Feko Solver Stuck on "Calculation of submatrices A1x"

Jizheng He
Jizheng He Altair Community Member
edited June 3 in Community Q&A

Hi all, I am trying to simulate a dome reflector (see image below). I am using course-graded curvilinear meshing that breaks the dome into ~19k triangles and using MoM with higher order basis functions (HOBF). However, my solve is stuck for hours on the step:

 

Calculation of the elements of MoM submatrices

Calculation of submatrices A1x (100% of the full matrix)

 

From this post switching to MLFMM should converge faster. However, if I use MLFMM that does not work with curvilinear meshes, I am facing 1.2 Million triangles in my course-graded mesh because of high frequency bands... How can I make my model run faster?

 

image

Tagged:

Answers

  • Jaehoon
    Jaehoon
    Altair Employee
    edited April 2

    Hello Jizheng,

    I would recommend you to use "Characterised surface medium" to get a fast simulation result.

    (I assume the dome reflector consists of homogeneous medium.)

    The steps are 

    1) Calculate/Save T&R coefficients by using Infinite Planar multilayer or PBC (Periodic Boundary Condition) + Plane wave source with many incident angels.

    2) Define "Characterised surface medium" with the saved .tr file.

    3) Apply the characterised surface medium to the dome reflector curvature (no thickness needed.)

    You can see similar example (H.5 Characterised Surfaces for FSS) in the Feko example guide.

    Regards,

    Jaehoon

     

  • Jizheng He
    Jizheng He Altair Community Member
    edited May 5
    Jaehoon Kim_22316 said:

    Hello Jizheng,

    I would recommend you to use "Characterised surface medium" to get a fast simulation result.

    (I assume the dome reflector consists of homogeneous medium.)

    The steps are 

    1) Calculate/Save T&R coefficients by using Infinite Planar multilayer or PBC (Periodic Boundary Condition) + Plane wave source with many incident angels.

    2) Define "Characterised surface medium" with the saved .tr file.

    3) Apply the characterised surface medium to the dome reflector curvature (no thickness needed.)

    You can see similar example (H.5 Characterised Surfaces for FSS) in the Feko example guide.

    Regards,

    Jaehoon

     

    Hi Jaehoon, thanks for your answer and sorry for the delay. I constructed a dome curvature surface model but am stuck on the "Calculate/Save T&R" part. According to the H.5 guide I still need to define an incident plane wave source, and seems like running the solve with the T&R request still attempts to calculate the full A1x matrix (and never finishes). Is this intended?

  • Alexander Efremov
    Alexander Efremov New Altair Community Member
    edited June 3
    Wow???? I had same problem. Big thanks my friend

Categories