Message by Aldo:
Hello again.
I’ve runned succesfully experiments with an Al cylinder hovering on Condutive Lunar Dust with reasonable computational time (20 min). The automated Dt=0.1s for a 150s experiment.
The problem is when the cylinder is made of Kapton or any other dielectric material, the computational time rise to 7 days aprox, beacuse automatic time steps become as small as Dt=1E-4s.
Is there any way to reduce this computational time?
Thank you .
Message by Sébastien Hess:
Hello,
SPIS limits the time step to avoid large change of potential from one step to the other.
When you have a conductive material, the code first integrate the current over all the conductive surfaces and then roughly solves dV=Sum(J_i*Surf_i)/(C_sat*Surf_total)
When you have a dielectric material, the code solves the circuit for each dielectric surface element dV_i=J_i/C_diel
Two things happen: (1) the potential evolves with the dielectric capacitance rather then the Csat capacitance (you may increase the relative dielectric constant RDC of the material to increase C_diel, so it behave the same way than with Csat)
(2) the preliminary integration over the conductive surfaces increases the noise to signal ratio, so that the time steps are limited by the noise level. You may increase the number of particles to reduce the noise, smooth the current (Global parameter smoothingI =1 or 2), increase a bit the validityRenormalisation up to 2 should be OK.
Sébastien