Greenlight Loop — open demo

Spec a part, right here.

The Loop couples design, simulation, and workmanship into one system instead of three isolated ones. Below are the two stages you can run cold, with real physics under the hood — classical laminate theory, a Tsai-Wu strength screen, and correlated uncertainty propagation. Nothing here is faked, and nothing claims more precision than the data supports.

GUIDED TOUR

Sarah’s Journey

Sarah — a structural engineer qualifying a brand-new material for a composite bracket — walks every stage with you, one screen at a time. Old way: months of guesswork. Greenlight way: several weeks, with traceable data at every step.

01 · Ingestsign in required

A new material’s datasheet or test report becomes structured, cited data — every value traced to a source, page, and vendor-stated-vs-assumed flag.

SARAH Sarah has a datasheet for a material the library has never seen. She drops the PDF into Ingest; mAI (Machine AI) returns structured rows, every value tagged vendor-stated or assumed, with its source page. She saves it to her private save space — it’s hers.

Stage 1 of 6
Try it — Digital Twin
Digital Twinsimulated

A guided walkthrough, not a lab instrument

This is a fixed demonstration of what the Digital Twin does — build, watch, validate, remember — not a true engineering system yet: no material picker, no editable load case, just the four beats in order. The freeform version (any material, any load, any temperature window) is a signed-in Loop tool, not this demo.

Third-party product names (e.g. TC1320 PEKK) are used for representative demonstration purposes only.

01 · Build startssimulated

The validated layup goes to the TapeLayer™ inside a sealed, environmentally controlled build chamber.

[0/±45/90]ₛTC1320 PEKK · quasi-isotropic, the canon recommendation

Effective laminate modulus Ex = 51.82 GPa going into the build — the same number the Design tool above would recommend for this material.

Stage 1 of 4
Try it — Cost
Cost

Per-part cost calculator

Transparent, user-adjustable — every line recomputes live from the inputs below. This is the one tool in the Loop where dollar figures are shown; Design and Digital Twin stay cost-free by design.

Part & process

Autoclave-class cure is industry-reported at ≈60–75% of total processing energy; this calculator uses the 67.5% midpoint to split out a cure-energy line below.

Cost assumptions

Greenlight couples economics directly into the engineering workflow — cost is a design variable, not an afterthought.

Typical 100–500 kWh/cycle; >1000 for large aerospace.

Part area × plies + cutting scrap — set your own.

Consumables (per part)
Breakdown
Fixed vs. consumable line items, this part
LineTypeCost
Machine energy (9 kWh × $0.12/kWh)fixed$1.08
  of which cure energy (≈68%, industry-reported)memo$0.73
Autoclave process energy (250 kWh/cycle × $0.12/kWh)fixed$30.00
Layup labor (1.5 hr × $45.00/hr)fixed$67.50
Raw material (prepreg tape)consumable$360.00
Glovesconsumable$1.20
Cutting bladesconsumable$3.50
Vacuum bagging filmconsumable$4.00
Release filmconsumable$2.75
Maintenance reserveconsumable$6.00
Fixed subtotal$98.58
Consumable subtotal$377.45
Total per part$476.03
Cost per ply$59.50
Cost per kg$1,133.40
Try it — Design
Design

Requirement → material → ply-by-ply layupsimulated

Design-time prediction from classical laminate theory + correlated Monte-Carlo uncertainty. No physical build has happened yet — see the Digital Twin tool for as-made validation.

Third-party product names (e.g. IM7/8552, TC1320 PEKK) are used for representative demonstration purposes only.

Environment

Start from where this part has to live — not a bare minimum-modulus number. Check every environment that applies; each flows onto your printable requirements card below.

Environment-based material screening arrives with environmental property data — recorded on your requirements card today.

Describe it to mAI (Machine AI) simulated

Describe what you need in your own words — e.g. “a 40 cm bracket that can’t deflect more than 2 mm under a 50 kg load at the tip.” mAI reads it back to you; nothing is applied to the form until you say so.

Start from the structure, not the datasheet: modulus alone is a poor selection criterion — stiffness is the structural property, and it depends on thickness. Enter the panel and the requirement; the tool works forward to a layup.

Modulus vs. stiffness: modulus E (Pa) is a material property. Structural stiffness — axial k = EA/L (N/m) or bending stiffness EI — depends on geometry as well as material. This tool derives the structure from material + geometry; it never recommends on modulus alone.

CAD bounding-box intake

Drop a STEP file here
or click to choose · .step / .stp

Processed entirely in your browser — the file is never uploaded or stored. Reads the bounding box only (no holes, thickness maps, or load features yet). IGES is not supported in this demo.

Demonstrates CAD-file recognition only — not an engineering analysis. For demonstration purposes only.

Coming soon

Full feature extraction — holes, thickness maps, load features

Reading the bounding box is a start; mAI reading actual load-bearing features (fastener holes, thickness variation, cutouts) directly from the model geometry is coming soon, alongside the hardware that will validate what it extracts.

Results below the floor are labeled exploratory, never silently dropped or promoted.

Carbon fiber / PEKK (UD tape)

TC1320 PEKK

Matrix: PEKK (thermoplastic) · ply thickness 0.14 mm — representative value (assumed)· density 1590 kg/m³ ±3.9% band at 95% confidence (vendor) (ASTM D792 (typical — not vendor-stated))

Tg 159 °C ±3.9% band at 95% confidence (vendor) (ASTM D7028 (DMA) (typical — not vendor-stated)) · Tm 337 °C ±3.9% band at 95% confidence (vendor) (ASTM D3418 (DSC) (typical — not vendor-stated))

Derived service ceiling 159°C — Tg exactly — semi-crystalline matrix carries load through the amorphous fraction's glass transition (model-derived, unless otherwise stated).

Ply-level property card — provenance per value
PropertyValueBasis
E1 — fiber-direction tensile modulus (GPa)
ASTM D3039 (typical — not vendor-stated)
134.00 GPa ±19.6% band at 95% confidence (assumed)assumed
E2 — transverse modulus (GPa)
ASTM D3039 (typical — not vendor-stated)
9.20 GPa ±19.6% band at 95% confidence (assumed)assumed
G12 — in-plane shear modulus (GPa)
ASTM D3518 (typical — not vendor-stated)
5.00 GPa ±19.6% band at 95% confidence (assumed)assumed
ν12 — major Poisson ratio (—)
ASTM D3039 (typical — not vendor-stated)
0.310 ±19.6% band at 95% confidence (assumed)assumed
Xt — fiber-direction tensile strength (MPa)
ASTM D3039 (typical — not vendor-stated)
2050.00 MPa ±19.6% band at 95% confidence (assumed)assumed
Xc — fiber-direction compressive strength (MPa)
ASTM D6641 (typical — not vendor-stated)
1100.00 MPa ±19.6% band at 95% confidence (assumed)assumed
Yt — transverse tensile strength (MPa)
ASTM D3039 (typical — not vendor-stated)
62.00 MPa ±19.6% band at 95% confidence (assumed)assumed
Yc — transverse compressive strength (MPa)
ASTM D6641 (typical — not vendor-stated)
190.00 MPa ±19.6% band at 95% confidence (assumed)assumed
S — in-plane shear strength (MPa)
ASTM D3518 (typical — not vendor-stated)
130.00 MPa ±19.6% band at 95% confidence (assumed)assumed
Recommended layup
[0/±45/90]ₛ8 plies95.8% confidence of meeting the target

Why this one (others below are feasible too): Lowest areal mass (1.78 kg/m², 8 plies) among the 60 candidate layups meeting the stiffness target at ≥90% confidence; ties broken by least overshoot.

Ex — laminate fiber-direction modulus, GPa (predicted)51.8 GPa · ±19.6% band at 95% confidence
Gxy — laminate shear modulus, GPa19.8 GPa
Laminate thickness, mm1.12 mm
Derived stiffness & weight — k = Ex · h · W / L (computed from laminate modulus + your geometry; stiffness is not a material property)
k = 36.3 kN/mm · ±19.6% band at 95% confidence · areal mass 1.78 kg/m² · 22% lighter than an equal-stiffness aluminum 6061-T6 (representative: E 69 GPa, ρ 2.70 g/cm³) panel (derived comparison)
Explicit ply-angle sequence (bottom → top; the “s” mirror is in reverse order about the midplane)
[0, +45, -45, +90, +90, -45, +45, 0]
Ply-layup diagram — bottom → top
ply 1
ply 2+45°
ply 3−45°
ply 490°
ply 590°
ply 6−45°
ply 7+45°
ply 8
+45°−45°90°

8 plies total. Both outer surfaces are .

What the numbers mean: a confidence % is the modeled probability of meeting the stated targetgiven the material’s published property band — not how “sure” the tool is. A band like “±6.2% band at 95% confidence” means 95% of modeled outcomes fall within ±6.2% of the stated value — it is a spread, not a probability.

One-Page Requirements Card

A one-page summary you can hand off — “okay boss, this is what we need.” Printing this page prints only the card below.

Requirements card
MaterialTC1320 PEKK (Carbon fiber / PEKK (UD tape))
Layup[0/±45/90]ₛ8 plies, 1.12 mm thick
Structural size400 mm × 250 mm panel, axial stiffness target 30.0 kN/mm
Environmentservice temperature range 0 to 60°C requested
Environment-based material screening arrives with environmental property data — recorded on your requirements card today. (The service temperature range is genuinely applied wherever this tool screens candidate materials.)
Confidence95.8% ±19.6% band at 95% confidence
ProcessConsolidation — melt above Tm, controlled cool through recrystallization. Vendor-class profile only; develop your own consolidation cycle for real part geometry.
Cost estimate
Coming soon

Per-part cost, tied to this layup

Requires calibrated machine energy, cycle time, and consumable rates from an actual TapeLayer build of this part — arrives with hardware. Use the standalone Cost tool for a manual, transparent estimate today.

Time-to-make
Coming soon

Estimated build time

Depends on machine layup rate and part geometry, both measured from real builds — arrives with hardware, not modeled here.

By Greenlight mAI. 2026-09-17. SIMULATED — beta software.
Ranked candidates
Top 8 of 82 enumerated symmetric, balanced layups (8–16 plies) for TC1320 PEKK. Stiffness k is derived per candidate (k = Ex · h · W / L); the Ex Monte-Carlo band (400 draws, one systematic factor per property set — never tighter than the input band) propagates through unchanged. Ranked: feasible first, then lowest areal mass.
Layup notationPliesDerived stiffness k (kN/mm)Areal mass (kg/m²)Confidence of meeting targetStatus
[0/±45/90]ₛ recommended836.3 kN/mm · ±19.6% band at 95% confidence1.78 kg/m²95.8%feasible
[0₂/90₂]ₛ850.4 kN/mm · ±19.6% band at 95% confidence1.78 kg/m²100.0%feasible
[0₂/±45]ₛ853.6 kN/mm · ±19.6% band at 95% confidence1.78 kg/m²100.0%feasible
[0₃/90]ₛ872.3 kN/mm · ±19.6% band at 95% confidence1.78 kg/m²100.0%feasible
[0₄]ₛ893.8 kN/mm · ±19.6% band at 95% confidence1.78 kg/m²100.0%feasible
[0/(±45)₂]ₛ1036.1 kN/mm · ±19.6% band at 95% confidence2.23 kg/m²95.5%feasible
[0/±45/90₂]ₛ1039.1 kN/mm · ±19.6% band at 95% confidence2.23 kg/m²99.0%feasible
[0₂/90₃]ₛ1052.0 kN/mm · ±19.6% band at 95% confidence2.23 kg/m²100.0%feasible

What the numbers mean: a confidence % is the modeled probability of meeting the stated targetgiven the material’s published property band — not how “sure” the tool is. A band like “±6.2% band at 95% confidence” means 95% of modeled outcomes fall within ±6.2% of the stated value — it is a spread, not a probability.

Coming soon

As-made / Step-1 data tightening these bands

Every band above comes from published or assumed material property spreads today. Once a TapeLayer machine builds coupons of this exact layup, the Digital Twin tool’s as-made flywheel feeds real Step-1 workmanship + measured properties back into this material’s private save space — the same recommendation, but with the band shrunk toward the machine’s own measured precision instead of a vendor or assumed spread. Coming soon.

INTERACTIVE — try the pipeline
Prefer the real thing?

The walkthrough below runs on a shared demo material. Sign in and Ingest becomes real — drop your own datasheet and mAI (Machine AI) extracts it for you.

Walk the pipeline yourself — Ingest live. Sign in. →

You’ve read the Loop above — now click through it, one stage per screen. Every number below is computed live by the same NCAMP-pinned engine that powers the tools on this page; anything not computed is a labeled input. Use the stage buttons or the ←/→ arrow keys to move through the walkthrough.

Third-party product names (e.g. TC1320 PEKK) are used for representative demonstration purposes only.

01 · Ingestlive — sign in required

A datasheet becomes structured, cited rows

Input — a datasheet arrives
TC1320 PEKK datasheetCarbon fiber / PEKK (UD tape) (illustrative fragment; values from the engine’s material card)
Tg 159 °C · Tm 337 °C · ply 0.14 mm — representative value (assumed) · ρ 1590 kg/m³
E1 134 GPa (typical) · Xt 2050 MPa
mAI (Machine AI) reading the datasheet — extracting values with per-value provenance…
Stage 1 of 6