Programme site · build 0.4 · published progressively
Production site Contact the lab
ENGASPRINT Research & Development Talk to the lab

Applied research unit · Bielsko-Biała, Poland · EU

Applied research for print, packaging and the process behind them

ENGAS PRINT is the research and development unit of a folding carton manufacturer. We build and test the layer above production: machine vision and AI-assisted inspection, instrumented process control, compliance-by-design data, and sustainability that is measured rather than asserted.

This site is being built in the open. What follows is the working programme — the tracks, the roadmap and the method — not a finished brochure.

Build Site under construction. Programme pages ship progressively through 2026 and 2027.

Read the programme Partner with the lab
05
Research tracks
2026
Programme year
EU
Regulatory scope
0.4
Platform build

01 / Research tracks

Five tracks,
one production floor

Every track is anchored to a real machine, a real order and a real cost line. Research that cannot be run on a live shift does not enter the programme.

  1. 01

    AI & machine vision

    PROTOTYPE

    Inline capture and classification of print defects, colour deviation and label content. The question is not whether a model can classify a defect — it is whether it can do so at line speed, on coated stock, under real lighting, without stopping good work.

    • Inline sheet capture rig
    • Defect family classifier
    • Colour deviation tracking
    • Label and code verification
    • Human-in-the-loop review

  2. 02

    Process engineering

    PILOT

    Instrumenting the press and the finishing line so that setup time, spoilage and drift become data instead of anecdote. Waste that can be attributed to a decision can be designed out; waste attributed to a shift cannot.

    • Makeready modelling
    • Waste attribution
    • Drift and stability analysis
    • Changeover sequencing
    • Shop-floor feedback loops

  3. 03

    Compliance by design

    RESEARCH

    Packaging that carries its own evidence. Specifications, material declarations and traceability records structured, versioned and exportable from the moment an order is quoted — not reconstructed when an auditor asks.

    • Machine-readable specifications
    • Material declarations
    • Chain-of-custody records
    • Digital product passport readiness
    • Audit trail by default

  4. 04

    Sustainability & materials

    RESEARCH

    Fibre-based substitution, coating reduction and recyclability testing, with a cradle-to-gate footprint calculated per SKU from production data rather than from category averages.

    • Substrate substitution trials
    • Coating and varnish reduction
    • Recyclability assessment
    • Cradle-to-gate footprint per SKU
    • Energy and water instrumentation

  5. 05

    Optimisation & planning

    PROTOTYPE

    Imposition, run length and scheduling optimised against the real cost curve of a specific plant, including the costs that planning systems usually ignore: changeover, drying, storage and rework.

    • Imposition and nesting
    • Run-length economics
    • Schedule optimisation
    • Predictive maintenance
    • Demand-driven replenishment

02 / Programme

What we are building,
and when

The roadmap is deliberately short. Each entry has one owner, one measurable outcome and a date at which it is either working or abandoned.

Period Workstream Outcome Status
2026 · Q1 Data foundation Machine, job and material data unified into a single time-series store. CLOSED
2026 · Q2 Vision rig v1 Camera and lighting rig for inline sheet capture; first labelled defect dataset. CLOSED
2026 · Q3 Defect model v1 Supervised classifier for recurring defect families, validated offline against held-out runs. ACTIVE
2026 · Q4 Compliance schema Machine-readable specification and material declaration format, versioned per order. ACTIVE
2027 · Q1 Footprint engine Cradle-to-gate calculation per SKU, driven by live production data. QUEUED
2027 · Q2 Planning assistant Run-length and imposition optimisation against measured cost curves. SCOPING

Dates describe intent, not commitment. Entries move, and entries are dropped — that is what a roadmap is for.

03 / Method

How work gets
from idea to shift

A small unit inside a manufacturing business cannot afford open-ended research. The method is built to kill weak ideas early and to make strong ones survive contact with a production schedule.

  1. 01

    Frame

    Every track begins as a written problem statement with a measurable target and a defined failure condition. No target, no project.

  2. 02

    Prototype

    The smallest viable rig or model, built on real production data rather than a public dataset. Weeks, not quarters.

  3. 03

    Validate

    Blind runs against held-out data and live shifts. Results are recorded internally in full, including the ones that disappoint.

  4. 04

    Industrialise

    Documentation, a named owner, monitoring and a rollback path — established before anything touches a live order.

04 / Standards

The frameworks we
design against

Reference frameworks that shape how the lab specifies, documents and validates its work. Some are held as certifications by the manufacturing site; others are design references for work that is still in development. Current certification scope is stated in commercial documentation, not on this page.

Quality & product safety

  • ISO 9001
  • BRCGS Packaging Materials
  • ISO 12647 process control
  • SMETA social audit

Environment & materials

  • ISO 14001
  • FSC® chain of custody
  • ISO 14040 / 14044 life cycle assessment
  • 4evergreen recyclability guidance

European regulation

  • Packaging and Packaging Waste Regulation
  • EU Deforestation Regulation
  • CSRD / ESRS reporting
  • Digital Product Passport

Data, AI & security

  • ISO/IEC 42001 AI management
  • EU AI Act
  • ISO/IEC 27001
  • GDPR
  • GS1 identification standards

05 / Instrumentation

What the lab
measures

A track earns its place when it moves one of these. Each signal is instrumented at source on the production floor; none of them is a self-reported estimate.

  • Waste per 1 000 sheets

    sheets

    INSTRUMENTED

  • Makeready time per job

    minutes

    INSTRUMENTED

  • First-pass yield

    percent

    INSTRUMENTED

  • Colour drift in run

    ΔE00

    IN CALIBRATION

  • Energy per 1 000 m²

    kWh

    INSTRUMENTED

  • Certified fibre share

    percent

    INSTRUMENTED

  • Unplanned downtime

    hours

    IN CALIBRATION

  • Specification changes per order

    count

    IN DEVELOPMENT

06 / Lab notes

Notes from
the floor

Short entries from the working log. Published when they are useful to someone outside the building.

  1. Lighting beat resolution

    On coated stock, moving from diffuse to low-angle raking light did more for scratch detection than doubling sensor resolution would have. The inexpensive change won, which is not the outcome the budget request had assumed.

  2. Waste needs an owner

    Attributing spoilage to a specific setup decision rather than to a shift changed the conversation on the floor within two weeks. The measurement did not improve; the attribution did.

  3. Compliance is a data problem

    Most of what the Packaging and Packaging Waste Regulation and the digital product passport will ask for already exists in our systems. It is simply not structured, versioned or exportable — and that, not the regulation, is the actual project.

07 / Contact

Work with
the lab

We are open to pilot partners, material and equipment suppliers, and research institutions across the European Union. Commercial print enquiries are handled by the manufacturing site.

Write to the lab