[Home] [APPW 2004] [Journal papers]

peer reviewed logo


M.S. Sidhu, A. Lahouaoula, M. Schijf, A. Hird, R. Vyse and T.H. Steele

Presented at the 59th Appita Conference, Auckland, New Zealand, 16-19 May 2005



The use of thermo-mechanical pulp (TMP) for paper making provides significant economic benefits over chemical pulps. However, the TMP pulp must be of uniform and high quality to ensure high paper machine efficiency and good final paper quality. For example, uniform pulp freeness improves the drainage properties of the paper machine and hence has a direct effect on paper machine runnability. Furthermore, pulp properties such as brightness have a significant impact on the final paper quality.

In the first half of this paper we discuss an advanced control strategy for the main-line refining process. The advanced control strategy is based on the model-based predictive control (MPC) paradigm that is able to provide tight control of the refining process. The control strategy is capable of handling multiple refiner lines that empty into a common latency chest. The final pulp quality is controlled within an operator-defined window. As a result, the mill has been able to improve paper machine runnability and reduce kraft consumption. In the second half of this paper we discuss the implementation of an on-line sensor for measuring the pulp brightness in a TMP bleaching plant. The sensor technology is based on optical principles that utilize fibre optics to measure the reflectance properties of the pulp. The sensor is installed directly in the process stream, hence eliminating the need for a pulp sampling system. The sensor is able to measure the pulp brightness (% ISO) over a wide range. The ability to measure pulp brightness has allowed the second mill to reduce pulp brightness variability and chemical consumption.


Thermo-mechanical pulp (TMP) is widely used to manufacture a variety of paper products. The final paper properties are inherently linked to the quality of the pulp. For example, TMP freeness is critical to the drainage properties of the paper machine and pulp brightness has a direct impact on the final paper brightness. Therefore, it is essential to optimize TMP pulp quality parameters such as freeness, fibre length and brightness to ensure efficient paper machine operations and high final paper quality.

Mechanical pulps are also referred to as "high yield pulps" and as a result offer an obvious economic advantage as compared to chemical pulps. In order to meet the papermakers' desire to optimize the paper machine and in some instances replace the more expensive chemical pulp, the mechanical pulp must be both uniform and of a high quality. In order to provide the papermaker with a continuous advantage, tight quality control of mechanical pulp is essential.

The design and application of advanced control techniques provide the ability to maintain the process at the optimal operating point. Advanced control applications are designed to accommodate the complicated and interactive behaviour of the process. New sensor technologies are also incorporated to improve the measurement of critical process variables, as lab tests are typically inadequate for closed loop control. In this paper we present an advanced control strategy and novel sensor technology that have been applied to two different unit operations in two separate TMP plants.

In the first application, advanced control of a main-line refiner process is considered. As the pulp quality is ultimately determined by the operation of the main-line refiners [1], it is imperative that the refining process be under tight control. Closed loop control of a TMP refiner system is one of the most complex and challenging control problems in a pulp mill. The process is inherently multivariable, exhibiting strong interactions. In addition, the bandwidth of the process outputs is spread over a wide frequency range. For example, the open loop responses from the primary refiner plate gap to the primary refiner motor load and to the final pulp freeness are approximately two minutes and ninety minutes, respectively. The refining process is also complicated by nonstationary process dynamics due to wear of the refiner plates [2].

We first present a control strategy for the main-line refining process that is based on the multivariable model-based predictive control (MPC) paradigm [3]. Since process dynamics are spread over a wide frequency range, a single MPC executing at a fixed frequency would not be able to provide adequate control of both fast and slow dynamics. In order to mitigate this problem, a two-level control strategy is proposed. The first level of this control strategy consists of a MPC controller that is designed to handle the fast dynamics of the refiner line. The fast MPC controller is referred to as the Stabilization Controller as it controls the primary and secondary refiner motor loads and blow-line consistencies. The second level also consists of a MPC controller and is responsible for controlling the slow process dynamics. As the second level is associated with the pulp quality variables, it is referred to as the Quality Controller. The two levels are then integrated and coordinated by using an Optimizer to enhance constraint handling [4]. With this two-level control strategy the execution frequency of the Stabilization and Quality Controllers can be independently selected.

We then discuss the implementation of an optical sensor to measure the final pulp brightness. The sensor is implemented in a TMP plant that uses a single stage bleaching process to brighten the pulp. In this peroxide bleaching process, chemical bleaching agents are added prior to the bleaching tower. In most mills the bleaching process is operated in manual control and lab samples are collected for analysis. Based on the lab analysis, the operator manipulates the rate of chemical addition to maintain the pulp brightness at the desired level . The manual operation of the bleaching process introduces significant brightness variability to the paper machine. Due to infrequent and delayed information, operators tend to apply excess bleaching chemicals to ensure that the final brightness target is met. However, overbleaching erodes process economics by significantly increasing the cost of producing pulp.

By contrast, the novel brightness sensor is installed directly in the pulp stream and hence eliminates any sampling requirements. The sensor uses optical technology to detect pulp brightness and is capable of measuring brightness in realtime.


The process considered in this paper consists of four parallel refiner lines that empty into a common latency chest. Pulp freeness and mean fibre length are measured by an on-line sampling system at the exit of the latency chest. Each refiner line consists of two Sunds CD 70 refiners arranged in series.

The main objectives of the control strategy are:

  • Control refiner motor loads
  • Control blow-line consistencies
  • Attenuate wood chip density variations
  • Maximize production rate
  • Control pulp quality

The refining process is inherently multivariable and thus it is challenging to satisfy the above objectives. When one process input is manipulated in a multivariable process, the impact is realized by several process outputs. Figure 1 illustrates these process interactions for the primary refiner.

Advanced control figure 1

Figure 1: Process interactions

The process interactions can be described as:

  • Motor load - Motor load increases as the fibre feed rate is increased. Increases in the plate gap decreases the load due to reduced compression of the fibre pad as well as energy transfer from the refiner plates to the wood fibres. Finally, increasing the dilution water flow decreases the motor load by increasing total flow through the refiner. Assuming that the volume in the refiner is fixed (that is, the gap is constant) the residence time of the chip in the refiner is reduced resulting in fewer bar impacts on the fibres.
  • Blow-line consistency - When the chip feed is increased, the blow-line consistency also increases due to the increased mass throughput. As the plate gap increases, the motor load drops resulting in less steam generation and therefore a lower consistency. Finally, as dilution flow is increased the blow-line consistency is obviously decreased.
  • Pulp quality - If blow-line consistency and motor load are held constant, varying the screw speed changes the specific energy (kw-h/t), and as result the final pulp quality is altered. Dilution flow rate has a strong influence on the refining intensity which is strongly correlated to the pulp quality. Finally, the plate gap is responsible for determining the degree of energy transferred during the refining process. If the plate gap is too tight, fibre cutting can occur which results in degradation of pulp quality [5].

Disturbances to the refiner system come in the form of changes to species and wood morphology, chip moisture variation, bulk density differences of the chips, and plate wear [6].

The advanced control strategy is able to incorporate the multivariable nature of the process. Figure 2 considers two of the four parallel refiner lines and illustrates how the advanced control strategy interacts with the process manipulated variables (MVs) and controlled variables (CVs).

Advanced control figure 2

Figure 2: Controller configuration

The control strategy is able to control motor load and blowline consistency on both the primary and secondary refiners. Controlling the motor load is a great benefit to the mill as the mill has a peak power consumption limit. With the ability to control motor load, the mill is able to use maximum power without the danger of power excursions and thus avoid severe penalties during peak time. Control of the blow-line consistency stabilizes the refining intensity that is required to produce high quality pulp. The control strategy also controls the final pulp quality as measured by the PQM at the exit of the latency chest. The primary refiner plate gap is manipulated on each refiner line to control the final pulp quality. As the primary refiner gap is manipulated by the Quality Controller, the Stabilization Controllers work in concert to automatically adjust the specific energy applied to each refiner line. In this controller configuration, the requirement of implementing an independent specific energy loop is not required. Furthermore, the Optimization layer is able to coordinate the operation of both Stabilization Controllers and the Quality Controller for enhanced constraint handling. The operators are able to define a quality window by specifying the upper and lower limits for pulp freeness and mean fibre length. For a detailed discussion of the control strategy, please refer to Sidhu et al. [7].

On average, the advanced control strategy was able to reduce motor load variations by 53% and blow-line consistency variations by 86%. The most important advantage was the ability to define a pulp quality window for the main-line refiners. The pulp freeness and mean fibre length variability were reduced by 41% and 21%, respectively. Figure 3 illustrates the before and after pulp quality results.

Advanced control figure 3

Figure 3: Closed loop pulp quality control

With the production of low variability pulp from the main-line refiners, the downstream unit operations such as screening and reject refining also experienced significantly stable operations. Ultimately the final pulp quality to the paper machine was improved, assisting in reducing kraft consumption.


In this section we consider a single stage hydrogen peroxide bleaching process. In this process a questering agent is used prior to introducing the bleaching chemical to remove transition metals. A cascade mixing tank is used to mix hydrogen peroxide with sodium hydroxide and silicate. The bleaching chemical is then introduced into the pulp stream prior to an in-line mixer. After passing through the in-line mixer, the pulp is stored in the bleaching tower to ensure sufficient residence time to accomplish the desired brightness gain. The pulp from the tower is soured using sulphur dioxide for pH control. The pulp is then pumped into a storage chest for paper machine consumption.

The incoming brightness to the TMP bleaching process is quite variable due to upstream changes. For example, changes in wood species or water quality can cause significant disturbances to the bleaching process [8]. As the incoming pulp brightness changes, the amount of bleaching chemical applied must also change to maintain a constant final pulp brightness. The use of sampling techniques to determine pulp brightness is inadequate for tight control as the information is not frequently available. Brightness measurements from a sampling system also introduce significant delays due to sample transportation and batch analysis. Such infrequent measurements are difficult to incorporate into an on-line control strategy and provide limited information to the operator.

Advanced control figure 4

Figure 4: Brightness sensor schematic

Honeywell's Precision Optical Sensor® is implemented directly in the pulp stream, eliminating any sampling requirements. The sensor is able to provide real-time measurement of the brightness of the pulp stream. The brightness sensor is based on optical technology that incorporates fibre optics to detect the reflectance characteristics of the pulp. Figure 4 shows a schematic of the fibre-optics that are used to direct the light from one of two halogen light sources into the pulp stream. The probe is 2.25 inches in diameter and is inserted at a 70-degree angle into the path of the pulp. With this angle, the contact of the pulp tends to keep the window free of residue contamination. The window through which the light passes is constructed of sapphire, which has proven to be both scratch and chemical resistant. For extreme cases where scale deposits are a problem, an ultrasonic cleaning device can be attached. The device has one source path and two reflectance paths. A third bundle of fibres is used as reference feedback. With opticalbased devices, the intensity of the spectrum can change over time as the filament ages. Such drifts are compensated for by the feedback path. The compensation feedback signal is also subjected to the same environmental conditions as the measurement signal. The reflected light is fed back through selected band pass filters to an array of solid-state photo sensors.

After the sensor has been installed, lab samples are collected under various process operations for sensor calibration. The samples are processed in the lab and the % ISO brightness values are then used to generate a calibration curve. The calibration curve is developed using linear regression techniques. The calibration curve is downloaded directly to the sensor processor board that uses embedded software architecture. This embedded software approach allows the sensor to be a single entity capable of performing the measurement, signal conditioning and processing, and transmitting the final brightness value as an analog signal that can be incorporated into any DCS.

The implementation results of the brightness sensor can be seen in Figure 5. After the initial calibration, additional samples were collected for validation. The validation data provided an excellent R2 value of 0.95. The final brightness target is dictated by the final paper requirements, the brightness sensor is able to successfully measure brightness over a wide range.

Advanced control figure 5

Figure 5: Brightness sensor measurement vs. lab measurements

The mill has successfully incorporated the sensor brightness measurement to operate the bleach plant more efficiently. The operators are now able to reject process disturbances quickly due to the availability of brightness measurement in real-time. As a result, the brightness variability and chemical consumption have been reduced. Efforts are also underway to use the brightness sensor to construct process models and commission an automatic control strategy to further optimize the bleaching process.


The advantages of an advanced control strategy for the mainline refiners and a novel sensor for measuring pulp brightness have been demonstrated. The two-level control strategy is based on multivariable MPC control law to simultaneously stabilize and optimize the refining process. The modular controller design can easily handle multiple refiner lines that empty into a common latency chest. Ultimately the final pulp quality was improved, leading to improved paper machine operations and assisting in reducing kraft consumption. Likewise, the use of the brightness sensor at another TMP mill has also improved paper machine operations. The operators use the valuable information from the brightness sensor to run the TMP bleach plant more efficiently. Efforts are underway to incorporate the sensor output as part of an automatic control strategy.


[1] Stationwala, M.I., Atack, D., "The effect of control variables on refining zone conditions and pulp properties", IMPC 1979.

[2] Roche, A., Owen, J., Miles, K. and Harrison, R., "A Practical Approach to the Control of TMP Refiners", Control Systems 1996.

[3] MacArthur, J.W., "RMPCT: A new robust approach to multivariable predictive control for the process industries", Control Systems 1996.

[4] Lu, J.Z., "Challenging control problems and emerging technologies in enterprise optimization", 6th IFAC Symposium on Dynamics and Control Process Systems, June 2001.

[5] Owen, J., Roche, A. and Miles, K., "A Practical Approach to Operator Acceptance of Advanced Control with Dual Functionality", Control Systems 1998.

[6] Kortelainen, J. and Nystedt, H., "Disturbance analysis of the TMP-process", IMPC 1995.

[7] Sidhu, M.S., Van Fleet, R., Dion, M.R., Anderson, D.W., and Weger, B.W., "Modelling and Advanced Control of TMP Refiner System", Control System 2004.

[8] Dence, C.W. and Reeve, D.W., "Pulp Bleaching Principles and Practice", Tappi Press, 1996.

Author contact details:

M.S. Sidhu†, A. Lahouaoula†, M. Schijf‡, A. Hird‡, R. Vyse†, T.H. Steele†, Honeywell Process Solutions

† 500 Brooksbank Avenue, North Vancouver, BC Canada

‡ 679 Victoria Street, Abbotsford, VIC Australia