UTM V2.1: curved numerical material responses

9 November 2026. Model `utm-v2.1`; existing route `/simulations/tensile-test`.

Reason for the change

V2.0 already generated approximately 400 points. It evaluated a piecewise-linear law, so the points were collinear between control events. Increasing sample density alone could not curve those branches. V2.1 changes the authored constitutive law before calculating force, extension, graph coordinates or exports. The renderer still plots the actual data points; it adds no spline or Bezier smoothing.

Branch equations

On a branch between (εa,σa) and (εb,σb), define t = (ε−εa)/(εb−εa), 0≤t≤1, and σ = σa+(σb−σa)H(t).

LawH(t)Integral from 0 to t
Lineartt²/2
Decreasing-tangent rise2t−t²t²−t³/3
Progressive changet²t³/3
Saturating cubic rise1−(1−t)³3t²/2−t³+t⁴/4
Smooth endpoint transition3t²−2t³t³−t⁴/2

All H values are monotonic from 0 to 1. No branch exceeds its endpoint stress range. The elastic branch always remains σ=Eε, and the ideal steel yield plateau stays constant. Brittle polymer and upright FFF retain linear-elastic responses followed by fracture.

These are monotonic illustrative educational relationships, not fitted experimental curves or certified material datasets. There is still an idealised knee between linear elasticity/yield and plastic branches. No measurement noise, strain-rate sensitivity, temperature response, machine compliance or grip slip has been added.

Integration and outputs

The grid retains 400 intervals plus exact branch endpoints, the proof point and current time endpoint. Full steel tests contain 402 points including the zero-force fracture drop. A cumulative trapezoidal sum over the actual force–extension dataset supplies every exported energy value and the result-table work. An independent analytic integral of the polynomial branches verifies integration error. Units, area calculations, physical/playback speed, motion and fracture state are unchanged.

For the default steel specimen (10×5 mm, 50 mm gauge):

QuantityV2.0V2.1
Stress at ε=.085340 MPa385 MPa
Stress at ε=.20380 MPa405 MPa
E200 GPa200 GPa
Yield250 MPa250 MPa
UTS / maximum load430 MPa / 21.5 kN430 MPa / 21.5 kN
Pre-fracture stress / load330 MPa / 16.5 kN330 MPa / 16.5 kN
Loaded model fracture strain25%25%
Sampled external work217.609375 J231.525653545673 J

The V2.1 analytic steel work is 44453/192 J = 231.526041666667 J. The sampled error is approximately 0.000388121 J. The thermoplastic analytic work is 2771/48 J = 57.729166666667 J, sampled 57.729116555222 J. The brittle polymer stays exactly 1.5 J. Aluminium Rp0.2 becomes 208.860609961810 MPa at ε=.005026965361765368 under the new hardening law; the change is an illustrative model revision, not a newly measured property.

Reference values in `tests/fixtures/utm-v21-reference.json` were calculated independently using Python Fraction and Decimal arithmetic without importing simulator code. Tests check quarter/mid-branch values, every profile's monotonicity/bounds, endpoint strength, a zero-tangent UTS transition, exact elastic/brittle responses, proof intersection, integration and export consistency.

Saved-record compatibility

The current model version is `utm-v2.1`; the record schema remains version 2. The validator accepts both `utm-v2.0` and `utm-v2.1` and preserves the incoming model version and historical dataset. V2.0 material notes, linear laws and offset solver remain available through version-aware material selection. Read-only reopened graphs use their stored model version for event annotations, and reports use the matching historical explanation.

A frozen pre-upgrade V2.0 steel investigation is retained in `tests/fixtures/utm-v20-investigation.json`. Pure and isolated account-API tests verify exact dataset/property preservation and owner-only retrieval alongside V2.1 records. No production learner record or membership permission was changed.

Observed release checks

The publication gate remains a successful production build, retained regression suite and relevant browser checks. No claim is made that a smoother educational curve validates a real specimen or ISO-compliant experiment.

Open laboratory · Technical notes · Historical V2.0 release report