TappiJournal

A systems approach to pressure screencontrol in the pulp mill

Dan P. DumdieProcess control supervisorDaishowa America Co. Ltd.P.O. Box 271Port Angeles, WA 98362

Reprinted from Tappi]ournal, Vol. 74, No. 11, November 1991.Copyright 1991 by TAPPI, and reprinted by permission of the copyright owner.

ScreenControl.

A systems approach to pressure screencontrol in the pulp millDan P. Dumdie

A process control strategy for operating pulp mill pressure screens improuedpaper machine efficiency and Press room performance, significantly reducedrefiner power requirements, decreased chemical pulp usage, increased refinerpulp production rates, and Produced a more consistent and higher-jreenesspulp with substantially less debris.

The Daishowa America Port Angeles mill makes light-weight directory paper from refiner groundwood andpurchased kraft. Because of the lightweight sheet, papermachine runnability and press room performance are verysensitive to pulp mill shives and other furnish debris. Inan attempt to improve the relative competitive positionof the Port Angeles mill and to improve paper machineefficiencies, we installed Hooper pressure screens on boththe thermomechanical pulp (TMP) and refiner mechanicalpulp (RMP) lines. The new slotted screens replacedantiquated gravity Cowans, which provided poor shiveremoval efficiencies. The project exceeded all expectationsfor improved quality, lower costs, and higher efficiencies.Much of this can be attributed to the unique control systemthat was developed for the project.

Process backgroundThe purpose of a pulp mill pressure screen is to fractionatethe feed stream fiber into an accept stream containingquality pulp and a reject stream containing shives andother debris.

The feed enters at the top of the screen, where it is rotatedinside a cylindrical screening basket by a motor-drivenrotor. The basket may contain holes or slots for the acceptfibers to pass through. The rejects cannot penetrate thescreen. Instead, they travel to the bottom of the basketwhere they are forced out the reject line. The rotor containstips which rotate past the holes or slots to produce strongnegative-pressure pulses. This helps keep the screen fromblinding. As the fibers fractionate down the screen basket,they quickly dewater. Consequently, two dilution lines arerequired to help maintain uniform consistency near thebottom of the basket.

Screen performance can be defined in terms of shiveremoval efficiency (SRE). This is the weight percentage

Dumdie is process control supervisor at DaishowaAmerica Co. Ltd., P.O. Box 271, Port Angeles, WA98362.

1. Shive removal efficiency curves

<fi.u.iIX:(/)

60

40

20

o

REJECT RATE, %

of shives and other debris that is removed from the pulpby the screen:

where

SRE = shive removal efficiency, %

FR,FF = reject and feed flows, L/min

CR,CF = reject and feed consistencies, %

Reprinted from Tappi Journal, Vol. 74, No. 11, November 1991.Copyright 1991 by TAPPI, and reprinted by permission of the copyright owner.

(1)

2. Pressure screen control system

White water_---""'T----- dilution

Refinerpulp

(2.8%) - - --.IIIIIIIII

:--8---1I1I11

- --I

111

Accepts(1.1%)

II

••8

IIIIIIoo

RR ~SP

Io

IoI0-0-0-0-0-0-0-0-0-0-0-0-0-0-0-0-0-0-0-0

Rejects(2.0%)

CT = Consistency transmitterFT = Flow transmitterLT = Level transmitterPT = Pressure transmitter

CC = Consistency controllerFC = Flow controllerLC = Level controllerPC = Pressure controllerRC = Ratio controller

VPC = Valve position controllerDPI = Diff. pressure indicatorSP = Set pointRR = Reject rate

SR,SF = reject and feed shive contents, %

The SRE is a function of screen reject rate (RR), whichis the weight percentage of total dry fiber that is rejectedby the pressure screen:

Nelson and Bolton (1) have developed an empirical equa-tion relating shive removal efficiency to screen reject rate:

SRE = RR / [1.0 - Q(l - RR)]

A single coefficient, called the screening quotient (Q),determines the shape of the efficiency curve (Fig. 1). Thequotient can be determined for any screen from a fewexperimental data points and a least squares curve-fittingprogram. The screening quotient varies from 0.0 to 1.0.A value of 0.0 represents a screen with no efficiency forshive removal (450 line), and a quotient of 1.0 representsa perfect screen (100% efficient).

The Port Angeles mill has implemented on-line rejectrate control to help operations manage screening efficien-cies for various paper machine grades. This puts controllersetpoints in terms of actual reject rates and shive removalefficiencies (engineering units) rather than in terms of

98 November 1991 Tappi Journal

flows and consistencies that are used in conventionalregulatory control systems. This has had a very positiveeffect on the training of operating personnel.

Process control(2) There are a number of important control objectives for

operating a pulp mill pressure screening system. First,it is essential to achieve a high SRE to produce qualityaccepts while keeping the amount of rejected material toa minimum. The rejects should be concentrated withshives, chop, and other debris. This keeps the amount ofrework (reject refining) to a minimum and helps reduceoperating costs. In addition to maintaining a consistent andhigh level of accept cleanliness, screen controls should alsoproduce a pulp with uniform fiber fractions and a constantfreeness. Screening process quality is strictly a functionof accept cleanliness and freeness. Finally, the controlsystem must be able to balance screen throughput withprocess runnability. The controls must be capable ofhandling maximum pulp production while keeping screenblinding, reject line plugging, and other operatingproblems to a minimum.

All of these quality and operating considerations can beaddressed using the control strategy presented in Fig. 2.

(3)

3. Typical screening efficiency curve

100

80

60

~Wa:en40

20

oOptimum 40 60 80

REJECT RATE, %

100o

The system consists of one supervisory and sevenregulatory loopsfor each screen. These loopscan be placedinto one of three logical categories for discussion: rejectcontrol, inventory control, and feed control.

Reject controlA typical screen SRE curve is shown in Fig. 3. Optimumpressure screen operation occurs at a constant RR on theknee of this curve. At the optimum, pulp debris is removedwhile long-fiber fractions are retained. This results in auniform freeness and cleanliness of screened accepts. Athigher than optimum RR, quality long fiber is lost, andreject refining costs can become extreme. Some installa-tions typically run at excessive RR because this providesgood screen runnability at higher pulp production rates.Excessive reject rates are typical in applications wherethe screens are undersized. The problem, however, can alsooccur if dry chips, worn refiner plates, or other abnormalconditions cause high shive levels in the screen feed. Atlower than optimum RR, debris is not adequately removed,and freeness variability can be quite large even with smallvariations in RR.

We have implemented three modes of supervisorycontrol. The first is reject flow control, which must be usedduring startup with white water while consistencies are

outside their respective transmitter ranges. The secondand third modes are described below and provide a meansof directly regulating RR and SRE.

With adequate instrumentation, supervisory control canmaintain a constant RR on the screens. The supervisoryequation comes from the material balance (Eq. 2) and isused to compute the desired reject flow rate:

(4)

The reject flow is simply a function of on-line instrumentdata and the RR which is supplied by the operator as asetpoint. Once computed, FR is used as a supervisorysetpoint for the reject flow controller, as shown in Fig. 2.

Taking this system one step further, the SRE curve(screening quotient) can be put directly into the controlcomputer. This allows operations to enter its setpoint asan efficiency. In this case, the RR used in Eq. 4 is readdirectly from the screen's efficiency curve. It is oftendesirable to have operators enter setpoints in terms ofengineering units. This helps promote a better understand-ing of the process as well as the controls.

Pressure screen RR control (Eq. 4) obviously requiresan accurate and reliable measurement ofboth process flowand consistency. A magnetic flow tube is adequate for theflow measurement, and the optical Cerlic ACM/CT20/50has proven satisfactory for consistency measurement.

Before selecting a consistency sensor, we ran anextensive two-month trial on both optical and mechanicaltransmitters over a consistency range of 1.0% to 2.0%.During this period, the Cerlic unit outperformed all othersensors in terms of linearity, signal-to-noise ratio andcalibration drift. By combining all 35 ACM data pointscollected over the two-month trial and running aregression analysis, a correlation coefficient of 0.95 anda standard error of 0.07% consistency were obtained.Standard errors for the mechanical sensors were nearlythree times that of the optical unit. The mechanicals alsoshowed a very poor signal-to-noise ratio in this 1.0%to 2.0%consistency range.

Inventory controlThe main objective for inventory control is to prevent thescreen feed tank from running empty or from overflowingduring normal operation. There are a number of ways toachieve this control. However, the method selected mustbe compatible with other objectives such as maintainingscreen throughput and accept quality.

Port Angeles uses a simple accepts flow loop cascadedto the surge tank level controller, as shown in Fig. 2. Thisoptimizes throughput and keeps the overall freeness dropacross the screening system at a minimum.

It is quite common for a screen basket to slowly blindduring a period of several hours. Regardless of this partialblinding, the accepts flow loop (Fig. 2) will maintain aconstant process throughput. In contrast, a more conven-tional approach to inventory control is to manipulate thescreen differential pressure (DP). This can successfullyregulate feed tank level; however, it also significantlyreduces throughput when the screen partially blinds (i.e.,as the screen slowly blinds, the accept valve closesto main-tain DP). On the other hand, flow control opens the acceptsvalve during a partial blind to maintain throughput. The

November 1991 Tappi Journal 99

only lost time with accepts flow control is the 20 s necessaryto flush the basket each time it completely blinds.

Still another method of inventory control is to recyclea portion of the screen accepts to the feed tank. Becauseof the recycling, this method can result in a significantreduction in accepted long fiber and freeness. Moreover,as the recycle flow needed to control the level varies, sowill the freeness and pulp quality vary. Finally, recyclingaccepts will decrease screen throughput. All of theseproblems can be eliminated by manipulating accepts flowto control the feed-tank level.

Feed controlFeed stream pressure and consistency play an importantrole in the overall operation of a pressure screen system.They are the only process variables that require controlin the screen feed.

The feedstock pressure has no effect on screeningquality, efficiency, or throughput; however, feed pressurecan have an impact on process runnability. The feedpressure is controlled using a simple feedback loop, asshown in Fig. 2. It is tuned much the same as a flow loop.The setpoint of this controller establishes the internalpressure of the screen, which in turn determines thepressure drops across all screen system valves (feed,accepts, rejects, and dilution). Depending on process stocktemperature and screen throughput, the feed pressuresetpoint must be adjusted to prevent valve cavitation andthe runnability problems associated with it. In this respect,proper valve sizing for all anticipated production rates isa critical step in the overall process design of the screeningsystem. We commonly raise feed temperature to improvescreen runnability and increase throughput. Under theseconditions, the valves are even more prone to cavitationif improperly sized.

The feedstock consistency requires very tight controlbecause it does influence screen throughput and runnabil-ity, and it can also affect accept quality. A change as smallas 0.1% consistency can have a large impact on pulpthroughput and the frequency of screen blinding. Typically,an increase in feed consistency will increase screenthroughput; however, runnability (blinding) will becomea problem if the consistency gets too high. Consequently,there is an optimum feed consistency, and it changes aschip furnish and refiner plate conditions change. Finally,a large variability in feed consistency can upset the screenreject rate and cause a significant disturbance to efficiencyas well as pulp quality and freeness.

To adequately control feed consistency, trim dilution isrequired at the suction of the pump. Port Angeles usesan advanced control strategy with adaptive tuning (2), asshown in Fig. 2. The strategy regulates consistency bymanipulating the ratio of dilution flow to total feed flow.This not only provides uniform consistency but also ensuresthat the dilution water is turned off during an automatedflush cycle when the screen blinds (i.e., dilution flow isratioed to feed flow, which is turned off during the flushcycle). Good consistency regulation is also importantimmediately after a screen blind to aid in process recoveryand to avoid subsequent blinding. The adaptive controlssatisfy this need quite well.

Not all of the pulp dilution needs be done at the pumpsuction. Typically, some dilution water is added at the

100 November 1991 T"ppi journal

surge tank with the incoming stock. However, the amountof dilution at the pump should be enough to lower feedconsistency by at least a 0.4%to 0.5%range. A valve positioncontroller (VPC) can then be used, as shown in Fig. 2, tokeep the pump suction dilution valve at mid range. TheVPC is a proportional integral (PI) controller whichmanipulates dilution flow into the screen surge tank.

Application resultsThe pulp mill pressure screen project at Port Angelesexceeded all expectations for improved quality, efficiency,cost savings, and ease of process startup. The mill attributesthis success to the control system and to the improvedtechnology associated with pressure screen operation.

Quality and efficiencyPaper machine and press room efficiencies improvedsignificantly after startup. Fiber cutter (shive) problemswere reduced from the No.1 cause of breaks on the papermachines to the No.3 cause. Press room performanceexperienced a similar improvement. Shive relatedproblems on the presses were decreased from the No.1cause of web breaks to No.4 (0.2% of web breaks due toshives). Ray cells in the screen accepts also were reducedby 33%.

In addition to these benefits, the shive content in thepaper machine furnish was cut in half while the CSF wasincreased by 10 points on the RMP refiner line and by20 points on the TMP line. The additional long fiberimproved paper machine drainage and significantlyenhanced the physical strength properties of the sheet.

Cost savingsThe improvements in long-fiber fractions and papermachine drainage resulted in a reduction of kraft usageby at least 3% of virgin furnish (saving US$ 1 millionannually). The refiner production rates were alsoincreased by 25% on both lines. This accommodated theincreased paper machine efficiencies very well and pavedthe way for several machine speed-up projects. Theincreased production also reduced refiner power consump-tion by 12%, which currently saves us US$ 1.2 millionannually.

It was impossible to precisely determine the total dollarsavings associated with the pressure screen project.However, there is no doubt that the project payback wasless than two years.

Screen startupThe new screening system was brought on-line withouta major pulp mill down. The extensive instrumentationwas very revealing of the process and provided informationthat allowed operations to complete the startup within 12h. A process simulator developed in the Bailey Network-90 Distributed Control System (DCS) was used to trainoperators several months before startup. This alsocontributed to a successful project, making the transitionfrom simulation to on-line operation very smooth. Thesimulator also proved invaluable for check out of the DCSconfiguration.

One of the most important steps during startup was toget the feed consistency calibrated. Port Angeles converts

the 4-20-mA transmitter signal to a percent consistencyusing a linear regression equation in the DCS. Laboratorydata for the regression is collected by averaging theresults of triplicate samples taken at each of three dif-ferent consistencies.

Finally, smart transmitters were used where possibleand also contributed to the speedy startup. In particular,smart pressure transmitters could be reranged quicklyand on-line without the need for taking the process downor removing in-line instruments. The higher cost for thesetransmitters was more than paid for during startup, andthe smart field devices continue to pay dividends inpreventive maintenance.

Implementing the control system in Fig. 2 obviouslyrequires a DCS and extensive field instrumentation.However, the benefits associated with advanced DCSprojects typically justify the capital expense. This wasclearly the case for our screening project.0

Literature cited1. Nelson. G. L., 1975 International Mechanical Pulping

Conference Proceedings, TAPPI PRESS, Atlanta, supp!. p, 41.2. Dumdie, D. P., Tappi J. 71(7): p. 135(1988).3. Torborg, R. H., Pulp & Paper 54(3): 135(1980).4. Hawkes, S., 1990 Pulping Conference Proceedings, TAPPI

PRESS, Atlanta, p. 893.

Special recognition goes to Scott W. Kelly for his role in the project andto Rex Springer for his process knowledge. The author also wishes to thankBob Ketterling. Brad Blyrnyer. Guy Madison. Bruce Raymond. JohnHanson, Rob Buckingham. Susan Collicott, and other mill personnel whocontributed to the project. In addition. the author acknowledges MarkJackobson for the Bailey DCS configuration. and Fran Willard for ownerrepresentation.

Received for review May 3, 1991.

Accepted July 5. 1991.

Appendix

Automatic blind detection and flushAutomatic blind detection for a pressure screen can bedone in the DCS by monitoring the differential pressure(DP) across the screen basket. In a Hooper system, therotor provides a pumping action or pressure boostacross the basket. The DP in this case is defined asaccept pressure minus feed pressure. As this differen-tial drops from a positive to a negative value, blindingoccurs, and it becomes necessary to flush or clean thescreen.

Other screen manufacturers use foils, which do notboost pressure. The DP for these screens is defined asfeed pressure minus accept pressure. In this case, theDP increases as the screen blinds. It is much easier todetect blind conditions on a Hooper screen since the DPchange is so large (from +70 kPa to -55 kPa).

Once a screen blind has been detected, an automatedflush cycle can be completed as follows:

1. Closethe feed and accept valves and turn off the feedpump.

2. Open the reject and dilution valves 100%.3. Continue running the screen motor and flushing the

basket for about 20 s.4. Put all controllers back on automatic and start the

feed pump.

Some pressure screens do not use dilution water andmay require more than a 20-s cleaning cycle.

Automatic startup and shutdownThe Port Angeles screening system on both refiner linesuses primary (PI) and secondary (Sl) screens config-ured in a cascade arrangement. A small surge tank is

used between the PI and Sl screens. The automatedstartup sequence for this configuration is as follows:1. Ensure the cleaner system is running on white

water (permissive).2. Open the dilution valves to the secondary screen

100%.3. Open the seal water valves and satisfy any other

utility requirements for the secondary screen.4. Start the secondary screen motor and wait 20 s to

monitor the motor load and alarm any problems.5. Start the secondary screen feed pump.6. Put all secondary screen regulatory controls on

automatic and run on white water.7. Put the secondary screen supervisory control loop

in flowmode to maintain a constant reject flow rate.8. Repeat steps 2-7 for the primary screen.9. Start up the refiner mill and begin screening thick

stock.10. When both the feed and reject consistencies for PI

and Sl increase and enter the range of theirrespective consistency transmitters, switch thesupervisory control loops to reject rate mode orshive removal efficiency mode.

The automated shutdown sequence is simply a logicalreversal of the startup sequence. This ordering of stepswas designed for Hooper screens and may require mod-ifications when applied to other manufacturers' screens.

appendix continued next page

November 1991 TappiJournal 101

Miscellaneous project notesThere were several miscellaneous points of interestregarding the Port Angeles screening project that areworthy of mention. They relate to screen throughputand runnability, quality, operations, process design,and the controls:1. Primary screen throughput and runnability were

initially a problem at higher production rates onboththe TMP and RMP lines. These problems were solvedby raising the screen motor horsepower to increaserotor speed. This significantly improved screenthroughput without increasing slot size or compro-mising quality.

2. The original system piping was designed so that sec-ondary screen accepts could be combined withprimary screen accepts and sent forward to the clean-ers. This configuration helped ease initial throughputproblems without significantly affecting quality.

3. Consistency control at the feed pump suctionrequires a large dilution line to ensure the in-pipevelocity does not exceed 1.2 mls (3). In addition, theconsistency transmitter should be placed 10-15 sdownstream from the dilution line. This works wellwith the adaptive consistency controls and ensures

adequate rmxmg while keeping the control loopdeadtime to a minimum.

4. The automated startup sequence made it possible touse a very small surge tank between the PI and SIscreens. However, the control strategy in Fig. 2 canbe easily adapted to accommodate a directlyconnected PI-Sl as suggested by Hawkes (4).

5. As is the case with most advanced supervisorycontrol systems, ongoing DCS attention is necessaryto maintain peak system performance and quality.

6. Properly designed stand pipes with vacuum break-ers were used on all accept and reject lines to ensurea constant backpressure on the valves and to guardagainst cavitation.

7. Although upper and lower dilution water is neces-sary for proper Hooper screen operation, tightcontrol and flow instruments are not required. Wefound no correlation between dilution flow andaccept quality.

8. As expected, the screening system was found veryeffective in removing shives from the pulp while thecleaners efficiently removed chop.

102 November 1991 Tappi journal