Yesterday's step in the hanger-codesign programme found a confound. In our 3D cloth sheet, the shear springs were tied to the stretch stiffness EA, so every "EA" sweep also swept shear. On a simulated drape test, shear carried most of the effect. The day before that, an EA sweep had moved the best arm width of a shallow 3 cm hanger arm by 1.5 cm. Was that shear too?
This step set the shear spring to each fabric's measured KES shear rigidity (26 to 89 N/m) and swept EA on its own. It ran 126 converged cases across four fabrics, with predictions committed before the run.
- Stretch is still most of it. EA 40 to 800 N/m moves the best width by 1.0 cm on both the cotton jersey and the suiting (+2.0 to +1.0 cm). Shear accounts for the remaining third of the old 1.5 cm.
- Above about 800 N/m the choice settles. At EA 800, 3200 and the stiffness derived from the KES data, every fabric picks +1.0 cm, with one exception: the shirting at 800 N/m picks +1.5 cm. At those stiffnesses the margins are 0.02 to 0.08 cm, well clear of the solver's own wobble (at most 0.008 cm).
- That is the 2D chain's answer. At the KES-derived stiffness, the 3D sheet and the 2D chain pick the same width on all four fabrics. The half centimetre of disagreement we chased for three steps is a soft-cloth effect.
Where did the KES-derived stiffness come from? The table that gave us bending rigidity, weight and shear rigidity also lists tensile linearity (LT) and tensile energy (WT). With the tester's standard 500 gf/cm maximum load, those two numbers give the strain at that load, and a secant stiffness of 2,400 to 15,500 N/m. The load is an assumption, because the paper does not state it.
But a garment hanging on a robot's hanger pulls on its shoulder with a few N/m, not 490. A power-law curve that matches both LT and WT puts the stiffness at that load near 1,300 to 1,900 N/m for the suiting and over 5,000 for the elastane jersey, so their choice is settled. The cotton jersey (about 250 to 370 N/m) and the shirting (700 to 1,050) land right in the band where the choice flips between +1.0 and +1.5 cm.
So the one fabric measurement still missing is not another KES summary number. It is the toe of the tensile curve: how much a strip of T-shirt jersey stretches under the weight of a T-shirt. A spring scale and a ruler can measure that.
Simulation only, produced with AI assistance; nothing here is a measurement of real cloth. Code, numbers and the full log are in research/hanger-codesign.