Kamyr digester chip level poses a difficult control problem, because the chip column dynamics are complex and it is difficult to measure the chip-liquor interface. Control of the chip level is required to smooth digester operation and to stabilize the residence time. Conventional methods rely largely on fixed parameter controllers which are unable to compensate for changes in the process dynamics. Manual controller retuning is often required.An adaptive chip level controller based on Generalized Predictive Control (GPC) is presented. Closed-loop time series identification methods were used to develop the process model and to design the controller. A significant reduction in chip level and P-number variability has been observed since commissioning the controller on an industrial digester in September 1988. The adaptive controller has eliminated the need for manual retuning, reduced the need for operator intervention, and demonstrated the potential to predict hang-ups.

This paper investigates the development and industrial application of a multi‐input/single‐output adaptive‐predictive Kamyr digester chip level controller based on Generalized Predictive Control (GPC). The controller simultaneously manipulates the blow flow and the chip meter to control the chip level. The chip meter is manipulated in a way which tends to maintain the desired production rate, while still leaving some room for control. The idea was to use the chip meter as a second input variable to reduce blow flow manipulation and, thereby, reduce disruptions to the chip column movement. The two‐input strategy decreased blow flow manipulation by more than 50% (compared to an otherwise equivalent one‐input scheme) without sacrificing output performance. However, the anticipated improvements in permanganate‐number variability were not achieved. The multi‐input strategy was discontinued after six months of operation.