In specialty paper production (anti-counterfeit watermark paper, security paper), the paper machine is more than a large piece of machinery. It is a complex physical system in which multi-point drive speed differences, the physics of vacuum dewatering, the thermal regime of the drying cylinders, and online quality indicators are tightly coupled.
For a specialty paper mill, data has never been scarce. The real challenge is that while the stock preparation, approach flow, forming, press, drying, and winding sections all stream out large volumes of time-series data in real time, the floor still struggles to quickly answer a few critical questions. Where do the faint first signs of a vacuum leak appear? How does the periodic drift of multi-stage drive roll speed differences propagate downstream? Is the current problem a localized web flutter, or a major production crisis that threatens sheet moisture, defect rate, and the risk of a web break and emergency stop?
This is where paper-machine monitoring needs to become diagnostic. What a mill needs is not more isolated tag curves stacked on a monitoring screen, but data that reconstructs the complete physical process: vacuum degradation, insufficient press dewatering, compensating dryer load, and sheet moisture over spec; drive speed-difference drift, a sudden surge in web tension, and a web break and emergency stop. This is what moves the control room from passive firefighting toward proactive warning and diagnosis.
Specialty paper lines: operations pain points under multi-drive, tightly coupled physics
Specialty paper production is a typical high-precision, high-speed continuous line. It mainly comprises the stock preparation area (refiners, consistency chests) and the papermaking area (headbox, wire drive roll, cylinder mold, suction couch roll, vacuum press roll, front and rear dryer groups, and the reel). Sheet formation and transport depend on draw and speed coordination across seven drive rolls on the line, and the core control mechanism rests on a tightly coupled triangle of speed, speed difference, and load.
Under normal continuous papermaking conditions, the machine runs the wire drive roll as the reference line speed (400 m/min). The downstream drive rolls must hold a precise small positive speed difference (0.5% to 2.5%) so the wet web keeps proper tension between sections. If the speed difference runs too high, the web can break quickly; if it runs too low, the wet web can slacken, pile up, and wrinkle. At the same time, vacuum in the press section (vacuum pump, suction couch roll, press roll) sets the dewatering rate of the sheet before drying. If vacuum dewatering falls short, the sheet carries excess water into the dryers, forcing the dryers to raise steam pressure and motor load to compensate.
In a system where multi-point drives and the vacuum-thermal regime interlock, single-point equipment monitoring cannot untangle the cause of an anomaly. When the online scanner (QCS) catches reel moisture over the limit or falling tensile strength, operators often spend hours going through DCS reports before they can tell whether the vacuum pump air-water separator is leaking or a dryer-section drive encoder has feedback latency.
Plenty of alerts, slow diagnosis: the pain points of running a papermaking floor
The same swings can be monitored, but the cause is still hard to pin down
At different machine speeds and basis weights, thermal and mechanical parameters move across a dynamic range. Take the dryer group’s rising motor load rate. It could come from added resistance as speed rises, or from falling vacuum in the press section upstream, where the wet web carries too much water and forces the dryers to run hot and under heavy resistance. If only dryer load is monitored, operators cannot see an early vacuum pump leak and miss the maintenance window before wet-web defects worsen.
A single-point alert has triggered, but its downstream impact is hard to judge
Fault propagation in specialty papermaking is cascading. Aging vacuum pump seals may only show up as a drop in negative pressure, but the effect amplifies layer by layer along the chain of vacuum pump → suction couch roll/vacuum press roll → front/rear dryer group → QCS scanner. In the same way, a small drift in the front dryer drive controller carries speed-difference deviation along front dryer → rear dryer → reel, so web tension swings violently at the end and eventually trips a full-line emergency stop. Conventional single-point alerts treat these stages in isolation, so operators cannot anticipate a major event such as a web break.
The business impact is not abstract: sheet downgrades and web-break stops are costly
Slow response shows up directly in cost per ton and in yield.
- In the vacuum leak anomaly (reels R-205 through R-208), vacuum pump vacuum fell from 68 kPa to 43 kPa, pushing sheet moisture from 5.2% up to 8.2%, well over the limit. Four consecutive specialty paper reels were downgraded for exceeding moisture and defect limits, while the front dryer group compensated by running steam pressure beyond spec, causing serious per-ton energy loss and quality downgrade damage.
- In the drive-anomaly web break (reels R-215 through R-218), drive speed-difference swings pushed web tension past the 1850 N physical limit very quickly, the wet web tore, the wire drive roll line speed dropped to zero, and the whole line stopped in an emergency. Every break not only rejects the current reel; the re-threading and speed-up after a stop also wastes large amounts of stock, power, and steam.
From passive monitoring to proactive judgment: reorganizing the data chain
Object-oriented modeling: scattered tags return to one operational view
To give time-series data process context, sensor signals have to be turned into physical objects. TDengine uses Industrial Ontology modeling to organize 19 devices, from refiners, headbox, drive rolls, and vacuum pumps to QCS scanners and plant meters, into a tree-like asset system that runs through the whole upstream-to-downstream line.
Figure 1: A tree hierarchy organizes plant assets and measurement points into a unified operational view
Under this structure, every real-time tag automatically carries its section and process attributes. Control room staff no longer need to memorize complex tag codes. By following the process flow, they can pull full-line curves for the same reel batch, laying the groundwork for correlating multi-source data.
Real-time analysis and event linkage: alerts return to full process context
On top of the asset model, the TDengine real-time analysis engine runs online stream computation over composite metrics such as speed ratio, draw/speed-difference chain, vacuum differential rate, and energy per ton. When a metric leaves its normal process window, the system captures the anomaly, pulls data from the related upstream and downstream tags, builds an event context, and pushes it to the screen for the on-duty operators.
The system’s smart-query and proactive-suggestion services let an operator simply type “Why is the current reel’s moisture over spec?”, and the system automatically analyzes how far front dryer steam pressure and vacuum pump vacuum deviate from normal, points out the compensation chain triggered by vacuum pump seal leakage, and cuts troubleshooting time dramatically.
Figure 2: AI-recommended real-time analysis
Figure 3: General information settings for real-time analysis
Figure 4: Trigger conditions for real-time analysis
Process analysis and AI insight: judgment shifts from experience-driven to evidence-driven
Once multi-source time-series data is objectified and anomalies are correlated, floor-level troubleshooting moves beyond traditional rule-of-thumb experience to data analysis grounded in evidence.
For complex web-break or downgrade anomalies, the system supports cross-shift comparison in the process workbench. The AI automatically pulls data from the abnormal reel, overlays it against a benchmark reel (such as R-190), quantifies the accumulated deviation in the draw/speed-difference chain, traces where the vacuum degradation started, and generates a fault report with the causal logic, so process improvements have a documented basis.
Figure 5: AI interpretation and data mining on the analysis panel
Figure 6: AI event root-cause analysis
Closed-loop analysis in typical anomaly scenarios
Anomaly case 1: vacuum system leak pushes sheet moisture over spec and forces dryer energy compensation
During production of reels R-205 through R-208, the vacuum pump seals aged further. In the R-205 stage, vacuum pump vacuum fell from 68 kPa to 60 kPa (below the 45 kPa alert line), suction couch roll vacuum dropped to 48 kPa, and sheet moisture began to rise. By the R-207 stage, vacuum pump vacuum had fallen to 45 kPa, suction couch roll vacuum had plunged to 32 kPa (triggering the 35 kPa critical alert), and vacuum press roll vacuum had fallen to 38 kPa. To dry the sheet, the front dryer group’s steam pressure was raised from 300 kPa to 350 kPa (near the 370 kPa limit) and the load rate climbed to 94%. In the end, the QCS measured sheet moisture at 8.2% and defect rate up from 0.05% to 0.22%, causing heavy downgrade losses on off-spec reels.
After the QCS flagged over-spec moisture, the process staff in the control room used the process analysis workbench to trace the fault as follows:
Step 1: Confirm the quality degradation and the energy compensation. Look at the trend curves for QCS moisture and front dryer steam pressure. The moisture curve starts climbing clearly at R-206 and reaches 8.2%, while front dryer steam pressure is pushed to 350 to 370 kPa over the same period. This shows the drying system has made maximum thermal compensation and still cannot hold sheet moisture down.
Step 2: Locate the dewatering failure upstream. Overlay the vacuum curves of the vacuum press roll in the press section and the suction couch roll in the forming section. Both rolls’ vacuum drops sharply and in sync during R-206 (the couch roll falls to 32 kPa), confirming that the wet web carried far more water than the process standard before entering the dryers, and that the root of the moisture over-limit sits in the press dewatering stage.
Step 3: Trace back to the root cause. Pull up the vacuum pump vacuum and motor load rate data. The vacuum pump vacuum was already below its normal baseline as early as the start of R-205 (down from 68 to 43 kPa), and the motor load rate also fell, exposing the root cause: the vacuum pump’s own seals were damaged and its suction efficiency had degraded.
Figure 7: Judgment path for vacuum pump leak and couch roll negative-pressure drop causing sheet moisture exceedance
Through these three steps, the process team pinned down the failure logic: aging vacuum pump seals leak → negative pressure cascades down across the vacuum system (couch roll and press roll) → the wet web is under-dewatered in the press → the front dryers compensate beyond steam pressure spec → drying capacity hits its ceiling and the sheet is downgraded for over-limit moisture and defect rate.
Anomaly case 2: front dryer drive speed-difference anomaly causes tension loss and a web break and emergency stop
During production of reels R-215 through R-218, the front dryer drive controller output drifted. In the R-215 stage, the front dryer group’s speed difference rose from 2.0% to 2.5%. In R-216, it drifted further to 3.5% (triggering a critical alert), the rear dryer speed difference was forced up to 2.5%, the reel speed difference was pulled from -1.0% down to -3.5%, and reel web tension rose from 1000 N to 1400 N. By R-217, the front dryer speed difference swung violently between 0.5% and 4.5%, web tension shot to 1850 N and broke past the physical limit, and the web tore in an instant. The wire drive roll line speed plunged to 0 m/min within a minute, the line stopped in an emergency for protection, and it recovered only gradually after cleanup and re-threading.
For this serious web-break shutdown, the operations team retraced the fault through the process workbench:
Step 1: Pin down the web-break event. Look at the line speed and load rate curves of the wire drive roll and the reel. The wire drive roll line speed falls off a cliff from 400 m/min to zero and load across the line drops in sync, showing the system stopped in an emergency under mechanical or electrical protection during R-217.
Step 2: Correlate end tension with speed-difference evolution. Overlay the reel web tension and reel speed difference curves. In the run-up to the break, web tension oscillated violently and climbed from 1000 N to 1850 N while the reel speed difference swung far into the negative at the same time. This proves the end of the web was under abnormal tensile force, and the tension loss was the direct physical trigger of the break.
Step 3: Trace the drive speed-difference drift chain. Expand the speed-difference comparison across all seven drive rolls on the line. The analysis shows that the rear dryer and reel speed-difference swings lag and follow the front dryer group, whose speed difference had already deviated from the 2.0% golden baseline back in R-215, exposing the fault origin in its drive controller and encoder.
Figure 8: Judgment path for dryer speed-difference periodic drift causing web tension overload and paper break
Through this analysis chain, the site confirmed the root cause: front dryer drive controller output drift or encoder latency → front dryer group speed difference climbs abnormally → downstream dryers and the reel are forced to follow and adjust → reel web tension swings and breaks past the physical limit → the web tears → the wire drive roll trips the web-break protection emergency stop.
Data-driven decisions for specialty paper lines
In a high-speed, high-tension specialty paper process, the core value of an intelligent system is to turn complex drive and vacuum time-series data into an intuitive physical diagnosis chain as soon as an anomaly first appears, so operators can make the right process intervention before a web break or downgrade happens.
By organizing scattered equipment tags into a unified process ontology and linking local swings into complete physical process analysis, TDengine helps lower the risk of unplanned paper machine downtime and gives papermakers a data foundation for improving quality and efficiency.
TDengine bundles a high-performance, distributed time-series database, Industrial Ontology modeling, and an Industrial Agent Runtime, providing a full-stack solution for industrial data streams from collection, storage, and real-time analysis to visualization, event management, and root-cause analysis. To learn more about TDengine, visit www.tdengine.com and try it for free.
Try it yourself
Install and deploy TDengine
Visit the TDengine Download Center, select TDengine All-in-One, choose the deployment platform and architecture that matches your environment, and follow the guided steps to complete the installation.
Load the sample data
On first activation, choose Paper Process Quality Control on the sample data loading screen and wait for it to finish loading.
If you have already activated the product, click your avatar in the top-right corner, select Management Console, choose Sample Data on the left, then select Paper Process Quality Control. Wait a few minutes for the data to load.
Paper Process Quality Control
