In an exclusive interview with Paper Mart, Mr. SVR Krishnan, CEO, Sripathi Paper and Boards Private Limited, shares that the next phase of operational excellence in the pulp and paper industry will be driven by intelligent automation rather than conventional process control. He explains how intelligent automation, Artificial Intelligence (AI), Machine Learning (ML), predictive analytics, and Advanced Process Control (APC) are helping mills to reduce process variability, enhance product quality, and optimise the use of water, energy, and chemicals. He stresses that investing in robust digital infrastructure and intelligent control technologies will enable mills to build more efficient, resilient, and sustainable manufacturing operations, as mentioned while discussing the challenges of legacy systems, data silos, and the skills gap.
Paper Mart: Where does process variability remain most persistent in paper manufacturing today, and how does it affect quality and operating performance? Please tell us the process stages or parameters where tighter stability can deliver the greatest measurable gains.
SVR Krishnan: Most of the variability for any paper mill starts with furnish/raw material control, followed by wet-end controls, short circulation, the steam and condensate system, and tension & draw control up to the pope reel of the paper machine.
In the current era of cost savings, addressing the issue of process variability has become most important. There are two types of process variations.
Common-Cause Variation/Chance Variation: This involves reasons such as ambient temperature, equipment wear and tear, and fluctuations in the quality of raw material. A process with only common cause variation is acknowledged as statistically stable.
Special-Cause/ Assignable Cause Variation: This occurs due to identifiable causes such as a wrong mix of raw material, operator error, incorrect machine setup, sensor failure, wrong application of chemicals, or an unplanned change in operating conditions. All these causes make the process unstable and unpredictable.
Variability in the process significantly affects quality and operating performance as it will drift the output specifications out of limits, and create inconsistency from batch to batch or unit to unit, causing non-conforming production. Also, variability reduces process capability, meaning the process may not be able to consistently meet specification windows for the final product.
Further, the key quality parameters used for tracking variability of paper/paperboard quality include basis weight, moisture, caliper, ply bond, bulk, and stiffness. Among them, weight and moisture control are critical stages where tighter stability can deliver the greatest measurable gains.
Transversal variability amounts to a significant portion of overall process variability. Quality Control System (QCS) scanners provide 2-sigma values for dry weight and moisture; however, effective control involves more than simply adjusting cross-sectional spindles. It also requires interactions between the type of fibre being used, jet/wire ratios, and various head box adjustments.
Longitudinal variability in weight and moisture is always present during production and varies longitudinally. This is mostly controlled by the paper machine’s control loops. However, the root causes of the problems come from stock preparation.
By ensuring effective wet-end controls, including chest levels, flow rates, consistency at every stage, and the automatic dosing of specialty chemicals, we can significantly improve process stability and deliver measurable gains in product quality and operating performance.
[bsa_pro_ad_space id=16]
Watch: Top Paper Companies 2023
PM: How has the approach to controlling process variability changed in recent years? In other words, how are mills moving beyond conventional automation and feedback control to achieve greater process stability? Please explain with examples.
SK: A big shift has taken place from process experts manually adjusting the machine by analyzing laboratory reports to online Quality Control System (QCS) measurements, which automatically control process parameters. Such control uses feedback-driven predictive analytics to generate virtual measurements of key process parameters and quality properties that are impossible to measure manually in real time. Based on these measurements, the platform executes timely adjustments to key control elements, thereby optimising production and improving efficiency.
Proactive and accurate process control is critical for optimising plant performance and reducing quality variation. AI-driven technology provides real-time process improvement recommendations and optimal control parameters, thereby enhancing production line efficiency, revenue, and throughput.
Also, in recent years, mills are moving to auto sampling and measurement systems, and Real-time process monitoring & Quality Management (RTQM) systems, from the conventional laboratory sampling and periodic measurements.
For example, mills are moving from traditional feedback systems to feedforward systems & controls, where earlier-stage measurement systems and the values give predictive controls to the successive stage for correcting the process. This predictive approach helps to reduce off-quality production during the manufacturing process.
PM: How are Advanced Process Control and Model Predictive Control being applied to reduce process variability? Where have they delivered measurable gains?
SK: AI-driven technology leverages real-time and historical data to accurately control chemical dosages, minimizing overfeeding while maintaining quality standards that ultimately reduce energy and raw material usage. Predictive analytics with feedback systems help to anticipate quality issues before they occur to ensure consistent and reliable output that meets production standards.
Advanced AI and machine learning models learn from the mill’s historical data, prescribing real-time adjustments and executing corrective action to meet unique production specification needs. The measurable gains include improved machine runnability; lower steam, water, and energy consumption; savings in process/specialty chemicals; and a reduction in off-quality generation/the Cost of Poor Quality (COPQ).
PM: What practical role are AI and machine learning playing in predicting and correcting process deviations before they affect production or quality?
SK: AI and Machine Learning (ML) act as real-time predictive controllers in the paper industry, identifying anomalies and adjusting process variables minutes before physical defects appear. By processing high-frequency sensor data, these technologies transition mills from reactive troubleshooting to autonomous, proactive optimisation.
ML models analyze vibration, temperature, and acoustic data from jumbo rollers and bearings. The systems detect micro-deviations weeks before a catastrophic mechanical failure occurs, enabling preventive maintenance during planned shutdowns rather than unplanned emergency stops.
Process Optimisation and Control
Advanced Process Control (APC): Reinforcement learning models continuously adjust chemical dosages, steam pressure, and valve positions based on historical data from the best-performing production runs (golden batches).
Web break prediction: Deep learning models analyze historical data to predict process deviations such as vacuum fluctuations, consistency variations, flow changes, and chemical dosage variations to warn operators of web breaks on the paper machine 15 to 30 minutes in advance.
Speed and tension balancing: AI dynamically adjusts the web tension, torque, machine draws between the sections, and machine speed to stabilize the wet-end and drying sections, up to the pope reel.
Real-Time Quality and Strength Prediction
Virtual sensors (soft sensors): AI calculates critical product metrics like tensile strength, tear strength, and basis weight in real time.
Eliminating lab delays: Algorithms replace traditional destructive laboratory testing, which normally takes hours. AI also analyses fibre length, kinking index, curl index, and predicts fibre behavior with respect to twist angle and Tensile Stiffness Orientation (TSO) profile.
Instant recipe adjustment: The system instantly detects fibre attributes deviating from target specifications and adjusts the process accordingly.
Resource and Energy Efficiency
Chemical dose optimisation: AI calculates the precise amount of bleaching agents and retention aids needed based on incoming wood or secondary fibre pulp quality.
Steam and energy reduction: Neural networks, with monitoring at various sections from the press to the pope reel, optimise the thermal energy profile in drying cylinders to optimise fuel consumption.
Water recycling loops: Predictive models manage wastewater treatment chemistry through the automatic dosing of the required chemicals to maximize process water reuse.
PM: How are online measurement, smart instrumentation, machine vision, and QCS improving real-time detection and control of process variation? Where do significant measurement or data gaps still remain?
SK: As explained earlier, these control systems bring stability to the wet end of the paper/board machine and enable precise control of the steam and condensate system, resulting in smooth machine runnability with optimum quality at pope reel.
However, gaps still exist between AI-generated data and the departmental data stored in various silos, resulting in a lack of seamless process control. This practical gap between AI and human intelligence (HI) is due to the poor understanding of AI systems among the operators, as well as inadequate validation of AI models through human expertise.
PM: Drives and motion control directly influence speed, tension, synchronization, and machine stability. How are newer drive and control technologies helping mills reduce variability across the paper machine and associated processes?
SK: New drive systems and control technologies, including advanced frequency drives, sensorless vector drives, and direct-drive motors, eliminate mechanical backlash and slippage. By using high-speed, real-time communication networks to synchronize dozens of machine sections, these systems achieve precise speed matching and tension control, significantly reducing web breaks and product variability across the paper machine.
Direct-Drive Technology:
Mechanical drive components such as gearboxes, belts, and couplings are prone to backlash and wear. Many modern rebuilds and new paper machines use Permanent Magnet Synchronous Motors (PMSMs) directly coupled to the rolls. Direct-drive motors completely eliminate motor slip, providing precise rotational accuracy and higher low-speed torque.
Digital Twins & Simulation:
Mills can digitally map out mechanical dynamics and test drive tuning in virtual environments to prevent instability before physical implementation. This will reduce the real-time commissioning time and accelerate the stabilization of new projects.
In recent years, mills are moving to auto sampling and measurement systems, and Real-time process monitoring & Quality Management (RTQM) systems, from the conventional laboratory sampling and periodic measurements.
PM: How are automation, advanced control, and integrated process data helping mills achieve more consistent paper quality? Which quality parameters have benefited most?
SK: Real-Time Feedback Loops: Advanced sensors and scanners, such as QCS, continuously monitor sheet properties. If a deviation is detected, automated systems instantly adjust factors such as thick stock and thin stock flows or steam pressure to correct the issue before off-spec paper is produced.
Soft Sensors and Virtual Measurements: As some critical parameters, such as Kappa number in pulp digestors, are notoriously difficult to measure physically, Advanced Process Control (APC) modules use soft sensors to estimate these values and make immediate and even-cooking corrections.
Integrated Process Data: By connecting process control with Manufacturing Execution Systems (MES), mills centralize end-to-end data. This enables operators to implement data-driven, closed-loop strategies that maintain ideal operating conditions with limited fluctuations in raw material quality.
Benefits to Quality
Real-time control systems have significantly reduced the standard deviation of key sheet properties, allowing mills to operate closer to their optimal process limits.
Caliper: Automated controls reduce variations in paper thickness, resulting in a more uniform surface finish, improved runnability on the carton converting lines, and higher productivity.
Basis Weight & Moisture: Controlling the flow of fibre stock and managing steam pressure ensures consistent paper weight and moisture content. Once grammage and moisture are well controlled, most of the customer-related quality problems are addressed.
Colour & Shade: Colour accuracy benefits directly from APC
Sheet Ash & Opacity: The standard deviation for sheet ash is routinely reduced by over 50%, allowing mills to maintain higher average ash levels while minimizing raw material costs.
PM: As control systems become more intelligent, how is the operator’s role in process control and decision-making changing?
SK: With the help of AI and ML, the operator’s role is evolving from reactive to predictive decision-making. Operators are consistently guided by the intelligent control systems for their decision-making to allow deviations within acceptable limits or to indicate when action needs to be taken.
Without automation, the same decision will take more time or could even be missed. Operators are no longer just fixing paper breaks or off-spec issues as they happen. Instead, advanced control systems and AI algorithms predict web defects or equipment failures hours before they occur, allowing operators to take proactive steps to avoid the breaks, avoid equipment damage, and reduce the Cost of Poor Quality / COPQ.
PM: What typically prevents a mill from achieving the full benefits of advanced automation and intelligent control? Please tell us more about issues such as legacy systems, data quality, system integration, instrumentation, workforce capabilities, etc.
SK: Legacy infrastructure, poor data quality, and fragmented system integration are the primary barriers that prevent paper mills from achieving the full benefits of advanced automation.
Legacy Infrastructure: Many mills operate with very old assets. These machines rely on proprietary Distributed Control Systems (DCS). Older controllers lack the computational bandwidth required to process the advanced algorithms needed for predictive AI model control. Replacing a functional DCS costs heavily, and requires extended machine stoppage, along with substantial investment, in the current business scenario.
Data Quality and Silos: Mills generate massive volumes of data. However, much of it is trapped in isolated historians without proper context or time-synchronization. Data points often lack metadata. Without tags explaining the paper grade or wood species being processed, the data is meaningless for machine learning. Mills often mix technologies from different eras and vendors, making system integration challenging. Bridging proprietary communication protocols creates significant integration issues. Operational Technology (OT) on the mill floor is traditionally isolated from Information Technology (IT) networks.
Workforce and Cultural Barriers: Existing mill operators excel at manual, reactive tuning. However, they often lack the data literacy required to manage automated control systems. Operators frequently override automated suggestions. They lack knowledge and required skills to validate AI Models. Process engineers understand the chemistry, while automation engineers understand the code. A lack of cross-disciplinary collaboration stalls deployment.
Also Read: Paper Mart Emagazine Aug-Sep, 2026
PM: Can you share a recent implementation that measurably reduced variability or improved quality consistency? What results were achieved, and where should mills prioritize investment next?
SK: Furnish recipe control systems, stock and chemical flow control systems, and stage-wise consistency controls have resulted in better stability of the whole process.
AI tools with various end devices connected to the process chemical system have improved the performance of the process chemicals with respect to consistent wet-end process parameters and reduced chemical consumption.
Significant improvements have been made in the effluent treatment plant through AI-based models, which monitor the vital parameters of liquid effluent and control the mill influent for various production grades.
AI-based models are being developed for the order management systems to provide better visibility beyond the pope reel up to despatch, and efforts are now being made to extend this visibility up to the customer’s premises.
Mills should prioritize investment in basic end devices and sensors for measuring, monitoring, and controlling key resources like water, steam, and power on a section-wise basis, which will certainly make them much more efficient and sustainable in the business.
Mills should prioritize investment in basic end devices and sensors for measuring, monitoring, and controlling key resources like water, steam, and power on a section-wise basis, which will certainly make them much more efficient and sustainable in the business
