Published benchmark + Solver B
MacNeal–Harder twisted beam
The initially twisted beam is loaded in-plane and out-of-plane. Tip displacements are compared with the published benchmark and a separate Solver B solution.
Understand the analysis
A twisted cantilever, loaded at its free end.
The strip is already twisted in its initial geometry. One end is fixed and the other is loaded, first in-plane and then out-of-plane in separate runs. The comparison measures the free-end displacement.
In-plane loading: the final recorded state of the 8 × 48 shell mesh.
Full size ↗Out-of-plane loading: the matching 8 × 48 shell run.
Open video ↗Both views use the saved benchmark runs. Displacements are enlarged 200×; the colour bar shows the original displacement magnitude. The graphs use the hold-window mean, while the snapshot shows the final recorded state.
On the 8 × 48 mesh, MinuteSim reaches 0.9983 of the in-plane reference and 0.9975 of the out-of-plane reference. Solver B gives 0.9987 and 0.9976, respectively.
Two load directions. Three shell meshes.
Download plotted data ↓
Benchmark conditions
A quasi-static explicit run
Reported tip displacement is the mean over the 0.20–0.25 s hold window at the middle tip node. These results come from explicit runs under quasi-static loading, using the selected warped Quad4 configuration.
A close view of the differences
All panels use the same truncated vertical range, 0.98–1.003. Published reference displacements are 0.005424 in-plane and 0.001754 out-of-plane, in the benchmark’s unit system. The recorded four-decimal normalized values are plotted.
Minor differences between ratios recomputed from rounded raw displacements and the plotted ratios are below 0.0001. Solver B precision is not specified in the supplied comparison record.
Reference
R. H. MacNeal and R. L. Harder, A proposed standard set of problems to test finite element accuracy, Finite Elements in Analysis and Design 1(1), 3–20 (1985).
Run conditions: MinuteSim FP32 on an NVIDIA GeForce RTX 4060. Source results: 5 October 2026.
SUPPORTED · Numerical comparison record for the stated configuration. Broader validation and capability limits are described on the Evidence page.