At Blandin Paper Co., a model-based adaptive controller has allowed closed loop control of groundwood pulp brightness, resulting in less variability and chemical overdosing
	Blandin Reduces Variability in Pulp Brightness with Adaptive Controller
	By DALE LILLFORS, CHRIS BASSET, RAY AKKERMAN, SAVA KOVAC

Pulp and paper producers are under tremendous pressure to make quality products, and one indicator of quality is product brightness. Mills achieve the specified brightness by mixing stock with chemical bleaching agents, and the process time from the start of the bleaching process (chemical mixing) to the final brightness measurement can take hours. Process times of three hours or more are common in typical bleaching processes, but because proportional integral derivative (PID)-based algorithms cannot control processes with long time constants, the brightness targets are achieved using open loop control. However, this method can cause high variability in brightness and overuse of costly chemicals.

The Blandin Paper Co., a subsidiary of UPM-Kymmene located in Grand Rapids, Minn., produces lightweight coated groundwood paper products. The mill needed to reduce the variability in the brightness of its groundwood pulp. Blandin's bleaching process mixes premixed bleach chemical (mainly hydrogen peroxide) with pulp stock to achieve final brightness targets. The existing brightness control was open loop since the bleaching cycle times of 1.5 to 4.5 hours could not be controlled with PID-based algorithms. However, the open loop control contributed to high variability in brightness.

In June 2001, Blandin initiated a project to reduce variability in the final brightness of its pulp. A new model-based adaptive controller was selected to close the brightness loop. This controller is capable of handling long process time delays (in hours), non-linear process characteristics, and changes in process gains and time constants. With these difficult control issues resolved, the final brightness target was achieved with closed loop control. This closed loop control gave Blandin the ability to produce consistent quality products while reducing production costs.

OPEN LOOP CONTROL OF THE BLEACHING PROCESS. Before installing the new adaptive controller, refined, unbleached 4%-consistency hardwood pulp was pumped from the feed chest to the bleach press where the pulp consistency was increased to 15% (Figure 1). Pulp was then mixed with H2O2 bleach via a mechanical mixer. The injection rate of H2O2 bleach to the mixer was controlled via a bleach-dosing flow controller, designated as FRC-4. The operator manually set the bleach-dosing rate using FRC-4 based on information obtained from the existing brightness algorithm and the final brightness measurement, known as AI-7.

FIGURE 1. Before the installation of the new adaptive controller, pulp brightness control was open loop.

Next, bleached pulp passed through a screw conveyor where a residual measurement (AI-5) and pH measurement (AI-6) were taken and fed back to the existing brightness algorithm. Bleached pulp then passed through the bleach tower and the bleach chest, where consistency was decreased back down to 4%. At the bleach chest outlet, the AI-7 brightness meter provided indication of final brightness for the bleached pulp. The length of time between mixing the pulp with the bleaching chemicals and the final brightness measurement (AI-7) was a function of the production rate of the plant and the level in the bleach tower. The changing production rate created process times from 1.5 to 4.5 hours.

ADAPTIVE CONTROLLER CLOSES THE LOOP. In July 2001, a BrainWave controller from Universal Dynamics was installed to implement the closed loop control. No mechanical or process changes were required, but some field instrumentation was not functioning correctly, and repair and recalibration of these devices were required.

The new controller, designated AC-7, is a model-based adaptive controller. This controller uses the existing instrumentation and information from the distributed control system (DCS) to control pulp brightness, as outlined in Figure 2. The final after-tower brightness measurement (AI-7) is used as the process variable (PV) for feedback to the controller. Production rate information in the form of retention time calculated in the DCS is used as a feedforward signal (FF-1) to the controller. The control output variable from the controller is tied as the setpoint of the existing FRC-4 dosing rate controller. The existing control brightness algorithm and the AI-5 residual analyzer are not required to control pulp brightness.

FIGURE 2. The new model-based adaptive controller, designated AC-7, helped Blandin Paper close the loop on brightness control.

CONSISTENT BRIGHTNESS, LOWER CHEMICAL COSTS. Dale Lillfors, Blandin Paper's engineer responsible for installation of the new adaptive controller, says that the mill "is producing a more consistent brightness, and there are chemical savings. Before BrainWave, we had to set a higher brightness target to allow for the dips. Another benefit realized from BrainWave's ability to control the brightness loop is that the operator is free to focus on fine tuning other parts of the pulp mill."

In Figure 3a, open loop control by operators is compared during a 14-day period with closed loop control using the new controller. The results shown are after recalibration of the existing field instruments. Improvements from the new closed loop control are apparent. Within a few days of taking control, the new controller had learned the process and significantly reduced the brightness PV. Another improvement was a reduction in chemical usage. The average chemical dosing rate (CO) required to meet the brightness target is now lower, indicating a reduction in chemical costs. Figure 3b illustrates the significant variations in production rate that alter the brightness loop process dynamics.

FIGURE 3a&b. After the brightness loop was put in closed loop control with the adaptive controller, there was less brightness variability, and the chemical dosing rate set by the new controller was below the previous rate set by the operator (3a). During the same time frame, variations in production rate resulting in fluctuating retention times altered the brightness loop process dynamics (3b).

Parameters shown in Figures 3 through 6 are:

• CO - Control output (chemical dosing rate in % from FRC-4 controller)
• SP - Brightness setpoint (actual value)
• PV - After-tower brightness (actual value from AI-7 brightness meter)
• FF - Retention time in minutes (bleaching process time constant from FF-1 signal)

During the first seven days (Figure 3a), the brightness loop was under operator open loop control. The plant operator manually set the bleach-dosing rate, the brightness setpoint was constant at 83, and the retention time (Figure 3b) varied from lows of 100 minutes (1.5 hours) to highs of 250 minutes (4.5 hours). It was difficult for the operator to maintain the brightness target as indicated by the brightness value. The operator made step changes to the dosing rate to correct brightness errors, however, the adjustments were not large enough or quick enough to maintain the brightness target.

The operator actions can be seen clearly in Figure 4, which is a close-up of operator control. The brightness was below setpoint, and the operator reacted, changing the dosing rate to 100% of range. The process reacted, and the brightness value reached the 83 target on August 26. The chemical dosing rate was left unchanged, and the brightness value overshot the setpoint. The increase in brightness continued for 12 hours before an operator took action. The operator was cautious and made a number of small changes to correct the error in brightness. It took 17 hours before these small changes brought the brightness PV down to the 83 target. The 12-hour delay in corrective action, combined with the 17 hours of cautious changes in the dosing rate, contributed to a total of 29 hours of unnecessarily over-bright product, as well as overuse of chemicals.

FIGURE 4. Under open loop operator control, efforts to correct the brightness value resulted in 29 hours of over-bright product and overuse of chemicals.

The brightness loop was put in BrainWave closed loop control on August 28, as Figure 3a shows. The process conditions are similar to the conditions under operator control. The brightness setpoint continued to be maintained at 83, and the retention time shown in Figure 3a varied from lows of 130 minutes (2.1 hours) to highs of 250 minutes (4.1 hours). The new adaptive controller responded quickly to process changes limiting the variability in brightness. The average dosing rate set by the new controller was well below the previous rate set manually by the operators, indicating a reduction in chemical usage.

Figure 5 shows a close-up of BrainWave control. The controller's ability to respond quickly to process changes was evident on September 1. At that time, the brightness was right at the setpoint target of 83. However, a rapid change in the process retention time occurred, ramping up to 170 minutes before settling back down at 140 minutes. The adaptive controller quickly made the required changes to the bleaching rate, limiting the brightness variability and the over-application of chemicals.

FIGURE 5. Under closed loop control, the adaptive controller changed the bleaching rate, limiting the brightness variability and overuse of chemicals.

With the improvement in brightness control, Blandin can now produce product at various brightness specifications. Figure 6 shows brightness setpoint changes between lows of 82 and highs of 83.8 while under BrainWave control. The varying production rates between 1.5 and 4.5 hours are similar to the previous examples shown in Figure 3. The brightness setpoint changes are made at unusually short intervals, limiting the time the controller has to deal with the new conditions. The new model-based adaptive controller is able to maintain brightness value with a standard deviation of 0.3 under these difficult requirements.

FIGURE 6. Brightness setpoint changes between a low of 82 and a high of 83.8 while under closed loop control