Biermann Ch. Handbook of Pulping and Papermaking

Подождите немного. Документ загружается.

664 31. PROCESS CONTROL

the form of a current. One standard specifies that

the signal from a sensor be conditioned such that

a signal in the range of 4—20 mA be reported for

the range of the variable. The controller usually

converts this to a voltage by taking the signal

across a resistor (i.e., the IR drop). In pneumatic

systems, one common standard calls for a signal

output of 3—15 psi (20-100 kPa).

Response times of sensors

The time

response

of a sensor is a measure of

how the sensor responds to an actual change as a

function of

time.

For example, all thermocouples

have mass. This mass takes a certain period of

time to change temperature. Thus, when the

thermocouple at one temperature is immersed in a

liquid of a different temperature, the output of the

thermocouple will change with time until it finally

gives a reading that is within its normal error.

This might take a fraction of a second to many

seconds depending on the geometry of

the

thermo-

couple. The time constant of the sensor is usually

much lower than that of the system, so it should

not be a factor. The time constant, however, may

depend upon how the sensor is used; for example,

temperature measuring devices respond much

more quickly in a water bath (where heat transfer

is high) than in air (where heat transfer is low).

A first—order time response (such as the

charging of a capacitor, or, approximately, most

temperature sensing devices) follows the equation:

when

= 0, then: b{t) = bf[l - e-'"]

where b is the output value as a function of time,

r, initially, /, or finally, /. The symbol r is called

the time constant. If the initial output is zero then

after one time constant of time has passed the

sensor reaches only 63.2% of its final output.

After 2, 3, 4, 5, and 6 time constants, the sensor

reaches

86.5%,

95.0%, 98.2%,

99.3%,

and

99.75%

of its final value, respectively.

Second—order responses will cause an oscil-

lation in the output (as will be shown in Fig. 31-

4),

that is impossible to solve precisely, but can be

characterized mathematically.

Optical sensors

The principles of light have been presented in

Chapter 24, which (implicitly) provides some

details for measurement of

light.

Photoconductors

use a thin band (in a serpentine pattern to increase

the effective length) of a material that becomes

conductive when exposed to light of a minimum

frequency (energy level). Two common materials

are CdS (which requires light of wavelength 515

nm or less and has a time constant of about 100

ms) and CdSe (which requires light of wavelength

716 nm or less and has a time constant of about 10

ms).

Photovoltaic cells produce a voltage and

current that are a fimction of the light level inci-

dent to them. These have time constants on the

order of

to 100 /xs. For very low levels of light

(used in liquid scintillation counters to determine

radioactivity) photomultiplier tubes are used.

31.3 TEMPERATURE SENSORS

Resistance—temperature detectors

(RTDs)

The resistance of metals is a fimction of

temperature and increases with increasing tempera-

ture.

The relationship is fairly linear over a wide

temperature range, although second—order trans-

fer functions are often used. Resistance—temper-

ature detectors (RTDs) often use platinum or

nickel. The sensitivity of platinum is a function of

temperature, but is about 0.00385 QIWC (cali-

brated to the internationally accepted DIN 43760

standard.) (One common U.S. practice is to

configure RTDs at 0.00392 0/0/°C.) Nickel is

0.005/°C at room temperature. Thus, a 100.00

platinum RTD at 0°C will have a resistance of

100.385

at rC.

This small signal difference is conditioned by

using a Wheatstone bridge circuit which may

contain a

compensating

line to nullify the effect of

long leads to the RTD. The current used in RTD

devices must be low enough so that the RTD does

not alter the environment through self—heating.

RTD time constants are about 0.2 to 0.5 seconds

in flowing water to more than 10 times that time

in air. RTD devices can be used from —100 to

600°C (or 900°C with ceramic materials).

Thermistors

The resistance of semiconductors is also a

function of temperature, but semiconductor materi-

als (oxides of Ni, Mn, Fe, Co, Cu, Ti, etc.) can

be made whose resistance increases drastically

with increasing temperature. This fact is used to

TEMPERATURE SENSORS 665

Metal 1

Reference

Junction

Metal 2

Measuring

Junction "v'

Fig. 31-1. Diagram of a thermocouple where V

is the voltage generated by the circuit.

make thermistors. High sensitivities of 0.1

Q/Q/°C (or 30 mV/°C) are possible. Typically,

an uncoated thermistor has a time constant of 10

seconds in air and 1 second in air. Coating with

Teflon will increase these values by 2.5 x. They

are used in the range of —50° to 150°C. Therm-

istors are subject to self—heating and are often

used in Wheatstone bridge circuits.

Thermocouples

Thermocouples consist of a pair of wires

made of dissimilar metals joined together at one or

more junctions that generate a voltage when sub-

jected to a temperature gradient (Fig. 31-1). (This

phenomenon is known as the Seebeck effect, first

reported by Thomas Seebeck in 1821.) The sens-

itivity of type J thermocouples is about 50 /xV/°C.

Table 31-1 shows some characteristics of

thermocouples. Since low voltages are used,

signal amplification with differential amplifiers is

used. A reference junction must be used, but this

is easily accomplished in one of a number of

convenient methods. Noise reduction techniques

should be used to avoid interfering voltages that

may be otherwise generated in the system.

Type T and J thermocouples are inert and can

be used in oxidation, reduction, or vacuum envi-

ronments. Type J should never be exposed to

temperatures above 760°C because a magnetic

transformation can cause an irreversible change in

its calibration. Type K should not be used in

sulfiirous or reducing environments, nor in vacu-

ums for very long due to a calibration change.

Type E should not be used in reducing or vacuum

environments. Tungsten—rhenium thermocouples

can be used at temperatures up to 2760°C (but not

in oxidation

environments).

Thermocouples can be

made with time constants of 0.05 seconds (0.001

in. wire) and

second (0.005 in. wire) in still air.

Other temperature sensors

Integrated circuit temperature transducers,

which appeared in the 1980s, have similar advan-

tages and disadvantages of thermistors, but their

output is linear such as 1 ^A/K or 10 mV/K.

Bimetallic strips (where strips of two differ-

ent metals of differing coefficients of thermal

expansion are stacked and bonded together) have

been used in thermostats for years. They are

mechanically connected to a switch, such as a

mercury switch, to control heaters.

Optical pyrometry is used to measure the

temperature of objects. This relies on blackbody

radiation and is discussed in Chapter 24.

31.4 MECHANICAL SENSORS

Introduction

Mechanical sensors are used to measure

variables such as position, velocity, acceleration,

Table 31-1. Properties of the common types of thermocouples.

Thermocou-

ple type

Metal type 1

(positive)

Iron

Copper

ChromeF^

Chromel

90%Pt/10%Rh

87%Pt/13%Rh

Metal type 2

(negative)

Constatan'^'^

Constatan

AlumeF^

Constatan

Platinum

at 20°C

Color

code

Black

Blue

Yellow

Purple

Suggested

temp,

range

0-760°C 1

-200-370°C 1

-200-1260°C 1

-200-900°C 1

0-1480°C 1

0-1480°C

666 31. PROCESS CONTROL

force, pressure, levels (such as a liquid in a tank)

and flow. This section describes some sensors

used by the pulp and paper industry; more infor-

mation on these and other sensors is found in

Johnson (1993), which is the source of much of

the information in this section.

Electrical position sensors

Position sensors rely on a change of capaci-

tance, resistance, inductance, or reluctance . The

capacitance of a capacitor depends on the distance

between two plates or on how two plates overlap

A potentiometric displacement sensor uses a

wire of relatively high resistance with a wiper in

electrical contact with the wire. As the object

moves along the wire, the resistance of the circuit

changes. By measuring the resistance, the position

of the object along the wire is known.

Variable inductance is achieved by the use of

a ferromagnetic core in the shape of a rod that is

surrounded by a hollow coil of wire. As the core

enters the coil the inductance of the coil is

changed. With reluctance sensors, the magnetic

flux coupling between two or more coils is varied

by a ferromagnetic core.

Reluctance sensors, a type of transformer,

are the basis of the prevalent LVDT (linear vari-

able differential transformer) shown in Fig. 31-2.

The signal is easily conditioned to give a DC

voltage that in linearly proportional to position

over part of the range of motion of the core; the

sensitivity can be on the order of

0.001

mm move-

ment. It is possible to alter the design to measure

the angular displacement as the core is rotated.

Other position sensors

Ultrasound can be used as a position sensor

by measuring the time it takes a pulse of ultra-

sound to "echo" back to the transmitter. It is

similar in principle to a radar. This technique has

been used on early "self—focusing" cameras, some

consumer measuring devices used to measure

room sizes in houses, and other ranging devices.

Strain gauges and load cells

Strain is the change in length of a material

caused by a stress (force) on the object. For

example, tensile loads on a metal rod cause the

metal rod to elongate, while at the same time the

cross—sectional area decreases slightly. If the

force is limited to the elastic range of the metal

rod, the metal rod will return to its original shape

when the tensile force is released. A compression

force causes the metal bar to contract with a slight

increase in cross—sectional area.

Strain gauges use this phenomenon to mea-

sure strain by the change in resistance of metal.

As metal increases in length and decreases in

cross—sectional area, its resistance will increase;

however, the change in resistance is largely due to

the change in length. A metal strain gauge is

made of fine wire or foil that zigzags

(Fig.

31-3) to

give a useful resistance (often 120 Q, but common

values are 60—1000 Q). This design makes the

gauge sensitive to strain in only one direction.

The gauge factor is the relative change in

resistance divided by relative change in length

(i.e.,

the definition of

strain).

Metal strain gauges

have a gauge factor of about 2; thus a strain of

0.1%

will cause a 0.24 0 change in a 120 Q strain

gauge. The sensitivity of metal strain gauges can

be 10'^ strain. Specialized semiconductor strain

gauges based on silicon have gauge factors of —50

to —200, but these devices are nonlinear, making

them more difficult to use.

24Vdc 25mA

FROM

A BATTERY

OR OTHER DC

POWER SOURCE

SCHAEVITZ SERIES HR~DC LVDTs

MODELS FROM +/- 0.05" to +/- 5" FS

OUTPUT OV e CTR« +/- lOVfs

LyBLK"

’

SUPPLY

OSCIL

MAKES

AC FOR

XFRMR

DEMOD

MAKES

DC FOR

OUTPUT

:BLU

:GRN

CORE

POSITION

OUTPUT

Fig. 31-2. Diagram of a linear variable differential transformer (LVDT). Courtesy of H.R. Holbo.

MECHANICAL SENSORS 667

Of course, if the temperature of the metal

changes, this will have a confounding effect. A

second strain gauge is set perpendicular to the

direction of strain and counteracts the change in

temperature when they each make up separate legs

in a Wheatstone bridge (Fig. 31-3). Many other

configurations are possible to measure twist or

other parameters.

To make a load

cell,

one attaches one or

more sets (one to measure, one for temperature

compensation) to a material. For example, if the

load cells are attached to a support of a hopper,

then the weight in the hopper can be measured as

a corresponding strain. The ratio of the change in

weight to the change in strain depends on the size

of the support, with a large support giving a small

change in strain per change in weight. Thus, the

same strain gauge can be used to make highly

sensitive load cells or load cells capable of mea-

suring very large weights.

Pressure sensors

Pressure is force per area. If a fluid is at

rest, one refers to static pressure. If a fluid is in

motion, one refers to dynamic pressure, which is

a function of the motion of the fluid. The

relationship between gauge pressure is the differ-

ence between absolute pressure and atmospheric

pressure. For example, a flat tire on a automobile

is said to have a (gauge) pressure of zero, but it

actually has an absolute pressure of 14.7 psi. The

pressure ip) exerted by a liquid is a function of the

density of the liquid (p), the acceleration due to

gravity (g), and the height of the liquid {h),

P = Pgh

Pressure can be measuring by monitoring the

movement of a bellows with a LVDT. Strain

gauges can be mounted on diaphragms to measure

pressure; this method has been miniaturized onto

integrated circuits.

Pressure can be measured mechanically by

using a

Bourdon

tube, a coiled tube that is partial-

ly flattened (the pressure version of the coiled

bimetallic strip used to measure temperature).

Very low pressures, i.e., very high vacuums,

which indicates a gas phase, requiring specialized

techniques. The temperature of a heated metal

filament (or its resistance) depends on the gas

pressure, since the gas molecules can conduct heat

from the filament. Pressures below 10"^ can be

measured with ionization gauges.

Strain gauge Measuring circuitry for strain gauge

Fig. 31-3. Diagram of a strain gauge and its circuitry.

668 31. PROCESS CONTROL

Flow sensors

Flow rates of materials moved on conveyors

can be determined by weighing a portion of the

conveyor (dividing by the length of

the

section that

is weighed) and multiplying by the speed of the

conveyor to give units of weight (or mass) per unit

of time. This method can be approximated by

using an LVDT to measure the sag of the belt.

Flow rates in pipes are often measured

indirectly by measuring the pressure (or head) that

causes flow in a pipe. Flow in pipes can be

measured by placing a restriction in the pipe and

measuring the pressure drop (Bernoulli's principle)

in the restriction. Rotameters use an obstruction

in the flow path (a ball that rises in a tapered,

vertical column); the flow provides the force to

raise the ball against gravity. A vane that is

pushed upward by the fluid is an indicator of fluid

velocity if the vane's position is monitored.

Magnetic flow

meters

use an insulated portion

of pipe where two coils (on the outside top and

bottom of the pipe) generate a magnetic field. The

movement of charged particles in the fluid gener-

ates a voltage perpendicular to the magnetic field.

The voltage generated between two electrodes

placed outside the two sides of the pipe is an

indication of the velocity. Knowing the cross-

sectional area of the pipe section gives the volu-

metric flow rate. The fluid must be at least a

partial conductor (i.e., water with electrolytes).

Level sensors

The level of liquid, slurry, or solid in a tank

or hopper can be measured in several fashions. A

float can be used in conjunction with an LVDT

core or other displacement sensor. An ultrasonic

position sensor at the top of the vessel can be used

to detect the top of the material. A pressure

sensor can measure the pressure the material

exerts at the bottom of the vessel. Similarly, load

cells can be used. Occasionally the electrical

properties of the material (resistance or capaci-

tance) can be used to measure the level.

31.5 CONTROLLERS

Introduction

Controllers respond only to errors; there-

fore,

some error will be present in the system for

control to occur; without error, a controller

would not be required. Sophisticated controllers

will minimize this error by tuning the controller to

the system, a process to be described later.

Regulation is a process of keeping the error within

acceptable limits. In the case of our house it

might be 22°C ±1°C.

A transient is an upset caused by a change in

the setpoint or some other factor that influences

the response. In our heated house, what happens

if we change the temperature (adjust the thermo-

stat) to 26 °C? The transient response is the error

as a function of time, which is to be minimized.

Fig. 31-4 shows an example of a transient

response curve. One control strategy is to mini-

mize the area under the error time curve relative

to the new (or continuing if there is an upset in the

system) setpoint. A second control criterion is to

have the amplitude of each peak one—quarter of

that of

the

preceding peak in the

quarter amplitude

dampening system, which was the industry stan-

dard for many years since its description by J.P

Ziegler and N.B. Nichols in 1942.

Let us consider the four individual control

strategies available. The simplest form is the on

or off control, which has been modeled above in

the case of heating a house. In this case, there

will always be cycling with the temperature going

appreciably above the setpoint and then below the

setpoint. In the remaining control strategies, one

has the choice of varying the actuator from off to

various levels of on to fully on.

(An on—off system can be used to approxi-

mate various intensities. In microwave ovens, the

megatron is always on or always off, but various

cooking levels and defrosting are achieved by

turning the megatron on for an increasing percent-

age of a short time cycle. The time cycle is usual-

ly on the order of about 2 seconds. Industrial

electric heater controls often employ this technique

so that devices are not required to partially turn on

huge loads. These times are usually urmoticeable

relative to the time constant of the system.)

Controller action for a continuous actuator

may be proportional, integral, or derivative.

Many controllers use a combination of all three

techniques and are called PID controllers. We

will consider a benchtop stirred hot water bath

with a heating element whose output (heat) can be

controlled continuously from 0 watts to 1000 watts

as an example of a device to be controlled.

CONTROLLERS 669

200% prop, band

100% prop, band

50% prop, band

1 \ 1 1 1 1 r

50 -40 -30 -20 -10 0 10 20 30 40 50

Error, %

Fig. 31-5. A proportional diagram with three

proportional bands.

Proportional controllers

Proportional controllers give an output to the

actuator that is a multiple of (proportional) to the

error; they respond to the size of the error. The

multiple is the gain (= A output/A input). When

the error is zero (the measurement equals the

setpoint), the output is 50%. Fig. 31-5 shows

some examples with various proportional band

values. When the proportional band is too high,

a large error is required to make the necessary

change. When the proportional band is too low,

a small error causes a large change in the output

that overcompensates; this can send the system

oscillating out of control, much like feedback from

a public address system. Gain is related to the

proportional band expressed as a percentage:

Gain = 100/(Proportional Band, %)

Using our water bath as an example demon-

strates one problem with this methodology.

Suppose we want to heat our water at 30°C and

room temperature is 25

°C.

It is quite likely that

our 1000 watt heater operating at 50% output

(when the error is 0) will tend to heat the water

much above out setpoint of 30

°C.

Proportional

controllers operate with an

offset;

that is, they try

to maintain the measurement at a value that is

different from the setpoint. A high gain will

minimize the offset (but has the problems men-

tioned above); an infinite gain is essentially an

off/on control. If one adds the integral controller

this problem is solved.

Integral controllers

(An obsolete term for integral controllers is

reset.) When an error occurs, the integral mode

causes a continuous change in the output until the

error disappears. This mode responds to the size

of the error and the length of time the error

occurs. In the case of our water heater, when

trying to control the temperature to 30°C with a

proportional controller, the temperature will

30 40 50

Time, arbitrary units

Fig. 31-4. Response of a (hypothetical) controlled system to a change in setpoint from 70 to 80

units at time 0.

670 31. PROCESS CONTROL

consistently overshoot the mark. The integral

mode would step in and continuously lower the

offset until the temperature goes below 30°C,

which might be a power level of 20%. The

curves in Fig. 31-5 would be in effect, except that

they would intersect at 20% output for this device

operating at this temperature.

The effect of integral controllers is measured

in units of minutes per repeat or its reciprocal. A

high nimiber of minutes per repeat indicates a low

action by the integral controller.

Derivative controllers

Derivative controllers respond to the change

in error with time. If the absolute value of error

is rapidly decreasing and the setpoint is near, it is

likely that the setpoint will be reached and the

error will change sign and increase in the other

direction. The derivative mode is useful in pre-

venting this situation, especially in systems with

large time constants. Derivative response is

usually measured in minutes; the time is a mea-

sure of how far ahead the derivative -h proportion-

al action is ahead of the proportional action only

in an open loop; the higher this value, the greater

the derivative action..

Since a spike (a large, but brief response in

the measured value) would cause a large action by

the derivative mode, where no action is appro-

priate, the derivative mode is not always appropri-

ate.

This is true with signals that are inherently

noisy (such as flow rate measurements) or systems

with long dead times. Sometimes the derivative

mode is only used when a change in setpoint

occurs. The derivative mode us usually well

suited for temperature measurements and control.

PID controllers

Proportional—plus—integral—plus—derivative

controllers are described (Anderson, 1980) as:

•-2\e + — fedt + D —

^\ RJ dt

where

D =

e =

g =

m =

R =

t =

derivative time

error

gain

controller output (0-

integral time

time

-100%)

31.6 ACTUATORS

Once the sensors determine the status and

once the controller determines what action is

required, actuators carry out the appropriate

response. Actuators take the result of control

signals and cause a change on the control element.

They are not the control elements. The term

actuators seems to imply movement, but that is not

always the case. An actuator may be an electronic

relay (such as a silicon—controlled rectifier, SCR)

that switches power to the control element.

Solenoids convert an electrical current into a

relatively small back and forth motion. When a

coil is energized, a metal plunger is drawn into it.

A spring may be used to return the plunger to its

original position. Motors convert electric signals

into rotary motion. Motors are, in turn, used for

movement, pumping of fluids, valve positions, etc.

Motors can also be driven by air (pneumatic) or

hydraulic fluid. Pneumatic actuators convert air

pressure into movement. Actuators designed for

valves are discussed on page 655.

31.7 ANNOTATED BIBLIOGRAPHY

Anderson,

A., Instrumentation for Process

Measurement and Control, 3rd ed., Chilton

Co.,

Radnor, Penn., 1980, 498 p. This

work is useful for controller theory.

Johnson, C.,

Process

Control Instrumentation

Technology, 4th ed., Regents/Prentice—Hall

Englewood Cliffs, New Jersey, 1993, 592 p.

The work is up to date and a good overall

reference on the topic. It includes analog

and digital signal conditioning, thermal,

mechanical, and optical sensors, process con-

trol, actuators, and controllers. A later

edition is available.

Omega, The Temperature Handbook, The

Data

Acquisition

Systems, pH and

Conductiv-

ity, Pressure and Strain, Heaters, and Flow

and

Level,

This is a series of catalogs by the

Omega Technology

Co.

(Stamford, Connecti-

cut) of products they sell; however, these

substantial catalogs (available at no charge)

contain invaluable practical information and

tutorials. Every mill engineering library

should have a set of these books.

CORRUGATED CONTAINERS

32.1 INTRODUCTION

Information on printing, paper used in corru-

gated boxes, paper production, paper testing, and

other topics related to the manufacture of corrugat-

ed boxes can be found in the appropriate chapters

of this book. Many of the same techniques de-

scribed in other sections of the book apply to the

manufacture of corrugated

boxes

including slitting,

scoring, and crowning of press rolls.

Brief history

The production of corrugated paper was

patented by Albert L. Jones in 1871 as a packag-

ing material. It was used for protecting glass

containers from each other during shipping. In

1874,

Albert Long patented the use of corrugated

paper glued to another piece of paper (single-

faced corrugated board). The Shefton machine of

1895 could make single—face or double—face

corrugated board and operated below 10 ft/min.

In the early part of

this

century, railroads had

a virtual monopoly on long—distance transporta-

tion. Many of the railroads owned large areas of

timber and had ownership in companies that

manufactured wooden crates. The railroads,

therefore, charged a tariff for products transported

Fig. 32-1. Some examples of corrugated board

including single—face

(top),

double—faced (mid-

dle),

and double—walled.

with fiber (solid paperboard) boxes and corrugated

boxes.

In 1914, the Pridham decision forced

railroads to rescind this

tariff.

This marked the

beginning of the large—scale use of fiber and

corrugated boxes in transportation. The amount

used (in billions of square feet) was 6.6 in 1923,

44.3 in 1943, and 251 in 1979.

Box types

Fig. 32-1 shows diagrams of several types of

corrugated materials. Corrugated paper glued to

one sheet on linerboard is called single—face

board.

It is used primarily as a spacer inside

packages. If a second sheet of linerboard is glued

to the other side the product is known as double—

face board, and this product accounts for the

largest percentage of corrugated board products.

Double—wall corrugated

boxes

are made with

three layers of linerboard separated by two layers

of corrugating medium. (A small amount of

triple—wall corrugated boxes is produced.)

Double—wall corrugated boxes usually use two

different sizes of flutes, and the smaller flute is

usually toward the outside of the box for several

reasons. The smaller flute gives an even surface

for the outer liner which provides a good printing

surface; it also provides higher puncture resis-

tance. (A common configuration for double—wall

corrugated boxes is a B flute on top of a C flute.)

Flute types

The size of the flutes controls the properties

of the final board. Small flutes increase the

puncture resistance of boxes, while large flutes

enhance the stacking strength of boxes. Page 162

gives the basic flute types, but the situation has

increased in complexity. Flute C was originally

42 flutes per foot, but is now 39 flutes per foot.

The C flute has replaced the A flute for most

purposes since it requires less paper and gives

similar properties to the A flute.

The E flute increased in use in the late

1980s in corrugated boxes designed to compete in

the folding carton industry. The F flute was intro-

671

672 32. CORRUGATED CONTAINERS

duced in 1991 (about 0.03 in. high and about 150

flutes per inch) as a smaller alternative to the E

flute and folding carton stock. Larger flutes are

recent developments to allow double—face boxes

to replace double—wall boxes. The K flute is

0.25 in. thick. The S flute is a jumbo flute.

32.2 MANUFACTURE

Introduction to the corrugator

The first part of the operation is

accomplished by the single—facer section (Fig.

32-2).

First the corrugating medium is unwound

and conditioned to the proper moisture content and

temperature with a preheater and steam shower.

Preconditioning rolls, if used, are usually 12 to 18

in. in diameter. The corrugating medium must

have enough water to make it flexible for proper

flute formation. The adhesive usually has enough

water to gelatinize the starch.

Splicers for the corrugating medium and

linerboards attach the end of one roll of paper to

the beginning of the next roll so production can be

continued without having to stop the machine.

Automatic splicers are activated by the (small)

diameter of the roll in use or by detection of the

paper end. Modern splicers have an overlap of

only a few inches, which causes little problem in

the production process or the final product.

The corrugating medium is sent between a set

of corrugating rolls (12 to 24 in. in diameter) to

flute the medium. After the corrugating nip,

adhesive is applied to the tips of the flutes on one

side of the sheet. Linerboard (which is preheated,

but not usually premoistened) is pressed against

the corrugating medium by the pressure roll (with

pressures of 300—500 pounds per linear inch)

while it is still supported by one of

the

corrugating

rolls;

the pressure roll, or single—facer, is

16—18 in. in diameter and heated to 177°C

(350°F). The initial bond that is formed is called

the green bond. Control of the moisture content

of the linerboard prevents curl of the board prod-

uct during its production. In some (older) corru-

gators, thin metal plates called fingers hold the

medium against the corrugating roll to maintain its

shape while the liner is applied to the fluted

corrugating medium. Most designs are now

fingerless and use a vacuum or air positive pres-

sure.

... steam

/|\ shower

Medium

from

conditioning

roU

Liner from

preheater rolls

Fig. 32-2. Diagram of a single—facer.

MANUFACTURE 673

The single—face product then goes to the

double—facer (alias double—backer) section to

apply a second layer of linerboard on the opposite

side of the fluted corrugating medium. Starch

adhesive is applied to the exposed tips of the fluted

corrugating medium, the liner is put in contact

with the medium, and the combination is placed

under a belt that travels over steam chests; the nip

at this point is extremely dangerous as it can drag

a person's arm into it with severe consequences.

(From here on out the board must be handled in a

flat condition since wrapping it around rolls will

cause permanent deformation.) The pressure

applied in the hotplate section (or double-backer)

cannot be so high as to crush the flutes. The

product goes through the hot plate section for

several seconds to allow the starch adhesive to set.

It goes through a cooling section where it is pulled

by the addition of a lower belt. The later section

is often called the traction section because it pulls

the web through as in the Langstron.

The adhesive application system had used

tapered speed drives at one time, but developments

in drive technology have changed this. For

example, the speed of the adhesive applicator

surface might be 20 ft/min slower than the paper

speed regardless of the paper's speed in order to

cause some wiping action; the more wiping

action, the more adhesive transferred. When the

machine is not moving, the applicator roll is

separated from the paper. The distance between

the starch applicator roll and the doctor blade is

about 0.15 mm (0.006 in.).

Miscellaneous aspects of

corrugators

Recent corrugators are 2500 mm (99 in.)

wide and operate at 6 m/s (1000 ft/min). The

corrugating medium may initially travel at 8 m/s

if the take—up ratio is 1.33 ( or 8 m/s divided by

6 m/s) as for a B flute. An 85" corrugator will

use 4—5 tons of steam per hour requiring a mini-

mal pressure of 175 psig.

Corrugating

adhesive

Initially starch (one part in about 15 parts of

water) was used as the adhesive, but it took a long

time to set, which caused numerous problems for

box makers. By 1920, sodium silicates (water-

glass) were the principal adhesives used in box-

making. This caustic material was strong but

brittle and had some major problems in its use.

In 1936 J. V. Bauer patented the use of starch

adhesive that was cooked on the box during

assembly (which, of course, requires that the box

be heated to set the starch adhesive). By 1960

starch regained its position as the principal adhe-

sive for boxmaking. Corn is the most common

source of the starch. The use of starch

just over

2 lb/1000 ft2 of double—faced board.

In the Stein—Hall process, cooked starch

(about 15—17% of the total starch) is used to

make a suitable formulation for the uncooked

starch. Sodium hydroxide is used in the formula-

tion to lower the temperature at which the starch

will gelatinize on the box; corn starch will gelati-

nize around 70°C, but NaOH lowers this to about

65°C for making corrugated boxes. Borax is used

to increase the tackiness of the adhesive and act as

a buffer for the NaOH. A preservative must be

used to prevent microorganisms from attacking the

starch because they can quickly alter the viscosity

of the formulation. Other materials in addition to

cooked starch can be used to make a suitable

formulation with the uncooked starch, especially in

water—resistant adhesives.

There must be adequate water in the starch

formulation and sufficient heating of the formula-

tion on the corrugator to get a suitable bond. The

cooking of the starch and the formation of the

bond take place within a few one—hundredths of

a second on fast—moving single—facer machines

(where high pressure between the corrugated sheet

supported by the corrugator and the linerboard are

possible). On double—backer machines, the time

for cooking must be several seconds since much

lower pressures are used, so a higher viscosity

starch solution can be used.

A recent practice at some box plants is to use

flexographic wash water as the water in the starch

adhesive. One should know exactly what ink

components are in this wash water and how they

will affect the system. DiDominicis and Klein

(1983) give much information on the starch—based

adhesive used in corrugated boxes.

Wiite boxes

White boxes can be made using a thin layer

of white pulp on the top liner (applied with a