HFSS mesh problem: simulating a big structure (4m) with a tiny component (1mm).

Hi all,

I am simulating a monopole antenna on the UAV. The UAV is about 4 [m] by 3 [m]. The monopole is put on the wing of the UAV, monopole is 70 [mm] long with a 1.5 [mm] diameter. Frequency is 1 GHz. When I try to mesh the structure, using the default mesh setting, an error occurred: "critical features could not be preserved correctly".

I googled this one, and people say it is because I am simulating a big structure with a tiny component. I am wondering how can I change the mesh setting?

Plus, the simulated UAV is a simplified version, only with a rectangular wing and a rectangular body, no curved or complex structure.

I tried to change the initial mesh setting, set the maximum surface deviation as 1/10 of the smallest dimension, maximum normal deviation 5 deg, and the aspect ratio 10, as shown in the attached pic. But seems that the meshing took a pretty long time... I just want to see the impact of the UAV body to the EM performance of the monopole, so I think I dont have to mesh the whole UAV body using the smallest dimension? Maybe just mesh the particular monopole carefully?

I am really not familiar with this mesh operation in HFSS. Usually I just keep the default mesh setting. Friends, any suggestions on the meshing operation will be appreciated!

Thanks.
Panou

I guess only option is to switch over to simulators such as ADF (Antenna Design Framework) wherein the complete platform + your antennas could be analyzed accurately and quickly.

FEKO could also be of help .. try it out !

Thanks! But it is not easy to purchase another EM solver...

What's the version of hfss do you use? if you used newer than v13, you can try the hybrid method in HFSS, such as HFSS-IE (Integral Equation), HFSS-PO (Physical Optics), Hybrid FEM-IE or Enhanced Hybrid IE-Regions.

You have to understand the problem nature, it is a very typical multi-scale problem, which means objects in the problem domain vary from meters to mm or even smaller. Such type of problem is a great challenge nowadays and ongoing research is still being performed. Hybrid method is the solution! However, implementing hybrid is not trivial, if you use HFSS-IE + HFSS-FEM, the boundary integral method for solver communication can be problematic and I am curious to know if ANSYS-Ansoft has resolved this issue perfectly. FEKO + Antenna Magus can be another solution, but to my best knowledge, I've never seen a true multi-scale example released to the public domain. None of them CST, FEKO, ANSYS, XFDTD, ADS .....

I suspect with manual intervention, you can make the mesh finer in the region of interest, and course elsewhere. I don't know how to do it, but I believe it can be done.

Also, there is software based on raytracing to simulate such applications
http://www.delcross.com/products-savant.php

Delcross has a Shooting and Bouncing Rays (SBR) code called Savant that can do this for high frequencies where the problem size is very large. They can import 3D full-wave EM models for isolated antennas, and then use that to stimulate a hybrid simulation of using SBR to analyze the installed performance of the antenna.

They can import antenna models simulated from CST Microwave Studio, HFSS, the FEKO FEM solver, and WIPL-D. They offer integration with CST to do this automatically, as well as for WIPL-D. You have to contact them to get MATLAB routines to convert HFSS or FEKO FEM models for import into their Savant solver. Here is the product page for Savant: http://www.delcross.com/savant/