Error 21109
This is another question in a line of questions on this forum. I am thankful for, & greatly appreciative of, the advice that is forthcoming.
Thank you Altair Communiy!
This one is about running Optfeko. An optimisation goal, after running for a long while, terminates at about 22% (I've run this about 2 or 3 times). The error (attached) refers to a previous detailed error message but there is none.
I would be grateful for some more advice.
Regards,
GS
Answers
-
Hi Gaurav
Please attach the cfx file.
1 -
-
mel_21333 said:
Hi Gaurav
Please attach the cfx file.
Hi Sir, trust you are well.
Hope you had a restful weekend.
Do you have some advice for me w.r.t Error 21109?
Best,
Gaurav
0 -
Thank you for sharing your file.
In the model you are using a MoM solver and have drawn the "Patch" structure and the "Feedline" structure as finite thickness PEC Cuboids. You have modelled the ground plane as a single face with material Copper and thickness = t.
I would encourage you to rather model the patch and feed line in the same way that you have modelled the groundplane. Modelling the finite thickness in the way that you have for the patch and feedline will require a much finer mesh for simulation to avoid errors - and it will result in very slow simulation and no improvement in results for the MoM solver.
I have quickly made these changes and the optimisation (Grid-search) as you have it set up then seems to run fine.
1 -
Brian Woods_21145 said:
Thank you for sharing your file.
In the model you are using a MoM solver and have drawn the "Patch" structure and the "Feedline" structure as finite thickness PEC Cuboids. You have modelled the ground plane as a single face with material Copper and thickness = t.
I would encourage you to rather model the patch and feed line in the same way that you have modelled the groundplane. Modelling the finite thickness in the way that you have for the patch and feedline will require a much finer mesh for simulation to avoid errors - and it will result in very slow simulation and no improvement in results for the MoM solver.
I have quickly made these changes and the optimisation (Grid-search) as you have it set up then seems to run fine.
Appreciate the suggestion, Sir.
If it's still lying on your Desktop, or Recycle Bin, could I request you to share the file?
Attached is my effort with your corrections implemented; this sadly still terminates at 22%.
Best,
Gaurav
0 -
Gaurav Sobti said:
Appreciate the suggestion, Sir.
If it's still lying on your Desktop, or Recycle Bin, could I request you to share the file?
Attached is my effort with your corrections implemented; this sadly still terminates at 22%.
Best,
Gaurav
Here is the file that I set up. Please note that I did not look too closely at the optimisation set up - but with the model parameterization and the setup that is there, there are certainly some questionable geometries created.
1 -
Gaurav Sobti said:
Appreciate the suggestion, Sir.
If it's still lying on your Desktop, or Recycle Bin, could I request you to share the file?
Attached is my effort with your corrections implemented; this sadly still terminates at 22%.
Best,
Gaurav
Some quick feedback on your model:
[Note: After making changes 1 and 2 I do not get any error in the optimization. I did not make changes 3 and 5, but I encourage you to consider them to get sensible results!]
1. I see that there was no excitation defined. I added a Voltage source to the port.
2. The mesh is very coarse on this model (you are using the "Coarse" setting in the mesh dialogue). I set this to "Standard", but for this patch, you will need to use a finer mesh to get good results - particularly in the slots and on the radiating edges of the patch (maybe something like I show below).
3. It does appear as if you intended this geometry to be symmetric? It is drawn asymmetrically at the moment, but if this were corrected, you could use symmetry to accelerate the simulation
4. Your optimisation is focussing on input impedance. For this case, it would be a good option to use a planar multilayer substrate (Greens-function) to model the substrate and a microstrip feed. This would (again) make the simulation/optimization considerably faster.
5. with the Grid-search setup there are some non-sensible cases (such as the one shown below). You should maybe just double-check that the model parameterization is as you intended.
1 -
Brian Woods_21145 said:
Some quick feedback on your model:
[Note: After making changes 1 and 2 I do not get any error in the optimization. I did not make changes 3 and 5, but I encourage you to consider them to get sensible results!]
1. I see that there was no excitation defined. I added a Voltage source to the port.
2. The mesh is very coarse on this model (you are using the "Coarse" setting in the mesh dialogue). I set this to "Standard", but for this patch, you will need to use a finer mesh to get good results - particularly in the slots and on the radiating edges of the patch (maybe something like I show below).
3. It does appear as if you intended this geometry to be symmetric? It is drawn asymmetrically at the moment, but if this were corrected, you could use symmetry to accelerate the simulation
4. Your optimisation is focussing on input impedance. For this case, it would be a good option to use a planar multilayer substrate (Greens-function) to model the substrate and a microstrip feed. This would (again) make the simulation/optimization considerably faster.
5. with the Grid-search setup there are some non-sensible cases (such as the one shown below). You should maybe just double-check that the model parameterization is as you intended. Greatly appreciate your detailed response. Sir. Very kind of you.
Today I've finally diagnosed the problem and I thought I'd share. Turns out that every file terminating optimisation at 22% was caused due to a limit on the student version of Cadfeko that I've access to.
It's a relief to know the cause of error, but disappointing at the same time!
Thank you anyway.
Regards,
Gaurav
0
Altair Employee