AI Soft Sensors: Predicting Wet-End Paper Web Breaks

This article provides an overview of how high-speed paper manufacturing machines utilize artificial intelligence (AI) soft sensors to prevent costly production downtime. By continuously analyzing complex chemical and mechanical data streams in real time, these virtual sensors detect the microscopic precursors to paper tearing. Readers will learn about the vulnerability of the wet-end stage, how AI processes physical sensor data to forecast web breaks before they happen, and the automated corrective actions taken to maintain continuous paper production.

In high-speed paper manufacturing, a "web break" occurs when the continuous sheet of paper tears while moving through the machine. The most vulnerable area for a break is the "wet end," where the paper begins as a slurry that is roughly 99% water and 1% cellulose fiber. As water is rapidly drained and pressed out, the fragile sheet lacks structural integrity. A break at this stage causes significant operational downtime, wasted raw materials, and the need for manual machine washdowns.

Physical sensors cannot be placed directly onto the fragile, fast-moving wet web without causing the very tears manufacturers are trying to prevent. Furthermore, traditional laboratory testing of pulp chemistry takes too long to stop an imminent break. To solve this, mills implement AI-driven soft sensors. A soft sensor (or virtual sensor) is an advanced software algorithm that fuses data from hundreds of existing non-invasive physical sensors to estimate unmeasurable variables, such as the real-time tensile strength of the wet paper web.

AI soft sensors rely on deep learning and machine learning models—such as Long Short-Term Memory (LSTM) networks and random forests—trained on years of historical mill data. The AI continuously monitors thousands of data points per second across the wet end, including:

• Pulp Chemistry: White water pH, conductivity, cationic demand, and fiber consistency.
• Mechanical Pressures: Headbox jet-to-wire ratios, vacuum box suction levels, and press roll loads.
• Machine Speeds: Micro-fluctuations in the speed of the forming fabric and rollers, which dictate sheet tension.
• Temperatures: Stock and shower water temperatures that affect drainage rates.

Tearing rarely happens because of a single catastrophic failure; it is usually the result of a complex chain reaction. For example, a slight drop in pH might subtly alter fiber bonding, which marginally slows water drainage, causing the wet sheet to stick slightly more to a press roll. While a human operator or traditional alarm system would see all these individual metrics as within "normal" operating limits, the AI soft sensor recognizes the multi-variable pattern as a definitive precursor to a web break.

By processing time-series data, the AI tracks how changes in the pulp vat will affect the paper sheet several seconds or minutes later as it travels down the line. When the soft sensor calculates a high probability of a tear, it triggers immediate action before the physical defect manifests.

In an automated closed-loop system, the AI soft sensor communicates directly with the mill’s Distributed Control System (DCS). To prevent the predicted break, the system automatically executes micro-adjustments. It may instantly inject retention-aid chemicals to improve fiber bonding, adjust vacuum pressure to optimize water removal, or slightly alter the speed of specific rollers to relieve tension on the weakened wet web. By resolving the root cause invisibly and instantaneously, the machine maintains its high speed, and the tear never occurs.