypothesis, I’ve looked for documentation on the topic and asked for advice from professors at the institute, but no luck so far.
I would like to be in touch with you via e-mail, I think we can exchange use experiences with SPIS.
My address is: aldo.vladimir@correo.nucleares.unam.mx