The accuracy of digital image-based finite element models.

Area of Science:

• Computational mechanics
• Biomaterials engineering
• Finite element analysis

Background:

• Digital image-based finite element meshing (DIBFM) provides a faster alternative to conventional methods for generating realistic 3D models.
• DIBFM is increasingly used for structure-specific biological models, but its accuracy at boundaries requires further investigation.

Purpose of the Study:

• To compare the solution accuracy of digital and conventional smooth boundary finite element models.
• To evaluate accuracy using theoretical solutions for 2D and 3D benchmark problems.

Main Methods:

• Finite element analysis of a 2D compression plate and a 3D circular cantilever beam.
• Comparison of digital and conventional meshing techniques at model/material boundaries.
• Analysis of solution accuracy based on theoretical predictions.

Main Results:

• Digital models exhibited local oscillations at boundaries, causing potential errors in individual elements.
• Digital boundary solutions oscillated around the theoretical solution, with averaging improving accuracy.
• Averaged Von Mises stress error was <4%, and free-end displacement error was <1% for the cantilever beam.

Conclusions:

• Averaging digital finite element solutions over several elements significantly enhances accuracy.
• DIBFM is a viable approach for complex structure modeling when boundary effects are managed.
• A minimum of 3-4 elements across the cross-section is recommended to mitigate numerical stiffening errors in bending analyses.