Paper board online gauge correlation asks whether a continuous supplier or converter reading can support decisions normally based on a laboratory method. The technologies may respond to different material features. Pairing an online number with a specimen taken at another time or position can create an apparent bias that is actually a sampling mismatch.
This article is a correlation-planning guide. It does not calibrate an installed gauge or replace the governing laboratory standard. The material supplier, gauge manufacturer and qualified laboratory must approve the methods and any operational use.
Diagnostic decision table
| Correlation risk | Control to add | Avoid concluding |
|---|---|---|
| Sample and signal are not synchronized | Record reel position, time and sample window | Gauge bias explains every difference |
| Only normal center values are compared | Cover the approved operating range | Agreement near the mean proves full-range fit |
| Methods measure different properties | Define measurands and conditioning | An offset makes technologies equivalent |
| Residuals change by grade | Model or review grades separately | One equation applies to every construction |
Define both measurement systems
Document the online gauge model, signal, averaging interval, calibration status and installation position. Separately define the laboratory property, standard, conditioning and specimen preparation. Use the same unit only after confirming what each system actually measures and reports.
ISO/TS 20460 provides guidance for linking paper and board online gauges to standardized laboratory measurements and notes that the technologies are commonly different. Treat the laboratory reference and online output as two measurement systems requiring demonstrated comparability, not as interchangeable columns because their names are similar.
Synchronize the physical sample and data window
Create a traceable method to link laboratory specimens with reel or sheet position and the corresponding online readings. Account for transport distance, process delay, web direction and any averaging window. Record splices, grade changes and stops that break a simple time relationship.
Define how much online data represents the laboratory sample. Avoid choosing the signal window after seeing which interval produces the best agreement. Preserve raw timestamps and position markers so a mismatch can be investigated. Where exact synchronization is impossible, quantify the limitation rather than hiding it in an adjusted offset.
Cover the intended material and process range
Collect paired observations across the approved range, relevant grades and operating conditions. Include independent production periods rather than many adjacent points from one steady reel. Plan the number and distribution of pairs with the analyst before estimating a relationship.
Do not rely only on correlation coefficient. Two methods can correlate while maintaining an unacceptable bias, and a narrow range can produce misleading conclusions. Review paired differences, residual patterns, uncertainty and grade effects. Keep unusual points with their process evidence unless a documented validity rule supports exclusion.
Separate calibration, validation and monitoring
Use one planned data set to estimate any authorized relationship and independent data to check its performance. An offset or equation chosen and evaluated on the same points can appear more successful than it will be in routine use. Keep software versions, coefficients and effective dates controlled.
Define ongoing checks against the laboratory reference and rules for drift, maintenance or grade changes. A gauge service event can change the relationship even when the process itself remains stable. Do not continually recenter the equation to absorb unexplained differences without investigating both measurement systems.
Limit decisions to demonstrated comparability
State which materials, properties, ranges and decisions the comparison supports. In a dispute, ISO describes the usual reference as laboratory testing unless the parties agree otherwise. Preserve the purchasing and quality agreement rather than allowing an operational dashboard to silently replace the contracted test method.
Send HANNAI the synchronized comparison, material identities and forming consequences when gauge data is used to explain machine behavior. This can reveal whether a board-property shift aligns with cup defects while keeping the inference separate from formal material acceptance and gauge calibration.
Related equipment and next checks
Review the 950 roll die cutting machine and the related guide to paperboard conditioning comparison. The Engineering Notes archive connects these checks with wider machine planning.
Technical reference
- ISO/TS 20460: comparability of paper and board online and laboratory measurements — general reference; confirm applicability to the installed equipment.
Frequently asked questions
Does a high correlation prove the methods agree?
No. Review bias, residuals, uncertainty and range coverage.
Can one offset be used for every board grade?
Only if the validated evidence supports those constructions and ranges.
Should laboratory samples be linked to gauge time and reel position?
Yes. Synchronization is central to a meaningful paired comparison.
Can online data replace the purchase test automatically?
No. Follow the governing standard and buyer-seller agreement.
Review the evidence with HANNAI
Send your machine model, component identification and the observations described in this guide. Include the drawing or material reference and any before-and-after measurements so the required next check can be identified.