Message by SHINYA NAKAMURA:
hello.
I did the simulations in the same shape model under same case but in different mesh size.
I had the results that have large difference on sunlight surface especially in daylight case.
I don’t know why there are so much difference (2000-4000 V) , but it is obvious that the reason is related with mesh size.
I would like to know how should I decide mesh size.
What from should I decide mesh size?
Please tell me this.
Shinya NAKAMURA
Message by Guillaume Tcherniatinsky:
Hello,
I had recently a SPIS formation and we were advised to take approximately halth the Debye length sqrt(eps0*Te,ev/(Ne*e)) for our mesh size (as long as it is not too heavy for the solver). The documentation tells us that the linear solver is stable only for mesh size smaller than Debye length and that you can have a mesh size larger only when you are using non-linear solver with Maxwell-Boltzmann distribution for electrons. You have to be careful though, as Maxwell-Boltzmann distributions is physically correct only if the potential is negative everywhere.
Note that changing the electron distribution to Maxwell-Boltzmann doesn’t switch automatically the solver to nonlinear. You have to switch linearPoisson flag as well.
Hope this helps,
Guillaume TCHERNIATINSKY
Message by SHINYA NAKAMURA:
Hi, TCHERNIATINSKY-san
Thank you very much for replying.
I am using non-linear solver (linearPoisson is 0 in Global parameter/Poisson equation ), and I simulate GEO environment (mainly worst-case).
When I use non linear solver,I should take half the Debye length sqrt for mesh size? I think I don’t have to take so small mesh size,so I take around 1/4 to 1/2 size of the length of each spacecraft parts for mesh size.
Shinya NAKAMURA
Message by Guillaume Tcherniatinsky:
Well this part in my explanation was not clear, I actually meant half the Debye length (not the Debye length sqrt).
When using non linear solver you don’t need to use a resolution that thin indeed (see the description for parameter linearPoisson in doc “Controlling SPIS-NUM from UI”). I don’t know about any criteria in that case, but if you get unpredictable results I would say you shouldn’t take a resolution too large compared to the Debye length.
Guillaume TCHERNIATINSKY
Message by SHINYA NAKAMURA:
Thank you for your advising. I understood what you meant,and I read the documentation “Controlling SPIS-NUM from UI”. I was able to know mesh size is related with the Debye length. really thank you.
I will try readjusting the mesh sizes on my model.
best regards.
Shinya NAKAMURA