PROCESS CONTROL

Model predictive control

of the chip level in a continuous
pulp digester, a case study
By T. Lindgren, T. Gustafsson, H. Forsgren, D. Johansson and J. stensson

Abstract: This paper describes the digester chip level control problem and shows a MPC system
implementation for the level control.
The objective with implemented chip level control system is to stabilize the transport conditions
in the bottom of the digester by using the chip feed and not the blow flow as the primary manip-
ulated variable. The results indicate a smoother digester operation and a reduction of the kappa
number variations compared with the original control system.

he continuous digester (The Kamyr The MPC scheme is straightforward regarding

T digester) is a complex tubular reactor

where the delignification of the wood
chips takes place through combined
chemical treatment and thermal
the handling of constraint variables in the control
problem. By combining the controlled and con-
straint variables, it is possible to set-up and tune
the MPC controller both to control the level to a
effects. The digester consists of three important set point and to have upper and lower limits to
zones: the impregnation zone, the cooking zone ensure the production level, i.e. to keep the chip
(delignification reaction) and the washing zone level at an average value (set point) but at the
where the counter flow of the bottom dilution same time allow the level to fluctuate within cer-
flow slows done the reaction and removes residu- tain limits (as a buffering functionality). The
al chemicals and lignin. The key parameter for MPC controller is implemented on a continuous
monitoring the digester system performance is digester at the M-real Husum mill, Sweden and
the kappa number, which represents the residual has been running since the end of 2003. The
lignin in the pulp. digester operators state that digester system has
The kappa number is influenced by the fol- operated more smoothly during this winter peri-
lowing variables, Svanholm [5], Lundqvist [10] od with the updated level control system, com-
Chip quality (humidity, mixture, species, size, pared to last years operation.
etc.) Another purpose of this work is to implement
The quantity of Effective Alkali (EA) in the a multivariable model-based controller, such as a
chips MPC controller, on a selected pulp and paper
T. LINDGREN The temperature that chips are exposed to process. The selected process (the digester sys-
Lule University of The time the chips are exposed to EA and tem- tem) and the MPC controller will be used as a test
Technology, Lule, Sweden, perature (the time in the delignification process) system for the development of a monitoring and
Network for Process Intelli- The statements above are valid when looking diagnosis system for multivariable and multi-loop
gence (NPI), -vik, Sweden at the delignification process at the chip scale, control systems.
Eurocon AB, rnskldsvik, but it is very important to also consider how the
Sweden variables above are behaving when the chips are THE CHIP LEVEL CONTROL PROBLEM
[email protected] transported through the digester process (the The primary objective of using a digester super-
digester scale). It is important to understand the visory control system is to minimize the variation
general sequence of transport and reaction pro- in the kappa number and to improve (stabilize)
T. GUSTAFSSON cesses that govern the overall digester operation. digester operation. The chip level control has an
Lule University of Stable transport conditions at every subpart in the important effect on that objective. It is impor-
Technology digester system are essential for having good con- tant to understand how the variations in the
Lule, Sweden ditions for the chemical and thermal treatment of chip level itself are influencing the chip cooking
the chips. The kappa number variability is related process, and also how the manipulated variables,
to variations in chip retention time in the used for controlling the level, are influencing
H. FORSGREN digester, caused by fluctuations in chip packing, the digester process.
Eurocon AB blow line consistency, etc. These variations can A number of papers have been written regard-
rnskldsvik, Sweden be, to some extent, detected by measurements of ing the control of the Kamyr digester chip level.
chip level, blow line differential pressure, and The following section is an attempt to review the
D. JOHANSSON bottom scraper amperage. control strategies, controller structures, Manipu-
M-real, Husum Mill In this paper a number of reports on digester lated Variable (MV) and Controlled Variable (CV)
Husum, Sweden chip level control have been analysed regarding the selection, etc., described in those papers.
control strategy and controller structure. A chip lev-
el control strategy (two-vessel Kamyr digester) is Fiber discharge rate as MV
J. STENSSON proposed, aiming to stabilize digester bottom con- Sastry [12] designed a Self Tuning Regulator
M-real, Husum Mill ditions. The strategy is implemented as a MPC con- (STR) which used the blow flow as the primary
Husum, Sweden troller scheme and as a PID-solution in DCS. MV. The chip flow and bottom scraper speed

46 106:11 (2005) T 239 Pulp & Paper Canada

 PROCESS CONTROL
TABLE I: Summary of reviewed papers. FF = Feed-Forward
Signal.

FIG. 1. Digester system at M-real mill, Sweden.

were used to deal with extreme level con- Chip feed rate and fibre discharge rate as consistency control were described.
ditions. No primary control of the blow MV Discussions with operating personnel
line consistency was presented. The Fuchs et al. [13] used the chip metering indicate that large blow flow manipula-
results showed a decrease in chip and speed as the primary manipulated vari- tions could disturb the chip column
black liquor flow variations to the digester able for the chip level control. They point- movement and change the mass and ener-
by manipulating blow flow compared to ed out that it is important to have a stable gy balances. The results showed a
the original control system. According to fibre discharge rate in order to stabilize decrease of 50 % in blow flow manipula-
Belanger et al. [9] this control system was the residence time in the cooking zones tions, but the expected improvements in
abandoned because of bad reliability of of the digester. The manipulations of the pulp quality were not achieved (a slight
the STR, but Belanger mentioned noth- blow flow were minimized, but a very slow increase in kappa variations). The MIMO
ing about any problems with the selected feed-forward signal was added to the fibre GPC controller was disabled after six
control strategy. discharge target (blow flow target) based month of operation. According to Allison
Belanger et. al. [9] made a comparison on the difference between the nominal et al. [6] the bad chip level measurement
between using the blow flow and the production and the actual production i.e. (strain gauge measurement) and the fact
digester bottom scraper as the primary the chip meter speed. Manipulating the that the mill had been experimenting
manipulated variable for the chip level digester bottom scraper, with some cor- with a number of new grades were con-
control (using a STR control structure). rection of the bottom dilution flow when tributing factors to why the GPC was
The two alternatives showed similar the scraper speed was at its limits, primar- switched off.
results regarding top separator amperage ily controlled the blow line consistency. Lundqvist [10] presents a chip level
(level measurement) variations but the No specific results were shown regarding control strategy that utilizes both the chip
digester bottom scraper (outlet device) the level control performance. meter speed and the blow flow as manip-
was selected because of different practical Al-Shaikh et al. [11] presented a control ulated variables depending on the size of
reasons. The operators also preferred the strategy where the chip meter speed was the chip density variations. At small chip
bottom scraper as the primary MV used as the primary manipulated variable density and wood property variations, the
because of the more steady blow flow. for short-term level variations, and the chip meter is used as primary MV and the
Controlling dilution flow, where the set blow flow ratio, for long term variations. blow flow is kept constant. If the chip den-
point was ratioed to the scraper set point, The chip meter speed sets the production sity varies considerably, the blow flow is
aimed to stabilize the blow flow consisten- target, with a limited bias added for the used as MV and the chip meter is kept
cy. The dilution controller output was level control. When the accumulated constant. The tuning of the mix of the two
slowly changing the blow flow set point. error for the chip meter speed (from the MVs should be done regarding the chip
The STR controller increased the stability nominal production speed) exceeds a level response characteristics and with the
of the production. certain limit the blow flow ratio target is kappa number chosen as the control per-
Allison et al. [7] designed an adaptive adjusted to maintain target production. formance criterion.
controller based on a generalized predic- The blow line constistency was not direct- Amirthalingam et al. [2,3] discussed
tive control structure (GPC). The blow ly controlled. how chip level control relates to pulp
flow was used as the primary manipulated Allison et al. [6] used both the chip quality parameters such as the kappa
variable. No clear motive was presented meter speed and the blow flow as manip- number. They proposed that the manipu-
regarding why the blow flow was selected. ulated variables. The chip meter and the lated variable for the chip level control
The blow flow set point is set by the GPC blow flow simultaneously reacted to level should be selected according to the origin
together with the production feed forward disturbances, but with the chip meter of the disturbance, e.g. if a change in the
signal and a proportional feedback signal slowly returning to steady state produc- digester chip level is caused by a change
from the digester top separator current. tion rate and with blow flow to compen- in chip bulk density, the right action
No primary control of the blow line con- sate for the persistent load disturbance. would be to change the chip metering
sistency was presented. The adaptive con- The GPC was set to constrain the chip speed. However if a disturbance in the
troller showed improvement of the chip meter speed about the production target chip level were caused by a change in the
level control by means of a closer mean speed. The motive for the use of the chip fibre discharge rate (change in blow line
value to level target and a reduction of the meter was to reduce blow flow manipula- pulp consistency), the right action would
standard deviation of the chip level signal. tions. No specifications about blow line be to change the blow flow (or digester

Pulp & Paper Canada T 240 106:11 (2005) 47

 PROCESS CONTROL
bottom scraper). They demonstrated the The fluctuation of the chip pile may cause The impregnation vessel chip level is controlled
proposed digester control structure changes in the driving force and further- by:
through a simulation model (fundamen- more variations in the chip column move- Manipulating the impregnation vessel
tal model). They also showed that a very ment and compaction. The chip column bottom scraper and the bottom dilution
tight control of chip level (via a well tuned is, however, elastic and compressible and flow.
PI-controller) actually disturbed the infer- the elasticity of the column may be The dilution flow (sluice) is used as the
ential kappa number control. reversible or irreversible [8] and the com- primary manipulating variable to main-
Table I is a summary of the reviewed paction degree of the chip column may tain the level close to target.
papers, showing the selection of manipu- fluctuate. The scraper speed is used to maintain
lated variables and the controller struc- If the chip column is moving quite the level within an upper and lower limit.
ture for the digester level control. freely through the digester, the residence When the level is within these limits no
time in the impregnation, cooking and action is made by the scraper. (MPC con-
PROCESS ANALYSIS AND washing zones depends mainly on the straint functionality, see below)
PROPOSED CONTROL fiber discharge rate [13]. It is then impor- The scraper speed is slowly moved to a
STRATEGY tant that the fibre discharge rate is stabi-
lized. Amirthalingam et al [2] pointed out
minimum speed target to move the aver-
age control action to the bottom dilution
Process description of the M-real digester that a good chip level controller structure flow and to prevent too high scraper speed.
system should react to the root cause of the level
The continuous digester at M-real Husum disturbance. CASE STUDY
mill is a conventional Kamyr digester sys- The choice of control philosophy (selec- M-REAL HUSUM MILL
tem, Fig.1, which consists of an impregna- tion of MVs and CVs, controller structure The proposed control strategy above is
tion vessel and a steam/liquor phase etc.) for the chip level control may be more mainly implemented as a Model Predictive
digester. The digester has four liquor cir- important than actually achieving a very Control scheme on a stand alone PC com-
culation flows (C5, C6, C7, and C8) where tight control of the level [2]. municating via an OPC link to the
the C7 is not used and the C6 trim circu- Siemens DCS system. A watchdog function
lation is only used for alkali measurement. Proposed control strategy for the OPC link checks that the commu-
The alkali is added in the impregnation The proposed control strategy for the chip nication is established and running. If the
vessel and in the C5 and C8 circulation. level control focuses on stable conditions communication is disrupted, a back-up
The production rate is about 1000 in the digester bottom. A conclusion control function is implemented in the
tons/day (softwood) and the kappa num- regarding the described strategies above is DCS system with mainly the same control
ber target is 30. The digester supervisory that excessive blow flow manipulation will strategy as the MPC implementation.
control system at M-real Husum mill of disrupt the chip column movement and Implemented control strategy
consists of following modules the digester operation. Blow flow manipu- A system structure of the implemented
1) Production rate change control lations will affect the washing liquor up control strategy is shown in Figs. 2 and 3.
2) Pre-steaming flow (dilution factor control), which will The digester chip level is controlled by
3) Chip level control of digester and disturb the conditions in the washing zone adding a production bias (+/- 2 t/h) to
impregnation vessel (Blow flow, gamma and change the position where the cook- the nominal production target, which is
source as level ind.) ing process (delignification) is disrupted. controlled by manipulating chip-metering
4) Wood/liquor ratio and circulation The following proposal for the chip speed. The digester chip level is also set
flow control level control strategy is based on up as a constraint variable (CT) with an
5) Pulp consistency and sluice flow con- [10,11,13], discussion with process engi- upper and lower limit. If a violation of
trol neers and digester operators at m-real these limits is detected, the MPC algo-
6) White liquor addition (concentration) Husum mill, and experiences from other rithm will make a powerful adjustment of
control digester chip level control system installa- the chip meter speed to bring the level
7) H-factor control tions made by Eurocon AB. This control within the limits again. When the level is
8) Kappa number prediction and feed- strategy also includes the level control of within the limits, the control actions made
back control the impregnation vessel. by the level constraint variable are dis-
All control modules are implemented abled.
in the Siemens DCS system (mainly with The digester chip level is controlled by: The actual production is used as a
PID controllers). The actual production Manipulating the speed of the chip feed-forward signal for the impregnation
load on the digester is about 50 % higher metering wheel. The control signal is vessel level control to improve the chip
than design figures, production per bot- added as a production bias to the nominal level control at production rate changes
tom square meter [ton/m2*day]. During production with upper and lower limits and to shorten the time delay for the chip-
the cold wintertime, snow, ice and frozen (+/- 2 t/h). metering wheel.
chips make it more difficult to get a Manipulating the digester bottom The blow flow to production ratio is
smooth digester operation compared to scraper controls the pulp blow consisten- very slowly adjusted to keep the produc-
summertime. cy. The pulp consistency measurement is tion bias within +/- 1 t/h. The implemen-
primarily based on the differential pres- tation of the blow flow control function is
Process analysis sure between digester bottom and blow still a matter of discussion and is, at the
The chip level gives an indication of the line. The in-line pulp consistency mea- moment, not implemented in the MPC
chip column movement and the resi- surement device (yield stress) and the controller. The differential pressure
dence time (the extent of the chemical load on the bottom scraper may be used between the digester bottom and the blow
and thermal treatment) for the wood to adjust the pulp consistency controller. line is controlled by the bottom scraper
chips in the digester. The blow flow is manipulated slowly to speed to maintain a constant pulp consis-
The downward movement of the chip move the actual production back to target tency and to stabilize the fibre discharge
column is due to the gravitational force, production. rate.
which mainly depends on the height of Depending on the performance of the The impregnation vessel chip level
the chip pile (bulk density, etc) that is pulp consistency controller the dilution (controlled variable) is controlled to the
above the liquor level and the density dif- flow could be used to improve consisten- level target by adjusting the set point for
ference between the chips and the liquid. cy control. the bottom dilution flow (FC5.2 in Fig. 2).

48 106:11 (2005) T 241 Pulp & Paper Canada

 PROCESS CONTROL

FIG. 2. Digester chip level control system. FIG. 3. MPC controller structure.

FIG. 4. Step response for the differential pressure. FIG. 5. Step response for digester level.

The impregnation vessel chip level is also which is used for the output predictions in the matrix contains the dynamic model
set up as a constraint variable with an and to relate the MVs to the CVs. The that relates the MVs with CVs and CTs.
upper and lower limit. If a violation of data used for the modelling and design The prediction horizon for the MPC con-
these limits is detected, the bottom scrap- has been generated by step response troller was set to 50 minutes (100 control
er will react powerfully, together with the experiments. The possibilities of using intervals). The prediction horizon for the
dilution flow, to bring the level within the more sophisticated data generation tools level constraint variables was set to 30 and
limits again. When the level is within the like PRBS signals were limited. Figures 4 20 minutes to make that control function
limits, the scraper speed is slowly brought through 6 show the step responses for the less restrictive. The MPC algorithms are
back to a target speed to move more of chip levels and the differential pressure. described in Maciejowski [1].
the average control action onto the dilu- The step response figures show that
tion flow (FC5.2). The impregnation ves- the digester and impregnation vessel level RESULTS AND DISCUSSION
sel scraper speed is both a manipulated process can be characterized as an inte- The MPC controller has been running
variable and a control variable in the MPC grator process with a time delay at the since start-up in mid December 2003. The
controller, Table II. input. The time delay is, however, longer back up level control system in the
The constraint variable can be seen, in than physically expected. That could be Siemens DCS system was installed in
this case, as a safety function to prevent explained by the time it takes to acceler- August 2003 and was running until the
plugging of the impregnation vessel. That ate the chip column in the impregnation MPC controller was installed. The original
allows a smoother control action for the vessel and the compaction of the chip col- digester level control system, with the blow
set-point control. It is not very critical to umn in the digester due to chip column flow as the primary manipulated variable,
have a tight level control of the impreg- elasticity [8]. The chip column elasticity is still available for the operators but has
nation vessel, and that also gives the pos- can possibly be modelled as a second not been in operation since the start-up of
sibility of using the impregnation vessel as order system. An integrating process with the new level control strategy.
a buffer for chip feed variations into the dead time was selected as the model struc- Table III shows some figures regarding
digester system. ture for the models needed in the MPC the standard deviation for the chip meter
design, except for the differential pres- speed [rpm], blow flow [m3/h], kappa
Step response modeling and MPC con- sure control, where a first order model number, digester chip level [m] and pulp
troller design structure was used. consistency [%]. The standard deviation
A key element in the design of the MPC Table II shows the controller design calculations are made from periods of
controller is the dynamic process model, matrix for MPC controller where each cell equal length and similar production rates

Pulp & Paper Canada T 242 106:11 (2005) 49

 PROCESS CONTROL

FIG. 7. Digester level control. The production bias is chang-

FIG. 6. Step response for impregnation vessel leve. ing between 2 t/h to + 2 t/h.

TABLE II: MPC controller design matrix. TABLE III: Comparison between old level control system
(2003) and new level control system (2004).

was used as the primary manipulated variable for the level con-
trol, instead of the blow flow, which was used in the original lev-
el control system. A comparison of production data for January
to February 2004 and the same period in 2003 indicates that the
new control strategy decreases the kappa number variations. A
smoother digester operation (stated by the operators) together
with an increased focus, from the operators, on the complete
digester mass balance control was probably two contributing fac-
tors for the reduction of the kappa number variations. Even if
the chip level variations did not decrease significantly, the reduc-
tion of the blow flow manipulations and pulp consistency varia-
(40 to 46 t/h) during January and February in 2003 and 2004 tions probably stabilized the chip column movement and the
(sample time 20 minutes). conditions in the digester bottom.
The results show a clear reduction of blow flow manipula- The MPC controller implementation has improved the con-
tions and pulp consistency variations. The reasons for the reduc- trol of the impregnation vessel chip level. With the constraint
tion in kappa number variations may need a closer investigation functionality in the MPC scheme it has been possible to combine
of the overall production conditions. The digester operators disturbances reduction with safety function control in the same
confidence for the new control system is high and they state that control algorithm.
the digester operation has been smoother this winter period The future work regarding the level control system will be to
than previous ones. Modify and improve the control function for the blow flow
Figure 7 shows a trend plot of the digester level control. The and implement that function into the MPC controller
production bias (chip meter speed) is manipulated to control the Improve the pulp consistency control by utilizing dilution flow
digester level. The blow flow set point is adjusted at time 16:48 and bottom scraper motor amperage.
when the production bias is outside the +/- 1 t/h control limits. Utilize gain scheduling for improved tuning at low production
Figure 8 shows how the blow line pulp consistency is controlled rates
by manipulating the digester bottom scraper speed. The goal is to Start-up of the digester and MPC controller performance
keep the pulp consistency within +/- 0.25 %. The digester bottom monitoring project.
dilution flow may need to be involved in the pulp consistency con-
trol function to improve control performance [13]. ACKNOWLEDGEMENT
The impregnation vessel chip level is primary controlled by The authors would like to thank the digester operators and pro-
the bottom dilution flow (FC5.2 in Fig. 2) but with the bottom cess engineers at M-real Husum Mill for their assistance, and the
scraper acting on the upper and lower limit. In the beginning of Network for Process Intelligence (NPI) at Mid Sweden Univer-
the trend plot, Fig. 9, the scraper speed increases to correct a sity, rnskldsvik, Sweden for their financial support. A special
detected violation of the upper constraint limit. When the level thanks goes to Capstone Technology, Seattle USA, for the use of
is back within the limits the scraper speed slowly returns to tar- their MPC software.
get speed, which is set to 3 rpm.
LITERATURE
CONCLUSIONS AND FUTURE WORK 1. MACIEJOWSKI, J.M. Predictive Control with Constraints. Harlow, UK. Prentice Hall,
331p. (2002).
This paper describes the implementation of a new control strat- 2. AMIRTHALINGAM, R., LEE, J. H. Inferential Control of A Continuous Pulp
egy for the digester chip level control at the M-real Husum mill. Digester In the Presence of Chip Level Variations. Proceedings Control Confer-
The objective of the new control strategy was to stabilize the ence (2000).
transport process in the bottom of the digester. The chip feed 3. AMIRTHALINGAM, R., LEE, J. H. Subspace Identification Based Inferential

50 106:11 (2005) T 243 Pulp & Paper Canada

 PROCESS CONTROL

FIG. 8. Blow line pulp consistency control. FIG. 9. Impregnation vessel level control.

Control Applied to a Continuous Pulp Digester. J. of 11. AL-SHAIKH, A., TU, F. Multilevel control of con- Digester Chip Level. Pulp & Paper Canada 79(5):T160-
Process Control 9:397-406 (1999). tinuous digesters: Applications and results. Proc. IFAC T163 (1978).
4. PELLETIER, S., EVANS, R., WESSLEN, T. New 7th World Cong., Helsinki, 239 (1978). 13. FUCHS, R. E., SINGLE, G. J. Advances in Com-
Digester Controls at Donohue Boost Kraft Pulp Qual- 12. SASTRY, V. A. Self-Tuning Control of Kamyr puter Control of Kamyr Continuous Digesters. Tappi
ity and Production , www.paperloop.com (1998). Journal 59(2):96-99 (1976).
5. SVANHOLM, C. Supervisory control of continuous
digesters. Certification of digester operators,
Markaryd Education, internal document Eurocon AB, Rsum: La prsente communication dcrit le problme du contrle du niveau des copeaux
(1997). dans le lessiveur et illustre la mise en oeuvre dun systme MPC pour le contrle du niveau. Le
6. ALLISON, B. J., DUMONT, G. A., NOVAK, L. H. but de ce systme est de stabiliser les conditions du transport au fond du lessiveur laide de lal-
Multi-Input Adaptive-Predictive Control of Kamyr imentation des copeaux et non par le soufflage comme variable de commande principale. Les
Digester Chip Level. The Canadian J. of Chem. Eng. rsultats font tat dun meilleur fonctionnement du lessiveur et dune rduction des variations de
69:111-119 (1991) lindice Kappa par rapport au systme de contrle initial.
7. ALLISON, B. J., DUMONT, G. A., NOVAK, L. H.,
CHEETHAM, W. J. Adaptive-Predictive Control of
Kamyr Digester Chip Level. AIChE Journal 36(7):1075- Reference: RLINDGREN, T., GUSTAFSSON, T., FORSGREN, H., JOHANSSON, D., STENS-
1085 (1990).
SON, J. Model predictive control of the chip level in a continuous pulp digester, a case study. Pulp
8. HRKNEN, E. A mathematical model for two-
phase flow in a continuous digester. Tappi Journal & Paper Canada 106(11):T239-244 (November 2005). Paper presented at the Control Systems
70(12):112-126(1987). 2004 in Quebec, Quebec, June 14-17, 2004. Not to be reproduced without permission of PAPTAC.
9. BELANGER, P. R., ROCHON, L., DUMONT, G. A., Manuscript received April 01, 2004. Revised manuscript approved for publication by the Review
GENDRON, S. Self-Tuning Control of Chip Level in a Panel on April 6, 2005.
Kamyr Digester. AIChE Journal 32(1):65-74 (1986).
10. LUNDQVIST, S. O. State of Art in Continuous Keywords: PROCESS CONTROL, CONTROL SYSTEM, CHIPS, CONTINUOUS PROCESS,
digester control. Proc. EUCEPA Symp. Control Sys. in CHEMICAL PULPING, FEEDERS, DIGESTERS.
Pulp and Paper industry, Stockholm (1982).

Pulp & Paper Canada T 244 106:11 (2005) 51