Message by Martin Tajmar: Hello,
I have a satellite to model with quite a lot of non-conductive surfaces. I would like to get the surface potential of this spacecraft, but if I use SPIS/NASCAP materials such as non-conductive black paint (BLKP, NP2K) the simulations takes forever. So I want to just deactivate the conductivity of those particular surfaces.
I tried to set the conductivities to zero/the resistivities to infinity in the materials local parameters, but it didn’t work. Also setting the local parameters SurfConduct/VolConduct/IndConduct to 0 did not change anything, and I don’t want to set the global parameters (surfaceConductivity…) to zero because I have other surfaces that are conductive and need these parameters to be activated.
So is there a way to simply deactivate the conductivity of a particular physical group?
Message by Pierre Sarrailh:
Hello,
A specific physical group cannot be set as conductive or resistive just with a flag.
But you can change that in the material properties, you have only to put the volume conductivity (named “BUC”) to a negative value (-1 for example) to have this material as a metal.
Just be carefull if you use this material on different groups, the material properties are shared by default between the groups.
If you want two groups with BLKP one conductive and the other not, you have to uncheck the checkbox to share the properties (just on the right of the panel in the group editor).
Pierre
Message by Martin Tajmar:
Hello Pierre,
This is good for metals, but I already was using metals for conductive parts. My interest is for the non-conductive surfaces: am I forced to use a given value for the conductivity/resistivity? Because this multiplies the computation time by a huge value. Isn’t there any way to do exactly as you said for metals, but for insulating materials?
Message by Martin Tajmar:
I will be more precise on my goals in case I was not clear before.
Instead of using a given conductivity/resistivity for each material (which will be the case eventually, but in the very final simulation, not for now), I want to use either FULL conductive or completely NON-conductive materials.
To do this, I already was using metals for FULL conductive surfaces as you suggested, but for the NON-conductive materials (i.e. insulators that should not be collecting or emitting any current) I still have not found any way to do this simply.
My objective would be, like for metals, to be able to completely deactivate the computation of the emitted/collected currents on these insulated surfaces, which is currently not the case if I use materials such as BLKP.
Message by Pierre Sarrailh:
I better understand now what you want to do.
(empty line)
Changing the conductivity of the material will not save CPU times in your simulation.
Because even if the conduction current is zero, the dielectric materials are floating.This capacitive effect requires the calculation of the current on each surfaces of a dielectric.
In fact the calculation of the conduction currents takes a negligible CPU time.
(empty line)
Non conductive dielectrics on the spacecrafts requires to solve on each surface element of the SC: C.dV/dt = I_net with Inet the net current on the local surface.
(empty line)
Most of the time non conductive SC simulations takes a lot of time due to the fact that the codes requires very small time steps to solve this equation on the dielectrics.Is it you case ?What is the time step (simulatioDt) computed by SPIS after several iterations ?
Message by Martin Tajmar:
I see, it is indeed my case. When I simulate all surfaces as conductive ones, my test simulation takes around 3 minutes to complete, and the simulationDt always increases beyond the second. But when I set a surface to non-conductive with BLKP for instance, the simulation is still at 0% after a few hours, and the time step is stuck at about 0.001s I think. Depending on the current emitted from the thrusters of the satellite, it took between 0.2s and 30s (time “inside” the simulation, that is incremented at every loop ending) for the floating potentiel to reach a steady state. So if after a few hours the simulation is only at 0.01s, it will never end.
Moreover as I said I tested this on a test case, which was very simple (only 55000 tetrahedra, 110000 faces), but in the end I would like to perform a simulation with a much more complicated geometry, so I guess even if by some miracle the computer could withold such a long simulation, it would take months to finish. Hence my question about deactivating/overlooking the conductivity for some surfaces (i.e. not even calculating the currents on these surfaces, just putting everything straight to 0).
Message by Pierre Sarrailh:
A small time step can be due to:
very high current collected locally
2. high numerical noise on the current collected (in case of PIC populations)
3. too small capacitance on the dielectric
There is a large number of solutions to reduce the CPU cost of such a simulation with SPIS. It depends on what is the reason of such small time step in your case.
This king of simulation of a thruster plume interaction with a dielectric spacecraft takes usually hours not months to run on such small mesh.
You may have to play with the numerical parameters of the different solvers to reduce the CPU cost.