Page 1 of 1

Convergence History

Posted: 20 Nov 2013, 07:36
by siti.nurhasanah
Hi,
when i running the solver, there is a convergence history monitor.
Can you explain the meaning of the graph, what is the meaning of magnitude in iteration step to relative change?
And the next problem, when i start the post processor Sometimes it is successfully work but sometimes it doesn't work and log window says:

ELMER_HOME: C:\Elmer7
ELMERGUI_HOME: C:\Elmer7\bin
ELMER_POST_HOME: C:\Elmer7\share\elmerpost
Starting ElmerGUI...
Process state changed
State: Starting
Process state changed
State: Running
Started
Load C:\Elmer7\bin/edf/egini.xml... done
Load tetgen plugin... not found
Tetgen functionality unavailable
Constructing ElmergridAPI... done
Load C:/Elmer7/bin/edf/edf.xml... done
Load C:/Elmer7/bin/edf/electrostatics.xml... done
Load C:/Elmer7/bin/edf/heatequation.xml... done
Load C:/Elmer7/bin/edf/helmholtz.xml... done
Load C:/Elmer7/bin/edf/k-epsilon.xml... done
Load C:/Elmer7/bin/edf/linearelasticity.xml... done
Load C:/Elmer7/bin/edf/meshdeform.xml... done
Load C:/Elmer7/bin/edf/navier-stokes.xml... done
Load C:/Elmer7/bin/edf/poissonboltzmann.xml... done
Load C:/Elmer7/bin/edf/resultoutput.xml... done
Load C:/Elmer7/bin/edf/statcurrent.xml... done
Initialize GL
Vendor: NVIDIA Corporation
Renderer: GeForce GT 620M/PCIe/SSE2
Version: 4.2.0
Unable to open file: file name is empty
Project directory: D:/CTECH Laboratories/output 1
Clearing model data
Loading project document...
Loading mesh files...
Loading elmer mesh files
Summary:
Nodes: 3782
Point elements: 0
Edge elements: 280
Surface elements: 7362
Volume elements: 0
Found 11143 edges on boundary
Bcs/materials on surface elements: 2
Bcs/materials on edge elements: 8
Bcs/materials on point elements: 0
Generating lists to display
Generated 14 lists
Ready
Bcs/materials on surface elements: 2
Bcs/materials on edge elements: 8
Bcs/materials on point elements: 0
Generating lists to display
Generated 14 lists
Reset model view
Loading geometry input file
Geometry input file: D:/CTECH Laboratories/output 1/variasi material.mphtxt
Selected elmergrid

ElmerGrid checking filename suffix for file: D:/CTECH Laboratories/output 1/variasi material.mphtxt
Filetype determined by suffix: 9
Initialized the filetype
Regenerating and saving the solver input file...
Ready
Solver emitted signal: QProcess::ProcessState: 1
Solver emitted signal: QProcess::ProcessState: 2
Solver started
Solver emitted signal: QProcess::ProcessState: 0
Solver ready
Loading ep-file
Ep file header says:
Nodes: 3782
Elements: 7642
Scalar components: 4
Timesteps: 2
scalar: Potential
vector: Electric.field
1 timesteps read in.
Process error
Error: Crashed
Process state changed
State: NotRunning
Finished
Exit code: -1073741819
Status: CrashExit

what is the problem with my simulation? I use elmer 7.0 under windows 7 :?


thanks before :)
best regards

Re: Convergence History

Posted: 20 Nov 2013, 10:15
by akirahinoshiro
The numerical solution searches the (local) minimum so you give a threshold to stop.
The convergence monitor shows you the state of your solution progress.
In this graph you can see how fast your solver converge.

I would suggest you to use paraview for post processing.
Just change your sif file to

Code: Select all

Post File = case.vtu
If you prefer the GUI:
model->setup->Post file->case.vtu

If you need more options use the Result Output solver or SaveLine/SaveScalars solver.

Re: Convergence History

Posted: 20 Dec 2013, 14:53
by VVS
Hi,

I changed the simulation a little.
I changed the top boundary condition from an absolute temperature to a flux.
In ElmerPost I get a reasonable result but it is static and I'd like to see the transient.
How do I see the transient in ElmerPost?

thanks
VVS

Re: Convergence History

Posted: 20 Dec 2013, 15:32
by annier
Hi VVS,
Visualization(both for transient or steady state simulations) can be done in Paraview too if you add an additional solver in SIF file. The solver's name is Result Output Solver.

Code: Select all

Solver n   !n is the corresponding number for this solver
 Exec Solver = after saving
 Equation = "result output"
 Procedure = "ResultOutputSolve" "ResultOutputSolver"
 Output File Name = "case"
 Vtu Format = Logical True
 Binary Output = Logical True  ! binary format is the default
 Single Precision = Logical True ! double precision is the default
 ! In the Vtu format all fields are saved if the user does not list them explicitely.
 End
This solver will generate "case.vtu" file in the directory where your mesh is located. With Paraview, you can visualize the required field variables.
Please refer to this post (it is for steady state simulation).
viewtopic.php?f=4&t=3258

I am not experienced with ElmerPost. Maybe some others may reply you for it.
Yours
Annier

Re: Convergence History

Posted: 20 Dec 2013, 16:59
by VVS
Hi Annier!

Thanks for the post.
I added the code you posted at the end of the SIF file.
But after the simulation there isn't any vtu file in the project folder nor in the mesh folder.

Thanks
VVS

Re: Convergence History

Posted: 20 Dec 2013, 17:02
by annier
Hi,
Are you using ElmerGUI?
If it is so, you have to use the ResultOutPut Solver in the same way you used for the another solver.
Yours
Annier

Re: Convergence History

Posted: 20 Dec 2013, 17:04
by annier
Hi VVS,
ElmerGUI will erase the manually added code in the SIF. So, you have to add the ResultOutPut Solver from the Equation section of the ElmerGUI.
Steps for Adding Result Output Solver in ElmerGUI based simulation:
  • 1. Model -> Equation ->Add ->ResultOutput
    in my guess, here you can choose the options as written in the code above
  • 2. Generate the SIF (including all your settings)
  • 3. Save the SIF
This will encode the result output solver in the ElmerGUI based SIF and thereafter generate the case.vtu in the directory.
Yours
Annier

Re: Convergence History

Posted: 20 Dec 2013, 17:31
by annier
akirahinoshiro wrote: I would suggest you to use paraview for post processing.
Just change your sif file to

Code: Select all

Post File = case.vtu
If you prefer the GUI:
model->setup->Post file->case.vtu

If you need more options use the Result Output solver or SaveLine/SaveScalars solver.
Hi VVS,
I do the simulation mostly without ElmerGUI and so in my first message, I had told you in the way i do.
Also, Akirahinoshiro, has clearly mentioned about two options for using Paraview with ElmerGUI.
1. For general settings: model->setup->Post file->case.vtu
2. For advanced visualization settings:model->equation->add->resultoutput - basically required for observing the field variables
Also, model->equation->add->SaveLine and model->equation->add->SaveScalars, can be added for visualising the scalars and other quantities.

In conclusion, Paraview visualization requires the file with the filename ending with .vtu i.e. "case.vtu".

yours
annier

Re: Convergence History

Posted: 07 Jan 2014, 12:37
by mzenker
Hi,
VVS wrote: How do I see the transient in ElmerPost?
I would also recommend ParaView, but if you want to use ElmerPost nevertheless, you can get the transient results by selecting "all" in the File->Open dialog (not totally sure if this is exactly correct, ElmerPost is not running on my Win7 machine, complains about missing glaux.dll).

HTH,

Matthias