How to define the most precise meshing for simulations?

Matheus Pessoa

I have a doubt: when it comes to simulating an antenna with wires only, with a given radius, how do I define ultimately if the simulation is precise? I ask this because, for instance, changing the mesh size changes the results of the simulation by a few Ohms in impedance, and as it gets finer and finer, I don't see convergence in my analysis.

 

For example, I've used the fine meshing setting for different maximum frequencies in my simulations, as shown below, as increasing the maximum frequency will change the number and the length of the segments in which the wires in the simulation are divided. But as you can see on the second image, where I take the 3GHz maximum frequency as the reference and compare all other maximum frequencies with it, there appears to be no convergence. Is the solver misbehaving when using smaller and smaller elements?

 

 

Mel

A slight shift in response is expected based on changing mesh sizes.

For high fidelity results, if practical, the wires could be replaced with cylinders meshed into triangles.

If you can attach the model or an image of the model, that would of course help.

 

Matheus Pessoa

Hello, Mel. Thank you for the message. Attached is the FEKO model, and as you can see, I try to simulate metal rods inside infinite soil; I am not completely sure how to proceed for using solid rods in this case, so your help could be of amazing value. Cheers.

 

Matheus

Mel

At 500 MHz, in the lossy soil, the wavelength is already 0.136 meters. This would then require a segment length of around 1/10th of that, so around 1.4 cm. The "Fine" setting in this case is not fine enough, as some segments are around 2 cm long, which for higher frequencies, is even more inadequate.

So the mesh is probably not converged yet.