Modal analysis problem with STKO - mesh size

[oguzhan.gumus]

Hello STKO Team,

I hope you are doing well.

I have a question regarding mesh size use in STKO vs other FE software.

I have a timber beam which is simply-supported. I would like to perform a modal analysis. However, the selected mesh size or pattern has a significant effect on the modal frequency results.

I've been analyzing the effect of mesh size on the first modal frequency and analysis time using STKO software. As shown in the attached Excel file inside the "mesh_study_results.zip" file, the frequency converges very slowly—requiring a mesh size as small as 1 cm to converge around 9-11 Hz. Beam dimensions: h=48mm, b=396mm, L=5.52m, rho=475.0 kg/m3, and E=11000 N/mm2. In comparison, the same model in other FE platforms reaches convergence at much coarser meshes (typically 5–10 cm). This raises concern about the modal accuracy and mesh sensitivity of the solver. Could you clarify why such fine mesh refinement is needed to achieve frequency convergence in your software? Are there any internal settings or solver formulations affecting this behavior? I attached the timber beam STKO model for your reference. This is crucial for my research studies about modal updating. Could you please assist me about this topic?

Please, let me know if you need further explanation about the issue. Thank you very much in advance.

Best Regards,
Oguz

[STKO Team]

It seems reasonable: you are modeling a beam with a thickness of about 5 cm.
If you use linear tetrahedrons (with a constant strain) you have to use at least 5 elements along the thickness.
Furthermore, when you use a mesh = 100 1) you have only 1 tetrahedron along the thickness, and 2) it's super distorted, leading to a very stiff result