[Archive] Particle source problem on windows v4

Message by Matias Wartelski:
Hello, I have a basic simulation with a spacecraft and two different sources which works fine in windows v3.6 and v3.7.
When I run it on windows v4 with exactly the same input, I have no exceptions raised but the simulation runs with absurd time steps (changing but allways around E-70). When I turn off the particle sources (sourceFlags=0), it runs normally, but as soon as I turn on sourceFlag1 and sourceFlag2, the problem appears. Does someone understand what is happening? Thanks a lot
This is the beginning of the log:
Time 0.0 s => POTENTIALS: 0.0 V on SC ground (node0), 0.0 V on node1 ground, 0.0 V on node2 ground, 0.0 V on top of node0, 0.0 V on top of node1, 0.0 V on top of node2, 0.0 Vmin onTopOfNode0, 0.0 Vmin onTopOfNode1, 0.0 Vmin onTopOfNode2, 0.0 Vmax onTopOfNode0, 0.0 Vmax onTopOfNode1, 0.0 Vmax onTopOfNode2,
Monitoring numerics of simulation at time 0.0
ThreeDUnstructVolMesh: 93033 tetrahedra, 188545 triangles, 111306 edges, 15796 nodes
Mesh quality (best = 1): min = 3.0169587E-4, max = 0.9892087, average = 0.5915012
histogram of mesh quality: number per interval between min = 3.0169587E-4 and max = 0.9892087 (unit -): 1122.0, 2097.0, 4283.0, 6749.0, 9802.0, 15510.0, 23212.0, 20358.0, 8762.0, 1138.0 PICVolDistrib ions1: 318393 particles, last numerical time step used = 0.0s (max allowed = 6.3390494E-6s), numerical speed up = 1.0
PICVolDistrib elec1: 318341 particles, last numerical time step used = 0.0s (max allowed = 1.4847186E-7s), numerical speed up = 1.0
PICVolDistrib source1: 0 particles, last numerical time step used = 0.0s (max allowed = 4.1936855E-6s), numerical speed up = 1.0
PICVolDistrib source2: 0 particles, last numerical time step used = 0.0s (max allowed = 8.572027E-9s), numerical speed up = 1.0
Last local volume potential change: 0.0V maximum change, and 0.0V average change (last plasma time step = 0.0s)
Time: 0.0, Dt = 4.286327E-7
Number of particles reflected on symmetry boundaries : 0
Number of particles reflected on symmetry boundaries : 0
Number of particles reflected on symmetry boundaries : 0
Number of particles reflected on symmetry boundaries : 0
Variation on node 0 : -0.0011697058798745275
Ratio of validity : 1.0000000150474662E30
Validity ratio too big : 1.0000000150474662E30
New time step : 2.1431633617487145E-37
Integration duration performed so far : 0.0
Validity Table : 0.0
Variation on node 0 : -1.1700966277001054E-33
Ratio of validity : 1.0000000150474662E30
Validity ratio too big : 1.0000000150474662E30
New time step : 2.1431633294995365E-67
Integration duration performed so far : 0.0
Validity Table : 0.0
Variation on node 0 : -0.0
Ratio of validity : 0.0
Variation on node 0 : -0.0
Ratio of validity : 0.0
Variation on node 0 : -0.0

Variation on node 0 : -1.4013132620366003E-36
Ratio of validity : 1.0000000150474662E30
Validity ratio too big : 1.0000000150474662E30
New time step : 2.143163297250355E-70
Integration duration performed so far : 2.381292588332814E-41
Variation on node 0 : -0.0
Ratio of validity : 0.0
Variation on node 0 : -0.0

and so on (I have not been able to finish a simulation)
Matias

Message by Jean-Charles Mateo-Velez:
Hi Matias,
wwhich SPIS v4 do you use ? 4.0, 4.1 or 4.2 ?
4.1 and 4.3 can take account of multi species on a single emitter.
Perhaps you have found a bug related to this new capability. I suggest you to send me your project at mateo@onera.fr and I will look at it asap.
Regards,
Jean-Charles

Message by Matias Wartelski:
Hi Jean-Charles,
My problem was in version 4.0 and I think it was due to a bug concerning the currentScaler of the sources. But this problem seems to have been fixed on 4.2RC3 as my project works in this version.
Thanks for your attention
Regards
Matias

Message by Jaime Perez Luna:
Hi Matias, hi everyone
I have been experienced similar problems with the simulation of an ion source with SPIS v4.0.0 (ratio of validity extremely high and integration times extremely, as described in Matias’ log), so I decided to download v4.2RC3 to see if everything worked.
Now, I can’t even start the simulation :). The mesher is unable to load the mesh from GMSH. I get the following log:
INFO ! MeshLogger ! - 3D meshing with Gmsh from C:\DOCUME~1\Jaime Perez Luna\Local Settings\Tempspistmp1271320372\tmp.geo
INFO ! MeshLogger ! - I am trying command… …

SpisUI
…\ThirdPart\Gmsh\Windows\gmsh-2.0.8\gmsh.exe -format msh1 -3 C:\DOCUME~1\Jaime Perez Luna\Local Settings\Tempspistmp1271320372\tmp.geo -o C:\DOCUME~1\Jaime Perez Luna\Local Settings\Tempspistmp1271320372/Tmp3D.msh > C:\DOCUME~1\Jaime Perez Luna\Local Settings\Tempspistmp1271320372\Tmp3D.mshout
INFO ! MeshLogger ! - Mesh 3D done.
INFO ! Task ! - Meshing file: C:\DOCUME~1\Jaime Perez Luna\Local Settings\Tempspistmp1271320372\Tmp3D.msh
INFO ! Task ! - Mesh C:\DOCUME~1\Jaime Perez Luna\Local Settings\Tempspistmp1271320372\Tmp3D.msh Importing. Please wait…
ERROR ! Task ! - Impossible to import C:\DOCUME~1\Jaime Perez Luna\Local Settings\Tempspistmp1271320372\Tmp3D.msh ! Maybe the file is corrupted or Gmsh is missing. Please check.

I downloaded the full Spis4.200RC3 carpet from the site and checked that GMSH works and is in the right location. However I also noticed that when I try to open a .geo file directly from the Geometry/CAD manager in SPIS, GMSH does not open the .geo file I specified for my project (it doesn’t open any file actually).
Any ideas ?
Thanks,
Jaime

Message by Jaime Perez Luna:
Hi everyone,
I am answering to my own post :
First problem : integration time droping down to 10-30 s (!) as soon as the simulation starts in a problem with a source. I realized that my emitting surface was parametered with a floating potential. I changed this and set a fixed potential to my source. Now the simulation goes on properly, with a normal time step.
Second problem: I have done this with v4.0.0 - I still can’t manage to use v4.2R3 (for windows) because SPIS is unable to mesh my problem via GMSH. I have no problems doing it directly with GMSH (or with version 4.0.0) but not directly from this version of SPIS. I don’t have any clue as to why this happens.
Jaime

Message by Nuno Loureiro:
Hi Jaime,
I am also having some problems related to the mesh of some surfaces with v4.2RC3: I am trying to use a circle as a source’s surface but the best I can get with this version is a square form. With v4.0.0 and with GMSH it actually works well, it is a circle.
Have you tried increasing the size of the elements you are trying to mesh, to see if that works?
Another thing I have noticed was that when I don’t use the circuit integration (electricCircuitIntegrate=0) the Dt used for the simulation is always constant and equal to simulationDtInit, no matter what value I insert for simulationDt (-1.0 or 0.0). If I turn electricCircuitIntegrate=1, Dt changes at each time step but I get an infinite ratio of validy (and so an infinite loop) after some time steps.
To set the fixed potential did you changed the variable electricCircuitIntegrate to 0 or changed the variable initPotFlag?
Nuno

Message by Nuno Loureiro:
Sorry, I forgot to tell that I am using a source of Xe+ and for the plasma environment I am using electrons with a GlobalMaxwellBoltzmannVolDistrib for the variable electronDistrib.
Nuno

Message by Jaime Perez Luna:
Hi Nuno,
I’ve given up using v4.2RC3 because I couldn’t manage to simulate even very simple problems (fixed potential sphere immerged in a plasma) due to the meshing problem I mentioned in my previous message. Again, the mesh goes on with no problems in GMSH but I don’t get any result via SPIS. I guess this is due to a problem in the interaction between SPIS and GMSH ?, but for now I don’t know SPIS well enough to understand more (and even less to fix it :slight_smile: !).
So I am using v4.0.0.
I am also simulating Xe+ sources in a Boltzmann distribution of electrons (n~10e16 m-3). Indeed, to fix the potential I don’t use the circuit integration and set electricCircuitIntegrate=0. However, my Dt is not equal to simulationDtInit. Typically I have a Dt of a few 10e-6 s, usually constant.And yes, if I set a floating potential for my source (electricCircuitIntegrate=1), I get stuck in an infinite loop with absurd time steps (10-30 s)…
Jaime