Resource Menu

posted by Raluca Cojocariu at Aug 19, 2014 12:39 PM
Dear Amin,

How did you manage to increase the speed of the simulation?

For both, LEO and especially GEO, my simulations lasts for too long and if it ends it doesn't really reach an equilibrium (the graphs/ the distribution around the 3D model are very messy) or if it's still running I have to stop it after a while for obvious reasons, nothing happens.

Any help would be appreciated.

Best regards, Raluca

posted by Amin Ghiasi at Aug 15, 2013 11:55 AM
Speed problem is solved, Thank you Pierre. There was a stupid mistake I prefer not to mention!
posted by Amin Ghiasi at Aug 13, 2013 12:40 PM
I don't know if I set the global parameter (plasmaDuration) correctly! Please tell me if I set it wrongly!
posted by Amin Ghiasi at Aug 13, 2013 12:30 PM
Dear Pierre

Firstly, I apologize for late reply. I was on a trip! Secondly, I tried to set plasmaDuration manually. I mean I manually added a new global parameter like this: Name: plasmaDuration, Type: double, Value: 1.0E-5, Unit: ! s !. But the simulation was at the same rate as before! Therefore I tried to simplify my geometry (my geometry is too complicated!). Surprisingly, as I try to start the simulation with the new geometry, I encounter an error! I will be very grateful if you take a look at my error and guide me. You can find an image of the error here: and the debug file here: Thirdly, I will definitely use your advice about the distribution of electrons and ions in LEO. Thank you!

Sincerely Yours, Amin

posted by Pierre Sarrailh at Jul 23, 2013 4:03 PM
Dear Amin,

A duration of simulation of 0.1 s in LEO is probably not sufficient to obtain a differential charging but it is a very long time in comparison to ion transport. If we consider a typical velocity of ion in addition to the orbital velocity of the SC, the order of magnitude is 1E4 m/s. This means that the ions take 1E-5 s to cross a typical simulation box thus it is not needed to integrate the trajectory of ions over 0.1s. A manner to treat efficiently this difference of time scales is to us the "integration duration" concept of SPIS. To do that, you have only to fill one parameter in the GlobalParameters: PlasmaDuration. In this parameter, you can enter the maximum integration duration of ion which is a time corresponding to the typical dimension of your simulation box divided by the typical velocity of ions in your simulation. Doing that you can gain a factor of 100 in the simulation duration. On a general point of view, the simulation CPU time cost is very dependent on the time step and duration parameters.

Another advise, if not already done, in LEO case, you may select PICVolDistrib for the ions distribution (ionDistribX) and GlobalMaxwellBoltzmannVolDistrib for the electrons (elecDistribX).

Regards, Pierre

posted by Amin Ghiasi at Jul 23, 2013 11:44 AM
Dear Benoit Thank you very much for your answer. I have run the simulation two days ago (duration: 0.1 second) but it has progressed just by two percent so far! I was wondering whether I should run it on a supercomputer, or there is something wrong with my simulation? Are there any hardware requirements to run the simulation for longer durations? Have you any experience with such difficulties in using the software? Regards, Amin
posted by Benoit Thiebault at Jul 22, 2013 9:22 AM
SPIS-GEO can indeed be used to model LEO environments (and most space environment in fact).

This version has just been tuned to select GEO environment characteristics by default, but the numerical kernel supports other environments.

Kind regards,


posted by Amin Ghiasi at Jul 20, 2013 3:47 PM

As you know there is a version of NASCAP called NASCAP/LEO which simulates the LEO environment. I was wondering if it is possible to simulate the LEO environment by putting the LEO parameters in SPIS? Or SPIS is designed exclusively for GEO?

Kind regards, Amin