Tutorial 21 Actuator driven with electrostatic currents

The graphical user interface of Elmer
Post Reply
Chris_DE
Posts: 10
Joined: 20 Nov 2011, 19:45
Antispam: Yes
Location: Deutschland

Tutorial 21 Actuator driven with electrostatic currents

Post by Chris_DE » 24 Nov 2011, 20:58

Hello all,

the questions are right down at the bottom but I need first to give some explanations.

I am learning how to use Elmer GUI. I try to reproduce the tutorial 21 Thermal actuator driven with electrostatic currents in Elmer GUI.

- In this prospect I make sure that the Static Current Conduction solver was in the edf directory.

- I found on the internet the required .iges geometry file

- I realise a 3D mesh with the freeware gmsh-2.5.0

Image

- I use Elmer GUI 6.2 and start to configure the solvers and define the main equation, material, body force, initial and Boundary conditions in accordance to the tutorial.

After these points I start the solver:

The problem converges after 24 steady state iterations (in the tutorial it was 27 but I don't use here the same mesh)
I notice in the solver console that I have two recursive errors in the iteration loops:

Code: Select all

ERROR:: ElementMetric:  Degenerate 3D element:    0.0000000000000000
ERROR:: IterSolve: Failed convergence tolerances
The first might be linked to the automatic mesh tool I use?
I understand the meaning of the second but what should I do now? modify the convergence tolerances?

Using the Postprocessor (VTK) I display the following results:

Potential:
Image

Volume Current:
Image

Temperature:
Image

At this point I was hoping that the results match the ones in Tutorial but it isn't the case :(
-> MyResut_T_Max= 4460K // Tutorial_T_Max = 1113K

Temperature in the Elmer Tutorial 21:
Image



PS: I also Edit the xml file statcurrent.xml to add a "Free text input"

I set Electric Conductivity = Variable Temperature

and write this code in the "Free text input" field. (it is also possible to edit the .sif file before running the solver)

Code: Select all

Real
	298.0 4.3478e10
	498.0 1.2043e10
	698.0 5.1781e9
	898.0 2.7582e9
	1098.0 1.6684e9
	1298.0 1.0981e9
	1683.0 1.0
	2000.0 1.0
End
Edit -----------------------------------------
I notice a possible error in the code given in the Elmer tutorial on page 105

Code: Select all

Nonlinear System Convergence Tolerance = 1.0-6
should be 1.0e-6 instead of 1.0-6

QUESTIONS:

1) Why does my simulation results for the temperature differ from the Tutorial ones?

2) How can I get rid with the error ElementMetric: Degenerate 3D element: ?


I would be grateful if someone who already reproduces this tutorial gives me some help.

Best regards
Chris

I join the project data in a Zip file
Attachments
actuator.zip
sif, msh ...
(257.35 KiB) Downloaded 167 times
Last edited by Chris_DE on 04 Dec 2011, 14:44, edited 1 time in total.

Chris_DE
Posts: 10
Joined: 20 Nov 2011, 19:45
Antispam: Yes
Location: Deutschland

Re: Tutorial 21 Actuator driven with electrostatic currents

Post by Chris_DE » 04 Dec 2011, 13:20

I don't manage to answer the two questions posted above.

Could someone explain me the role of the tick-box "Joule Heating" in comparison with the variable transfer "Equals Joule Heating" ?

Should I use both or ticking the box is enough?

In ".\Elmer6.2\tests\current_heat_control" the example doesn't require this variable transfer

Regards
Chris
Attachments
Joule-Heating_Actuator.PNG
A graphic description of my question
(48.4 KiB) Not downloaded yet

raback
Site Admin
Posts: 3498
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Tutorial 21 Actuator driven with electrostatic currents

Post by raback » 04 Dec 2011, 14:39

Hi

Ticking should be enough. It activates internal computation of Joule heating at the integration points. To use the computed field directly as heat source is suboptimal since the source should be discontinuous over different materials and the latter approach does not enable this.

-Peter

Chris_DE
Posts: 10
Joined: 20 Nov 2011, 19:45
Antispam: Yes
Location: Deutschland

Re: Tutorial 21 Actuator driven with electrostatic currents

Post by Chris_DE » 14 Dec 2011, 13:30

Thanks Peter for your explanation, I understand now better the difference between the two methods. It means that for this tutorial, it is not a matter to use the calculated Joule Heating as Heat source because there is only one material and therefore no discontinuity.

I have also found the reason for the difference between my simulation result for the temperature and the one given in the tutorial.

In the ElmerGUI postprocessor, I let as defaut "read all the timesteps" (=all steady state iterations). Obviously, the first iterations do not produce realistic result.

One should only load the last iteration to display the temperature after convergence.
It was simple but I didn´t catch first that all the iterations are saved in actuator.ep

I notice a possible bug in the timestep management using the postprocessor VTK.

In Elmer GUI: Steady State Max Iterations = 30

Elmer solver: Convergence is reached after 24 iterations

I read the result data "actuator.ep" in the Postprocessor VTK and it shows:

Timesteps: Start: 1 End: 31

If I try to load only a set between 24 and 31, for example 27 to 31, a failure occur and Elmer crash.

Why does the iteration for the displacement (performed after the simulation) is saved with the number 31 (30+1) and not rather 25 (24+1) ?

I understand that the set 25 to 30 has no sence since it contains no data. The same failure occur if I load the set 31 to 31.

Code: Select all

Process error
Error: Crashed
Process state changed
State: NotRunning
Finished
Exit code: -1073741819
Status: CrashExit
and one time I got more information about the failure

Code: Select all

ERROR: In C:\VTK\VTK\Common\vtkLookupTable.cxx, line 111
vtkLookupTable (050AEE20): Bad table range: [9e+099, -9e+099]
Thanks
Chris

raback
Site Admin
Posts: 3498
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Tutorial 21 Actuator driven with electrostatic currents

Post by raback » 14 Dec 2011, 15:39

Hi Chris

You can also set "Output Intervals = 0" and you'll only get the last timestep.

The problem with the number of steps is difficult since in the ElmerPost format the number of steps is in the first line but all the data is appended at the end, and the first line is not visited in the case of not using all steps. ElmerPost can deal with this, obviously VTK widget not.

You might also try ResultOutputSolve with binary VTU output and visualization with Paraview.

-Peter

Post Reply