Real-Time Fat & Sugar QC on a Hazelnut-Paste Line — Gürsoy, Turkey

A Century-Old Hazelnut Business, Checked in Real Time

Gürsoy produces hazelnut pastes, spreads, and crunchy preparations on Turkey’s Black Sea coast — in the hazelnut heartland, where raw material quality is a given and proven consistency is the product. Across three product lines, spec targets for fat and sugar sit between the company and its customers, and every batch has to land inside them.

Challenge | The Lab Was the Bottleneck — and the Pacer

As with most fat-based matrices, reference analysis is slow by design: solvent extraction, long runtimes, small daily throughput. In production that creates a familiar loop:

• Batches queued behind pending analyses
• Process adjustments made on yesterday’s numbers
• Safety margins carried in-process to cover the delay
• The lab busy with routine release work it was never designed for

Solution | M2000 at the Line, One Model per Product

Gürsoy runs the portable M2000 directly in production, with a dedicated chemometric model per product — covering both fat and sugar where required. Measurements are taken on the product itself, non-destructively, and logged on the Altum cloud platform for traceability.

In practice the workflow inverts:

• Prediction in production — fat and sugar checked at the point of need, in seconds, per batch
• The lab as reference — standard methods (e.g. Soxhlet fat determination) used for periodic verification, not for every release decision
• Model care as routine — periodic verification against reference values keeps each model honest across the full working range

The Run So Far (draft figures — pending customer sign-off to publish)

Since early 2025 the line has run 17 months of continuous production use — roughly 2,800 real-time fat and sugar predictions across three product specifications, on the order of a hundred and a half per month — with the wet lab retained for scheduled reference verification rather than release.

(The precise figures above come from internal records and will be confirmed in writing with the customer before publication.)

Results

• Batch release decided on the line, not in the queue
• Process adjustment on current data, reducing margin-for-lag
• One instrument covering multiple parameters across multiple products
• A documented, traceable measurement record for internal QA and customer documentation
• The laboratory freed for the work that only a laboratory can do

Looking for real-time composition checks in your own line? Talk to us, or read other customer stories including Det Norske Meieri and OFI.