PROCESS FAMILY One digital thread

POC HYPOTHESIS

The part may be distorted in free state. The system must predict the geometry after final assembly constraint—not merely copy a nominal path onto the scan.

COMMON FOUNDATION

Scan and qualify the free-state part

Establish reliable datums, a full-field as-built surface and a confidence map before any robot path is released.

G3 DIGITAL THREAD HOLD

PART C-017 · CURVED PANEL

VIEW SCANNED FREE STATE

Nominal Actual Approved path

RECIPE & MODEL

Scan controls

Representative sample Part sample

POC model

Conceptual prediction only. Calibrate with physical samples, final-assembly constraints and blind validation before releasing a real robot path.

Free-state peak warp 5.0 mm Part-specific scan

Predicted fit error 0.62 mm Includes model + process

Measurement confidence 88 % Occlusion-limited

Qualified coverage 94 % Rescan 2 regions

CALIBRATED PROCESS MODEL

Process response laboratory

Tune the active process on coupons, then test the model on representative geometry and held-out parts.

SURFACE MAP

Confidence map

LOWHIGH

RELEASE LOGIC

Automatic refusal rules

PRODUCTION ARCHITECTURE STUDY

One software platform. The right physical separation.

Compare a single enclosure, modular dry/clean zones and fully separate specialist cells using the POC evidence.

POC BASELINEMODULAR COMMON PLATFORM

Floor-space priority 70% Parallel throughput priority 65% Contamination control priority 90%

DRY / EXTRACTED ZONE Scan · Trim · Machine · Sand

Composite-dust extraction, tool exchange and protected metrology.

↔Datum-controlled transferClean + verify

CLEAN COATING ZONE Spray · Flash · Cure · Verify

2K preparation, booth controls and contamination release.

ASSEMBLY / FIT ZONE Constrain · Gap scan · Shim

Representative final boundary and installation-force evidence.

Decision rule

“One cell” may mean one coordinated digital line—not every process inside the same physical enclosure. The POC measures cleaning, changeover, contamination, utilisation and recovery time before the production layout is frozen.

BLIND-VALIDATED FEASIBILITY

Prove predictability—not one successful demonstration.

Characterise multiple demoulded parts, build the free-to-constrained model, then verify on held-out parts.

STUDY READINESS INDICATIVE

SAMPLE POPULATION

Validation design

Unique parts 15 Production batches 3 Blind parts held out 5 Reloads per study part 7

G0–G6

Evidence gates

CUSTOMER INPUTS

Required before release

Nominal part, mould and assembly CAD with datums Final constrained geometry, fastening sequence and fit limits Multiple representative parts across batches / demoulding periods Edge, hole, pocket, undercut and no-process definitions Current process recipes, tools and fixture / template method Coating TDS/SDS, A:B ratio, pot life, viscosity and cure window Approved trial facility, extraction, booth and safety owner

G0–G1Freeze inputsParts · datums · truth · safety

G2–G3Build evidence modelMetrology · free → constrained

G4–G5Coupon + process trialsDry run · recipes · refusal

G6Blind demonstrationFit · limits · architecture