Reduction Of Twist On Board Machines

John Shakespeare
Senior Scientist, Honeywell
PL-1001, Viestikatu 1-3
70600 Kuopio, Finland
Tel: +358-207522824
E-mail: [email protected]

ABSTRACT
This paper presents some of the benefits of measuring a proxy for twist on multilayer board machines. In particular, it
presents the improvement possible in twist for machines making liquid packaging and folding boxboard using FotoFiber
sensors. The proxy for twist is defined as the difference in fiber orientation angles between the top layer and the bottom layer.
Measurements of the surface fiber orientation are required on both surfaces, which provide an angle and an anisotropy as
profiles. Although most paper makers know that the amplitude of a twist profile can be controlled (to some extent) by
changing the jet to wire ratio setpoint, some do not know that twist profiles can be controlled quite precisely by the top and
bottom slice lips.
A closed-loop CD control for the twist proxy is also possible, if the mill has retained motorized slice lip actuators with
positional feedback on the top and bottom headbox slice lip. Provided one of the headboxes is a dilution headbox, the fiber
orientation can be controlled independent of the basis weight or caliper to a considerable extent. Otherwise, the controls are
coupled quite tightly. If the mill has slice actuators, but they are not motorized, then manual control is possible. Results from
both of these are presented.
A consulting service is also described, which is especially useful if the mill has not retained motorized slice lip actuators. In
this, a mill has its twist profiles optimized, possibly including basis weight or a whiteness profile, and the operators or
engineers are taught to do so.

INTRODUCTION
Twist is a problem on some grades of paperboard. It may be a problem without a twist measurement or even without
knowledge that it is an issue [1]. Twist is caused by the tendency for paperboard to shrink when dried together with the non-
zero anisotropy of the paperboard. We may depict this shrinkage for a single layer as:

MD MD MD
original circular disk

disk after shrinkage

CD CD CD

anisotropy = 0 low anisotropy high anisotropy

Figure 1. Anisotropy of single layer of paper.

It is possible to approach the anisotropy=0 case only with hand-made paper sheets. On machine-made paper, the fiber
orientation anisotropy (hereinafter, FO anisotropy) is broadly in the machine direction, as depicted.

Twist
Paperboard is generally made in multiple layers which are spliced together. First, a definition of what twist is, and a proxy for
twist [2], [3]. The effect of shrinkage can be different between the surfaces, even with identical FO anisotropies:

MD
top layer
axis of greatest
shrinkage, top
CD

saddle shape
axis of greatest (twisted disk)
layer shrinkage
MD

CD

axis of greatest
shrinkage, bottom
bottom layer

Figure 2. Effect of difference in angle between layers - Twist.

The usual I-beam nature of paper stiffness causes the difference to be dominated by the surface layers. The fact that filler
layers are typically lower grade fibers helps, even if the filler layers significantly outweigh the surface layers. A proxy for
this is the difference in fiber orientation angles (hereinafter, FO angle). All that is required for twist to occur is that this proxy
is significantly nonzero.
Generally, twist is defined in terms of a 50% relative humidity test at 23°C:
∆𝜃
𝑇𝑤𝑖𝑠𝑡 = 2
∆𝑑
where ∆𝜃 is the change in angle in radians between CD chords separated by ∆𝑑 meters in the machine direction, and has
units of 1/m. Note that:
∆𝐿 100
𝛽=
𝐿 ∆𝑀
Where 𝐿 represents the length of a sheet, ∆𝐿 is the change in length and ∆𝑀 is the change in moisture percentage between
two states. 𝛽 is nearly a constant for each grade, but differs for the MD and CD directions. It is typically three times as large
for CD as for MD, and about ten times as large for thickness as for CD.

Curl
Curl is a related effect. This is caused by the anisotropies being different on the two surfaces. Typically, machine direction
curl is handled in the dryers (by limiting the rate of final drying). Cross machine direction curl is not normally a problem for
liquid packaging, but may be an issue for other paperboard mills. Folding boxboard mills usually are aware of the curl, and
may control it by setting rush-drag values appropriately. It is caused by a difference in the surface layer anisotropy, even if
the angles are identical:

MD
top layer
more shrinkage
on top layer
CD

cylindrical shape
(curled disk)

MD

CD

less shrinkage
on bottom layer
bottom layer

Figure 3. Effect of a difference in anisotropy between layers - CD Curl.

A proxy for the CD curl is the difference in FO anitotropies between the layers. Generally, for folding boxboard one wants
the package to be “bellied outwards” for ease of packing items (pencils, cigarettes, etc.) into it and for ease of the consumer
extracting them afterwards. In other words, there should be less shrinkage on the outside of the box.
Generally, curl is defined as the inverse of radius of curvature for a 50% relative humidity test at 23°C:
1
𝐶𝑢𝑟𝑙 =
𝑟
where 𝑟 is the radius of curvature. It has units of 1/m. Curl will tend to zero for flat sheets, and become large for curled
sheets.
Both twist and curl are combined in most paperboards, as twist-curl. In practice, one is more important than the other,
depending on the grade being produced. The type of fiber used may also have an impact – fines may be less rigid than long
fibers, but may shrink less. Among others, beating fibers of a particular quality has a greater effect on hardwoods than on
softwoods.
Curl is the tendency for a sheet to curl from planar into a cylindrical shape when the humidity is changed. The direction of
curl is generally either in the machine direction or the cross machine direction. CD curl is a form of curl in which the curl
direction is across the sheet. In other words, the front and back edges of a sample are deflected away from the center line of
the sample, and it is usually indicative of a difference in anisotropy between the two surfaces of the sheet. MD curl is a form
of curl in which the curl direction is along the sheet. In other words, the upstream and downstream edges of a sample are
deflected away from the middle of the sample, and it is usually indicative of a high rate of drying in the last dryer sections.
Twist is the tendency of a sheet to deform from planar into a saddle shape when the humidity is changed. It is usually
indicative of a difference in fiber orientation angles between the two surfaces of the sheet.
𝑇𝑤𝑖𝑠𝑡 𝑝𝑟𝑜𝑥𝑦 = ∆𝐹𝑂 𝑎𝑛𝑔𝑙𝑒
𝐶𝑢𝑟𝑙 𝑝𝑟𝑜𝑥𝑦 = ∆𝐹𝑂 𝑎𝑛𝑖𝑠𝑜𝑡𝑟𝑜𝑝𝑦
An example of the twist proxy is in Figure 4. This was a folding boxboard machine which claimed to have “no probem with
twist”, but on inspection had a major problem with twist despite – or possibly because of – getting their headbox slice lips
optimized by the headbox manufacturer the previous year. Such large amplitudes in the twist proxy or such a range of
profiles are rarely seen.

Figure 4. Twist proxy for a folding boxboard mill.

Note On Terminology
The terms in use by mills often involve “flatness” and “warp”, among others [4]. Flatness and warp are both useful in
describing the out-of-plane deformation of paperboard, but are ambiguous with respect to the process cause of that
deformation. These terms may refer to twist and/or MD curl and/or CD curl. For example, any absence of planarity due to
any cause may be described as warp.
These terms may be useful for describing product issues, but are deprecated for discussion of process issues, as warp is often
caused by MD curl instead of by CD curl or by twist. The appropriate control action for MD curl is inappropriate for CD curl,
for example. A measurement of FO on both surfaces may indicate which issue it is.
MANUAL CONTROL
Twist can be controlled manually, but a few things have to be correct. Also, the slice lips must be equipped with dials which
must be (at least approximately) calibrated.
First, the zero point for rush-drag must be established. This may need to be done for each grade being run. Given the zero
point, it is instructive to look at past records of rush-drag values (unless a headbox was changed). Next, the FO angle must be
visualized. The positive direction can be either towards the front of the machine or towards the back. It flips for any
coordinate transformation (front-back, top-bottom, upstream-downstream), and while an upstream-downstream and front-
back flip in combination preserve the sense of the FO angle, they do not preserve its profile. The twist proxy is defined in
terms of the difference in FO angles between the paperboard surfaces, so the top-bottom alignment is important. The third
item is a response model for the upper and lower slice lips. Given a true rush-drag value and a visualization of the FO angle,
it is possible to manually control the twist proxy. One must remember the definition of the twist proxy: top-bottom or
bottom-top in FO angles.
In some headboxes, it is also possible to control edge flows which influence the twist (or FO angle) directly. However, there
is a lack of useful models for these, which depend nonlinearly on total slice opening, machine speed, presence of vanes (and
vane tip geometry), geometry of headbox nozzle, etc.
The upstream-downstream flip preserves the FO anisotropy, but others do not. Having front-back reversed will change the
profile, while having top-bottom reversed may lead to entirely inappropriate results.

Broad Twist Issue

Generally, one starts with persistent broad twist issues and proceeds to narrower issues, keeping in mind the actuator
mapping and the spectral limits for twist control. The issue must be persistent, as a considerable amount of time is used for
each control action. The following manual procedure can be used for the slice lip:
1. At the center of the region to modify, leave the slice actuator in its current position.
2. Move the actuators progressively in greater amounts downwards from the center to one edge of the region to
modify.
3. Move the actuators progressively in greater amounts upwards from the center to the other edge of the region to
modify.
4. Beyond the edges of the region to modify, move the actuators progressively in smaller amounts, until reaching a
point where the movement amount would be negligibly small.
This is shown schematically on the left of Figure 5.
 Figure 5. Correcting a broad twist issue. Left: isolated issue. Right: opposite issues.
Often, a persistent positive twist issue will occur with a persistent negative twist issue elsewhere in the sheet. In this case, all
the actuators between the regions can move by a similar amount, as on the right of Figure 5.
It is possible to automate these actions, as otherwise they are very slow to make. The effect on the dry weight profile is not
shown but should be taken into account, unless one has a dilution headbox.

Edge Twist Issue

This is appropriate only if controllable edge flows exist (edge bleed valves should be used very carefully). If the twist is near
an edge of the sheet, and the edge flows can be moved, then one should make a change to both flow controllers at the front
(or back) of the headbox to reduce it. In general, one controller should be increased and the other reduced in value as in
Figure 6. Be aware that the response may not be as large as one hoped for, or as deeply penetrating into the sheet. Responses
to edge flows are especially problematic if one headbox has more penetrating responses than the other. They may be invisible
if they are trimmed off before the scanner or the scan is inadequate. They may be very narrow if the machine has a particular
headbox configuration.

Figure 6. Twist issue near edge. Top: upper headbox. Bottom: lower headbox. Left: deep. Right: shallow.

PROCESS EFFECTS
The effects of the process are implicit in the above, and can be divided into:
 Rush-drag
 Slice lip responses
  Edge flow responses (depends on headbox)
 Other (depends on equipment)

Rush-drag
Rush-drag is set to give the desired properties of paperboard, and most makers of paperboard know the appropriate values for
each grade they make. However, the values are often in error by perhaps several meters per minute (in one case, it was off by
about 50m/min on a wire speed of 800m/min). Since rush-drag values are generally in the range of ±10m/min, this can be a
significant error.
It is possible to use the fiber orientation measurement directly in calibrating the zero point for rush-drag. Essentially, one
does the following:
1. Put the scanner with fiber orientation measurement sensors into a fixed position. This should be near the middle of
the sheet (front-center to back-center), but should not be a “problem” spot.
2. Step the headbox pressure in increments over a range which is expected to extend from drag to rush, and includes the
nominal operating pressure. Record the surface anisotropy at each setting when steady state is reached. Allow for the
delay in reaching the scanner, if necessary.
3. Plot the surface anisotropy versus headbox pressure or nominal jet speed. The actual zero point (jet and wire are the
same speed) occurs when the surface anisotropy is a minimum. The calculated jet speed should be corrected by an
offset to match this.
It is possible to step the two surface headboxes together, so this procedure is faster and less labor-intensive than others.
Generally, the method is shown to mills, and they are keen to use it as soon as it is shown to agree with other methods. They
are keener still, when the other methods are shown to be slightly in error (unless made more laborious).

Figure 7. Sheet width and FO anisotropy vs nominal rush-drag at a liquid packaging mill.
In the case shown in a liquid packaging mill in Figure 7, all headboxes were kept at constant jet speeds, and the wire speed
was changed instead. The two FO anisotropy sensors (in fixed position at a typical location) disagreed by a couple of meters
per minute on when their respective layers were at the zero point for rush-drag. The sheet width measurement was an
aggregate of the top and bottom layers (and the middle layers) which are caused by rush-drag of the various forming units.
The zero points in Figure 7 were about ±0 and about -2 meters per minute on the top and bottom, respectively. The sheet
width measurement – which was standalone – indicated about -1 meters per minute. The discrepancy was small, but
significant for a mill where a few meters per minute of rush or drag is large.
Alternatively, the same can be done at a more leisurely pace (if production schedules allow), using averages of the surface
anisotropies. In a folding boxboard mill depicted in Figure 8, there was little point in making the sheet width measurement, as
the rush-drag values were quite different.
 Figure 8. FO anisotropy vs nominal rush-drag at a folding boxboard mill.

Slice Lip Responses

This is the primary means of controlling twist in paperboard machines. Broadly speaking, there are two patterns which are
used in manual control of the slice lip. One is the basis weight type (an isolated pulse in CD), the other being the FO type (a
pulse comprising a negative, a zero, and a positive in CD). These are generally made, the response is seen, then they are
inverted, the response is seen, and they are returned to zero.
Both can be seen in a liquid packaging mill. The lower headbox was considerably steadier than the upper headbox, so a basis
weight type pulse was possible for this headbox, as seen in Figure 9. Note that the basis weight type pulse gives an
asymmetric response, as detailed in [5].

Figure 9. Basis weight type CD pulse. Response in bottom FO angle at a liquid packaging mill.
The FO type CD pulse was required in the upper headbox, which was run closer to its zero point for rush-drag. Note that the
FO type pulse gives a symmetric response in FO angle, as seen in Figure 10, depite being inherently asymmetric.
 Figure 10. FO type CD pulse. Response in top FO angle at a liquid packaging mill.
The basis weight was on control using the dilution headbox in the middle layer. Note that the statistical methods are outlined
in [6] in which normal PCA is used, and [7] in which FDA is used.

Edge Flow Responses

While it was stated that there is a lack of useful models for FO angle responses to edge flows, they are repeatable under some
situations (especially if the same process operation is repeated). They can thereby be used.

Figure 11. Top edge FO angle responses on linerboard at 125gsm.

For the linerboard machine in Figure 11, the top headbox edge responses are usable because the top headbox always runs
about the same, contributing about 60-70gsm to the weight and with a slice opening of below 20mm. However, the bottom
headbox is more variable, contributing 60-110gsm to the weight, and with a slice opening which varies from 25-40mm.
 Figure 12. Bottom edge FO angle responses on linerboard at 135gsm.

Figure 13. Bottom edge FO angle responses on linerboard at 170gsm.

The bottom headbox has shifted to a shallower edge response than the top headbox (which retains the same parameters
throughout) at 170gsm. The slice opening for the lower headbox can be different on every grade run. It is substantially
different between the headboxes for 135gsm so that its response is already mixed. Moreover, at 135gsm the bottom headbox
has different responses between the front and back edge flows.

Other
There are many influences on twist in a paperboard machine, and the most common are treated above. Other effects include:
 Changes in the rush-drag across the jet. Typically, these are jet speed related, and are usually less than 1 m/min.
However, they can be significant if the rush-drag is only a few m/min.
 Header imbalance, especially without a stilling chamber in the headboxes. This may affect the angles and anisotropy
on one side more than the other, depending on the header optimization.
 Header optimization issues. This can affect the angle and anisotropy if the total thoughput of the headbox is
significantly different from what its header was optimized for.
 Headbox pressure, which may cause an improperly designed headbox (or one operated beyond its original design) to
belly outwards on pressure increase or inwards on pressure reduction.
  Tilt of the slice lip. In particular, the beam may be deflected differently in L/b ratio or in total slice opening.
 Thermal compensation, which may cause the headbox slice lip to deflect upwards or downwards in the middle.
However, gradients in the temperature can result in one side being affected more than the other.
 Partial plugging of the headbox. This should never happen, of course.
Most of these items can be confused with one another. An example of the FO angle response to thermal compensation is in
Figure 14.

Figure 14. Effect on FO angle of thermal compensation of bottom headbox of a liquid packaging mill.

Difficulty
Some of the difficulty with manual control can be gleaned from a result in bottom FO angle (on the left) and the bottom slice
lip (on the right) of Figure 15, which describes a single slice action at a liquid packaging mill. The action took more than 20
minutes to make, and the headbox was at a true rush-drag of -1 to -2 meters per minute.

Figure 15. Effect of a manual control action on the bottom headbox of a liquid packaging mill.
AUTOMATIC CONTROL
Automatic control is also supported for twist [8], [9], largely based on [10]. Exactly the same things must be correct for
automatic control as for manual control, but the actuators for the slice lips must be controllable from the CD control system
(i.e. have positional feedback and be remotely controllable).
At a folding boxboard mill, some tests were done for the jet-to-wire ratio.

Figure 16. Jet-to-wire ratio at a folding boxboard mill.

As a result of the data plotted in Figure 16, it was decided to leave the jet-to-wire ratio unchanged for the top layer, but to
make an adjustment of 0.02 (equivalent to 12m/min) in the bottom layer.
Then tests were done to measure the responses (both headboxes received the FO type of bump in this case), with results
plotted in Figure 17.

Figure 17. Responses for closed-loop control of FO angle at a folding boxboard mill.
The twist proxy was progressively reduced, so that mostly uncontrollable residual variations remained. The average twist
measured in the mill laboratory is 0.11/m when on control, while the uncontrolled twist (after control was on) is 0.32/m. An
example of an instantaneous profile is in Figure 18, in which largely uncontrollable disturbances remain.
 Figure 18. Twist proxy at a folding boxboard mill.

CONCLUSIONS
A measurement for fiber orientation has been described [11]. If two such measurements are on opposite surfaces of the sheet
on the same scanner, then twist and curl can also be measured. Twist is a complex phenomenon, mostly involving the FO
angles on the surface layers. It is not intuitively obvious how to control twist, as it is subject to inversion in changing between
rush and drag. Control of twist depends on the individual true rush-drag states of the headboxes. Curl involves the FO
anisotropy on the surface layers, and is somewhat better understood.
A consulting service exists for twist and other fiber orientation issues, based on the sensors in [11]. Some paperboard mills
are very conscious of their issues with twist and conscientious regarding suppressing those issues. However, at least some of
the mills which claim not to have an issue with twist are in denial. On almost every mill that we have seen, there are issues
with twist.
Use of consultants is generally a good idea, as mills can be thin in experience (and often are), especially in twist. Moreover,
the application of experience takes time and effort, which are both in short supply. Consultants may have a deep knowledge
of fiber orientation and related issues such as twist and curl.

REFERENCES
[1] Kirwan, M. (ed.) Paper and Paperboard Packaging Technology. Blackwell Publishing, 2005.
[2] Sasaki, T. and Sano, H. and Yamamoto, J. and Ono, K. and Ochi, T. “On-Line Fiber Orientation Measurement and
Control”. Pulp & Paper Canada, T32-T36, March/April 2010.
[3] Ketoja, J. “Dimensional Stability”. In Niskanen, K. (ed.) Paper Physics, 2nd ed. Paperi ja Puu, Helsinki. (2008)
[4] Paavola, A., “Improving Board Quality and Flatness”, Pulp, Paper, & Logistics, 2(12): (2012).
[5] Shakespeare, J. “Tutorial: Fibre Orientation Angle Profiles – Process Principles and Cross-Machine Control”. In
Proceedings of TAPPI Process Control Electrical and Information ’98 (16-19 March 1998, Vancouver Canada), pages
593-636, March 1998.
[6] Basilevsky, A. Statistical Factor Analysis and Related Methods: Theory and Applications, 2nd ed., Wiley, New York
NY, 2008.
[7] Shakespeare, J. Identification and Control of Cross-machine Profiles in Paper Machines: A Functional Penalty
Approach. PhD thesis, Tampere University of Technology (Tampere Finland), December 2001.
[8] Chu, D. and Gheorghe, C. and Backstrom, J., “Advanced Multivariable Fiber Orientation Control and Twist
Optimization”, In Proceedings of PaperCon 2011 Conference (Covington Kentucky), pages 2402-2411, 2011.
[9] Gheorghe, C. and Shakespeare, J. and Jantti, J.-P. “Performance and Benefits of Fibre Orientation CD Control”. Journal
of Science & Technology for Forest Products and Processes, 5(6) 17-23, 2016.
[10] Backström, J. and Gheorghe, C. and Stewart, G. and Vyse, R. “Constrained Model Predictive Control for Cross
Directional Multi-Array Processes”. Pulp & Paper Canada, T128-T132, May 2001.
[11] Shakespeare, J. and Paavola, A. “Online Measurement of Surface Fiber Orientation”. In Proceedings of the Papermaking
Research Symposium 2009 (1-4 June 2009, Kuopio Finland), June 2009.
Reduction of Twist
on Board Machines

John Shakespeare, Ph.D.

Contents

• What are Fiber Orientation, Twist, Curl

• How to Control Edge Flows and the Slice Lip
• How to Control Twist
• Twist on Liquid Packaging and Folding Boxboard
• Manual vs Automatic Control
Online Camera-based Fiber Orientation
Measurement
• Measures surface Fiber Orientation (FO) distributions
• Determined from image measurement, acquired at 25Hz
• Characterized as FO Angle, FO Anisotropy, gray mean, gray sigma
(for contrast), goodness of fit, etc.
• Every 1° evaluated for FO angle, FO anisotropy
• Fitted curve, and residual
• Profiles (of all characterized quantities) available on sheet
• On single layers:
• Comparable to TSO measurement
• Possibly, a single FotoFiber will do (if the grade is light enough)
• On multiple layers:
• Not comparable to TSO measurement
• Dual FotoFibers are always needed
Fiber Orientation Comparison with top TSO
(when layers are split)
FotoFiber vs. Lab TSO, top layer

25
20
FotoFiber
15 Lab TSO
FO angle [degree]

10
5
0
0 0,2 0,4 0,6 0,8 1
-5
-10
-15
-20
-25
Fractional position [0=front, 1=back]

• Peeled layer analysis by TSO (top & bottom)

• CD strips separated by about 10m in MD
Fiber Orientation Comparison with bottom TSO
(when layers are split)
FotoFiber vs. Lab TSO, bottom layer

25
20
15
FO angle [degree]

10
5
0
-5 0 0,2 0,4 0,6 0,8 1

-10 FotoFiber
-15 Lab TSO

-20
-25
Fractional position [0=front, 1=back]

• Peeled layer analysis by TSO (top & bottom)

• CD strips separated by about 10m in MD
Fiber Orientation

• The FO Angle is the average angle in the distribution. Range -90

to +90 degrees.
• The FO Anisotropy is the extent to which angles are aligned with
the average angle. Range 0 (essentially random) to 1 (all fibers
aligned at average angle).
• An anisotropy of 1 is impossible, of course, and an anisotropy of
approaching 0 is most unlikely. In practice, the range of values
seen is 0.05 to 0.25
• Anisotropy depends on the rush-drag value
Angles are more significant at high anisotropy
 Single-layer Fiber Orientation
CD CD CD
original circular disk
disk after shrinkage

MD MD MD

anisotropy = 0 low anisotropy high anisotropy

CD CD
original circular disk
disk after shrinkage

MD MD

small angle large angle

What is twist?
CD

top layer axis of greatest

shrinkage, top

MD

saddle shape
axis of greatest (twisted disk)
layer shrinkage
CD

MD

axis of greatest
shrinkage, bottom

bottom layer
Twist proxy

• Twist can be represented by a proxy, such as

Bottom fiber angle – Top fiber angle

• The proxy can be augmented by top and bottom anisotropies,

but on a case-by-case basis.
• Twist is (mostly) unaffected if anisotropies are similar.
• Twist is dominated by one surface if the anisotropy is near zero on
the other.
What is curl?
CD

top layer more shrinkage

on top layer

MD

cylindrical shape
(curled disk)

CD

MD

less shrinkage
on bottom layer

bottom layer
MD curl and CD curl

• Curl is nearly always dominated by either

• MD curl in which the MD edges of a rectangular strip are at a
different height to its interior (i.e. the curl axis is across the machine)
• CD curl in which the CD edges of a rectangular strip are at a different
height to its interior (i.e. the curl axis is along the machine)
• MD curl is usually caused by too high a drying rate
• Especially in the later stages of drying
• In single-sided dryers
• CD curl is caused by a difference in anisotropy
How to Control the FO angle
slice lip

jet flow

dry weight

jet angle

fiber angle
large rush

fiber angle
small rush
fiber angle
small drag

fiber angle
large drag
How to Control the Edge Flows

Using edge flows (if appropriate and response is deep enough)

Jet angle profile Jet angle profile

Edge flow responses depend on state

Jet angle profile

Jet angle profile

How to Control the Slice Lip (1)

A slope in the whole slice lip

Slice lip profile & jet Fiber orientation angle profile
rush
slice lip

jet drag
How to Control the Slice Lip (2)

A shape in the whole slice lip

Shrinkage profile Dry weight profile

at reel

on wire

Slice lip profile & jet

slice lip

jet

Fiber orientation angle profile

rush

drag
How to Control the Twist

Upper headbox slice lip & jet Upper headbox slice lip & jet
correction change in region 1 region 2
region slice lip

jet jet

Lower headbox slice lip & jet Lower headbox slice lip & jet
change in correction region 1 region 2
slice lip region

jet jet

Change in twist, upper-lower jet angle Change in twist, upper-lower jet angle

+ + +
- - -
-
Folding Boxboard Mill, Twist proxy
Folding Boxboard Mill, Top FO angles
Folding Boxboard Mill, Top FO anisotropy
Folding Boxboard Mill, Bottom FO angles
Folding Boxboard Mill, Bottom FO anisotropy
Folding Boxboard Mill
Neutral point for rush-drag – FO anisotropy view
Folding Boxboard Mill
Neutral point for rush-drag – FO angle view
Other effects (1)

• Header imbalance
• This is rare on some designs of headbox, and those designs are
being discontinued. So it is becoming more common.
• Header shape optimization
• The shape optimization may be inappropriate for the current
operating regime. The header shape may not correspond to the
throughput of the headbox.
• Insufficient rigidity in the headbox slice nozzle
• See next slide.
• Thermal compensation of the headbox slice
• See next slide.
Other effects (2)

• Thermal compensation of the headbox slice

• Slow dynamics, compensation is an issue
Positive bending (too hot) Negative bending (too cold)
slice support beam slice support beam

jet jet

Fiber orientation angle profile Fiber orientation angle profile

rush drag

drag rush

• Insufficient rigidity has a similar effect

• Faster dynamics, compensatable
Other effects (3)

• The response in a liquid

packaging mill to headbox
thermal compensation
• Thermal compensation
has to be approximately
correct. This is because it
can cause other actions to
lose their effectiveness
over time.
Liquid Packaging Mill (1)
Liquid Packaging Mill (2)
Folding boxboard mill
Closed loop control of FO angle and Twist
• It is possible to put a closed-loop control on the twist as two
interdependent FO angle controls with:
• FO angle (top and bottom) target = 0, across the profile
• Penalization of twist deviation across the profile

• Many mills have just one slice lip controlled, but

for twist
• The following slides had both slice lips controlled
Folding boxboard mill (1)
Folding boxboard mill (2)
Folding boxboard mill (3)
 Folding boxboard mill (4)

Tw ist RH 35% Tw ist RH 50%

avg min max avg min max

-0.04 -0.36 0.19 -0.02 -0.24 0.30

0.01 -0.27 0.25 -0.02 -0.34 0.25

0.10 -0.27 0.56 0.06 -0.72 0.68

Conclusions

• Measurement of FO angle and FO anisotropy is important for

making paperboard
• Very important for twist & curl
• Rush-drag (or jet-to-wire ratio) is critical
• There are adequate devices on the machine whereby the twist can
be controlled – if you have a measurement
• Losses with paperboard twist issues
• Lost production
• Returned production
• Lost customers
• Optimization of twist is tested and working
Questions?