Every number carries its pedigree.
This is a side-by-side: first, how composite parts are proven today — coupons, months, and margins no one can trace to a build. Then the loop’s way: every value shows where it came from, how uncertain it is, and what it is not allowed to claim. The ultimate composite is the one you can prove.
Click, use the arrow keys, or use the controls below to start with today.
How composites are qualified today.
Cut coupons from witness panels. Wait months for test campaigns. Absorb the remaining doubt with knockdown factors that no one can cite to the build that produced the part. Composite parts today are overengineered — and manufacturing them is too costly — because every uncertainty is paid for twice: once in scrap, once in qualification. Your datasheet was written in a lab; your part was built in a room.
Problem economics shown as industry-reported ranges, compiled 2026.
Three companies. Uncoupled feedback.
Feedback between the three exists — reports, datasheets, purchase specs — but it is uncoupled: nothing ties a test value to the build that produced the part.
All twin data must originate from the same set of equipment. A part built at one facility cannot be validated by a twin that lacks that facility’s provenance.
Provenance chips — read any value at a glance.
The loop’s way replaces untraceable margin with visible pedigree. Three colors, one rule: the treatment of a value tells you its reliability before you read it. Vendor-stated values print in ink. Assumed values print in blue with the assumption disclosed. As-made values — measured on a Greenlight machine — print in green.
Ink / black. The source document stated both the value and its spread. The chip carries the citation: document, page, condition. Greenlight never restates a vendor number without its source.
Blue. The vendor gave a value but no spread, so the disclosed default of ±10% is applied — and labeled assumed, always. An assumption you can see is an input you can challenge.
Green.Measured from a build on a specific TapeLayer™, keyed to machine, build, and process record. The spread is the measured one, not a default.
Values above are representative published values for an IM7/8552-class UD ply, shown to demonstrate the treatment. Alongside the chips, every Library row is categorized measured / model-derived / assumed.
The rules the software cannot break.
- 01
Every value shows an explicit ±
No bare numbers. If the vendor states a spread, Greenlight uses it. If the vendor is silent, the default is ±10% — applied openly, colored blue, labeled assumed. Metric primary, imperial in parentheses.
- 02
No recommendation below 90% confidence
Design will not issue a layup recommendation below a 90%confidence floor. Uncertainty is propagated with correlated draws, so output bands are never tighter than the input data justifies — the software cannot manufacture precision it wasn’t given.
- 03
The validity ladder: "needs FEA," never a faked number
Closed-form methods have domains. When a case steps outside them — a deflection beyond small-deflection theory, a geometry the model doesn’t cover — the answer is “needs FEA”, stated plainly. A refusal you can trust beats an estimate you can’t.
- 04
Simulated content is badged, always
Nothing simulated is ever presented as measured. Badges persist on screen — they don’t scroll away, and they export with the data.
- 05
Iterations are integers, rounded up
A plan that calls for 2.2 test panels is a plan for 3. Fractional test articles don’t exist, so they never appear in a Greenlight plan.
Built for the qualification world.
From CAD to coupon to part — with traceable data your customers can cite. Ingest and Library output is structured for contribution to CMH-17 Volume 1[2] (polymer matrix composites) or emerging composite qualification efforts, so the data a customer generates inside the loop is a deliverable attached to test results, not a dead end in a private format.
Coupon panels built on a TapeLayer™ are cut to standardized ASTM coupon geometries and tested by the customer or a third-party lab of their choice — Greenlight sells machines and the Greenlight Loop™ platform, not testing services.
Test-specimen handling — responsibility disclaimer
The process record ends when the part leaves the machine. Customers must document post-production handling, storage, and conditioning of test specimens; the validity of downstream test results is the customer’s responsibility. Greenlight says this up front because a provenance chain with an undocumented gap is not a provenance chain.
Full treatment for qualification engineers: the Qualification Playbook (GL-DOC-002), in the member portal — request access →
Judge it yourself.
The demo runs real laminate physics with every policy on this page enforced — watch what it refuses to tell you as closely as what it does.
Sources
- [1] Material qualification duration and cost — industry-reported ranges, compiled 2026.
- [2] CMH-17 Composite Materials Handbook, Volume 1 — polymer matrix composites.
- [3] Nilakantan, G. & Nutt, S., “Reuse and upcycling of aerospace prepreg scrap and waste,” Reinforced Plastics 59(1), 2015, pp. 44–51 — aerospace manufacturers discard 30–50% of prepreg during ply-cutting (nesting inefficiency, fixed roll widths, limited out-life). See also Nilakantan & Nutt, J. Composite Materials, 2018 (DOI 10.1177/0021998317707253).
Full source library — available to investors and Letter-of-Intent signatories: request access →
Greenlight Industries · Delaware C-Corp · Virginia, USA — problem economics shown as industry-reported ranges, compiled 2026.