Digital Twin in Primary: Turning variability into value

From good equipment to truly optimized Primary

Traditional optimization in Primary often looks like this:

• Operators detect issues in Secondary.
• They trace the problem back to Primary.
• They manually adjust set points – often based on experience and intuition – and observe what happens.

This works, but it’s slow, reactive, and heavily dependent on individuals. Meanwhile, key conditions in Primary are constantly changing: ambient humidity, tobacco moisture, raw material mix, and product specifications

What a Digital Twin in Primary actually is

At Körber, the Digital Twin for Primary is a sensor‑based AI solution that collects data from primary machines and their environment, monitors key production parameters continuously, uses machine learning to predict the optimal set points for those machines, and automatically adjusts settings in real time to match prevailing conditions – including humidity and tobacco characteristics. In other words, it’s not just a dashboard. It’s an active optimization layer on top of your existing line.

How it works: From data to decisions to action

The Digital Twin follows a clear, industrial logic:

1. Data collection & modeling
   Historical and live data from the primary process are collected and analyzed. Relationships between variables (e.g. moisture, temperature, throughput, set points) are investigated to understand their impact on yield and quality.
2. AI & ML algorithms
   Machine learning models are trained to predict the optimal combination of set points for the primary machines under different conditions.
3. Real‑time optimization
   - Monitors production parameters continuously.
   - Automatically adjusts to shifts in raw materials, operating conditions, and process changes.
   - Suggests or directly applies optimized set points, stabilizing output moisture, filling power, and overall factory yield.
4. Line‑wide integration
   The solution connects information from different machines and processes and can adapt to whatever equipment is installed, protecting existing investments.

The business impact: Numbers, not just promises

The Digital Twin is explicitly targeted to deliver a +1% yield boost with a path to +2% by optimizing primary processes and stabilizing OEE. In a real customer project, the journey looked like this:

• 2023 – Assessment
  A major local producer with a modern line and well‑trained staff underwent an assessment visit; a detailed report identified where variability and losses originated.
• 2024 – Installation & model training
  Software setup, minor mechanical adjustments, and intensive data collection/model training followed.
• End of 2024 – Stabilization
  Output moisture was stabilized, giving a solid foundation for yield improvements.
• Early 2025 – Yield uplift
  The factory achieved a +1% yield gain, and by go‑live, a +1.5% yield gain was reached.

For a Primary line handling large tobacco volumes, this scale of yield improvement typically translates into significant annual savings.

Hardware + Digital: Pairing Digital Twin with sorters

The story doesn’t end with software. When combined with Körber’s Pure and Pure-F sorters the value multiplies: The sorter family delivers tobacco savings up to 2% and guarantees exceptional product quality. With constant efficiency above 85%, Pure reduces maintenance costs by up to 66% while minimizing tobacco losses. Pure‑F is unique in being able to process entire leaves before threshing, preventing NTRM (non-tobacco related material) from breaking into many small pieces and providing significant cost, efficiency, and quality advantages versus manual processes. The Digital Twin adds a line‑wide, AI‑driven brain to this proven hardware, aligning process behavior with what the sorters and downstream processes need.

Why Digital Twin in Primary – and why now?

The patterns are clear in the data: Batch‑to‑batch variability in Primary is a major factor in secondary defects and underperforming yields. Optimal results are achieved when machine and process set points are dynamically aligned with the product’s condition. But doing that alignment manually, in real time, across all process steps, is beyond what any human team can sustain consistently. Digital Twin technology gives manufacturers a practical way to apply AI directly on the line, a way to turn existing data into continuous optimization, rather than sporadic analyses, and a scalable approach, proven from pilot plants up to large customers

Getting started: Pragmatic, not perfect

A Digital Twin project doesn’t require a “perfect” Industry 4.0 environment from day one. Successful implementations have followed a phased, pragmatic approach:

1. Assessment – Understand your current process, data landscape, and improvement potential.
2. Data & modeling – Set up data collection where needed, build and validate models.
3. Pilot line – Apply the Digital Twin on one primary line, focused on clear KPIs like yield and moisture stability.
4. Scale‑up – Extend to more lines and sites, moving into a subscription and continuous‑improvement phase.

For primary tobacco processing, a Digital Twin is not a futuristic concept. It is a sensor‑based, AI‑driven optimization layer for your existing lines. Field‑proven to stabilize key parameters and deliver +1 - 1.5% yield gains, with a path to +2%. Designed to connect strategy to the shop floor, turning targets for yield and quality into daily, automated decisions.

Landingpage Contact