Does a patch-averaged stress make the 2 mm laser-cut bracket region mesh-independent?

What it found

Preliminary, stylised 2D plane stress. On the 39 gauge-2 bracket designs that fail the stress limit on fine meshes under the element-centroid peak, a 3 mm patch-averaged stress changes by at most 2.2% between 1.0 and 0.75 mm meshes (the centroid peak rises up to 9.7%), and all 39 stay feasible at every mesh size, with a median of 84–86 MPa on the finest mesh. The verdict flips because the criterion is less strict, not because the model changed; whether a patch average is an adequate design rule was not shown, and no fine-mesh design search was run.

Paper, code and results

• README and method
• Experiments and results
• Code
• Tests

More in Simulation-first design

• Does the peak stress at the fillet tangent point of the L-bracket converge under graded mesh refinement?Does the L-bracket's peak stress at the fillet tangent point converge under graded mesh refinement? Yes for four designs: it flattens at 1.09–1.14× its 1.5 mm value (0.0–1.1% change over the last 4× refinement down to 0.025 mm), so the 2 mm region's mesh dependence is a resolution effect; the 16/15/2 design is marginally over 100 MPa when converged; stylised 2D, no reference solution.
• Is the least-mass laser-cut bracket the same on coarse and fine meshes under a patch-averaged stress rule?Under a 3 mm patch-averaged stress rule the least-mass laser-cut bracket is the same on a 2.5 mm and a 1.0 mm mesh (gauge 2: 32.31 vs 32.33 g), while the centroid-peak rule leaves no feasible gauge-2 design on the fine mesh; the patch value hardly sees the fillet radius; stylised, preliminary.
• Do cheap beam and 2D models rank candidate cantilever designs like 3D finite elements?Do cheap beam and 2D models rank cantilever designs like 3D finite elements? Yes on 3 seeds of 80 designs: Kendall τ 0.987–0.996 and the 3D-best design always ranked first, even though single-design merit is off by up to 14%; stubby sections agree least; stylised, one load case.
• Can a parametric L-bracket be sized by finite elements for minimum mass and come out printable?Parametric L-bracket sized by 2D finite elements (scikit-fem) for minimum mass under a stress limit: differential evolution 19.6 g vs random 25.6 g vs grid 31.4 g at equal budget, exported as STL with a printability check. Assumed material inputs. Stylised.
• Does greedy substitution search still work when the printer-redesign list grows past seven items?Does greedy substitution search still work past seven items (synthetic instances, assumed generator): 96–100% of the exact optimum on average but optimal in only 33–95 of 100 instances; 64 random evaluations fall to about 89% at 24 items.
• Does an aluminium rod change which printer redesign the penalty-budget search picks?With an aluminium rod option priced by sag, the penalty-budget search of the printer redesign picks the 12 mm aluminium rod and reaches 40.0% mass closure at budget 2.0 against 33.1% with the wooden rod (41.4% against 35.8% at full budget); the penalty curve is assumed and the metric favours heavy stock; stylised, preliminary.

All 64 Simulation-first design projects →

Other research topics

Proper scoring & elicitation28Effort & peer prediction9Verification games & slashing19Detection & tolerances14Markets as routers13Prediction-market design9Paying contributors8Decentralised training systems23Digital twins & sim-to-real164LLM chip design loops40Robot manufacturing loops53Ambient AI assistant46Self-replicating manufacturing34Simulation-first design64Engineered clothing45Task-specific AI accelerators40Algorithmic chip-design loops42Open-source purpose-built robots44The software-engineering loop43Adversarial human imitation41Open-source residential construction41Autonomous multi-angle capture34Sensing, imaging and 3D31Energy supply and dense energy storage39Robot societies among people31Agentic co-design of low-cost robots27Laundry-hanging robots25Machine learning on qualitative data23Robots on the building site27Civilizations & group psychology1History & human sciences1

Produced with AI assistance (Claude). Results are from the code in this repository and are simulated or stylised unless the abstract says otherwise; nothing here implies affiliation with or endorsement by any lab, researcher or company cited. MIT licensed.