# Kirsch Plate FemLite static shell benchmark for the Kirsch infinite plate problem. The benchmark data follow `models.sme.kirsch_plate.pdf` in this directory: - Geometry: quarter model of a 2 m x 2 m plate with a central circular hole. - Modeled domain: x in [0, 1] m, y in [0, 1] m. - Hole radius: 0.1 m. - Thickness: 0.1 m. - Material: E = 2.1E11 Pa, nu = 0.3. - Load: distributed stress sigma_x = 1000 Pa on the right edge. - Constraints: symmetry on x = 0 and y = 0. - FemLite representation: linear shell membrane model, with all out-of-plane translations and rotations constrained to form a plane-stress benchmark. Generated decks: - `krisch_tri.inp`: S3 triangular shell mesh, 1025 nodes, 1920 elements. - `kirsch_quad.inp`: S4 quadrilateral shell mesh, 1025 nodes, 960 elements. Regenerate the decks: ```powershell Python\.venv\Scripts\python.exe examples\Kirsch\generate_kirsch.py ``` Run the static cases from the repository root: ```powershell New-Item -ItemType Directory -Force examples\Kirsch\results\tri_static | Out-Null build-win-gcc-16\femlite.exe examples\Kirsch\krisch_tri.inp --output-dir examples\Kirsch\results\tri_static --static-linear ``` ```powershell New-Item -ItemType Directory -Force examples\Kirsch\results\quad_static | Out-Null build-win-gcc-16\femlite.exe examples\Kirsch\kirsch_quad.inp --output-dir examples\Kirsch\results\quad_static --static-linear ``` The PDF uses the Kirsch analytical stress along the vertical symmetry line: ```text sigma_x(y) = 1000 / 2 * (2 + (0.1/y)^2 + 3*(0.1/y)^4) ``` Therefore: - Near hole, y = 0.1 m: sigma_x = 3000 Pa. - Away from hole, y = 1.0 m: sigma_x = 1005.15 Pa. The PDF reports finite-plate errors relative to the analytical solution: | Location | PDF finite plate error | PDF infinite plate error | |---|---:|---:| | Near hole | 1.1% | 0.2% | | Away from hole | -4.0% | -0.1% | FemLite comparison below uses global plane-stress sigma_x recovered from the computed nodal displacements at the nearest elements touching the x = 0 symmetry edge. This is a discrete element-point comparison, not a line-probe postprocess with nodal extrapolation. | Deck | Sample point | Element | Sample y | FemLite sigma_x | Analytical sigma_x at PDF point | FemLite error vs PDF point | Analytical sigma_x at sample y | FemLite error vs sample y | |---|---|---:|---:|---:|---:|---:|---:|---:| | `krisch_tri.inp` | near hole | 1874 | 0.100958 | 3030.09 Pa | 3000.00 Pa | 1.00% | 2934.44 Pa | 3.26% | | `krisch_tri.inp` | away from hole | 1919 | 0.977874 | 965.05 Pa | 1005.15 Pa | -3.99% | 1005.39 Pa | -4.01% | | `kirsch_quad.inp` | near hole | 937 | 0.101437 | 2968.95 Pa | 3000.00 Pa | -1.03% | 2902.75 Pa | 2.28% | | `kirsch_quad.inp` | away from hole | 960 | 0.966813 | 971.47 Pa | 1005.15 Pa | -3.35% | 1005.52 Pa | -3.39% | The S3 result closely reproduces the PDF finite-plate errors at both benchmark locations. The S4 result is also within a few percent; its far-field drift is consistent with the finite-plate trend reported by the PDF.