Error encountered after 85% progress of execution

Sam_21831

Torben Voigt

Hi Sam,

I could imagine that this is memory related. If it is MLFMM you may try reducing the number of parallel cores to save a bit of memory.

I see a warning in the picture (which is moost likely not related to the crash). Is it about usiing double precision or is it about not to tuse MLFMM?

If you want you can also share your model here to let  me have a look.

Best regards,
Torben

Sam_21831

Hi Torben,

 

Thank you for you reply. Could you please have a look at this model and let me know if you could help.

An error occurred which does not let runfeko to execute.

 

Thanks 

Torben Voigt

Hi Sam,

• The error indicates that the impressed current sources intersect with mesh elements. You can change this by either forcing mesh vertices at the right points ("Imprint points", that's what I did) or by activating "Connect the end point to closest mesh vertex" in the source dialog.
• The critical error from your first post is certainly related to ADAPTFEKO (continuous frequency range). The principle of the continuous frequency range is that Feko calculates as many frequencies as are necessary for a good representation of the entire frequency range. However, if there are jumps in the result, the solver will try to simulate in ever smaller frequency steps at these areas, which can then take a very long time. Since your model is extremely small electrically at 1 kHz (you must use double precision!), accurate results are very difficult to achieve. I would advise you to use discrete frequency range instead of continuous, otherwise the solver will always have problems.
• Even with double precision it's hard to say if the results at the low end of your frequency range are stable. The freespace wavelength at 1 kHz is 11,802,853 inches, the size of the model is 24 inches...
  I would recommend to turn on "Low frequency stabilisation" at least to validat the results. If you see that results are similar with and without LFS, you can turn it off.

Best regards,
Torben

Sam_21831

Thank you, Torben. I will consider your recommendations for the next simulation test.

 

Regards,

Sam

Sam_21831

Torben Voigt_20420 said:

Hi Sam,

• The error indicates that the impressed current sources intersect with mesh elements. You can change this by either forcing mesh vertices at the right points ("Imprint points", that's what I did) or by activating "Connect the end point to closest mesh vertex" in the source dialog.
• The critical error from your first post is certainly related to ADAPTFEKO (continuous frequency range). The principle of the continuous frequency range is that Feko calculates as many frequencies as are necessary for a good representation of the entire frequency range. However, if there are jumps in the result, the solver will try to simulate in ever smaller frequency steps at these areas, which can then take a very long time. Since your model is extremely small electrically at 1 kHz (you must use double precision!), accurate results are very difficult to achieve. I would advise you to use discrete frequency range instead of continuous, otherwise the solver will always have problems.
• Even with double precision it's hard to say if the results at the low end of your frequency range are stable. The freespace wavelength at 1 kHz is 11,802,853 inches, the size of the model is 24 inches...
  I would recommend to turn on "Low frequency stabilisation" at least to validat the results. If you see that results are similar with and without LFS, you can turn it off.

Best regards,
Torben

Hi Torben, 

Thank you for the previous support. The simulation can be successfully executed but there is something incorrect with the simulation results. I am interested actually in calculating the current over the line created in the time domain due to an impressed current (injection time pulse was defined).

Well, I guess an unexpected mistake could occur do I went back to calculate the current split between these two lines in this simple model I create (see figure below) but the simulation result does not show that KCL is valid. Why? I hope you could have an answer for me to better understand this. Thanks

 

 

Sam

Torben Voigt

Sam_21831 said:

Hi Torben, 

Thank you for the previous support. The simulation can be successfully executed but there is something incorrect with the simulation results. I am interested actually in calculating the current over the line created in the time domain due to an impressed current (injection time pulse was defined).

Well, I guess an unexpected mistake could occur do I went back to calculate the current split between these two lines in this simple model I create (see figure below) but the simulation result does not show that KCL is valid. Why? I hope you could have an answer for me to better understand this. Thanks

 

 

Sam

Hi Sam,

Time analysis is not necessarily trivial. It is noticeable that the currents at t=0 are not zero. Your time signal may be too short. Could you attach the model and POSTFEKO session here?

What is "KCL"?

Best regards,
Torben

Sam_21831

Hi Torben, Happy New Year.

 

I sent you in late November in regard to a few questions related to a testcase I am exploring. I am still have a question which is not fully explained. Let's assume that I have this flat plate stroke by lightning and I am interested in calculating the current magnitude (in time domain) along the surface of this plate (see fig. below). I learned that (from previous communications with Altair support) using the edge port with V=0 voltage source is a proper way on Feko to identify the current across the edges in time domain. 

When extracting the simulation results, I got this (please note time signal = input current waveform, current magnitude in green is Current 1, and current magnitude in red is Current 2). The question is why does the current magnitude decrease as we move from one side to the other side of the plate? I expect to see the same current everywhere along the surface. Could you please help if you have an explanation. Thanks.

 

Regards,

Sam

Torben Voigt

Sam_21831 said:

Hi Torben, Happy New Year.

 

I sent you in late November in regard to a few questions related to a testcase I am exploring. I am still have a question which is not fully explained. Let's assume that I have this flat plate stroke by lightning and I am interested in calculating the current magnitude (in time domain) along the surface of this plate (see fig. below). I learned that (from previous communications with Altair support) using the edge port with V=0 voltage source is a proper way on Feko to identify the current across the edges in time domain. 

When extracting the simulation results, I got this (please note time signal = input current waveform, current magnitude in green is Current 1, and current magnitude in red is Current 2). The question is why does the current magnitude decrease as we move from one side to the other side of the plate? I expect to see the same current everywhere along the surface. Could you please help if you have an explanation. Thanks.

 

Regards,

Sam

Hi Sam,

Happy New Year to you,

Would it be possible to attach the model (.cfx) and POSTFEKO session (.pfs)?

Best regards,
Torben

Sam_21831

Hi Torben,

 

Here are attached. Thank you.

 

 

Regards,

Sam

Sam_21831

Hi Torben,

 

Have you had a chance to look at this problem?

 

 

Thanks,

Sam

Torben Voigt

Sam_21831 said:

Hi Torben,

 

Have you had a chance to look at this problem?

 

 

Thanks,

Sam

Hi Sam,

please excuse me, unfortunately I have completely lost sight of this, I will look at it as soon as possible.

Best regards,
Torben