How can I fix this meshing error in Union1? FEKO 2023.1 version

sswh2

feko_bending_aps_KD_fixed_half_cylinder.cfx

Hello, I am getting two problems here-

1. how can I move my patch to the center of this half cylinder? [cadfeko is attached here]

2. Here during meshing I am getting the error in union1, can anybody kindly fix this error so that I can run it? My probe radius is fixed 0.65 mm.

Torben Voigt

Sorry that I overlooked the fact that you are still using Feko 2023. I will attach a model for you later. I'm not quite sure what you mean by "center of the half cylinder". Something like this?

Or something more like this?

Torben Voigt

Note, the warning "radius to length is very large" refers to the wire thickness only. So no need to overmesh the whole structure. It is also noticeable that the region of the half cylinder is defined as PEC. I assume it should be the dielectric medium, right?

sswh2

@Torben Voigt sir center of the cylinder means the second figure you sent, and yes the medium between the ground and patch should be dielectric which I mentioned in my medium dielectic as TMM10i, kindly send me the updated model when you can. thanks.

Torben Voigt

Hi Srabonty,

I have recreated the model to get a better overview. Please check that everything has been implemented correctly.

feko_bending_aps_KD_fixed_half_cylinder_alt.cfx

Best regards,
Torben

Torben Voigt

Hi Srabonty,

That surprises me. At 3.5 GHz, the simulation on my computer takes just 2 seconds and requires only 50 MB of RAM. Please attach your modified model here.

If you don't mind, I'd also be interested to know what you plan to do with the model. I could simply customize it to change the position of the patch with a variable, for example.

Best regards,
Torben

Torben Voigt

I had already written that the warning about the ratio refers exclusively to the wire, so no need to overmesh the whole structure. For the solver a wire segment (meshed part of a wire) should not be shorter than 5* the radius. It's only a warning, not an error.

I'll take a look at your model and try the parameter sweep.

Torben Voigt

Hi Srabonty,

Here is the model. Note, I have already defined an optimisation in case you want to use OPTFEKO instead of parameter sweep.

newly_created_feed_can_be_rotatedcfx.cfx

Best regards,
Torben

Torben Voigt

Hi,

The error message usually indicates that you may not have enough memory for meshing. I am not sure why you are getting this error. In my case, I can mesh the model, but I get other errors. This is because there appear to be (at least) two surfaces extremely close together:

This also explains why the mesh looks a bit strange there:

In fact, two faces seem to be almost identical here. But not quite identical, otherwise they would no longer be defined as two surfaces after the Union operation:

If I remove one of the two, the simulation works without problems

Note, you should use normal MoM instead of MLFMM, as the model is not electrically large. I would recommend MLFMM from a model size of 3 wavelengths. On the contrary, the model is electrically rather small, which is why you should activate double precision. This means that twice as many decimal places are used, which significantly increases the accuracy for electrically very small problems.

I also think your chosen mesh size of lambda/50 is clearly exaggerated.

See if the attached model solves your problems.

feko_bending_aps_KD_fixed_half_cylinder_alt.zip

Best regards,
Torben

sswh2

@Torben Voigt sir I can't open your model, as I mentioned in my question that my FEKO is 2023.1 version, kindly send me this version model please, and I used lambda/50 in meshing because I was getting error that the ratio of probe radius to length is very large, my probe is fixed 0.65mm radius so I put lambda/50 and that error left. Now kindly upload your 2023.1 file so that I can run that.

sswh2

@Torben Voigt Sir second thing is kindly move my patch to the center of the half cylinder in your file, I can't move the patch to the center of the half cylinder. Thanks.

Torben Voigt

Sorry that I overlooked the fact that you are still using Feko 2023. I will attach a model for you later. I'm not quite sure what you mean by "center of the half cylinder". Something like this?

Or something more like this?

Torben Voigt

Note, the warning "radius to length is very large" refers to the wire thickness only. So no need to overmesh the whole structure. It is also noticeable that the region of the half cylinder is defined as PEC. I assume it should be the dielectric medium, right?

sswh2

@Torben Voigt sir center of the cylinder means the second figure you sent, and yes the medium between the ground and patch should be dielectric which I mentioned in my medium dielectic as TMM10i, kindly send me the updated model when you can. thanks.

Torben Voigt

Hi Srabonty,

I have recreated the model to get a better overview. Please check that everything has been implemented correctly.

feko_bending_aps_KD_fixed_half_cylinder_alt.cfx

Best regards,
Torben

sswh2

@Torben Voigt Sir thanks, this is the design I wanted, can you kindly tell me how can I decrease my simulation run time here? Because my plan is to vary probe location and check the bandwidth. to run 1 GHz it is taking more than one hour.

Torben Voigt

Hi Srabonty,

That surprises me. At 3.5 GHz, the simulation on my computer takes just 2 seconds and requires only 50 MB of RAM. Please attach your modified model here.

If you don't mind, I'd also be interested to know what you plan to do with the model. I could simply customize it to change the position of the patch with a variable, for example.

Best regards,
Torben

sswh2

@Torben Voigt sir now the timing issue fixed. But I want to do parametric study by varying the probe location, I changed my design to full cylinder, can you kindly check my model and tell why parametric sweep is not working? it is showing sbatch run error msg.

feko_bending_aps_KD_fixed_a_decrease(perfect_layer_parametric).cfx

sswh2

@Torben Voigt sir can you kindly check this error for parametric sweep?

Torben Voigt

Which variables should be changed in the parameter sweep? Do you want to find the best spot for the feeeding wire? Or do you want to move the patch (incl. the wire)?

Torben Voigt

You have again set an extremely fine mesh (lambda/50). Why?

sswh2

I want to change feed location, so First I want to change Xp, and then Want to change yp. Thanks.

sswh2

I choose lambda/50 again otherwise I am getting error that the ratio of wire segment radius to legth is too large.

Torben Voigt

I had already written that the warning about the ratio refers exclusively to the wire, so no need to overmesh the whole structure. For the solver a wire segment (meshed part of a wire) should not be shorter than 5* the radius. It's only a warning, not an error.

I'll take a look at your model and try the parameter sweep.

sswh2

@Torben Voigt sir thanks for your help! I am getting error not warning if I don't put lambda/50 in mesh, Second thing is this parametric sweep error, this model is taking 5 hrs approximately to finish one simulation, so manually changing the parameter will waste my time, so I need parametric sweep. Thanks, I will wait for your parametric sweep model.

Torben Voigt

Hi Srabonty,

By changing Xp and Yp the wire will move alongh the x- and y axis, but it will obviously not follow the curvature of the cylinder:

This could work for a certain range of values, as there is a "tolerance" in the union operation. This means that the connection between the inner cylinder surface and the patch still exists even if the wire actually ends slightly above / below it. Nevertheless, I would suggest that you do not move the feed wire, but rotate it so that it always maintains the exact connection.

Would you like me to help you with this?

Best regards,
Torben

sswh2

@Torben Voigt sir, yes please send me one model like how by rotating I can vary my feed probe?

Torben Voigt

Hi Srabonty,

I don't know what you are doing wrong, but when I set the mesh setting to "Standard", I don't get an error message. With lambda/50 and 30 frequency points, the simulation naturally takes a very long time.

I'll attach a new model here later, which you can use.

Question: are you using the Sweep parameter to optimize the feed (e.g. minimize s11)? Or do you want to do some kind of study with it? Because you could simply use the OPTFEKO optimizer to optimize the feed.

Best regards,
Torben

Torben Voigt

As already written, you move the wire in your model by changing Xp and Yp instead of rotating it. At a certain value, a small piece of the wire will protrude through the patch. I guess this is the moment when you get the error message about the ratio. Because this piece of wire is of course extremely short, but still 0.65 mm thick.

Wait for my model, this won't happen there.

Torben Voigt

Hi Srabonty,

Here is the model. Note, I have already defined an optimisation in case you want to use OPTFEKO instead of parameter sweep.

newly_created_feed_can_be_rotatedcfx.cfx

Best regards,
Torben

sswh2

@Torben Voigt sir thank you.