Thesis 1993 Lau
Thesis 1993 Lau
Thesis 1993 Lau
Items in Figshare are protected by copyright, with all rights reserved, unless otherwise indicated.
PUBLISHER
© A.M. Lau
PUBLISHER STATEMENT
This work is made available according to the conditions of the Creative Commons Attribution-NonCommercial-
NoDerivatives 2.5 Generic (CC BY-NC-ND 2.5) licence. Full details of this licence are available at:
http://creativecommons.org/licenses/by-nc-nd/2.5/
LICENCE
CC BY-NC-ND 2.5
REPOSITORY RECORD
Lau, Andrew M.. 2019. “On the Creasing, Folding and Gluing of Corrugated Fibreboards”. figshare.
https://hdl.handle.net/2134/27793.
This item was submitted to Loughborough University as a PhD thesis by the
author and is made available in the Institutional Repository
(https://dspace.lboro.ac.uk/) under the following Creative Commons Licence
conditions.
VOL~NO~------- ___CLASS
~ __ ~_o ~;) ,
M-ARk--------------- ----
',. , '"
ON THE CREASING, FOLDING AND GLUING
OF CORRUGATED FIBREBOARDS
by
AMLAU
MSc, DIC, BEng, ACGI, AMIMechE
A Doctoral Thesis
submitted in partial fulfilment of the requirements
for the award of
31 December 1993
© by A M Lau 1993
.. ..
, "" ,
Loughborough University
of Technology Library
t---~"""' ..,.,--, - - - - - I
Date 0~ <\""
.... - .. "..... _ - - - 1
Class
To Grandma with loving memory.
SYNOPSIS
A case conversion process has been investigated with a view to understanding and
ultimately improving the case quality. In particular, fishtailing, or the squareness
of case, has been studied in relation to creasing, folding and gluing. The process
has been optimized in tenns of fishtailing on the individual panels. Uni-directional
fishtailing results have been attributed to the existing method of folding. To this
end, the folding mechanism has been studied and two innovations, generic to any
continuous folding operations, have been suggested: the coil cam and the twin cam.
The design of the coil cam has been based on the screw theory of three bodies in
relative spatial motion; it aims at transferring the folding force from the front edge
to a point on or near the centre of the face of the panel. A unique method of
computer aided design and manufacture has been proposed whereby a high-level
software program is written to produce the cam profile, the results of which are
then fed into a proprietary CAD software as macro commands. The twin cam, a
more compact design, has been synthesized with the aim of producing rolling
contact between the cam and the panel. Perfect rolling, however, cannot be
achieved due to computational and manufacturing errors.
The feasibility of a twin roll creaser design, which enables boards of various
thicknesses to be processed without the need for roll replacement, has also been
examined. A comprehensive literature survey has been conducted on previous
research into creasing. Theoretical models for simulating the creasing mechanism
have been shown inaccurate by means of an optical technique. .As a result, an
experimental approach has been adopted, and which has led to the development of
special creasing and folding test rigs. Three experiments have been conducted on
the twin roll creaser so as to examine the effect of various creaser parameters on
the quality of fold. This has led to better understanding of the creasing mechanism
and a robust twin roll creaser design. Further research has been recommended in
areas which may have generic applications in creasing, folding and paper related
processes.
i
ACKNOWLEDGEMENT
I would like to thank my supervisor Mr. T.R. Davies for his magnanimous
guidance, enthusiasm, constructive criticism and unique sense of humour. Thanks
also to Mr. B. Rardisty of the SCM Container Machinery Ltd., and Mr. T.
Fountain of the UK Corrugated Ltd. for their excellent technical support. I am also
indebted to Mr. P.D. King, the project manager, and my colleagues Miss E.
Rushforth and Mr. T.C. West for their active part in the research. Special thanks
are due to Mr. J. Petzing for his practical and theoretical proficiency in the optical
experimentation, Mr. A. Baggott for his expertise in CNC machining, Miss Heather
Jennings for the most efficient inter-library loan service, Mr. K.W. Topley and Mr.
v. Scothern for their photographic workmanship and Mr. P. Legood for his sound
engineering advice. Suffice it to say that without all the technical assistance
provided by the Department of Mechanical Engineering, the three years of research
would never have come to fruition.
On a more personal note, I am grateful to Alex and Ada Chung for having literally
provided food and shelter, and Adrian Fietcher and Jeff Pitchers for their
unswerving friendship. Above all, I wish to express my deepest gratitude to my
family for making a dream come true.
ii
CONTENTS
Page No.
1. INTRODUCTION ............... ....................................................... 1
3. FISHTAILING .................................................................................. 18
3.1 Problem synthesis .......................................................... 18
3.2 Experimentation ...................................................................... 20
3.2.1 Experimental Requirements .................................. 20
3.2.2 Control Settings .............................................. 21
3.2.3 Procedure .......................................................... 24
3.2.4 Results ........................................... ........................... 25
3.2.5 Discussion .......................................................... 29
3.3 Conclusion ...................................................................... 32
iii
5.2.1 Preliminary Study .............................................. 39
5.2.2 Model Generation .............................................. 42
5.2.3 Computer Aided Design .................................. 48
5.3 Design Optimization .......................................................... 49
5.4 Manufacture of Prototype .............................................. 53
5.5 Appraisal .................................................................................. 54
5.6 Conclusion ...................................................................... 55
IV
8.3.2 Objectives .......................................................... 98
8.3.3 Apparatus .......................................................... 99
8.3.4 Procedures 100
8.3.5 Results ...................................................................... 102
8.3.6 Discussion .......................................................... 106
8.4 Conclusion ...................................................................... 108
v
10.2 Folding Mechanism 146
10.2.1 Coil Cam 146
10.2.2 Twin Cam 147
10.3 Twin Roll Creaser 147
10.3.1 Creasing Analysis 147
10.3.2 Experimentation .............................................. 148
10.4 Remarks .................................................................................. 150
10.5 Recommendations .......................................................... 150
APPENDICES 152
2a. Tests for corrugated fibreboards and cases 154
3a. Case sample specification .............................................. 158
3b. Folding rail calibration .......................................................... 159
3c. Glue dispenser calibration .............................................. 161
3d. Tables of experimental data .............................................. 163
Sa. Determination of the third screw axis .................................. 170
5b. Cylindrical coordinate transformation .................................. 178
5c. CAD program listing .......................................................... 181
6a. Twin cam design optimization .............................................. 186
8a. Theoretical model on corrugation .................................. 188
8b. Determination of maximum twin roll gap ...................... 190
8c. Fourier harmonics for B-flute corrugation ...................... 195
8d. ESPI equation derivation .............................................. 196
9a. Folding rig design specification ............................... ............... 202
9b. L\8 orthogonal array .......................................................... 211
9c. Shear test .................................................................................. 212
9d. First experiment results .......................................................... 216
ge. Defective sample results ................................................... ....... 226
9f. SIN ratio ANOV A tables .............................................. 227
9g. Optimum settings .......................................................... 233
9h. Confirmation run results ........................... ............ .... ............... 235
vi
9i. Profile design for 2nd experiment .................................. 236
9j. Second experiment results .............................................. 238
9k. Third experiment results ................................................. ......... 242
vu
LIST OF ILLUSTRATIONS
Page No.
2.1 Single- and double-wall flute types .............................................. 4
2.2 Flute types and their relative size range ............................. ..... 5
2.3 Flattened case with panel dimensions .................................. 6
2.4 Front and plan view of a flexo-folder gluer ...................... 10
2.5 Standard creaser profiles .......................................................... 12
2.6 Case in the knock-downed and erected position ...................... 15
3.1 Case conversion product-process diagram .................................. 18
3.2 Fishtailing measure ...................................................................... 19
3.3 Case sample and dimensions .............................................. 21
3.4 Folding section ...................................................................... 21
3.5 Creasing section ............................................................. ......... 22
3.6 Damaged anvil due to wearing .............................................. 23
3.7 Combined SIN ratio plot ....................................... ................... 27
3.8 SIN ratio plot for separate panels .............................................. 27
3.9 Thickness hypothesis for fishtailing .............................................. 31
3.10 Distribution of fishtailing results .............................................. 32
4.1 Folding resistance characteristics .............................................. 34
5.1 Cylindroid of screw axes .......................................................... 38
5.2 Coil cam .................................................................................. 40
5.3 Prototype coil cam operation .............................................. 41
5.4 Screw path followed by a point 'P' on the panel ...................... 43
5.5 Cylindroid frame transformation .............................................. 44
5.6 Frame transformation algorithm .............................................. 45
5.7 Coil cam profile along its axis .............................................. 48
5.8 Coil cam with hyperboloidal core .............................................. 49
5.9 Coil cam with 'pear' cross-sections .............................................. 50
5.10 Optirnized coil cam design .......................................................... 52
5.11 Coil cam operation ...................................................................... 53
5.12 Coil cam machined mould .......................................................... 54
viii
6.1 Parabolic motion profIle ................... ....................................... 58
6.2 Cam-panel contact length ......... ............................................ ..... 59
6.3 Plan and side views of cam ......... ........................ ......................... 60
6.4 Cam geometry ............................................. ......................... 61
6.5 Twin cam design ...................................................................... 64
6.6 Twin cam assembly in operation .............................................. 65
7.1 B-flute corrugation with creasing locations .................................. 69
7.2 Experimental creaser profiles .............................................. 71
7.3 Edge-wise compression curve .............................................. 74
7.4 Force-angular displacement plot .............................................. 75
7.5 Compression ratio definition .............................................. 76
8.1 Single blade creaser ...................................................................... 80
8.2 Worst creasing scenario .......................................................... 80
8.3 Twin roll creaser ...................................................................... 81
8.4 Fold with two score marks .......................................................... 81
8.5 Constraint strategy ...................................................................... 85
8.6 Creaser profile model .......................................................... 95
8.7 Test jig for optical experiments .............................................. 99
8.8 Two possible 'c' locations .......................................................... 101
8.9 Test jig setup .................................................................................. 101
8.10 Video image of fringes .......................................................... 102
8.11 ESPI calibration plot ....... .......... ................ .............. .......... ............. 103
8.12 Creaser knife at location' A' .............................................. 103
8.13 Creaser knife at location 'B' .............................................. 104
8.14 Creaser knife at location 'c' 104
8.15 Creasing at location' A' with end constraints ..................... . 105
8.16 Flute shearing at one clamped end ............................................. . 107
9.1 Creaser roll radius derivation 111
9.2 Experimental creasing rig 112
9.3 Crease roll gap adjustment 112
9.4 Folding test rig assembly 113
9.5 Folding test rig control accessories 114
ix
9.6 Folding resistance plot 114
9.7 Twin roll creaser design 116
9.8 Fletcher's trolley for shear measurement ................................. . 120
9.9 Defective sample plot ......................................................... . 122
9.10 SIN plot on folding resistance 122
9.11 SIN plot on shear deformation 123
9.12 Pre-crusher profile (dimensions in mm) 130
9.13 Twin roll creaser main design parameters ................................. . 131
9.14 Defective sample plot ......................................................... . 132
9.15 SIN plot for folding resistance 133
9.16 SIN plot for shear deformation 133
9.17 Combined profile 136
9.18 SIN defect plot 138
9.19 SIN 1st peak gradient plot ........................................................ .. 139
9.20 Peak SIN plot ..................................................................... . 139
9.21 SIN plot for shear resistance 139
9.22 Folding resistance plot with no 1st peak 141
Appendix illustrations:
5a.l Two rigid bodies in spatial motion .............................................. 171
5a.2 Three axes screw system .......................................................... 173
5a.3 Angular velocity polygon .......................................................... 173
5a.4 Coil cam prototype ...................................................................... 176
5a.5 Coil cam angular velocity polygon ....................................... ....... 177
5b.l Coordinate frame ....................................... ............................... 178
5b.2 Cylindrical components in Cartesian frame ... ............................... 178
8a.l Full penetration of corrugation .............................................. 188
8b.l Twin roll design parameters .......................................................... 190
8b.2 Loading at location 'C' .......................................................... 191
8b.3 Twin roll gap at maximum opening .............................................. 192
8b.4 One-half of twin roll at location 'C' 193
8d.l Michelson interferometer 198
x
8d.2 ESPI out-of-plane interferometer 199
8d.3 ESPI in-plane interferometer .............................................. 200
9a.l Folding test rig design ............................................... .... ....... 204
9c.l Shear test measure on Fletcher's Trolley .................................. 212
9c.2 Calibration plot of shear vs. load .............................................. 215
9i.l Twin roll creaser design for 2nd experiment ...................... 236
xi
LIST OF TABLES
Page No.
2.1 Common board types and their properties ........... ....................... 5
2.2 Sloned case type classification ........... .......... ................ ......... 7
2.3 Case conversion function-operation diagram ...................... 11
3.1 Control variables and settings .............................................. 24
3.2 ~ onhogonal array ...................................................................... 24
4.1 Case blanks size range .......................................................... 36
5.1 Frame notation ...................................................................... 43
5.2 Spreadsheet results at twelve angle increments ...................... 47
6.1 Sample data for r(t), ~(t) and e(t) .............................................. 64
8.1 A summary on the twin roll gap settings and
the total creaser width .............................................. 82
8.2 In-plane displacements for the wide and
narrow sections of illumination 105
8.3 ESPI in-plane and out-of-plane displacement results 106
8.4 Finite element model in-plane and
out-of-plane displacement 108
9.1 Control factors and their levels of settings ................. .......... ....... 118
9.2 Defective samples count .......................................................... 121
9.3 Optirnized settings ...................................................................... 123
9.4 Theoretical SIN ratios 124
9.5 Confirmation run results 124
9.6 Design parameters and their settings 131
9.7 Defective sample count .......................................................... 132
9.8 F-ratio summary from the ANOV A tables .................................. 133
9.9 Control factors and settings .......................................................... 136
9.10 SIN ratios for pre-crushing defect rate ................................. . 138
9.11 ANOVA 'F' ratios summary ............................................. . 140
9.12 Experimental and predicted SIN ratio 140
xii
Appendix tables:
3d.1 Experimental results ................................... ............................ ....... 163
3d.2 SIN values for the various control factors .................................. 164
3d.3 Combined SIN ratios at various control operation levels 165
3dA ANOVA table for SIN ratio .......................................................... 166
3d.5 Alternative mean and variance analysis .................................. 167
3d.6 Confirmation runs 169
Xlll
CHAPTER 1
INTRODUCTION
Corrugated fibreboards and its conversion into cases by automatic case converters
have a history going as far back as some hundred and fifty years ago. As the
packaging industry evolves, more attention is paid to the case quality and the added
value in the final case products. Not surprisingly, case conveners today have
reached an unprecedented level of sophistication. Machine operations such as
creasing, printing, gluing, folding and palletizing (hence the name "flexo-folder
gluers") are carried out at speed in excess of four cases per second.
Creasing, folding and gluing are operations which have a direct influence on the
physical strength of a case. The way a board is creased will determine how well it
will fold. Once they are folded and glued together, the case panels may end up
lying skew to one another. Typically in the trade, the misalignment of bonded case
panels is known as fishtailing. Problems with automatic case erection and distorted
case appearance are attributed to fishtailing. So what causes fishtailing? How can
fishtailing be prevented? What part does the corrugation structure play in the
creasing process? Will innovative creaser and folder design be a remedial solution
to fishtailing? All these questions have prompted the following investigation.
Under the auspices of the LINK programme, a three-year project was jointly
funded by the Department of Trade and Industry and the Science and Engineering
Research Council for developing a new generation of case conveners. In addition,
two industrial collaborators were involved in the project; they were the SCM
Container Machinery and the UK Corrugated Ltd. Among the objectives were
finding the causes for fishtailing through looking at the creasing, folding and gluing
operations, and the study of innovations which will achieve less fishtailing and
hence better case quality for a diverse range of boards.
Chapter 2 commences with a description of the corrugated fibreboard cases and the
conversion process, in particular, the creasing, folding and gluing operations.
I
Fishtailing, a major quality characteristic in the case conversion process, is
examined in Chapter 3. By means of a Taguchi experiment on an industrial case
converter, the relative significance of creasing, folding, gluing and machine speed
is put into perspective. Results from the experiment also highlight areas for further
investigation.
Chapter 4 looks at the background of folding and reviews past literature on folding
mechanisms. Two innovative folding cam mechanisms for folding are proposed.
In Chapter 5, a coil cam, which touches the case panel at a single point, is derived
by first of all applying the theory of screw motion to the case folding action. This
is followed by a design methodology and a discussion on the methods of
manufacture. Another non-oscillatory spatial cam mechanism in the form of a twin
cam is introduced in Chapter 6. The chapter provides a design methodology for
cams that roll on the panel without sliding.
A literature review in Chapter 7 outlines the theory on creasing, past research into
the design and optimization of creaser profiles and the ways of assessing crease
quality. The twin roll creaser design is studied in the geometrical and finite
element analyses in Chapter 8. Optical experiments using the Electronic Speckle
Pattern Interferometry, however, show that the finite element model is defective.
To compensate for the inaccuracy of a theoretical model, Chapter 9 describes a
series of experiments on the twin roll creaser design. Major findings from the
experiments have paved way for future work as outlined towards the end of the
chapter.
2
CHAPTER 2
CORRUGATED FIBREBOARD CASES
Case converters and corrugated fibreboards are essential for the manufacture of
corrugated fibreboard cases. This chapter provides some background information
on the corrugated fibreboards and the design of a typical case converter.
3
separated by a flat sheet and faced on both sides with a liner (i.e. 5 components in
all). For heavy-duty application triple-wall board is also available. Figure 2.1
shows the single- and double-wall flute types.
Historically, butchers' strawpaper was the first material used for the corrugating
medium. Today, corrugating medium must be able to fulfil the following criteria:
The fluting itself can be categorized into four size range, as shown in Table 2.1.
4
Table 2.1: Common board types and their properties.
• Take-up factor is the ratio of length of medium divided by the length of single-face linerboard for a
given length of combined board.
Figure 2.2 displays the relative size for the various flute types.
.. .. ,
I." • w., FLUTE PITCH
"
,
c
,
..fI."," ..
,
As a rule, corrugated fibreboards with A, B and C flute are widely used in the
transit of goods. E-flute boards, however, take on a more decorative role of
producing display cases when combined with high quality printed liners. Several
.types of paper can be used for the corrugating medium, and they are as follows:
Semi-chemical papers made by treating wood chips with chemicals to achieve pulp
of the desired properties.
Straw papers made from furnishes of between 25% and 75% straw with various
quantities and grades of waste pulp.
5
Kraft paper used in the UK on some grades of weather-resistant board.
Secondary fibre, chip or waste papers of various grades can be used in their own
right after treatment or be added to any of the materials above to give a
balance of properties at a moderate cost.
The liners can be divided into three main types: pure Kraft, Jute (test liner that
consists of a lower grade of pulp, frequently waste, backed with a pure kraft sheet)
and Chip (straw or bogus papers principally used for making fittings).
The packaging code of BS 1133 Section 7 covers a fair selection of fibreboard case
styles. Figure 2.3 shows a typical flattened case with the panel dimensions.
Fl F3
'H
I
L B L B
F4
6
The slotted case type is one which can be folded and glued together on an
automatic case-maker. A systematic classification of the slotted case style is given
in Table 2.2:
Code Ft F2 F3 F4 Feature
0203 B B B B
The slotted case type mainly comes as one piece with a stitched, taped or glued
manufacturer's joint, and top and bottom flaps. The cases are shipped flat, ready
to use and require closing of the flaps. Fishtailing, or the misalignment of glued
panels, is often associated with extremely light or heavy cases, and cases with a
narrow body panel (i.e cases with dimension 'H' being much smaller than 'F' as
illustrated in Figure 2.3).
7
suffer during the conversion stage. Fibreboards are therefore recommended to be
stored under cover on dry flat board pallets in atmospheric conditions and relative
humidity of between 50 and 70 % prior to usage.
Test methods for the corrugated cases and liners can be found in the British
Standard and the TAPPI publications from the United States. Some of these tests
do however bear limitations in that they may not be sufficiently precise for use in
case of dispute. Test methods are becoming increasingly popular as customers
demand for higher quality or value added products. Appendix 2a outlines some of
the more common test procedures.
The structure of the fluting medium resembles that of a sine wave. For the same
amount of paper material, it is mainly the laminated design which gives the
fibreboard its enhanced stiffness over its paperboard equivalent. However, the
orientation of the corrugations is critical, not only to the bending stiffness, but also
the crease performance.
There is often less problem with perpendicular creasing than parallel creasing, as
suggested by Vogelpohl [51]. In perpendicular creasing, force is transmitted
directly from the creasing tool onto the corrugation. This is less probable in
parallel creasing when the position at which the creasing force is applied cannot be
pre-determined. The creasing tool may compress the corrugation by means of
tensile forces within the creased liner - a situation which encourages tearing of the
liner. However, it is debatable whether the resulting board behaviour from
creasing is attributable to the corrugation structure or to the paper material itself
(see Chapter 8). Analysis on other composite material (e.g textile), as indicated by
Kawabata [21], suggests that its mechanical properties must be considered from a
structural rather than a continuum standpoint. Cone et al. [6], on the other hand,
8
have put emphasis on the material by stating that paper ought to be treated as a
discontinuous network of fibres.
In the case conversion process, flat blank sheets of corrugated fibreboards are fed
into the machine and will emerge as knocked-down cases ready for automatic
erection. Figure 2.4 shows the front and plan views of a standard flexo-folder
gluer (acknowledgement given to SCM Containers Machinery for the photograph
reproduction).
9
Gluo applice.tor Printiro units
Slottor Ct8II!l8er
-
Sodon
10
Table 2.3: Case conversion function-operation diagram.
2.2.1 Creaser
In a flexo-folder gluer, creasing takes place after the flexo-printing stage. Creases
parallel to the corrugation are produced by four pairs of rotary creasers forming the
four edges of an erected case. To ensure a continuous flow of cases, pull rolls are
11
positioned before and after the crease roll. Roll synchronisation is therefore vitally
important in avoiding undue tearing. The use of pre-crush rolls is recommended
when creasing boards of thicker caliper.
The creaser itself comprises a rotary creasing head in the form of a circular blade.
The creasing head acts against a cylindrical anvil, and the two combine to give the
creasing action.
Creaser Components
Broadly speaking, there are four design parameters which give the creaser its
functional characteristics: creaser profile, creasing roll gap, material and roll
diameter.
The selection of creaser profile is entirely subjective as the need varies from one
case manufacturer to the next. Figure 2.5 shows a few standard creaser profiles.
The blade width also determines the amount of pressure that the liner is likely to
receive from the blade. Obviously, too sharp a blade will tend to tear the liner,
whereas too blunt a blade will not produce a definite crease line at all. In general,
the application of wider profiles is often reserved for thicker boards.
The crease roll gap is the separation between the creaser and the anvil.
Adjustment of the gap is wholly empirical, and its precision has never been taken
seriously. Normal practice is to crease at progressively smaller clearance until
cracking becomes perceptible on the outer liner. The crease roll is then backed off
12
a fraction prior to the actual production run. Typically the amount of backing-off
comes from the operators' own experience.
Regarding the material, over the years, practical trials have shown that a
chrome-plated steel male creaser together with a polyurethane anvil form the best
combination. The chrome-plated surface is designed to reduce wear, thereby
retaining the shape of the creaser profile. The elasticity of the polyurethane anvil,
on the other hand, compensates for the differences in board thickness and any
possible damage that would have been caused by a steel anvil on the liner surface.
Pressure exerted on the liner and the corrugation will depend on the roll diameter
of the creaser blade. From the geometrical viewpoint, a larger roll diameter will
extend a longer contact length over the liner. Less pressure is therefore induced in
the vicinity of the point of action, and the liner will be less likely to tear as a
consequence. Vogelpohl [51] realized the importance of the creaser roll diameter,
but did not investigate any further on the basis that any change in the shaft
diameter would alter the design of the mechanical coupling. Werner [54],
however, regarded the roll diameter as a function of the machine width. Since the
roll was supported on both ends, the weight of the roll would cause it to sag in the
middle, thereby producing the catenary effect.
Two pairs of pull rolls on either side of the crease rolls provide a gripping action
on the fast-moving boards. When dealing with smaller boards, the pull rolls will
also act as interstage drives. A situation may occur whereby a small board is
13
suspended in mid-air on leaving one set of rollers, but not quite reaching the next.
Another problem which may lead to roll shearing is when the pull rolls and the
crease rolls are not synchronised properly.
Corrugated fibreboards, when more than one-board thick, are usually crushed over
a few flutes along the crease before arriving at the creasing stage. This kind of
pre-crease crushing is to ensure the effectiveness of a crease irrespective of the
board's thickness. Pre-crease crushing is also applied to the two edges where the
case will be folded by 180". A crushed section is much preferred on the grounds
that, on folding, less resistance will be resulted from the compaction of a larger
area of buckled flute.
Most damages occur during the acceleration and deceleration of the case, notably
in the feeding section of the case converter. Conventional grip feeding often
distorts the board caliper while accelerating the board from a stationary position.
Irregular board calipers not only imply uneven colour distribution from the
flexo-printers, but may also lead to a deterioration in case strength. For this
reason, vacuum suction is increasingly adopted as a better means of feeding.
2.2.2 Folder
Folding is a mechanical process for joining the glue flaps together. Most case
converters produce flattened, folded cases with only two 180" edges. It is not until
the time of erection that the remaining two edges experience, for the first time, a
fold of 90". Figure 2.6 shows a case sample in its flattened and erected positions.
14
Crease locations
I I
180· 90· 180· 90·
90° 180·
Erected case
'1
180° 90"
Folding is commonly divided into two operations i.e from a' to 90° and from 9a'
to I8a'. A standand case converter comprises folding bars and belts which
generate the mechanical motion necessary for the folding process. During the ftrst
stage of folding, the bars exert a force onto the leading edge of the case, the
direction of which runs opposite to the case travel. It is suspected that this force
component is ultimately responsible for fishtailing. In the second stage of folding,
folding belts help to minimize the relative motion between the folder and the case
panels. Hence, a good fold quality will be ensured by the absence of the backward
force component and a minimum friction component. It has been known that the
two stages of folding can be combined into one by having a single pair of folding
belts; however, the required belt length has rendered it impractical.
Seymo·ur and Terle [42] have further improved the concept of relative belt motion
by introducing the multi-T belt mechanism for the second stage folding. In
essence, T-blocks are attached onto the folding belts which are synchronized with
the cases' motion. By directing the folding force normal to the board panels, the
backward component of force, which is commonly experienced in the folding rail
concept, will be eliminated. As claimed by the manufacturer, the design is capable
of keeping fishtailing down to a minimum. Friction, however, is inevitable
15
between the panels and the lugs because of imperfect synchronisation and relative
motion between the two as directed towards or away from the folding hinge.
To complement the folding belts and bars, Rother [39) has introduced air fans at
strategic locations along the folding section. It is hoped that the forced air current
will produce less mechanical damage to the case panels.
Other novel but less well known folding techniques are the "Rotofold" and the
"inclined wheels" which have been documented by Shulman [43). Rotofold
technique conducts the second 90° fold by means of helical rotating rods touching
the case blanks at the centre. Inclined wheels, on the other hand, comprise forming
rolls with changing roll profiles. As the board progresses through the sequence of
rolls, it will be folded stage by stage from 0" to ISO".
All the above techniques, except for the air folding one, have one inherent
problem: the smearing of ink. Friction will prevail as long as there is relative
motion between the folder and the case blanks while they are in contact. This
explains why some of the more dynamic techniques, such as the T -blocks and the
"rotofold", are only designed for the second 90" fold, by which time the ink on the
case panels will have set. To lessen the effect of ink smearing, one industrial
practice is to dilute the ink so that it can be absorbed more readily by the case
liners.
2.2.3 Gluer
16
such as PV Ac and its derivatives. It is important for the applied film of adhesive
to have good wet tack property. Two ways of glue application are available,
depending on whether it is a down-folding or top-folding machine: they are the
glue wheel type and the nozzle extrusion.
In the glue wheel application, a knurled roll deposits a desired band of adhesive to
the lap at the same speed as the machine. Adhesive is constantly circulated and
fed to the wheel by a small pan feed or 'shoe' arrangement. Extrusion nozzle on
the other hand relies on electronic control for the glue flow adjustment. A
photocell detects the leading edge of the blanks, thereby monitoring the glue
application onto the lap only.
Vyse et al. [53] have found that the cost benefit from a glue extrusion system can
be deceptive when compared with the wheel application type. Since, by extruding
several thin beads of adhesive and spreading them on compression, the overall fIlm
area may be inadequate and lap lifting may occur as a result of the lack of wet
tack at high speed. Increasing the bonding area will help, though at the expense of
having to apply more glue, thereby off-setting the original cost benefit
Examples on test standards for adhesives used in the corrugated industry can be
found in Appendix 2a.
17
CHAPTER 3
FISHTAILING
(N.B. This chapter has been presented and published in the 42nd TAPPI Conference, Washington D.e.,
U.S.A., Oct.1992 under the authorship of A.M.Lau and T .H.Davies [24].)
Figure 3.1 shows a product-process diagram for the case making process.
~
CASE-MAKING
SIGNAL. ~ 1--_ RESPONSE
PROCESS (flSh.~)
(_10
=:01) ~
COtmlOL
(0.0 folding rail iocafun,
rrJc speed, ftlTlOUnl 01 glue,
avese ro(t gap)
18
The diagram, proposed by Phadke [33], is a simplified version of the "cause-and-
effect" diagram as developed by Ishikawa [16]; it provides a systematic way of
analysing those factors which may influence the quality of the final product. The
factors can be defined as follows:
Signal - acceptable criterion for assessing fishtailing, though a standard has yet to
be defined.
Noise - parameters such as the board quality or the aerodynamics of the board
folding action, both of which cannot be easily controlled.
Control - parameters such as the folding rail location relative to the folding hinge
on the case convertor, the machine speed, the amount of glue applied and
the gap setting between the crease rollers. These factors are usually related
to the settings of the machine, and are best identified through a brain-
storming session with the machine operators.
Process response - the fishtailing measure.
Fishtailing, a term often quoted in the corrugated industry, describes the skew
edges formed by the case panels during the case conversion. Figure 3.2 shows a
typical example of fishtailing.
19
is that results showing a combined value of zero may imply correct alignment, but
not necessarily with the right amount of overlap between the two panels. For the
present purpose, however, the measure is thought to be adequate in assessing
fishtailing.
3.2 EXPERIMENTATION
In accordance with BS3430, a sample size of ten boards was chosen with another
ten as spare. The type of board used was the Ginster 300 KY2 which comprised
20
an outer kraft liner, an inner oyster liner and a B-flute corrugation medium (see
Appendix 3a for further details on the case sample). Because of the short case
length, it was believed that this particular board type would most likely experience
fishtailing. Figure 3.3 shows the overall case dimension.
L. _______ _
104 Panel ..• Panel b'
r--------------------
Four control parameters were selected for the experiment, and they were: folding
rails, creaser roll gap, machine speed and glue dispensation.
a) Folding rails
Nylon rails in the folding section, as shown in Figure 3.4, are responsible for the
first 90" fold. The rails can be positioned near to or away from the folding hinge
21
by manually sliding the rail clamps along their suppon shafts (see Appendix 3b for
folding rail calibration).
On previous occasion, distonion of case panels was noticed to have already taken
place in the first 90° fold. The amount of distonion could be correlated with the
position of the nylon rails relative to the folding hinge. For example, if the rail
was positioned nearer to the hinge, then a larger backward force component would
be exened onto the case panel, since the moment arm had shonened and the
folding moment should remain the same.
It was found that the pan of the polyurethane anvil immediately below the knife
had been worn down over its operation life. For this reason, a meaningful gap
22
width could be measured only between the creasing knife shoulder and the
undamaged part of the polyurethane anvil, as shown in Figure 3.6.
c) Machine speed
Running the case converter at maximum speed is, in practice, not recommended on
account of factors such as vibration, print consistency and rate of wear. A slow
machine throughput is equally undesirable. There is therefore an optimum range of
speed settings beyond which it is not advisable to operate.
d) Glue dispensation
Glue is extruded onto the case panels by means of an electrically controlled lap
gluing system. The rate of application can reach up to 6 m/s. Speed compensation
is performed via a closed loop control system; but the expected linearity between
the amount of glue dispensed and the machine speed has not been verified to the
author's knowledge. The glue itself is an aqueous emulsion adhesive specially
formulated for paper and board bonding operations.
In theory, the glue flow should be controllable via the main control panel.
However, a series of glue tests prior to the experiment indicated that the flow
adjust had no effect on the amount of glue extruded from the system. Instead, a
pressure valve governing the air pressure of the flow line was found to be more
effective in controlling the glue quantity.
23
Table 3.1: Control variables and settings.
I 2 3
3.2.3 Procedure
Expt. no. A B C D
I I I I I
2 I 2 2 2
3 I 3 3 3
4 2 I 2 3
5 2 2 3 I
6 2 3 I 2
7 3 I 3 2
8 3 2 I 3
9 3 3 2 I
While planning the experiment, it was clear that column ' A' should be for settings
which were the most time consuming to change. In addition, the fIrst column of
24
the settings was taken as set on the machine so as to minimize the number of
setting changes.
In all, nine, experiments were performed with settings corresponding to those of the
~ orthogonal array. The case samples, twenty in all, were fed into the machine
following each setup. Using the smallest-the-best criterion, the signal-to-noise
ratios were calculated as follows:
A model of unit variation, as formulated by Taguchi [47], was adapted for the
present purpose. In particular, the absolute function for the dimensions was
essential if the above criterion were to remain valid. The smallest-the-best
criterion was satisfied by maximizing the signal-to-noise ratio or minimizing the
gap difference.
Once the signal-to-noise ratios were collected, they were processed using the
Analysis of Variances technique, ANOVA in shon. Optimized settings were those
with the most positive signal-to-noise ratios. A final test run using the optimized
settings should confmn the relative magnitude of the factor effects.
3.2.4 Results
25
5. Test folding rail adjustment.
6. Check board samples and record physical properties.
The four units of flexo graphic printer were taken off line as they were not
required during the experiment. Following the stages of creasing, slotting, folding,
gluing and slapping, the folded cases were then examined for fishtailing.
Dimensions 'a' and 'b' were measured in mm with an accuracy of ± 0.5 mm. (In
Appendix 3d Table 1, the first row represents the 'a' values, while the second row,
the 'b' values.) Each of the samples was then checked for foldability by manually
erecting and noting the final form of the cases. Cases with relatively high
fishtailing values were found distorted and difficult to erect.
Some of the observed defects (with the particular case sample in brackets) were
highlighted as follows:
The sum of Y and the SIN ratios for the nine experiments (in Appendix 3d Table
2) were added together and grouped in terms of the appropriate control factors (see
Appendix 3d Table 3).
The SIN level of effects due to the various control factors at different settings are
shown graphically in Figures 3.7 and 3.8. Note that the scales are different for the
two plots. The combined mean value, M, for the SIN ratio is -11.6. In terms of
the constituent mean values for panel 'a' and 'b', M" and Mo are -8.5 and -3.2
respectively.
26
The sum of squares, mean square and the variance ratio were then calculated using
the following equations:
Sum of squares = 3 x [(SINA1-Mf + (SINA2-Mf + (SIN A3-M)2]
Sum of squares
Mean Square =- - - - - - - - - - -
Degrees of freedom
Mean square
Variance ratio, F =-----------
Pooled error
•
-,
•
•
J
-~
I
-"
•
0 -"
~ -"
~
-"
-"
-.
-"
-...
-
., IIZ 11
'" .. '"
Fig. 3.7 Combined SIN ratio plot.
•
-,
,
•
•
•
~ ~ J
,•
••
~
~ -~
_n
•
• V \;
-u
-"
-"
."~
-
.Ir,
. '"
;, ,;, ~ .
,~
Q ~ Q
'" .. '"
~.
• ~,
27
ANOVA, as tabulated in Appendix 3d Table 4, highlighted the overall effect of the
control factors. A large F value implied a sizeable contribution to fishtailing due
to that particular control factor. Machine speed turned out to be relatively
insignificant when compared with the other factors; for this reason, it was used as a
pooling factor. Pooling was a computation method which would give an
approximate estimate of the error variance. To obtain a better estimate of error
variance, a significantly larger number of experiments would be needed, the cost of
which would outweigh the added benefit.
By selecting the factors with the lowest SIN ratios in Figure 3.7, the optimum
settings for the confmnation run were: Al B2 C2 D3
With a confidence interval of two standard deviation, the prediction error was
calculated as follows:
1
Err pred. = ±2 + - ) = ±2.9
n,
11111111
-=-+(---)+(---)+(---)
n. n nA n nB n nD n
28
Likewise, analysing in terms of individual panel results of 'a' and 'b' would give:
SINpred..• = -2.78 ± 8.21
SINpred.. b = -0.08 ± 5.32
A separate analysis on the mean and the variances also proved to be equally valid,
with the smallest combined mean values of l1(a,b) confirming the same optimum
settings (see Table 5 of Appendix 3d) .
Finally, two confInnation runs, labelled here as CRI and CR2, were conducted at
the optimum settings (results as tabulated in Appendix 3d Table 6). The reason for
having two runs was because for CRI, the squaring section had not been put back
together after an earlier jamming incidence. Taking the results from CR2, the
combined SIN ratio was calculated to be -6.3.
3.2.5 Discussion
Glue pressure has been identified to be the most critical factor in tenns of
minimizing fishtailing. This can be seen from the variance ratios F in the ANOVA
table. The result tabulation also reveals that, whatever the other control settings,
excessive glue pressure at BI will inevitably give rise to fishtailing.
Regarding the crease roll gap, it was found that a deep crease penetration did not
necessarily correlate with a good fishtailing response. Among the three gap width
settings used, 0.7mm (roughly a quarter of the board thickness) came out to be the
best. One reason could be excessive creasing resulting in the weakening of the
corrugation, which would in turn allow lateral movement to take place about the
hinge.
During the experiment, a slight asymmetry was noticed between the gap values on
the left-hand and the right-hand side of the crease roll in the direction of board
travel. This was detected during the crease roll gap setup, when the feeler gauge
29
was found relatively easier to slide into the gap on the left hand side than the one
on the right. However, the pairs of crease rolls, both left and right, were adjusted
with gap consistency, as demonstrated by the coherent response on the SIN plots in
Figure 3.8.
30
instant when the lowest case is being withdrawn from beneath the stack in the
squaring section. Where the shorter panel is glued to the flap of the longer panel,
the folded case is three board thick, as can be seen in Figure 3.9. It is in this area
that most of the weight of the cases above the lowest case acts on the short panel
of the lower case.
/.
•
Region of three board thickness
While being withdrawn in shingle fashion, the friction force exerted by the case
above this part of the lowest case causes the short panel to move backwards
relati ve to the remainder of the case. The resistance to this force, preventing
v
relative displacement and thereby the reappearance of fishtailing, could only come
from the glued joint, or if that is not set, the folded hinge. Note however that the
cases used in this experiment have hinges which are short by industrial standard,
and therefore provide poor resistance to lateral shear - an ideal condition for
fishtailing.
Finally, a plot of all the experimental data, as shown in Figure 3.10, gives a normal
distribution with the majority of values lying somewhere between -1 and +5 mm.
However, such tolerance may not be acceptable for an automatic erection machine.
31
FISH·TAlUNG MEASUREMENT DISTRIBunON
e , -__________~~~~~··~
..~____________,
n n ~n n n
-3 ~ _1 0 1 ~ 2 .. , • , • t ..: " ,~ " .. \i • t7 ...
measurement Imm
Fig. 3.10 Disoibution of fishtailing results.
3.3 CONCLUSION
Fishtailing has been defined geomeoically in order to express the. case squareness
quantitatively. An experiment conducted on a standard flexo-folder gluer case
converter has shown that the combined action of gluing and squaring has direct
influence on fishtailing. Placing the nylon rails nearest to the folding hinge helps
to minimize fishtailing, but the rails have been responsible for the uni-directional
fishtailing results. Creaser penetration, when set at the extreme range, can create
more fishtailing. On the whole, the effects due to the folding rail location and the
crease roll gap have been masked by those of the gluing and squaring. Fishtailing
is least dependent on machine speed.
32
CHAPTER 4
FOLDING CAM MECHANISMS
Folding devices can be viewed as spatial cam mechanisms, which differ from the
planar ones on account of the pennitted axial movement of the cam or the
follower. Examples such as the globoidal and camoid cams, are typically used for
indexing followers that rotate intermittently. Equally uncommon is the concept of
a rolling cam mechanism which does not experience rubbing between the cam and
the follower. This chapter describes the theoretical background of folding, and
provides a literature review on spatial cam mechanisms. A summary on the
technical specifications for the spatial folding cam mechanisms is included towards
the end of the chapter.
On the theoretical front, little is known about the folding process. Since folding
occurs after the creasing process, the quality of fold will go hand-in-hand with the
crease quality. Assuming that the fold is as prescribed along the crease, and that
there is no false crease in the vicinity of the fold, then the case will be subject to
folding resistance as a result of the following factors:
Air resistance - this is most prominent when the case panel approaches the final
1800 of fold. When air is expelled from the confined wedge as formed by
the closing panel, folding resistance will be experienced the magnitude of
which will depend on the rate of air expellation.
Compaction - folding relies on the breaking of the crease on the inner liner and
the buckling of the flute structure. As the inner liner and the fluting
medium crumple together, folding resistance will increase due to the
compaction of material. Suffice it to say that the thicker the board type, the
more will compaction contribute to the folding resistance.
33
Tension - the outer liner in the creased region experiences tension during folding.
Depending on the liner quality, a strong liner will resist folding, whereas a
weak one will simply tear away, thus lowering the overall folding
resistance. Low folding resistance is preferrable, but not at the expense of
reducing the case strength and distorting the case appearance as a result of
having torn outer liner.
Compaction and tension are interrelated since the corrugation is structured in such
a way that the outer liner will restrict the movement of both the inner liner and the
medium as they crumple together. Factors like the air resistance and case inertia
are machine speed dependent. In general, the folding resistance characteristics, as
construed in Figure 4.1, can be measured separately. There may be difficulties,
however, in simulating the high speed motion of the case panels and taking
measurements of the folding resistance at the same time.
Folding
.".;-.:e Air cxpeUrrtion
of fold
~
Folding
re:dsmnce Compoction '"
II:nsioo
Angle of fold
A low folding resistance implies less damage to the case as a consequence of the
interaction between the case panels and the folding mechanism. Previous chapters
have aiready discussed how the foldability of a case can be affected by fIshtailing.
Other machine-related factors may also affect the fold quality. For example, a
34
study by Edholm [7] has shown that crease quality will only be of secondary
importance if the case is not properly constrained during the folding process.
A typical planar cam mechanism consists of a cam disc and an oscillating follower.
The cam and follower axes are parallel to each other and often lie in the same
plane. The cam imparts an angular motion to the follower by means of a non-
cylindrical curve profile. Spatial cam mechanisms introduce a third dimension
which enables the simultaneous relative translation and rotation between the cam
and the follower.
Spatial cam mechanisms in the form of camoid and globoidal cams have been
mentioned by Rothbart [38] and Chen [5]. Such cams are primarily used for
indexing purposes. Fabication cost however has been a limiting factor for this type
of cams.
The concept of rolling cam or direct contact mechanisms has been well established
despite its lack of application. As Rain [10] has defined, direct contact
mechanisms apply to those with direct contact between two arbitrarily curved
bodies shaped in such a way that only pure rolling exists between them. Frictional
losses are greatly reduced by thus eliminating any sliding motion between the two
cams. Rain [11] has also proposed methods for designing the mating profiles of
35
rolling cams. Characteristically, rolling is possible only if the motion of the cam
mechanism is oscillatory, whether it be translational or rotational.
The technical specifications for a full size folding cam mechanism are as follows:
1) minimum backward force component on the cases.
2) rate of folding matching, if not higher than, the existing speed of around 4 m/s.
3) capable of processing a range of case blanks as typically shown in Table 4.1.
Height/mm Width/mm
36
CHAPTER 5
COIL CAM
A novel spatial coil cam is introduced as a better method for folding corrugated
fibre boards. Three dimensional cams are known to exist but the unusual feature of
the coil cam as proposed here is that the follower undergoes screw motion of finite
pitch. Even more unusual is the fact that more than one follower can be in contact
with the cam at the same time. The coil cam differs from the "Rotofold" design as
mentioned in Chapter 2 in that the former aims to confine contact to a single point
on the panel and has control over the panel's angular velocity, whereas the latter
operates like a helical bar.
37
5.1 THEORY
The present folding mechanism has a pair of nylon rails that gradually fold the
panels to their closed position. Each panel can therefore be regarded as a follower
in screw motion about the folding hinge; and the pitch of the screw is defined as
the ratio between the translational and angular displacements. The folding rail, in
turn, can be seen as a "static" spatial cam, whose folding action relies on the
relative motion of the follower. A "dynamic" spatial cam, on the other hand, is
one which rotates about an axis skew to that of the followers' screw motion.
Depending on the profile geometry, contact between the follower and the cam can
be chosen anywhere along the panel, which means that the folding force can be
transferred from the leading edge to some other position.
The cam is also in screw motion, though its pitch is zero as there is no translation.
The screw axes of the cam and the followers constitute part of a more general
ruled surface of motion generators called the cylindroid as shown in Figure 5.1
(with acknowledgement given to Hunt[15] for the reproduction of the figure).
The cylindroid, first discovered by PlUcker (1868-9) [37] and named by Ball (1900)
[1], is a twisted, symmetrical ruled surface upon which three screw axes of a three
body system reside when in spatial motion. The nodal axis of a cylindroid is a line
38
in space which intersects and is nonnal to the three axes. Phillips and Hunt [35]
have expounded on the properties of the cylindroid; their work has led to the
derivation of the three axes theorem for three bodies in spatial motion. The
theorem states that for three rigid bodies in relative spatial motion, their
instantaneous screw axes, ISA's, are uniquely defined by the angular velocities, the
intersection of the cylindroid and the instantaneous screw pitches 'h'. If the
properties about any two ISA' s are known, then those of the third ISA can also be
established.
Through studying a scaled prototype, the properties of the coil cam are first of all
generalized. Major design parameters are then estimated for setting up a
mathematical model. After a series of coordinate transfonnation, the cam profile is
generated and is graphically displayed onto a computer. The following section
describes in more details the design synthesis of the coil cam.
A cam prototype, as shown in Figure 5.2, has been manufactured for the
preliminary design study. The coil cam takes the fonn of a bed spring, the curve
of which resembles the intersection of two surfaces: a hyperboloid and a helical
conoid. A hyperboloid is essentially a surface generated by rotating a straight line
in space about an axis skew to it; its properties have been elaborated at length by
Phillips [36]. Except for the variable cross-section of the waist in the middle, the
overall shape resembles that of an augur.
39
~ .. -.,
The operation of the cam is such that for every folding action of 180·, the cam will
rotate through one-and-a-half turns (540"), as illustrated in Figure 5.3.
As already mentioned in the previous section, both the case panels and the cam
undergo screw motion. The screw axes of the case panels and the cam belong to
the same cylindroid instantaneously. This enables the location and properties of
the third screw axis to be detennined using the three axes theorem. (Parameters
for the third screw axis, in terms of dimensions taken from the scaled prototype,
have been calculated as documented in Appendix Sa).
Though not a prerequisite, the screw pitches for the cam and the panels are kept
constant in order to simplify the instantaneous kinematics of the problem. In
practice, however, a constant screw pitch and hence constant angular velocity will
be undesirable because of the infinite acceleration and deceleration towards the
beginning and end of the motion.
40
(-- - -
I
I
,,
\
,
I
x .,-.--"---
-/
/
CASE
{
/
PANEL
I
\
\
\
,,
,,
,-- - --
,
\
\,
,,
\
\
-~..:-
,,
{
/
,, I
I
,, ,, I
, I,
,, \
~ , \\
When viewed along the cam axis in the direction whose component is the same as
41
that of panel translation, then the cam is a right-handed helicoil and rotates
counter-clockwise. The velocity of the contact point on the cam is directed away
from the folding hinge and backwards. The friction force on the panel, as a result
of the shape of the coil, has components in the same directions.
The coil cam is expected to fold cases within a specified size range. For the
largest and smallest cases, the point of contact with the coil cam may not be at the
centre of the folded panel, but anywhere near the centre will be satisfactory for the
present purpose. Having arbitrarily decided that the cam will rotate through 540"
while folding a panel over ISO", the number of panels simultaneously in contact
with the cam is thus determined. With regards to the axial length of the cam, its
lower bound will be limited by the largest panel since there may be a danger of
touching the panel in more than one place. From the design specifications in Table
4.1, the height of an average case size is calulated to be 762 mm. Taking a scale
of 1:10, the panel height for the prototype cam is therefore 76.2 mm.
With reference to Figures 5.4 and 5.5, the following notation is adopted throughout
this section:
42
Table 5.1: Frame notation.
Body Description
2 coil cam
Before defining the dimensions and location of the cam, it is necessary to establish
the motion of the case panel. By symmetry of the folding action, only one of the
two case panels is considered here. The instantaneous screw axis with respect to
the machine frame (1) of the case panel (3), ISA", is aligned with the folding
hinge. Point 'P' is denoted as the mid-point on the panel of an average size case.
This is the location where cam contact is anticipated to take place, though it may
not be ideal for cases at the extreme size range. As the panel screws about the
folding hinge, i.e. ISA", the mid-point 'P' on the panel will follow a helical path
as shown in Figure 5.4. The path is helical only if ro l3 is constant: an impractical
proposition as has been explained previously.
~a
~-.~;.....--.;...:...,~ ~80
"
The location of P at the instants when the panel is in the 0°, 90° and 180° position
are labelled Po, P90 and P'80 respectively. Only three locations are mentioned here
for display purposes; as shall be seen later, more than three locations will be
required for the generation of the cam profile.
43
From the scaled prototype, the path in the x-direction along which the board will
traverse and fold is estimated to be lOOn mm long. The value is conveniently
chosen such that the board panel will undergo a screw motion of 1t about ISA 13
with a screw pitch of 100 mmlrad. A fixed reference frame is best located with
the origin at the centre point of the crease when the fold is at 90°; because it is
reasonable, as an initial guess, to suppose that the cylindroid nodal line will pass
through, or near to, this point. The fixed frame, therefore, is decided with x.
orientated along the crease in the direction of case travel, and z,. vertically upwards.
The orientation of y. is determined by the right-hand convention.
Four quantities are required to specify the location of a ray, i.e a line of indefinite
length, in space. This is true of the cam screw axis, ISA'2' The first variable is
the displacement 'a' from frame A to B along the x. axis, as shown in Figure 5.5.
NODAL AXIS
The next three decisions are concerned with the design variables Z,2' 1jI and U,2' Z,2
is the perpendicular distance between ISA 13 and ISA'2 (N.B the order of the
subscripts signifies the displacement from the screw axis of body 3 to that of body
2). 1jI is the angle measured from the z. axis to the Z. axis about the x. axis.
Estimate from the scaled prototype gives 1jI a value of _45°. U,2 is the angle from
the x. axis (lSA 13 ) to the Xd axis (lSA,,) about the Zb axis.
44
The length of the cam, and hence its angular velocity and pitch, is determined by
the number of boards to be folded at any instant. A design decision is that two
boards, though it does not have to be an integer, will be folded instantaneously.
But there is one moment when three boards will be in contact with the cam - one
starting, one at 90° fold and one finishing.
The next stage of cam generation is to transform the locus as followed by the mid-
point 'P' from the frame of ISA 13 to that of ISA 12• The transformation is shown
graphically in Figure 5.5.
In all, there are three frames in which the coordinates of'the point P will be
transformed from one to the other. The frames are denoted as 'A', 'B' and 'D'
with frame' A' being the frame of origin, Le the machine frame. An additional
frame 'c' is provisionally included should a need arise to examine the nature of
. the ISA 23 • Frames with more than one label indicate axes coincidence.
Characteristically, axes z", Zc and Zd all coincide with the cylindroid's nodal line z.
Referring to Figure 5.5, frame 'A' is first of all translated along ISA 13 to position
'B' at an offset displacement 'a'. The frame is then rotated by angle 'I' to become
frame B. Frame B is translated a distance Z:J2 along z" the nodal line and then
rotated by U,2 to become frame 'D'. Figure 5.6 shows the transformation
algorithms.
Frame A
I
FrameB
z., by z 32
l
Tmmlmc aIoog
Rotulabout xb byo<'32
45
transformation matrices. Since both rotation and translation are involved, it will be
reasonable, therefore, to compute the transformation in a mixture of Cartesian and
cylindrical coordinates. Equations used for the transformation are as follows:
x x -a
r b r • o
The transformation of point P from ISAn to ISA I2 , as already stated, requires more
than the three locations, namely those at Cf, 90° and 180°. The increment of panel
fold angle is represented by Bn.
Like any planar cam design, it is more convenient to regard the cam as fixed and
for the follower (3) to rotate about the cam (2). Subsequent positions of the
follower stroke can then be set out at fixed intervals of cam rotations. This
interval is conveniently labelled as B12 • The size of Bll will depend on the rate at
which the cam will rotate in relation to the rate of fold. In fact, the rotation
46
increments and the angular velocities between the cam and the board panel can be
expressed as follows:
It is not recommended for 00" to be constant as explained previously, except for the
purpose of demonstration here. The time function for 00" can of course be tailored
to the needs. For example, it may be worthwhile to make 00" small while work
done in folding is large. The average <0,.3' however, determines the length of the
cam; whereas 00,..0 will determine the average number of boards in instantaneous
contact with the cam. For example, a decision to make the cam rotate through
5400 for each 1800 fold will imply 00,..0 = 3. Hence, the average number of boards
in instantaneous contact will be oo,.atio divided by two, or 1.5.
The first set of spreadsheet results, based on the above transformations at twelve
increments of 'P' locations, can be seen in Table 5.2. Figure 5.7 shows a plot of
the cam profile as seen along its screw axis.
Pitch.. 100 a. 0
Rx.. ~O...:= -45
ph 3.14.15927
Angle a:(a) x{b)
j'.(aJ/ra.d )V(bJ/rad ",tb) p{b)/nd R%.{b) l' (b) P tb)
0-157.0796 -157.0196 00.785398163 28.28(271 2.963(38" 159.6058 45 169.79252
IS -130.8997 0.261799386 1.047197551 3".6"1016 2.9899764 132.41877
-130.11997 60 171.31303
JO -l04~ 719B 0.523598776 1.308996939 38.637033 3.0430513 105.23026
-104.7198 75 174.35399
45 -76.53982 -78.5]982
0.785396163 1.570796327 40"3.141592778.539816 90 180
60 -52.35988 1.047191551 1.632595715 38.6370H -2.946387 53.373556
-52.35988 105 _168.8156
75 -26.17994 -26.179H 1.308996939 2.09'-395102 34.64.1016 -2.489227 32.9,,5245 120 -142.6222
90 2.132E-14 2.132E-H 1.570796327 2.35619449 28.284271 -1.570796 28.284271 135 -90
10526.1799]926.1799391.8325957152.617993878 20 -0.923626 43.421069 150 _52.91984
120 52.]5967852.359878 2.094]95102 2.879793266 10.352762 -0.63572 65.072092 165 -36.4241
13578.53981678.539811. 2.35619449 ].141592654 4.899E-15 -0.471057 88.13911 180 -26.98955
ISO 104.71976 104.71976 2.617993878 3.40339204.1 -10.35276 -0.353462 111.6201 195 -20.25187
16S 130.89969 130.89969 2.879793266 3.66S191429 -20 -0.258707 135.4058 210 _14..82284
180 157.07963 157.07963 3.14.159265( 3.926990817 -28.28427 -0.178154 159.6058 225 -10.20748
",32.. SO
(llp~31. -45
"'(d) p' (d)/rad -~:(d) x ..p. (dJ/n.d 'f{d.) . Rx Y(dJ' y X
-21.71573 3.7488366 214.79252 _131.0721 -2.907518 -166.5885 93.625293 -166.5885 93.625293 0 -131.0721
_15.358983.7753746216.31303 _106.7022 -2.94818 -168.9183 79.907881 -123.9183 58.7536354.159767 _106.7022
-11.362973.8284494219.35399 _81.36856 -2.972922 -170.3359 67.666121 -80.33586 4.423892( 67.5434 -81.36856
-103.9269908 225 -55.53604 -2.963438 -169.7925 56.429171 -34.79252 _37.6089342.069222 -55.53604
-11.36297 -2.160989 -123.8156 -29.70352 -2.690747 -165.6276 45.777228 14.372391 _45.77079 -0.767689 -29.70352
-15.35898 -1.703829 -97.62221 -4.369877 -2.701942 -154.8099 36.065894 10.190099 _19.77058 -30.18801 -4.369877
-21.71573 -0.785398 -45 20 -2.315069 -132.6446 29.5224.l3 137.35519 16.484668 -24.4914 20
·-30 -0.138221 -7.919645 43.00691 -1.167643 -101.2185 30.590765 213.7215 28.688885 10.6180( 43.00691
-39.6(724 0.1C96777 8.5759026 64.344SH -1.330768 -76.24741 40.817419 283.75259 _0.242997 40.816696 64.344533
-500.31(3416 18:010446 83.820308 -1.071773 -61.(0806 56.944348 J43.59194 _49.(0468 28.316714 83.820)08
-60.352760.4.319364 24,74813" 101.)6856 -0.911961 -52.25152 16.327708 397.746(6 _69.54503 -31.(5485 101.36856
-700.526690830.177162 117.05496 -0.7994.11 -(5.80289 97.6364. 449.19711 _23.97468 -94.6(714 117.05496
-78.28(210.6072(393(.792515131.07207 -0.710033 -40.68189 120.09392 499.31811 70.430995 _97.27294 131.07201
47
11
Problems were encountered on deciding the best way to display the results
graphically. At the time, two proprietary software packages were considered; they
were AutoCAD and DUer. One way of displaying the results was to feed the
spreadsheet data into AutoCAD through modifying the IGES (Inter Graphics
Exchange System) file. An easier alternative was to write a special program which
would perform the transformation calculations. Data generated from the program
were then formatted into input files which the macro commands in DUer would
accept for automatic graphic display. The program, written in UNIX C, required
input for different design parameters such as the velocity ratio, offset displacement,
pitch P13' etc. (See Appendix 5c for the program listing.) .
The principle of DUer was based on the definition of a spline and the attachment
of transverse sections to the spline points. The sections in turn would be joined up
through the B-spline interpolation technique to form a duct surface. Besides being
an ideal display package, DUer was chosen on account of its CAD/CAM
capability. In essence, machining codes for the cam could be created on the same
48
package and downloaded directly onto a CNC milling tool.
In order to make the cam, it was necessary to specify a transverse section for each
of the spline point. A preliminary shape for the sections was chosen to be that of a
table-tennis bat. The semi-circular tip of the handle was where contact would take
place behind the cam and the panel. Through a series of scaling, rotation and
translation, all operated about the spline, the resulting duct, as shown in Figure 5.8,
was in the form of an auger with a core resembling the shape of a hyperboloid.
Interference was suspected between the hyperboloid core and the front edge of the
panel when folding large blanks. To verify this suspicion, a function in DUCT
called 'oblique duct' was resorted to. Essentially, the function would slice the duct
surface with a plane whose normal unit vectors coincided with those of the panel at
various points of contact. Any interference would show up graphically. To this
49
end, a subroutine was added to the source program for working out the normal
vectors of the point 'P' on the panel. The highlighted section on the cam, as
shown in Plate 1 (overleaf), did indicate interference and its exact location.
It was noticed that by translating the 'pear' sections radially from a straight spline,
part of the cam profile might be 'sheared', leading to a degeneration in the cross-
section. The situation was rectified by producing a duct surface with a curved
spline that followed the locus of the centrepoint 'P' on the panel. In addition, the
'pear' sections were replaced by circular ones to ensure single-point contact
between the cam and the panel.
50
PLATE I
DUCT 'slice' function on coil cam design
51
With regards to the optimization of design parameters, an Lt6 Taguchi experiment
was conducted on the following: a, '1'32' U,2' ~2 and r, of ISA 13 . The parameters
were estimated from the scaled cam prototype, and the response that the
experiment was aiming for was the best compromise between a visually balanced
cam and one with a minimum overall dimension. Obviously, a cam which required
a large work-space would be difficult to mount onto the folding section.
Parameters for the optimized cam were as follows:
Offset displacement, a =0
Section radius = 10 mm
Pitch P13 = 100 mm
Screw radius, r, = 40 mm
= _150
'1'32
~2 = 30 mm
U,2 = _60 0
Section increment = 12
52
The operation of the cam was as shown in Figure 5.11.
_~-+
---
__+_+F.=;;olding hinge
b-Q-r-..... Board panel
___ Direction
of travel
Axis of
rotation
Among some of the design considerations, a cam which allowed more boards to be
folded per turn and ran at a slower speed would be more preferrable.
Consequently, the 5400 cam was replaced by a 7200 one with <o"tio = 4, which
meant that on average two panels would be in contact at any instant, except at the
start or finish of contact for which there would be three (see Section 5.2.2).
DUCT, the surface modelling package, was chosen primarily because of its
capability to interface with a CNC machine tool on site. Ideally the cam would be
cut by rotating the workpiece with a fourth axis, while the cutter traversing in the
X-, y- and z- directions. In practice, however, it would be impossible to manipulate
the coil-like cam once the bulk of the material had been removed. As a result,
only half of the coil cam was produced as shown in Figure 5.12. Because of the
limitation associated with a 3-axis CNC machine and the shape of the coil cam,
undercutting was inevitable. Instead, a female mould was reproduced on the same
block of material, demonstrating the feasibility of manufacturing the cam by means
of casting.
53
Fig. 5.12 Coil cam machined mould.
Leung [25] has investigated ways of manufacturing the coil cam prototype, e.g
direct machining, casting, thermal fonning and stereolithography are among the
possible options. Casting, as Leung has identified, may produce poor surface
texture and the two halves of a die cast may not separate with a consistent split-
line. In thermal fonning, the difficulty is associated with the control of hot air
flow over the malleable material. Stereolithography relies on the hardening of
special block modelling resin by means of a laser. As the laser beam is directed
into the photosensitive liquid resin, solidification will take place layer by layer.
The capability of stereolithographic systems that are currently on the market is
limited by the size of the workpiece, and the material used - modelling resin -
which may not be suitable for direct engineering application. Short lead time and
good surface finish, however, make it an excellent tool for prototyping. Other fast
prototyping techniques can be found in a review on rapid prototyping by Sze et al.
[46].
5.5 APPRAISAL
In its present form, the cam will fail on account of an initial assumption that the
screw pitch for ISA'2 is constant. What this implies is that the acceleration and
deceleration at both ends of the cam are infinite. For a rigid panel, there will be
54
an impulsive force acting at the beginning of fold, to be followed by a crushing
force at the end, both of which are detrimental to the case panel. A gentler motion
profIle (e.g trapezoidal velocity profile) can be implemented by changing the
function of 0013 with respect to time. A more sophisticated motion profile will be
one which takes into account the folding resistance of a typical board sample.
The design of the coil cam has not been pursued any further because of the
following reasons:
1) difficult to manufacture
2) a similar device, i.e Rotofold, has been patented, though its design is not
as refined as the coil cam.
3) backward component of friction may still contribute to fishtailing
4) cam whirling at high speed
5) difficult to balance the cam dynamically
6) over-sized cam structure, hence the difficulty in mounting of the coil cam
onto the folding section
5.6 CONCLUSION
Based on the screw theory of three bodies in relative spatial motion, the coil cam
has been designed with its folding force directed at a single point normal to the
case panel. In this way, less ink-smearing and panel distortion will be made
possible, when compared with the folding rail design. A methodology which
utilizes a high-level software program to run a proprietary CAD software by means
55
of macro commands has also been demonstrated as a useful design technique. A
prototype mould has been manufactured and other means of producing the cam has
been reviewed, but the design has not been adopted on account of its practicality.
56
CHAPTER 6
TWIN CAM
This chapter describes the synthesis and the design of the first twin cam folding
mechanism which imparts rolling contact onto the follower surface.
6.1 NOTATION
The following notation, as illustrated in Figures 6.1 to 6.4, has been adopted
throughout the theoretical synthesis of the twin cam folding mechanism:
9(t) - angle between the folded panel and the horizontal; it ranges from fJ' to 18fJ'.
0)0 - maximum angular velocity of board panel during folding.
v - board translational velocity in the machine direction.
I - horizontal displacement on the panel of the instantaneous point of contact with
the cam and the first contact point.
c - distance between the first contact point and the folding hinge.
r(t) - instantaneous cam radius.
e(t) - instantaneous eccentricity at periphery of cam for minimizing friction contact
(see Figure 6.3).
57
b - datum line distance as defined between 'B' and '0' (see Figure 6.4).
x,y - local cartesian coordinates in the plane of the cam.
[3(t) - cam angle from moment of contact.
",(t) - instantaneous angle between y-axis and r(t=O).
s(t) - accumulated peripheral length of cam.
In the design of cam mechanisms, the first task is to specify the motion of the
follower. During the folding operation, the board panel can be seen as the follower
which 'screws' along the folding hinge. A preliminary step will therefore be to
choose a motion profIle for the cam follower, and express it in terms of some
known design parameters.
The translation velocity 'v' of the panel is not subject to choice since it is a design
specification. However, the angular displacement 9(t) can be chosen arbitrarily. A
parabolic function has been selected for the angular displacement of the follower;
though this would imply infinite acceleration and deceleration at the beginning and
end of the motion. Figure 6.1 shows the parabolic motion profIle expressed in
terms of 9 and its higher derivatives.
e e
w. 1::1 .. 11
2't
oLLl 'T 21"
58
For the first 90" fold, the motion equations are as follows:
e = ex
e = ex.t
e= '!'ex.t 2
2
and for the second 90" fold, applying the boundary conditions will give:
e = -ex
e = Wo + ex{"t - t)
e= wo(t - "C) + ex ("C 2 + t1
ex."C.t - 2" +-
11
2
where,
11
"C = -
v
Note that the right-hand side of the basic equations can be adapted for functions
other than the parabolic one, e.g. trapezoidal, cycloidal, etc ..
If the x-direction is defined as opposite to the board travel, then the design
parameter 'I' will correspond to the x-component of the total contact length
between the board panel and the rolling cam. The value of 'I' is equal to the
height of the shortest case less some tolerance, as shown in Figure 6.2.
Direction of travel
59
The first cam is derived as follows: consider Figure 6.3, which shows the plan and
side views of the proposed cam,
I
panel t--c _ _ '.
I
:----. 1~50-S
• J ,
i
I
cS; .
p(l
'-
!
;'1
- '- ./
S ( ......
/ """ folding hinge
/
/
!
",.
the cam is orientated at 45° to the horizontal. Cartesian coordinates on the cam are
defined such that the contact point made by the cam and the board panel is
enclosed within the x-y plane. In order to avoid rubbing, the plane varies
instantaneously along the cam axis, the motion of which will be generated by an
offset dimension 'e(t)' in the z-direction (hence the term quasi-planar). Point 'Q'
lies along the length 'I' as projected from the point 'P' along the same line, and is
defined by the intersection between 'I' and the y-axis. Expressing the kinematics
in terms of 'x' and 'y' will give:
-i = OQ.
:.V = r(t).cos[~(t)
*(t) .
- ljr(t)].Ijr(t) ........... (1)
y
PQ.*(t) =
:.c.6(t)cos[45° - 6(t)] = r(t).sin[~(t) - Ijr(t)].*(t) .........•. (2)
60
On examining equations (1) and (2), there are 4 unknown quantities, namely,
The last two quantities can be established by means of the finite difference method:
where 'i' represents the ilb term in the fmite difference table. Finite difference is
resoned to instead of analytical methods, since cam variable dependency pennits
solutions to be derived only by such means.
One more equation is required in order to solve for the four unknown quantities.
Referring to the cam geometry as shown in Figure 6.4,
-- "/
"--"·L
-",
x'
QF = c sine
FD = c cose
61
BG = a cos45°
OH = r(t) cos[/3(t)-'I'(t)]
GJ = r(t) cos[/3(t)-'I'(t)] sin45°
QH = e(t)
FI = e(t) cos45°
Also,
OC = OB = a
CH = OH - OC
HI = CH sin45°
QF = HI + QH cos45°
On the condition that the cam axes are orientated at an angle of 45°, hence the
terms 'sin45°' and 'cos45°', 'e(t)' and 'a' can be eliminated by adding equations
(4) and (5):
which provides the 4th equation with the same unknown r(t), /3(t) and 'I'(t). Since
/3(t) always appears with 'I'(t), the term '/3(t)-'I'(t), can be conveniently replaced by
$(t).
0
tancj)(t) = c.6(t).cos[45 - 6(t)]
v
hence $(t). Substituting into equation (6) will give r(t), which in turn will
determine the camshaft speed i.e. the fIrst derivative of'l'(t) from equation (I).
Incidentally, the camshaft speed must be variable if slipping is to be avoided. 'I'(t)
62
itself can be calculated from the finite difference equation (3), and e(t), from
equation (4).
For each of the twin cams, a spreadsheet with 300 time increments was set up to
calculate the following parameters:
63
Spreadsheet data for the parameters r(t), ~(t) and e(t) are as shown in Table 6.1.
Figures 6.5 and 6.6 show the twin cam design and arrangement respectively.
Rol..~inq
.--. ...., oontaat
/~ '~
(' I ' . '"
. . __ ~ __
l( "-
\ \. I' /------ -- ~
,,'-..,... /
''1
Cam for second
Ko.L.L~ng
9010lil t:ontact
64
Fig. 6.6 Twin cam assembly in operation.
6.4 APPRAISAL
The twin cam folding mechanism has been demonstrated to be a feasible folding
alternative. Further improvements, however, are necessary for a fully operational
cam system.
At present, a parabolic motion profile has been adopted for the cam design.
Unfortunately, this kind of profile is responsible for sudden acceleration changes
and is therefore prone to vibration. When running at high speed, not only the
nominal acceleration but also its time derivative will become critical in the overall
dynamic integrity of the cam system. The significance of this time derivative, or
the 'jerking effect', has been a subject of investigation (refer to Olmstead [31],
Hrones [14]), and the subsequent recommendation has been the modification of
motion profiles which will minimize this jerking effect (see Chen [5]).
In addition to making provision for the high and impulsive acceleration changes,
the cam's motion profile can be correlated with the folding resistance profile (see
Figure 7.4 in later section), leading to a more sophisticated motion profile design.
The idea is to dampen the buckling effect at the folding peaks (refer to Section 7.3)
by means of a gradual cam motion: the result is a more controlled folding action.
65
The distance between the contact point and the folding hinge, i.e dimension ' c', has
been chosen, though arbitrarily, within a reasonable scale range. For a prototype
design, the dimension will need to take into consideration the largest possible blank
size to be folded. The reason being that for larger case blanks, the centre of
gravity in the case panel will be shifted further away from the folding hinge. If
the location of the centre of gravity lies beyond some critical point, a shon
dimension ' c' together with a fast folding action will lead to accidental panel
buckling at the point of cam contact. The choice for the length of ' c' will
therefore have to be assessed in terms of the size and mass of the largest case
blanks.
6.5 CONCLUSION
The twin cam, another innovative design, has been examined as a folding
alternative. In the twin cam design, the folding operation is divided into two
stages, each of which is responsible for a fold of 90". Rolling contact, an
imponant feature if ink smearing is to be avoided, has also been incorporated into
the design.
66
CHAPTER 7
REVIEW ON CREASING
Creasing is one of the fundamental stages in the case conversion process. How
well the case is creased will directly affect the fold quality, and hence the strength
of the case and its appearance. Essentially, creases are score marks along which
folding will preferentially take place. Pan of the creasing mechanism involves the
weakening of the corrugation. Sufficient weakening is necessary for the board to
fold along a predefined line; but when in excess, the case liners may tear, one
consequence of which will be the lowering of the case compressive strength. It is
not surprising therefore to find that much interaction exists between the creaser and
the corrugated fibreboard.
Sporadic publications since the early 50' s (see chronological summary at the end of
this chapter) suggest that research into creasing has always been driven by the need
to innovate. In particular, new creasing profiles have been developed and tested
with increasing sophistication. Optimization is often achieved through
experimenting with the appropriate design variables, such as the profIle geometry
and the depth of penetration. The lack of standard procedures, however, in
assessing the crease quality has no doubt downplayed the importance of this one
process.
This chapter examines the theory on creasing through looking at the interaction
between the creaser and the case. This is followed by a review on previous
research into the design and optimization of creaser profiles. The chapter is
concluded with a survey on crease quality assessment.
67
top and bottom openings of a case, and are normally produced on the corrugator.
Parallel creases, on the other hand, do not appear on the case until the conversion
process, where the four vertical edges of a case will be defined by means of a
rotary creasing blade. Here, only parallel creases are considered, as they are the
ones produced in a case convener.
The mechanism for creasing relies on the compression of the corrugated structure
and the definition of a score mark about which the case will subsequently fold.
Compression enables the section thickness to be reduced, thereby, resulting in a
compaction of the buckled flute medium between the liners. On a case converter
or, more specifically, a flexo-folder gluer, this kind of mechanical straining will
take place between a pair of creasing rolls. The score mark is in turn produced by
the shape of the creaser, the design of which will be discussed in more details in
the next section.
The corrugation structure derives its strength from a good stiffness-to-mass ratio. In
other wonis, for an equivalent mass of board material, the corrugated board design
will provide a higher bending stiffness than that of a normal sheet. The orientation
of the corrugations is critical not only to the bending stiffness of the board, but
also the crease performance. In parallel creasing, depending on the position of the
crease relative to the flute structure and the creaser profIle, different forces can be
transmitted from the creasing tool to the corrugation.
68
CREASI>13 LOAD
As Vogelpohl [51] has pointed out, the creasing load, depending on the profile
geometry and the point of load application, may compress the corrugation in such a
way that sufficient tensile forces will lead to the tearing of the inner liner. On
inspection, the ideal position for creasing would be at location 'A' - the mid-point
between the two fluting peaks. This permits the creased liner to stretch evenly
without undue tearing. Another crease location which bears symmetry is that of
'B', though a greater damage will be imposed onto the flute structure as a
consequence of being compressed over a wider area. Location 'C' is considered
the worst in that an asymmetric load pattern may lead to an unpredictable buckling
behaviour. Both Vogelpohl [51] and Grebe [9] have acknowledged the high
tendency to tear when placing the creasing load at location 'C'.
Since creasing and folding are inter-related, a word on the folding mechanism will
clarify the role of a crease. Essentially, folding involves the buckling of the
corrugation along the crease. As the fold approaches 180°, i.e when the two case
panels come close together, there will be a potential risk of crease wandering. This
has been suggested by Grebe [9] who attributes the cause to the cramming of the
flute material and the simultaneous stretching of the outer liner. The result will be
a redistribution of stress between the inner and the outer liners, leading to a lateral
deformation of the case panels.
Creasing mechanism also depends on the physical properties of the liners such as
the Young's modulus, flexural stiffness of the board, bending stiffness of the liner
component and friction of coefficient. But by far the most important is moisture
69
content. Vogelpohl [51] has alened to the fact that a loss of moisture content in
the liners will increase the possibility of tearing during the creasing stage. The
behaviour of liners which are composed of shoner fibrous material (e.g chip liner)
will be more susceptible to the fluctuation in moisture condition than that of liners
with longer fibrous material (e.g Kraft liner). Not surprisingly, Vollmer [52]
recommends a relative humidity of 60% as the creasing condition for the chip
liners, and 55% to 60% for the Kraft liners.
Moisture condition can also affect another property of the liners - coefficient of
friction. Runnability is a term for describing the physical behaviour of the flute
medium when being passed through a labyrinth in order to be shaped into the
corrugation profile. In the investigation into the problems associated with
runnability, Thomas [50] has discovered that the coefficient of friction will go up
by raising the moisture content in the liners. Whitsitt [55] has emphasized how
important the coefficient of friction is to the runnability in a corrugator. More
precise still, according to Sprague [45], the coefficient of friction actually decreases
with surface contact temperature. Subsequently, Sprague concludes that the total
effect of heat and moisture on the medium properties is very complex and has been
poorly quantified in the past.
The high speed nature of the case-making process has been addressed briefly by
Vogelpohl [51], who suggests that case liners may tear due to large accelerations
and inertial forces. Quoting from other research evidence, a creasing speed of up
to 15000 cases per hour (cph) is, according to Vogelpohl, still not high enough to
cause tearing of the liners. Experimental evidence collected by McKee and
Altmann [26] has shown that the creasing speed has no significant effect on the
foldability and the crush resistance of cases. Present research evidence as
70
described in Chapter 3 also confmns the relative insignificance of the machine
speed in the control of fishtailing.
The choice of creasing profiles is often a subjective matter, and there is no such
thing as a single standard profile. Among the various profiles which are
commercially available, the single blade creasing profile is by far the most
common in the corrugated industry. The blade is normally made of chrome-plated
steel, and its female counterpart is moulded from polyurethane. The choice of
having two different types of material for the creasing unit is such that, ·when the
creasing blade acts on the corrugation, the polyurethane counterpart will provide
compliance for the paper to form into the shape of the blade.
Other profiles have also been developed but their application has not been as
widespread. A range of profile examples can be seen in Figure 7.2. The single-V
profile is a forerunner of the single blade one, and has been extensively studied by
Vogelpohl [51] and Buchanan [3]. Double-V, 3-point and 5-point profiles find
their application in the creasing of thicker board types. Asymmetric profiles have
also been reported by Vogelpohl [51], though it is not known to what extent the
profiles have been successfully adopted in industry.
l----....-J ~
l J
11 ~ ~
Double-V 3-polnt S-point
Single-V
71
width thus created will be detenmned as a function of the board caliper. On
imparting two score lines, the case will fold with two 90" corners with the tensile
stresses in the outer liner being markedly reduced. There are problems, however,
related to this type of creasing, e.g dimensional inaccuracy, case strength and
appearance. Two sets of creasing shafts, as quoted by Shulrnan [43], have been
used for producing wide creases on case blanks, but further development has been
abandonned on account of weakened case corners. Experimental evidence in
Chapter 9 will shed some light on this subject matter.
Crushing prior to creasing is another way of creasing cases which are more than
one-board thick. Rolls with rounded off edges, or in the form of barrel-like shapes
are the preferred norms. From his experiments, Vogelpohl [51] has concluded that
a symmetric barrel-shaped crush profile gives the best performance. It is also
recommended that crushing must not be too intense, or else subsequent creasing
will not produce a sharp fold line. The trade-off with pre-crease crushing,
however, is a reduction in the compressive strength of the case. Evidence put
forward by Buchanan [3] has highlighted the fact that crushing before creasing
with a 3-point profile can reduce the case compressive strength by as much as
25%.
Past research into the development of creasing profiles took the form of
experimenting with various shapes and profile combinations. Single- and double-
V, 3-point, 5-point, round, asymmetric profiles were among those investigated by
Mckee [26], Buchanan [3] and Vogelpohl [51]. The profiles were all standard in
that they had been accepted by industry and were readily available on the market.
Paradoxically the existence of more than one "standard" profile was a controversial
matter and could only be resolved by the customers themselves.
In 1980, Grebe [9] broke away from the approach of comparative study on
standard creaser profiles. Instead, through a series of factorial experiments, Grebe
managed to identify those factors and geometrical dimensions that were critical to
the crease performance. Single creaser profiles were tried out with and without
72
female rubber counterfonn. Factors such as the slot width on the female
counterfonn, the depth of the slot and the compression ratio were assessed in tenns
of the accuracy of fold at 90· and 180·, the fonn and the consistency of the crease.
A crease would be considered inaccurate when excessive straying took place
between the crease and the line of fold. Crease consistency was expressed in tenns
of the extent of damage due to tearing. Edge exactness and curvature, collectively
known as the fonn of a crease, were judged by visual inspection. Results showed
that the quality of the board liners was equally important in as far as the crease
perfonnance was concerned. Only boards with high quality liners were found
capable of withstanding the tearing forces induced by the creaser profile.
As yet, there is no standard method for assessing the crease quality of a case.
However, tests on crease quality can be categorized into the following subgroups:
strength, damage-related and appearance.
73
stiffness, on the basis of comparing the individual contribution to the overall case
compressive strength.
Compression Iood
/lbf
[000
500
Displtlremonl
c __ ["
Compression distance
Fig. 7.3 Edge-wise compression curve.
74
Anal< of fold
00
The first peak, normally lies between 80 and 160 , measures the force required to
break the crease; its magnitude is a function of the corrugation type and the
flatness of the material. Buchanan [3] has been able to show that this peak value
increases with liner caliper and decreases with the depth of creaser penetration.
The first peak, as discovered by Vogelpohl [51], also provides information on the
crease formation, the material type and its dampness. The folding resistance then
drops to a constant level before it picks up again at around 1350 •
The second peak lies towards the end of the folding process, i.e between ISO" and
1650 , and its magnitude is much greater than the fIrst one. Vogelpohl suggests that
this second peak depends on where the corrugation is attacked by the creaser. An
account by McKee et al. [26] has shown that the folding resistance will increase as
soon as the board is folded to the point where the inside liners on each panel begin
to butt against each other in the creased region. The folding resistance continues to
go up until the weaker of the two panels becomes unstable. Consequently, a
redistribution of stresses will be accompanied by the deformation of the inner liner
and the flute medium. A high second peak is undesirable, as it will encourage the
constant reopening of the glued flaps immediately after they have been bonded
together.
Quality assessment methods which are damage-related can be grouped into those of
liner tearing and false creasing. Creasibility has been defIned by Vogelpohl [51] as
the minimum gap width at which the inner liner of a certain board caliper will not
tear. An alternative measure can be expressed in terms of the compression ratio as
75
shown in Figure 7.5. Note that the definition will also take into account any gap
reduction due to pre-crease crushing. However, negative gap width is inevitable as
the male creaser may dig into the female polyurethane rolls.
Case appearance is vital from the aesthetic and the strength point of view. Position
of edges, cracking of liners, pressure ridges on narrow folds and crumpled internal
rib on wide folds are all examples of case appearance criteria. Apart from the
position of edges, which can be assessed using the fishtailing measure as defined in
Chapter 3, case appearance can only be treated in a subjective manner.
As a concluding remark, it must be emphasized that the crease quality also depends
on other preconditions. For example, Edholm et al. [7] have found that a crease
will lose its effectiveness if creasing a piece of warped corrugated fibreboard.
Furthermore, there is no point in producing a good crease when the folding
mechanism does not match up in terms of quality standard.
76
Chronological summary of major creasing research to date:
1956 McKee and Alttnann [26] compared different creasing wheel contours and
their effect on the foldability and compressive characteristics of corrugated
boxes. The quality of the creased samples was quantified in terms of the
folding resistance.
1963 Buchanan [3] experimented with several types of creaser profiles, fluting
medium and liners. The compressive strength and the foldability of cases
with both parallel and tranverse creases were compared in a series of
factorial experiments.
1980 In Munich Fachhochschule, Grebe [9] also resorted to factorial design for his
3-point creaser profile experiment. The properties monitored were the fold
accuracy (relative to the crease location), no-tear consistency and the final
form. Grebe then quantified the fold quality by means of a special folding
rig, which had the capability of measuring the rate of folding, the angular
displacement and the folding moment.
1987 Major research into creasing was conducted by Vogelpohl [51] at the Munich
Technical University. The research was funded by companies in the
packaging industry and the Research Council for Packaging and Food
Technology. Vogelpohl investigated the conditions responsible for better
creases. Among other design parameters, creasers with different number of
knife edges were tried out on B- and C-flute samples. Once creased, the
boards were then folded on a pneumatic folding rig. The results were found
to be in good agreement with those obtained by Grebe.
77
CHAPTER 8
CREASER DEVELOPMENT
The development of new creaser profiles has traditionally relied on the empirical
and comparative study of profile configurations. Any theoretical analysis without
taking into account the complex behaviour of paper will be penalized on the
grounds of over-simplification. But to what extent is the creasing effect due to the
corrugation and the inner liner individually?
This chapter commences with a geometrical analysis of the single blade and the
twin roll creaser profiles. The analysis proceeds with the hypothesis that the
fluting medium is significantly more elastic than the inner liner, hence ensuring the
inner liner to be the first to fail as a result of tearing. Design parameters generated
from the analysis are then evaluated in the finite element model. Results from the
theoretical and the finite element models are subsequently validated using the
Electronic Speckle Pattern Interferometry (ESPI). The chapter is concluded with a
summary of how well the theoretical analysis and modelling have correlated with
the creasing mechanism in reality.
Crease quality depends on the mechanical creasing action and the material
properties of both the liners and the fluting medium Variation in the crease
quality may arise as a result of the rheological properties and the fibrous
composition of paper. What is predictable, however, is the mechanical interaction
78
between the creasing profile and the corrugation structure, the latter of which may,
under cenain circumstances, be treated like a truss.
8.1.1 Assumptions
ii) Creasing conditions such as temperature and relative humidity are kept constant,
which means that any fluctuation in the paper properties due to external
factors can be ignored.
Note that the above assumptions do not, in any way, concern the physical
properties of paper. i What the assumptions try to achieve is to address the creasing
mechanism from a structural rather than a material viewpoint. Only an
experimental verifIcation, as described in Section 8.3, will prove whether or not
this is a justifIable approach.
79
8.1.2 Single blade creaser
Creaser profiles come in various styles, but the most common is the single blade
profile. Figure 8.1 shows the design of the single blade creaser.
~\ fijJtf1\ 7 \
Fig. 8.1 Single blade creaser.
The single blade creaser is extensively used in the creasing of the single- and
double-wall boards, and pre-crushing is often a pre-requisite for the latter.
The worst scenario for the single blade design is when it penetrates right down to
the trough of the corrugation at location 'A'. As shown in Figure 8.2, if the flute
structure remains intact during the creaser penetration, then the inner liner will
have split long before it reaches the bottom of the corrugation (see Appendix 8a
for a simple theoretical treatment). In practice, however, as the creaser comes
down onto the corrugation, the flutes will collapse due to the crushing action of the
shoulders, before any tearing can occur in the inner liner. The importance of
crushing will be emphasized repeatedly in later sections.
, .
Fig. 8.2 Worst creasing scenario.
80
8.1.3 Twin roll creaser
The variable gap twin roll creaser, as shown in Figure 8.3, is an inversion of the
single blade creaser. Instead of actively creasing the corrugated board by means of
a blade, the twin roll creaser relies on the gap between the rolls to perform the
creasing operation. As the conical shoulders of the twin roll come down onto the
corrugated board, the edges which form the twin roll gap will imprint two score
marks onto the board.
According to Werner [54), part of the creaser profile must possess at least one
acute corner for good crease definition. Sharp creases, however, do not necessarily
guarantee a good fold, as uncontrolled flute buckling may arise due to an
insufficient crease width. In this case, the crease width is determined by the
separation between the two score marks. One immediate question is: how to set
the twin roll gap? Should it be based on the board caliper or the flute geometry?
When a piece of board is folded with a crease width, or in this case the score-to-
score dimension, equivalent to the board caliper, the case edge will be chamfered
as shown in Figure 8.4.
1\ Ilf~:~\ ;"
Fig. 8.4 Fold with two score marks.
81
The ease of fold from having two score marks may, however, be offset by a poor
case appearance and extra material cost. If, on the other hand, the score-to-score
dimension is a function of the flute geometry, then the interaction between the twin
roll creaser profile and the corrugation will be the deciding factor. Assuming rigid
suppon for the corrugation, it can be shown that there is a maximum twin roll gap
beyond which tearing of the inner liner will take place (see Appendix 8b). The
angle for the conical shoulders on the twin roll is seen to be a function of the liner
strain rate and is independent of the flute pitch; its value is found to be
approximately 10". Table 8.1 summarizes the design parameters as derived
geometrically for the twin roll creaser. Theoretical values for the creaser gap
width with or without slipping are based on the maximum permissible depth of
creaser penetration beyond which tearing becomes inevitable.
Table 8.1: A summary on the twin roll gap settings and total creaser width.
Flute type Gap width without slipping Gap width with Creaser width
slipping
There is a strong argument for slippage to occur between the inner liner and the
creaser. For if there is no slippage, tension within the inner liner will not be
evenly distributed. In this way, the current single blade design will not be able to
penetrate to its effective depth without inducing tearing - the effective depth being
the protrusion of the blade as measured from the creaser shoulder level. This is
true provided that creasing is conducted at ambient temperature and relative
humidity. Increasing the moisture content of the inner liner, however, will raise
the friction coefficient between the creaser and the liner to the extent that slippage
may not take place. The effect of moisture content on friction coefficient can be
found in various publications by Thomas [50], Vogelpohl [51] and Vollmer [52]
(see Section 7.1).
82
8.2 FINITE ELEMENT ANALYSIS
This section describes a finite element analysis on the single blade and twin roll
creasers using a proprietary software package called Mechanica [29]. The
objective is to examine the effect of placing a single blade creaser on each of the
three locations 'A', 'B' and 'C' along the corrugation as defined in Section 7.1.
Results will then be compared with those from the twin roll creaser.
83
Properties for the final model were as follows:
Structure - B-flute corrugation; 9 flutes in all (10 troughs and 9 peaks)
Element type - 2D shells
Average shell size - 0.16954 mm x 0.2 mm
B-flute board caliper - 2.5 mm
Liner & flute thickness - 0.2 mm
Young's modulus E - 1.3 x 1<1 Nm·2
Liner density - 0.625 kgm·3
Poisson's ratio - 0.3
N.B. The material properties were based on those of the 125 gm·2 chip liner.
Locations ' A' and 'B' could easily be identified in the model, since they were both
symmetrical in relation to the corrugation. With location 'C', however, a point had
to be arbitrarily chosen between 'A' and 'B'. A position some 7 elements (or
1.186 mm) away from the nearest peak 'B' was subsequently defined for location
'C'.
The loading condition was such that a single blade of 0.6 mm wide would be
represented by 4 elements, all being subjected to a downward displacement of 0.2
mm. Loading for the twin roll creaser was slightly more sophisticated in that the
conical shoulder was represented by a 7-element long 2D shell. One end of the
element, being the tip of half a twin roll creaser, was constrained in both the y-
and the rotational z-directions. The former provided the required downward
displacement, wbi.le the latter enabled the angle of the conical shoulder to be
modelled at an inclination of H1'. The twin roll was separated with a gap width of
0.59 mm (see Table 8.1), equivalent to approximately 4 element length. Slippage
was allowed by freeing the loaded elements in the x-direction.
84
8.2.2 Procedures
In order to simulate the reaction due to the creasing action, the base of the
structure, i.e the outer liner, was restrained in the vertical direction. The present
software could only simulate the constraint as if the outer liner was glued to the
lower roll. To overcome this limitation, which was essentially a boundary element
problem, a constraint strategy was devised as in Figure 8.5:
-
ComoraIn
one trough
Once the constraint conditions for the outer liner were met, the corrugation was
ready for loading. The ftrst analysis modelled a single blade penetrating to a depth
of 0.2 mm at locations 'A', 'B' and 'C'. This was followed by another analysis
performed under the same conditions but with the twin roll creaser at a gap width
of 0.59 mm.
85
,
From both the displacement graphs, the loading and constraint conditions were
checked if they had been applied correctly. Membrane strain energy, representing
the work done on the 2D shell elements by the displacement load, would indicate
locations where buckling would most likely take place. Due to the nature of
loading, bending rather than tensile stress would constitute the main component in
the maximum principal stress. Where there was a high level of stress
concentration, tearing would be anticipated.
8.2.3 Results
Results, in the form of colour graphical plots, can be found in the following
labelled plates:
A 2 lowCA
B 3 lowCB
C 4,5 lowCC
A 6 twin_A
B 7 twin_B
C 8 twin_C
A blown-up y-displacement plot for 'load_C' is also included for showing the end
deflections of the corrugation. Scale for the deflections is 1: 1. The sign
convention on the graphical display complies with the cartesian coordinate system.
86
IU ... a _ '(looa--, ~
~f ___ O"'.,LA<ol -.t A + _ JI.-o~
...... " .......
kr.." .t.U.fa-"
Dol _ _ o,rl..,l.. ol _1
..... ClU• • 2.0010..... Il0l_ ClI..q +J.un... u
... h I.uooao" . ...100 1.000 .... 00
Load
-
-lUH·"r
*** - .... •
.
1- •• 915.- .BSE -0'
/ P' -9.14!1l-05
* { ~ -< 1
*.
'---------- ~
G -l.13Z£-04 G -2.6271-06
Constrained elements
J -1. 7831-04
....... . . ....
.... ol.onu-Ol
.t.,._o "". 'u_
...... tl.un.·n
M,i.. . . . . . . . or... oo IU n -I.Udlt-tft
D . fo~_ o..~1 ... 1 ~ Orj,.,J~d_1
.... o.u, -Ol.Off •• -1t
.ou. 1.lOohtOO -+9 . 8321-03 +1.251&+01
+8 . "/.foOE-03 +1.1161+01
+1.647&-03 +9 . 151!+06
+2.711&+06
N
i'
+1.0921:-03 "'1.1031+06
IUA ~l.OOOU.~U IOU .. -1. u"l-n
fIo<It.~_ 0.."" ...
..... Du,. ~2.U"a..~
1 IIM,e.l o.f.~ Or~"'-l ~
. . . OUp .2 . .. 06 ... 04
...u. 1 . • 0UlhU .. -.1 .. 1."0".01 A 5.10Si:-DS
Load
~ltsl[·t>.r
** * )to *** tI
D -l.9atE-OS
*
,....,.-
....
! -6.6538-05
..*
* " *......* *
! - 2.9741-08
F -9.l22E-05 r -1.430£- 05
Constrained elements
I -1.7331.-04 I -5.7lU-05
"'1.1368+01 +6.555&+07
2
_ _ _..,,..~~_ _ _ _ _ _............_ _ _ _ _ _..,,.....~_ _ _ _ +8.5181"'00 ... 4.820E+07
ZS ZS ZS: ~7.0i81+00
+5.6791+00
"'3.953&+07
....3.085E+07
"'.2591+00 +2.2181+07
11 1
+2 . 8391+00 +1.,3501+07
Load
D +5.191.&-05 D +3 . 653&-05
.* ~~ ~ *. j;> . . .
ft
I
Ii: '1"1.492.&-05 •• E +3 .02(1.£-05
* *
;
*-
.,* - ~ * ......... ~ ..... * F +2 • • 031-05
'------------- ------
Constrained elements
+J ' H "l (-or
+llr H -or
I -1 . 5"1Q1!;-O • I +5.28£1-06
H.381£+OO +6.2611+0"1
+1 . 0J(jE+00 +4..660£+07
zs: z~ sz
" +1 . 631.... 01
" +J.8591+0"l
. +6 . 9051- 01
h
+3 . 0511+07
"0
+5 . 1'191-01 +2 . 251!+O"l r
;J>
11
+3 . 452.1-01 +1. '5'71+07 @
.j>.
+1 . 726&- 01 +6.5591+06
Max +2.2985E-04
Min -2.0000E-04
Deformed Original Model
Max Disp +2 . 2985E-04
A +1.869E-04
Scale 2.0000E+OO
I I oo H ·o~
~* .
CD
o
t / E +1. 492E-OS
Constrained elements
F -2.80GE-05
I -1.S70E-04
..... H.UIi~n ........"lur.··II.
IoUn - • • • UU_U IUa _::UUIl·II'
O.t... _ OU,lna1 _ , Oo~ • . . - Oc1-t1Jod ......1
.... D~ .~_JUn.._G4 ..... Du,. .'.nua-u
.... 1. 1.00001..00 " 8, !l36IH)6 'ul. 1, o...a.oo •
Load
lJLL Cl of .4281.-0"
* *
It -1..0.0&-0-4
* * *
F -1..315£.-04 F -5_4871--0'
~----~---~--
Consrrained elements
I -2.0SlB-04
L I -2.186£-06
...............
..... ,',OlU&-.)
u ...... ....... ~
...... l.)u~~n
""' • • 0.0000 ... 0' IU.OI • ... UU~IZ
0. 1 • . - - DU", ... ! _ . 1 ~f.'- _,,,u.ol ......1
.... ~ DU., .I.nu..·,. ..... Dut> ·a_*n •• _Ot
1.010 1.10001-900 t2.125&-03 •••1. 1.0000&'"
+5.861£+01
+-2.120£- 03 +5.1342.+0.
+1.817£-03 +4.401£+0.
/ +3.661&+0.
+2 . 93.£... 07
lA
+1 •• 6'1£.... 01
+3.028£- 0.
L +"1,334£"'06
d
t . ~._"
lUoa ...... 111:-01 IkJo . 1.IUU\-Cl!I.
IUA ·.2 . unz..Ot .... n -'7.U4fr.-U
~(._ o ..:I.,J.n.d. _ . \ tI. f .. ~ or~,~n..-1 .......1
Ma_ tu_.,. .~. U41~O. w.. 0.1. . .2.un._H
a.u .. t.ClOOO&+U . . . 1- I.uoa.. oo .. .... 1&
Load
-*
lJLl o +1.S21~-O!
*
..,
-* 14HIf'-* ,'* £ -1.34"-E-05
* #
*
*
... ""* * F -1.0U£- 0 4 * ... * * F -1. 533£- 05
Constrained elements
H.2978.+01 ·1.000£+08
..1.135£+01 '1-8.702£+07
s: Z 'S Z s:
+9.730£+00 '1-1.400&+01
7 7
" " , , +8.108£.+00
+6.481£+00
+4.865E+OO
..
/
? .~
,
-, Z " " " ' \ +6 . 091£+01
H.795£+07
+3.493£+0'1
"tl
, l'
>-
+3.24.3E+OO +;:.191£+0'1 ~
Lx ~ ..1n_.
+1.622£+00
Lx ~n"_'
I
Hi.1'I82£+06
-..J
11• • • !. U1 ...·U ,..•• '.,ffu.--O\
11.\... ·.I."H"·O~ ...a ·•. u.n-,'
o.l .. ~ Qo,~~ ... ' ....... ; o.f • . - . I 0.,.... J. ... 1 _ 1
. . . Ol...... 2,lOUa.-U
Mo.. gUll' 'J.UU"·U
....1. I. OOOU.OO A + .fl6l.·:'4 ".U I .•• U .... OO ".'._3&-5
Load
1J Ll H ' ''!lf·o~
* .. [) 3. ,;;O£-O~
-- **""'* *
o +3 . 18oiE-0:5
.
E +2 . 63~E-O'i
"*
*
* ** F - 5 . 3 0f,£- O~
* * F +2.0':flS-O!>
'-----------------~-----
Constrained elements • j jft'5[-oI
""'un ...... .~ . . . . . . . .
h i ...
M. . . 1.ISU.·Ol If. . _1 . 14101;.01
Ill. . . 0.0000 ...... IfU _'.Uur. .. 05
o.fo~ Od.,h-.1 _ 1 ~h~_4 (I~ll1';. .. d Ma""'l
Ilaa Du,. .l.lOUS-U .... (It. . . a.J04U·Oo
"'-.1. I.OIIOU.• OO +7.]36&-01 .... h \.0000.+10 ..6.591&+01
+6. 521E-01
"5.109&+01
='= 7~-~:=
+4.365£+07
7
~ '3.'2"'O'
~ / L .3.2m-Ol +2.877£+07
+2.44SE-Ol .,.2.133&+07
, HI ,
+8.1511-02
8.2.4 Analysis
Before interpreting the results, the finite element model can be verified by means
of simple checks. For example, the y-displacement at the point of load application
should be the same as the blade penetration. In cases where the load had been
symmetrically applied relative to the flute, the x- and y-deflections should also
exhibit symmetry.
Referring to plot 'load_A', it can be seen that a mid-span load will lift the sides of
the corrugation off from the lower roll as a result of bending moments' about the
two flute peaks. The flanks of the flute will be compressed downwards and pushed
away from each other. Compression at location 'B', as shown in plot 'load_B',
enables the two troughs to act as pivots, as the two neighbouring peaks are being
pulled towards each other. Again, two sides of the corrugation will leave the lower
roll, as indicated by the displacement contours at the base of the corrugation. But
compared with 'load_A', the flanks of the loaded flute will deform more, as the S-
shape becomes more prominent.
Plot 'load_C', on the other hand, provides some insight into the asymmetric
loading behaviour of the corrugation. From the enlarged y-displacement plot, one
side of the corrugation is found to tilt upwards, while the other side curves
downwards, despite the suppon as provided by the lower roll. Except for the base
elements, the x-displacements are all positive, implying a uni-directional shear
along the whole corrugation. One possible explanation is as follows: as a bending
moment is applied about the flute peak which is nearer to the point of load
application, that pan of the flute above the line 0-0 will be given an anti-
clockwise twist. On superposing with the depression due to the downward load,
the net result will be a lateral shift in the positive x-direction.
94
The highest membrane strain energy is recorded in 'load_B', whereas ,10ad_A '
shows a distribution of strain energy over a much wider area. Both ,10ad_A ' and
'load_B' exhibit even distribution of strain energy, which suggests symmetrical
buckling behaviour, as compared to an uneven buckling failure in 'load_C'.
When loading at location 'B' and 'C', the maximum principal stresses are of
similar order of magnitude. High stress concentrations can be found at the tip of
the single blade creaser. Lower maximum principal stress in ,10ad_A • may be due
to the ability for more elements to bend freely.
In the case of 'twin_A', the sides of the corrugation show no sign of an upward
tilt. Instead, the flanks can be seen to have been pushed symmetrically away from
location 'A' in the x-direction as in ,10ad_A '. The magnitude in the x-
displacement, however, is less than that of the single blade creaser. This may be
due to the way the inner liner deflects when being loaded with a single blade, and·
can be explained as follows: consider the deflection of the inner liner due to a
single blade load, as shown in Figure 8.6.
0.2 mm
blade penetration
~
95
orientating a long element about the tip of the creaser. Consequently, the bending
moments responsible for pushing the flanks apart as in 'loruLA', will be
counteracted by the pulling in of the neighbouring flute peaks. The net result is a
smaller x-displacement compared with that of the single blade.
The same line of reasoning can be used for explaining the significant difference in
the maximum principal stresses. In the twin roll arrangement, the elements within
the gap are subject to higher bending stresses because of the larger profile
transition (10" conical profile compared with 3.80 for the single blade). Much
lower membrane strain energy is experienced by the inner liner and the flute in the
case of the twin roll creaser. Intuitively, the inner liner within the twin roll creaser
gap will· tend to tear; whereas the preferred mode of failure is flute buckling when
a single blade is applied at location 'A'.
On loading at location 'B', evidence from the plot 'twin_B' suggests that more
upward deflection and higher maximum principal stress are experienced than in the
case of the single blade loading. Again, this can be due to the large profile angle
transition of 10" as explained previously. The x-displacement plot reveals a more
distoned S-shaped flank, but the magnitude for the membrane strain energy is
similar to that of the single blade load.
The x- and y-displacements for the loading condition in 'twin_C' bear resemblance
to those of the single blade loading. It is interesting to note that, when creasing at
location 'C', the maximum principal stress for the single blade is, for the first time,
higher than that for the twin roll creaser. One possible reason is due to the shon
section of liner between the position 'C' and the peak, giving a sharper profile
transition. Elements at the tip of the creaser will, in the case of load_A, be forced
to wrap over a larger angle and constrained by a higher bending stress.
To summarize, the twin roll creaser is considered bener than the single blade
creaser in terms of producing lower membrane strain energy in the corrugation
structure. A lower maximum principal stress, the main component of which is due
96
to bending, is also expected from the twin roll creaser when creasing at location
'C'. However, when it comes to creasing at locations 'A' and 'B', a higher
maximum principal stress can be found in the section of the liner between the twin
roll gap than in the case of the single blade. Where there is a high stress
concentration, the liner will be more susceptible to tearing.
Important questions have been raised from the theoretical analysis and computer
simulation of the corrugation structure. For example, how localized is the straining
effect on the inner liner of the corrugation? Does sliding take place under the
creasing blade? Certain assumptions adopted in the finite element model, such as
the elasticity of paper, may also need to be scrutinized. This section describes the
use of the Electronic Speckle Pattern Interferometry in the study of the planar
displacement behaviour of corrugated fibreboards when subjected to a creasing
load.
97
the images electronically using the frame store facility on a video system, fringes
will thus be correlated and ready for analysis.
1
d1 = -nA
2
nA
cl,. - - -
2sin6
where A. corresponds to the wavelength of the laser, and 6 is the angle of incidence
between the laser illumination and the surface normal. (For further details on the
theoretical derivation, refer to Appendix 8d).
8.3.2 Objectives
98
8.3.3 Apparatus
The experimental jig, as shown in Figure 8.7, comprises two loading mechanisms:
one for pulling the sample laterally and the other one for simulating the creasing
action.
The micrometer which controls the lateral strain has a resolution of ± 0.0002 mm
and the range of travel is set to 10 mm, well above the experimental requirement.
Resolution on the smaller micrometer, which is responsible for the vertical
adjustment of the creasing blade, is limited to ± 0.01 mm on account of its
available size range. Because of the limitation as imposed by the optical
arrangement (essentially, clearance for the laser illumination), the smallest
micrometer was chosen for the jig design.
B-flute board samples of grade 400 KT were cut into spatula shape by means of a
500 W CO2 laser. Liners towards the ends of the samples were detached from the
flute to allow for attachment. Normally, a more uniform surface reflectivity could
be obtained by spraying the sample matt white; but in the case of paper samples,
such action might significantly alter their mechanical propenies.
Two lasers were employed in the experiment: a 10 mW HeNe laser (A. = 632.8 nm)
99
and a 200-500 mW Argon ion laser (A. = 514 nm) for the out-of-plane and in-plane
displacement sensitive interferometers respectively. (Arrangements for the two
interferometers can be found in Appendix 8d).
8.3.4 Procedures
A preliminary step was to establish whether fringes could be produced from the
surface of a corrugated board sample. To this end, an in-plane displacement
sensitive interferometer was set up with a board sample loaded in the lateral
direction only. The mirror together with the blade assembly were positioned near
to the clamped end of the inner liner, where displacement was most likely to be
found.
The in-plane interferometer was calibrated with respect to the test jig by loading a
piece of liner sample. The spatula sample, with an effective length of lOO mm,
was cut from a piece of Kraft linerboard (210/300 g). Once the sample was
clamped and loaded under tension, it was stretched incrementally. Readings from
the micrometer were then compared with the in-plane displacements as calculated
from the number of fringes.
The second objective proceeded by obtaining fringe counts for the in-plane
displacements of a datum point some 10 mm away from the locations 'A', 'B' and
'C'. The sample itself was held by friction under the creasing blade and its ends
were left unconstrained. In-plane displacements were derived from counting the
fringes across the illuminated section, and substituting the value into the
appropriate equation (see Section 8.3.1). The test was then repeated but with the
datum restricted to within one flute span. Position 'B' obviously could not be used
as it would sit right at the end of the flute span. The only possible loading
locations would be 'A' and 'C'. Two arbitrary locations, however, became
available as the blade could be situated away from or near to a particular peak;
these locations, defined as 'Co' and 'C;, are as shown in Figure 8.8.
100
Direction Of laser
Illumination
The finite element model could be verified by comparing its in-plane and out-of-
plane displacements with those of the optical experiment. Both the in-plane and
out-of-plane interferometers were operated such that the fringes could be captured
simultaneously under identical conditions.
A sample was held by friction under the creasing blade at location 'A'. Since the
jig was lying on its side, as shown in Figure 8.9, care must be taken to ensure that
sufficient friction would prevent the sample from falling.
Displacing the blade by 0.2 mm all at once would overwhelm the sensitivity of the
interferometers. For this reason, the load was applied incrementally. Fringes were
then recorded after each load increment. The fringe count for the in-plane
displacement was taken from the illuminated section between the blade and a point
some 11.5 mm away from the blade. As for the out-of-plane displacement, two
sections, and hence two sets of readings, were considered for extrapolation
101
purposes. The sections were bounded by the blade and two points 11.5 mm and
43.5 mm away from the blade. The total number of fringes would represent the
displacement of the datum point from the blade reference.
8.3.5 Results
Figure 8.10 shows a typical video image of fringes within the illuminated section.
The calibration plot, as shown in Figure 8.11, compares the strain measurements as
obtained from the micrometer readings with the ESPI in-plane displacement results.
102
ESrl Calibration I
(Cumulatiue plot)
0.016
8.811
8.812
8.810
.:
~ ILBBS
"
co
8.8Bb
8.801
B.80Z
6 11 16 21 Zb 31 36 11 16 51 56 61
Ddta. number
-+-l1icro. data -- ESrI data
Loading conditions at locations 'A', '8' and 'C' are shown in Figures 8.12,8.13
and 8. 14 respecti vel y.
103
Fig. 8.13 Creaser knife at location 'B'.
104
Figure 8.15 shows creasing at location 'A' with the sample being constrained at
both ends.
•
Fig. 8.15 Creasing at location ' A' with end constraints.
With reference to the first objective, as described previously, Table 8.2 gives the
in-plane displacements and fringe counts at various loading positions for the wide
and narrow sections of illumination. The wide section is 10 mm long whereas the
narrow one is composed of just one flute span.
Table 8.2: In-plane displacements for the wide and narrow sections of
illumination.
105
The in-plane and out-of-plane displacements as derived from ESPI for the
verification of the finite element model are as shown in Table 8.3.
8.3.6 Discussion
In the preliminary test, it was noticed that by loading the creasing blade at location
'C', fringes would move closer to each other in increasing number, and the overall
movement was towards the blade. This behaviour, known as the "hollowing
effect", represented the downward displacement of the inner liner. On adding a
106
lateral load, i.e. pulling the inner and outer liners in opposite directions, it was
found that the fringes would propagate in the same manner. But the spacing
between fringes would widen for a few seconds after the load had been applied.
Under normal room temperature and humidity, it would be inappropriate to explain
the fringe separation over such a time period in terms of the relaxation behaviour
of the liner material. Slipping would be a more plausible reason.
Figure 8.11 shows the consistency between the strain readings as obtained from the
micrometer and the ESPI in-plane displacement results.
The localized strain effect of the liners is more apparent in the photographic
evidence than the optical experiment. Figure 8.16 shows a board sample subject to
a lateral load. The end where the inner liner is clamped displays shearing action of
the flute.
- shearing action
/
In-plane displacement results in Table 8.2 exhibit a decreasing trend for locations
'C;, 'A' and 'Co'; which indicate that the longer the liner span, the more
effectively it will pull.
After allowing for a load offset due to the blade thickness, the finite element model
has been recalculated for creasing at location 'A'. At a point 11.5 mm away from
the blade, the displacement results are as shown in Table 8.4.
107
Table 8.4: Finite element model in-plane and out-of-plane displacement.
Mode Displacement I m
On comparing the displacement results of those between the optical experiment and
the finite element model, there is discrepancy in orders of magnitude. The
significant difference cannot be accounted for geometrically (as the confounding
effect of one displacement direction due to the other can be shown to be
negligible); nor can it be explained by the sliding phenomenon. At present, no
explanation is available for the result discrepancy, other than that which attributes
to the material property, namely, the fibrous nature and compliance of paper.
8.4 CONCLUSION
The starting position of this chapter has been prompted by one question: "how
much of the creasing effect is due to the corrugation structure and how much is
due to the inner liner?" A hypothesis, wrong as it may seem, is then made on the
assumption that the creasing of a corrugated board can be compared to the loading
of a homogeneous and isotropic truss structure. Major assumptions such as small
loading displacement and constant creasing conditions have been made in order to
substantiate this claim. Based on this hypothesis, design parameters have been
derived through the finite element analyses on the single-blade and the twin roll
creaser profiles.
In the finite element analysis, only the pre-buckling mode of load application has
been considered. This is where paper can still be assumed to behave elastically.
On increasing the load, however, paper will behave inelastically up to the point of
buckling. Any theoretical treatment beyond this point will be rendered invalid due
to the stochastic buckling behaviour. In addition, the proprietary software is found
108
incapable of modelling the boundary interaction between the corrugation and the
creasing blade. Instead, the creasing boundary conditions have to be simulated by
directly displacing the liner elements. Nonetheless, the model has provided better
understanding in the way the strain mechanism works for a piece of homogeneous
and isotropic corrugation.
ESPI has been effective in analysing the displacement behaviour of the corrugated
board samples when subject to the creasing action. Sliding of the inner liner under
the creasing blade has been demonstrated by the moving fringes. The localized
strain effect is more apparent from photographs on the experimental set up than
through interpreting the ESPI results. Results from ESPI, however, have shown
displacements of orders of magnitude higher than those derived in the finite
element model. The only explanation available relates to the fibrous, and hence
discontinuous, nature of paper - a property which has not been accounted for in the
geometrical and the finite element analyses. An experimental approach, as
described in the next chapter, is therefore needed in the assessment of the twin roll
creaser.
109
CHAPTER 9
EXPERIMENTATION
The geometrical and finite element analyses in the previous chapter have been
based on the model that paper behaves within the elastic regime. But as soon as
inelasticity appears, there will be an element of unpredictability with regards to the
mode of failure. Paper failure is difficult to model, as it involves the random
distribution and bonding of pulp fibres on a microscopic level. The true
performance of the twin roll creaser will therefore need to be evaluated
experimentally.
This chapter commences with the design of the creasing and folding test rigs. The
first experiment investigates the effect of various control and noise factors on the
overall crease performance. This will be followed by a study on the geometry of
the creaser profile. Twin roll creasing with pre-crushing will be covered in the
final experiment. The chapter is concluded with major findings from the
experiments, which will help towards designing a feasible twin roll creaser.
110
LINER STRETCH v .• CREASER RADIUS
Uner stretchl %
a rc length.a
3.0
2.5
~ W-t effe~tive
. ,. I
2.0 onglnal ength. 0
1.5
a -0
stretch = - -
o
1.0
o.s
0
10 20 :lO 40 50 70 Radius I mm
The creaser roll gap is accurate to within ± 0.05 mm. The inability to position the
creaser accurately in relation to the flute, as Vogelpohl [51] has realized, implies a
high level of sampling in order to select the few which will satisfy the position
requirements - namely, locations' A', 'B' and 'c' (as already described in Section
7.1). Fine flute positioning and flute alignment are therefore vital features in the
experimental rig.
111
Fig. 9.2 Experimental creasing rig.
A close up photograph in Figure 9.3 shows the creaser unit and the dial gauge
attachment. As the creaser moves vertically, the dial gauge, acting against a fixed
datum at the top, will provide a direct measurement for the creaser roll gap.
112
P"R'T t. 1ST
--. eRD
THIRD ANGLE PROJECTION 1. Will, • • • • • mDl'
,.
.. Sp I ' l
C•• ,ng
r I no • • • • mb "
.... 5.""" •
,. Sp I ,t 'uD'
8r ••• tlll,.1
0" I Cl.
~:=--_-0
22. Mlcromlt.r 0'''01
r
I r
" I r r ~. I. :
L J CAIS. CAll ,anl'lft , . ' ..... r le
-<'-: -;:- ::~ -,
r ."'1111 •• , I. . . . . . . . . . . . . , •••
n••• 't "luI'" '1",,,,,'11
....
t.
r ",1'
" I 1ft
"
I~' 11 •• , • • • • , •• , • • •
I
"
- A"'.V "1"1
li0ii(11 Cr . . . . '
.I.,.',
co.
9.2 FOLDING RIG DESIGN
The basic mechanical features comprise a counter-balanced folding leg the end of
which is fitted with a folding edge. With the board sample placed on top of the
principal support bar and one end of it sitting on the edge of the cantilever, the
folding leg is now ready to be latched into position. To ensure correct folding, the
crease on the sample must align with the folding edge. Strain gauges on the
cantilever enable it to act as a force sensor. As soon as the d.c motor is switched
on, and the board panel is folded to slightly over 90", the folding leg will be
unlatched automatically. The whole folding leg assembly is then allowed to swing
to its balanced position, thereby clearing way for the board sample. The d.c motor
will be deactivated once the board is folded through ISO". Signals from the force
sensor and the shaft encoder are sampled continuously at a rate of half a degree for
every strain sample reading. In this way, the test results are made independent of
the rate of folding. Figure 9.5 shows the control accessories for the folding test
113
rig. Detailed specifications for the rig can be found in Appendix 9a.
As a board sample is folded from ('f' to a closed position of 180°, the folding torque
can be measured in relation to the angular displacement. The result is a torque-
displacement or force-displacement (if the moment arm about the folding hinge is
constant) diagram which provides a continuous history of folding. Figure 9.6
illustrates a typical plot of the folding test.
Force / N FOLDING RESISTANCE PLOT
, .....
114
-~. no I THIRO ANGLE PROJECTION --®-cr. ..
PART LIST
I • at I I 10. P I lIlt
,.
...... ¥....
Or I~' Inaft r. D c It d •
Dt , • ~ I1 t • •••• I
loa, , Ill •
.. '"
L
~
8 r,
0 CI I ft g
•• b··',"O
, '"D
I. F0 I • I "t laot
I. SIr. It Clnll"W"
~
". SU"lft
I~ , ".
.'" "
SU"OII
I
JNd I I + I ... -, e
11.
,. .
LIIC~I"O
L'U~ •• 'c
"' ,-~ ".
~
S~."ld.r d I le
I I 11. S ~. 11 connlClor
I n. ,
<W
B. 11 " C It .It " h t
...
I
I D
I
~- Jl ~ ".
I I 1111. tU,"
i , ~ <-> ~;~~-~- >I. e I. It
:~ (~~
>l r:;:,
~
~f
. . . ,~
-I
b-,
"d I
~~ " , J eJ
I~ ~
The 1st peak folding resistance represents the force required to break the crease on
the inner liner (see Section 7.3). The rate at which the crease is broken can be
evalutated from the gradient of the curve between the origin and the 1st peak. This
gradient will give a fair indication to how well a crease has been defined.
However, the gradient must be interpreted in the light of the 1st peak folding
resistance, since the absence of a crease may also produce a steep gradient. The
2nd peak (or peaks) of the folding resistance provides information on the closing of
the board panels. Essentially the inner liners on either side of the crease are
brought into contact, leading to further flute buckling in the process. The type of
board tested can be categorized in terms of the total folding work done.
Design parameters such as the creaser profIles at different shoulder angles, the twin
roll gap and the creaser roll gap, as shown in Figure 9.7, were derived
geometrically in the previous chapter. Optimization in this context would imply
the consistency of crease quality irrespective of any noise effect. The noise factors
inherent in the case conversion process could be simulated as control factors during
the design stage. Examples of noise factors were the crease location and alignment
relative to the flute structure, the moisture content in the fibreboards and the time
lapse between conditioning and creasing. Creasing with hot rolls was also included
in the investigation, as the liner was suspected to become more resilient to
mechanical scoring at elevated temperature.
115
TwIn ",11 crDMGr gop
9.3.1 Objectives
9.3.2 Procedures
In all, eight factors were thought to affect the creaser performance. Four factors
were parameters at the designers' disposal; they were the creaser profile, creaser
temperature, twin roll creaser gap and creaser roll gap.
The creaser profile was defined in terms of the angle of attack and hence the
sharpness of the profile. Three cone angles of 10°, 15° and 30° were chosen in
order of increasing sharpness (see Figure 9.7). Creaser temperatures ranging from
room temperature of 25° C to 55° C were chosen as preliminary settings. Values
for the twin roll gap were based on those as derived previously in Section 8.1.3.
The range for the twin roll gap settings was adjusted upwards at fixed increments
of 0.5 mm. Settings for the creaser roll gap must take into account the liner
116
grades, as boards with weaker inner liners were found to be more vulnerable to
tearing.
The remaining four factors were noise factors which would normally be subject to
random variation; here, they were simulated as control parameters. The factors
were the time delay, relative humidity for sample conditioning, crease alignment
and crease location relative to the flute.
The time delay was taken from the end of the conditioning period to the beginning
of the test. Immediate creasing took place soon after the samples were taken out
of the environmental cabinet; whereas samples which had been left under
laboratory condition for more than three hours after conditioning would be
considered as relaxed. The samples were conditioned over 24 hOUTS in the
environmental cabinet to relative humidity levels of 30 %, 50 % and 70 %. To
simulate the condition of creaser misalignment, the guiding rail was set to an
arbitrary angle of IS. Crease location was defined in terms of the trough (A),
peak (8) and the in-between (C) positions (see Section 7.1). It must be
emphasized that the noise factors were included here purely for investigation
purposes, as distinct from design optimization.
117
Table 9.1 Control factors and their levels of settings.
Factors \ Levels t 2 3
B. Relative humidity % 30 50 70
D. Creaser temp.·C 25 40 55
• Two levels were set for the single- and double-walled boards.
An LtB (2t x 37 ) orthogonal array (see Appendix 9b) was chosen for the
experiment. Apart from the first column, which was a 2-level one, all the factors
had three levels of variation, which would allow for the detection of any
monotonicity within the preset range.
Three responses were measured during the experiment; they were the percentage of
defective folded samples, the minimum force required to break: the crease (i.e the
first peak: on the folding resistance plot) and the resistance to shear.
One precondition for fishtailing was the presence of shear forces acting on the case
panel while it was being folded. In order to simulate this effect, shear was
deliberately introduced by applying a constant moment about the hinge of the
creased and folded panels. The subsequent shear deformation would be recorded in
terms of the distance between the corner of the overlapping panel and the edge of
the base panel (see Appendix 9c for more detail on the shear test).
The signal-to-noise or SIN ratios for both the folding resistance and the shear
118
deformation were based on the same criterion of "smallest-the-best", and could be
calculated as follows:
Both the percentage of defective samples and the SIN ratios were then plotted
against the control factors at their individual settings. Only the SIN ratios for
samples which had not failed were included in the plots, as the defects criterion
would take precedence over all the other quality characteristics. Optimized settings
were those with the most positive SIN ratios and the lowest percentage of defective
samples.
The SIN ratios were then examined using ANOV A. From the ANOVA tables, the
relative importance of each control factor could be identified. Finally, the
optimized settings for the design parameters were verified in the confinnation runs.
In order to test the robustness of the new creaser, a representative range of board
samples was chosen for the experiment. The types of board were as follows:
230 CC(B) - low liner strength
250 TH(B) - medium liner strength
400 KT(B) - high liner strength
400 KT(Gemini) - C-flute caliper with double flute lamina
250 TCH(B/C) - double wall heavy-duty
119
For each type, a batch of five was used in each of the eighteen experiments. The
board samples were conditioned according to BS3431. Permission was kindly
granted by the Institute of Polymer Technology and Material Engineering for
conditioning the samples in their environmental cabinet.
The main equipment featured in the experiment was the creasing and folding test
rigs. A Fletcher's trolley, as shown in Figure 9.8, was also converted into a test
platform for measuring the shear deformation. Other accessories such as
micrometer, steel rule, bunsen burner and gas supply, 50 g weights, hygrometer
and psychrometric chart, and digital pyrometer were also used in the experiment.
9.3.4 Results
The experiments took a total of three days to complete. During this period, the
relative humidity in the laboratory fluctuated at around 50 % (± 5%) and the room
temperature, between 25°C and 27°C. (All test data can be found in Appendix 9d.)
Two modes of failure, as a result of creasing and folding, were observed in the test
samples, and which could effectively alter the case property. Principally, the two
120
modes were tearing of the inner liners and secondary flute buckling (or more
generally known as false creasing). Tearing was noticed to have been caused by
the tip of the creaser profile and its shoulder edge.
Results on the defective samples can be found in Appendix ge. A summary of the
number and types of failures, and the number of missing samples is shown in the
Table 9.2.
230 CC (B) 3 57 1
250 TH (B) 1 30 1
400 KT (B) - 1 3
400 KK (GEMINI) 14 2 I
Figure 9.9 shows the percentage of defective samples plotted against the control
factors at their various settings.
121
P ( DEFECT ) PLOT ~.
)0;' UoOCCtBI
O·uoTo.(1'IC1
• 2.:iO fH (Ill
4 • 4<10 >ll<.CCi
• . _"'T(81
::7 J P\x.
i~ f:::
"
--..-.. ' \ I
"'M
~
~"""""-J';
--- .........
I
O$41U
~U'
I
Tt .... ·C
I
----..
!. ~. ,\. :. ,t, ,I. . .,
~,
Likewise, the SIN ratios for both the folding resistance and the shear deformation
are as plotted in Figure 9.10 and Figure 9.11 respectively.
., .---'
•
,A •
• ~~7 --- ~ ' /
.. ,L.
..-'
p
<.QC! H(03)
""" .... ,q
""""Io..I.-;,u
>0 IS
--'.------.<.--.--.-.-.-~.-.-~-
...
T£ . .··C
" "
~
." "'
......... ~ ... ,
.... '. "
..
."<'«-.51"",..
'D
«:'-'-~
"
......
• 8 "
122
------ --
"
.
."
f{--4
~.~
/' @ Ji
~ ~~~(§8~
m
x.~ ..... ..IC
d.
.
~ ...-"
x ...... otT (Ill
(J
o
'~n!(II)
, ·ooo.;,oqq
ANOV A tables for the folding resistance and the shear deformation SIN ratios can
be found in Appendix 9£.
Factors 'C', 'D', 'F' and 'G' were optimized as they were the only design
parameters in the experiment Optimum settings were those whose SIN ratios were
the most positive on the scale (based on the 'smallest-the-best' criterion). Apart
from the percentage of defective samples, which were not subject to the SIN ratio
criterion, factors from both the folding resistance and the shear deformation could
be ranked in order of significance. The best compromised set among the three
responses was then selected as the optimum (see Appendix 9g). Optimized
settings, as shown in Table 9.3, were then tested in the confirmation runs.
230 CC (B) GI D2 C2 F2
250 TH (B) G2 C3 D2 F2
400 KT (B) Gl C3 . .
400 KK (C) C2 G2 D3 F2
123
The theoretical SIN ratios in Table 9.4 were calculated using the equation:
Only two sets of confInnation runs were perfonned with the above optimum
settings. The board samples chosen were the 250 TH and the 250 TCH, and the
results (see Appendix 9h) were summarized in Table 9.5.
9.3.5 Discussion
The confIrmation runs showed that the shear resistance had been improved, albeit
at the expense of raising the folding resistance or lowering its SIN ratio. No
conclusion could be drawn on the rate of defects since only fIve samples were used
in each of the confIrmation runs.
Several areas of interest have been highlighted as a result of the experiment, and
they are as follows:
124
a) Modes of failure
In general, samples failed because of tom liners or secondary flute buckling.
Tearing must not be allowed to happen, since the sandwiched structure relied solely
on the liners to bind the corrugation into position. Any damage to the liners might
lead to crack propagation and jeopardize the integrity of the corrugation structure.
Tearing could occur any time while the creaser was in contact with the sample.
The middle of the panel was just as likely to tear as the leading edge where the
creaser would fIrst come into contact with the board sample. Causes for tearing
failures could be due to a combination of factors such as a small creaser roll
diameter (as compared to a full-size one), a sharp creasing profIle (at both the tips
and the edges) and an incorrect creaser roll gap. The suggestion that the twin roll
was the main cause for tearing must not be ruled out, when the creaser shoulders
could stretch the part of the liner which was within the twin roll gap. Stress
induced in this part of the inner liner could be lessened by having a wider twin roll
gap. However, setting the twin roll gap too far apart might encourage the two
halves of the creaser to behave independently as if they were two single blades.
b) General trends
Referring to the previous plots on defective samples and SIN ratios, the distinct
levels of results on all three plots clearly reflected the importance of board grades
on the crease quality.
125
Some of the noise factors had hardly made an impact on the folding and shear
resistance. An example was the instant of fold which served as a pooling factor in
every analysis of variances. The crease location, however, showed direct
correlation with the folding resistance. From the folding resistance plot, position
'B' or the flute peak (referring to Figure 7.1) could be seen to be the least
favourable, i.e it was the position where the highest resistance to folding was
experienced. One interpretation could be that extra effon was required to buckle
not just the inner liner, but also the flute situated immediately below position 'B'.
By comparison, less resistance to folding was experienced at positions' A' and 'C',
when only the inner liner was buckled to form the fold. It was not until the folded
panels came to a close, however, that boards with creases at positions' A' and 'c'
would experience flute buckling for the first time. For boards which were creased
at location 'B', flute compaction would have been well underway upon closing of
the panels. The folding resistance plots would therefore be expected to show a
higher first peak (as demonstrated by the SIN plot on the folding resistance) and a
lower second peak for creases at position 'B' than those at position' A' and 'C'.
Moisture content in the board samples, a noise factor by nature, also exhibited
some form of correlation with shear as shown on the shear deformation plot.
Definite trends were also found among some of the design parameters. An
example was the correlation between the folding resistance and the crease roll gap.
By having a small roll gap, and hence a deep creaser penetration, the folding
resistance would decrease substantially. This applied to all the single-wall boards,
whereas for the double-wall boards, the roll gap effect was marred by other factors.
Optimized settings in Table 9.3 would suggest that sharper creasing profiles were
more suitable for single-wall B-flute boards with stong inner liners. But as soon as
a weaker liner or a thicker board is creased, a gentler creasing profile would be
more appropriate. Sharper creasers too could lower the folding resistance and the
shear deformation, though the risk of tearing would increase with board calipers.
Shear resistance for the 230 CC samples, however, proved to be an exception, as
their behaviour was more sensitive to the moisture condition than the creaser
sharpness.
126
Similar trends for individual factors such as the twin roll gap setting could be
observed between the 250 TCH and 250 TH samples. It was noticed that both
types of boards shared the same inner and outer liners, namely the T - and H-liners.
It would be presumptious to conclude that the similar trends in all three responses
were entirely due to the same liner constituents. But the importance of the liner in
relation to the creasing, folding and shearing reponses could not be overlooked.
When creasing the double-wall boards, the optimum profile, being the shallowest
of all, was behaving more or less like a pre-crease crusher. This reinforced the
idea that the thickness of the creased region must be reduced to such an extent that
the subsequent creasing force could be transferred directly onto the liners. With
regards to the shear resistance, the twin roll gap setting for the double-wall boards
was identified to be the most significant factor, more so than in the single-wall
boards.
The physical behaviour of the 230 CC board samples was extremely sensitive to
the moisture condition. The ideal creasing condition was when the samples had
been prepared at a relative humidity of 50%. Any deviation from the median value
by 20% would drastically alter the performance of this type of boards. Equally
important, the creaser must penetrate deep enough or else the board might fail due
to secondary flute buckling.
d) Result inconsistency
One major problem still to be resolved was the inconsistency in the crease
alignment result. The settings were chosen such that the first two levels were the
same, i.e both were set with the crease parallel to the flute. In theory, the same
127
two levels should yield identical results. The three plots, however, revealed
inconsistency at the two levels, suggesting that either the rig design was fallible or
some kind of interaction had not been accounted for.
Another curious phenomenon was observed from heating the creaser roll to the
mid-range temperature of 40" C. Under this condition, the fIrst peak: of the folding
resistance reached its highest level for the double-wall boards, whereas the least
shear deformation was experienced in the 230 CC. This was surprising because
one would anticipate a monotonic trend for the temperature effect
On the whole, it must be emphasized that any results in this experiment must be
interpreted in the light of a small sample size together with a high defect rate.
9.3.6 Conclusion
An optimization experiment was conducted on the novel twin roll creaser. The
objectives were to identify the optimum design parameters for the creaser and
investigate the relative effect on creasibility due to a range of noise factors.
Three quality characteristics were measured during the experiment; they were the
percentage of defective folded samples, the fIrst peak: folding resistance and the
shear deformation under a constant load. Results showed that boards would fail on
account of liner tearing and secondary flute buckling. Tearing was mainly due to
the tip of the creaser and the conical profile edge. Board grades were seen to have
a signifIcant effect on the creasing and folding performance. Both the 230 CC and
the 250 TCR, belonging to the light- and the heavy-duty categories respectively.
were identifIed to have caused the most problems. Trends from the result plots had
revealed correlations between the various factors (both design and noise) and the
responses. Shear resistance for the selected samples was improved in the
confIrmation runs, though their folding resistance had not reached the desirable
level.
128
The experiment has prompted further investigation into the geometrical properties
of the twin roll creaser: in particular, the creaser shoulder edge responsible for the
majority of tearing failures and the pre-crushing function when creasing double-
wall boards. From the theoretical as well as practical viewpoints, any future
experimentation will have to aim at minimizing the number of defects. Other
measures such as the second peak of folding resistance and the gradient of the first
peak are to be included in the analysis. Second peak folding resistance will
indicate the ease of panel closure, while the first peak gradient will show how well
the crease has been defmed. On reflection, the ranges for the relative humidity and
temperature have been set too close for them to produce any observable difference
in the results. But since the present prerogative is to optimize the twin roll creaser
design, any investigation into the temperature and moisture effect on liners will
have to be left as future work.
The first experiment has revealed that different creaser profiles are needed for
different board thicknesses in order to avoid tearing. For example, a shallow
profile is more suitable for the double-wall boards; whereas for single thickness
boards, except when the inner liners are weak, a steep profIle will be more
appropriate. Experimental evidence, however, suggests that the majority of tearing
has been due to the sharp edges on the creaser. Hence, the objective of the second
experiment is to study the creaser shoulder effect and the tip radii on the twin roll
creaser.
9.4.1 Theory
129
34 % reduction in caliper. In doing so, the creasing force can be transferred more
effectively into the already compressed region.
13
19
'----45 - - - '
One hypothesis is that if the total width of the creaser amounts to more than the
width of the pre-crusher, then, for the same depth of creaser penetration, damage
may spread over a much wider section leading to higher shear defonnation. This
hypothesis, at present only a speculation, will be valida.ted in the following
experiment. The width of the twin roll creaser, however, is designed to within that
of the pre-crusher, such that depending on the twin roll gap width, the overall
creaser width can be adjusted to within or beyond the pre-crusher width.
Radial shoulders and creasing tip edges, as shown in Figure 9.13, are the main
design parameters in this experiment. Both 'R' and 'w' have been chosen to be
within the pre-crusher width constraint (see Appendix 9i). Gap values 'g' wider
than those used in the fIrst experiment are tested in order to study how the
separation between two score lines is related to the board calipers (see Section 8.1).
130
~I
9.4.2 Method
The control factors and senings were as shown in Table 9.6, and an L, orthogonal
array was adopted.
Factors\ Levels 1 2 3
In all, nine pairs of creasing rolls were manufactured to represent the nine possible
combinations of 'r' and 'R' (see Figure 9.13).
The samples 230 CC (B) and 250 TCH (B/C) were chosen because they had been
found to have caused the most problems in the previous experiment. The number
of defective samples, the fIrst peak of the folding resistance and the shear
deformation were responses measured during the experiment.
131
9.4.3 Results
The laboratory conditions at the time were 22.5 °C and 50 % relative humidity.
No samples were discarded as a result of human error during the experiment.
Table 9.7 shows the number of defects recorded in the experiment. Note that a
total of 45 samples was tested for each type of boards.
230 CC (8) I 21
A high rate of tearing had meant that neither the folding resistance nor the shear
deformation responses could be evaluated for the double-wall boards.
Tabulated results for the 230 CC (B) samples can be found in Appendix 9j. A
result summary on the rate of defects and the SIN ratios can be seen in Figures
9.14,9.15 and 9.16.
.....
"
"
"
.. ~
" .1
I
"
1
R1
I-+---+--+----t--------t-i-t-
R.1 IlJ 9' .t ,1 ~'o.1.J
132
SIN_
2
•
,
.
.
I I I I I I I I I I
r1 r2 r3 Rl RZ RS gl g2 IP h1 h2 1'3
......
"
'"
IS
I I I I I I I I I I I I
r'I rt r3 RlR2iQ 1f112 P IIlh2h3
ANOV A tables for both the folding resistance and the shear deformation can be
found in Appendix 9j. Table 9.8 summarizes F ratios for the four control factors.
133
9.4.4 Discussion
As in the first experiment, the mode of failure for the 230 CC (B) was
predominated by secondary flute buckling; and for the 250 TCH (B/C), tearing of
the inner liners. A majority of 43 out of a total of 45 failed among the double-
wall samples. Failures could not be prevented even with a creaser roll gap as large
as 2.5 mm (or a 59 % reduction in board thickness). This clearly demonstrated the
discrepancy in the initial design criterion, namely that of a wide rounded-off proftle
as a replacement for the pre-crushing and creasing functions.
With regards to the 230 CC (B) samples, the creaser roll gap 'h' appeared to have
the most significant effect on the folding resistance. Again, as in the first
experiment, the trend for the roll gap was monotonic and the narrowest gap was
found to be the most desirable. In terms of shear deformation, a large shoulder
radius and a small twin roll gap were the most preferred combination. A small
twin roll gap had the effect of producing a clearly defined crease. The advantage
for having a large shoulder radius, however, was not immediately obvious.
Increasing the shoulder radius will gradually flatten out the curved proftle, hence
enabling the latter to act like a shallow conic proftle. As described in Section
8.1.2, a 10° conic proftle is recommended on the basis that tension within the inner
liner will be relieved by the time the proftle starts to depress the neighbouring flute
peaks. The shear deformation results for the 230 CC boards in the first experiment
have also cofmned that a 100 proftle angle will give the least shear.
9.4.5 Conclusion
In an attempt to develop a single stage creaser for both the single- and double-wall
boards, the shoulder and the tip radius effect of the twin roll creaser were
investigated. An unacceptable number of defects, however, suggested that pre-
crushing for the double-wall boards was indispensable with the current twin roll
134
creaser design. For a twin roll creaser whose width was limited to that of the pre-
crusher, it was found that a large shoulder radius would give low shear
deformation. More important was the width of the crease which would be
governed by the twin roll creaser gap. The twin roll creaser would succeed in
crushing and creasing the double-wall boards only if its width was extended to
beyond that of the pre-crusher. Although the objective of a single stage creaser
and crusher could be achieved with low folding resistance and shear deformation, a
reduction in the section thickness over a much wider area would imply lower
structural stiffness and hence the case compressive strength.
Previous experiments have shown that a single stage twin roll creaser will cater for
the double-wall boards only if the crushing profile is wide enough for a sufficient
creaser penetration. However, the width of the creaser must not be wider than the
present pre-crusher, or else a larger area of flutes will be crushed and the case
strength substantially reduced. A compromise will, therefore, be to check if the
twin roll creaser, with a total width less than that of the pre-crusher, will operate in
conjunction with a pre-crusher on double-wall boards.
135
'---- " ----'
All d.m.n.,ons ,n mm.
1II0t to 'Cl'.
9.5.1 Methods
An L., orthogonal array was adopted for the experiment, and the control factors and
settings were as shown in Table 9.9:
The 10" and 20· profiles were the same as those in the first experiment. The type
of board chosen was the double-wall 250 TCH, and a sample size of ten was
assigned to each of the nine experiments. All the board samples were conditioned
to 50 % RH and 23 ·C prior to the experiment.
Below were the responses measured in the experiment, and which would be
136
maximized in the subsequent analysis. For notation purposes, '11' was used to
represent the SIN ratio.
• 2
E yj
'1..,..
.. = -lOlog 10 ~
n
SIN ratio for the gradient of the first peak, 'D', using the 'larger-the-better'
criterion:
3.) Shear
The defmition for shear deformation was same as before. The load used on
the Fletcher's trolley weighed 1813.8g. SIN ratio for shear deformation
using the 'smaller-the-better' criterion:
•
Ey/
'1 = -lOIog10~
n
137
9.5.2 Results
All test results have been tabulated and can be found in Appendix 9k. SIN ratios
for the defect rate due to pre-crushing is as shown in Table 9.10.
A2 8.36
A3 9.54
SIN plots for the defect rate as a result of creasing, fIrst peak and its gradient and
second peak of the folding resistance, and the shear defonnation are as shown in
Figures 9.18 to 9.21.
""'
1. .... DEFECTS SiN PLOT (CCI
•
• -/- -
I I
A1 A2. It3 Ot 82 83 Ct C2 Cl Ot D2 03
138
1at PEAK CtRADtEHT SIN PLOT
I I
Al A2. A:S 81 fI2!!O Cl C2: C3 Dl [» D3
·7
.10 ~ 2rdpeoi<B
·11
I I I I
A1 A2 A1 81 82 B3 Cl C2 C3 D1 D2 DJ
32
31
30
27
I
AI A2. A3 61 B2 83 Cl C2 C3 01 D2 03
139
From the results, the optimized levels for each response and control factors were
obtained and ranked in order of significance (through comparing the F ratios). A
summary of the ANOVA 'F' ratios is as shown in Table 9.11.
The settings for the confrrmation run were as follows: D1, B2, C2, A2. Table 9.12
shows the SIN ratios predicted for these setrings together with the acceptable
standard deviations and those which were derived from the confIrmation run.
9.5.3 Discussion
The 1st peak folding resistance was found to have been ignored by the software for
some of the test samples (see tabulated results in Appendix 9k). A sharp yield
point which would normally be associated with the initial stage of folding had,
instead, become a point of inflexion as illustrated in one test sample in Figure 9.22.
140
This phenomenon, according to Nordman et al. [30], could be explained by the
material flow within the board samples.
3 ~ - - - - - - - - - - - - - - - - - - - - - - ..---:::: - - - - - - - - - - -
~
~.~~ .•. ,
" ,--
-- .~~-.- ... -.~-
2 --------------------~---------------
,-- ,.
,~-
.....
1
On comparing the predicted and the experimental SIN ratios in Table 9.12, an
improvement has been achieved in the defect rate and the gradient of the 1st peak
folding resistance. Shear deformation is found to lie just outside the acceptable
tolerance. In fact, it is difficult to optimize both the defect rate and shear
deformation, as shall be discussed later in this section. The rest of the
experimental SIN ratios are not far from the predicted ones. The fact that the
ratios fall within the tolerance band suggests the adequacy of an additive model.
Present rate of defect however is still too high for the creaser design to be of any
practical value.
Pre~crushing was proved vital in the creasing of double-wall boards. A high defect
rate during the pre-crushing stage might imply a fault in the experimental design,
or more to the point, an incorrect range roll gap settings for the pre~crusher.
Enquiries with a machine user, one of the industrial collaborators, revealed that the
roll gap should normally be set to between 3.5 mm and 4 mm. The magnitude and
141
gradient of the 1st peak folding resistance were also found to be correlated with
pre-crushing. Less pre-crushing would result in a steeper gradient, or a more
definite crease, and a higher 1st peak of folding resistance.
Among the various control parameters, twin roll gap setting was identified to be
the most effective in controlling shear deformation. For minimum shear, the
lowest gap value would be the most appropriate. However, there was a conflict of
interest as the least defect due to creasing would correspond to the highest twin roll
gap value. Evidence from the finite element model in the previous chapter also
suggested that the highest principal stress could be found in that part of the inner
liner which was within the twin roll gap. Lengthening this section of the liner,
hence widening the twin roll gap, would diffuse the principal stress over a much
larger area and therefore minimize the chance of tearing.
The creaser roll gap had a significant effect on the gradient and the magnitude of
the 1st peak folding resistance. As in the previous experiments, monotonicity was
expected between the peak resistance and the roll gap. As for the gradient, there
was no immediate explanation for the mid-range gap value being the most
preferred one.
The creaser profile was also found to be crucial in determining the rate of creasing
defects, the 2nd peak of folding resistance and, to a less extent, the shear
deformation. The 100 profile, being the shallowest of all, was demonstrated to be
the most robust, though a sharp profile would minimize shear deformation and
therefore fishtailing.
9.5.4 Conclusion
The third Taguchi experiment has been conducted with the aim of optimizing the
twin roll gap creaser for the double-wall boards. A 10" profile was found to be the
best among the three choices. Creaser roll gap, be it the pre-crusher's or the
142
creaser's, has a significant impact on the gradient and the magnitude of the 1st
peak folding resistance. Reducing the roll gap will lower the 1st peak folding
resistance, but the relationship between the roll gap and the 1st peak gradient does
not appear to be monotonic. One possible explanation may be due to the property
of a double-wall board. Apart from the discrepancy of setting the pre-crush roll
gap closer than required, results achieved with the optimized factor levels are still
less than satisfactory. Defect rate can be reduced by widening the twin roll gap,
hence allowing more liner to be stretched under the same stress level. But the
disadvantage for having a wide twin roll gap is an increase in shear deformation.
The creasing and folding test rigs have both been proven indispensable in the
experimental study of creaser profiles.
In the first experiment, it was found that the crease location had direct correlation
with the folding resistance. Sharp creaser profiles too were shown to have the
effect of lowering the folding resistance and the shear deformation. Unfortunately,
the risk of tearing due to sharp profiles would increase dramatically with board
calipers. Furthermore, boards were found to have failed on account of liner tearing
and secondary flute buckling (or false creasing). From observation, tearing seemed
to have initiated from the tip and the shoulder edge of the creaser profiles.
Board grades were shown to have a significant effect on the creasing and folding
performance. The 230 CC and the 250 TCH, being the light- and the heavy-duty
cases respectively, were identified to have caused the most problems. In an
attempt to develop a single stage creaser for both types of boards, the shoulder and
the tip radius effect of the twin roll creaser were studied in the second experiment.
An unacceptable number of defects, however, suggested that pre-crushing for the
double-wall boards was a prerequisite in as far as the twin roll creaser design was
concerned. The twin roll creaser would succeed in crushing and creasing the
143
double-wall boards only if its width was extended to beyond that of the pre-
crusher. But the benefit from combining the creaser and crusher into a single stage
,
would be outweighed by the reduction in section thickness over a much wider area.
The net effect would be a lower structural stiffness for the final case.
The third experiment demonstrated that a 10" profile for the twin roll creaser was
the best of the range. Creaser roll gap, whether it was the pre-crusher's or the
creaser's, would determine the gradient and the magnitude of the 1st peak folding
resistance. Tearing could be reduced by widening the twin roll gap, though at the
expense of increasing the shear deformation.
144
CHAPTER 10
CONCLUSIONS & RECOMMENDATIONS
Research has been conducted on improving the performance of the case conversion
process. In particular, the study has investigated the effect of creasing, folding and
gluing on fishtailing - a quality characteristic which describes the squareness of a
case. Through perfonning an industrial experiment on a standard "flexo-folder
gluer" case converter, important factors have been identified and which have led to
new innovations for the correction and reduction of fishtailing. Design
methodologies for the folding innovations have been described in depth.
The feasibility of a twin roll creaser design, which processes boards of different
thicknesses without the need for roll replacement, has also been examined.
Analytical and experimental studies have resulted in better understanding of the
mechanism of creasing, and a functional twin roll creaser design.
The following subheaded conclusions refer to the set objectives aimed for in this
study.
145
10.2 FOLDING MECHANISM
The present method of folding takes the fonn of a stationary and twisted rail which
guides the case panel through the fIrst 90" of fold. Inevitably, a backward force
component is introduced onto the leading edge of the case panels, thereby causing
fIshtailing. This is consistent with the uni-directional fIshtailing results obtained in
the industrial experiment.
Based on the principle of three bodies in spatial screw motion and the aim to
eliminate the backward force component in folding, two folding innovations were
investigated; they were the coil cam and the twin cam.
The coil cam was designed with the principal objective of transferring the folding
force from the leading edge to a point on or near the centre of the face of the case
panels. Minimization of ink smearing and panel distortion due to a single point
contact was inherent in the design.
The design methodology involved the derivation of kinematic equations for the
cam proflle. Using the screw theory, the kinematic properties of the cam system in
spatial motion were established. A high-level software program was written to
146
perfonn the transfonnation calculations. Results were formatted into macro
commands and fed into a proprietary CAD software for the generation and
automatic display of the cam profile. Computer Aided Manufacture of a prototype
was proved feasible on downloading the design data into a CNC machine.
A design more compact than the coil cam was synthesized in the fonn of a twin
cam. The cam was unique in that it divided the folding operation into two stages,
each of which was responsible for a 90" fold. Rolling contact was designed into
the cam with the intention to prevent ink smearing. It was realized, however, that
computational and manufacturing errors would make perfect rolling impossible to
achieve.
The requirement for shorter batch runs without increasing the downtime through
replacing creasing rolls for various board thicknesses has prompted an investigation
into a separable twin roll creaser design. A comprehensive literature survey has
been conducted on previous research into creasing. In particular, the mechanism of
creasing in relation to the physical properties of the corrugated fibreboards, the
approaches adopted in designing creaser profiles, and the methods for crease
assessment have been reviewed. From this, a creasing analysis and
experimentation have been perfonned on the twin roll creaser.
Geometrical and finite element models were set up in an attempt to simulate the
creasing mechanism for the standard single blade and the twin roll creasers. The
147
models were shown inaccurate by means of an optical technique (Electronic
Speckle Pattern Interferometry). Sliding phenomenon was also demonstrated
between the creasing blade and the inner liner of the board samples. Inability to
model the fibrous and compliant nature of paper thereby gave way to an
experimental approach.
10.3.2 Experimentation
Special creasing and folding test rigs were designed for studying the twin roll
creaser. The creasing rig was capable of providing a roll gap accuracy to within ±
0.05 mm, and fine flute positioning and board alignment relative to the flute.
Some features of the folding rig were the continuous force sensing over 1800 fold
and the instant display and characterization of the folding resistance. In all, three
major experiments were conducted on the twin roll creaser.
a) Principal Experiment
Three sets of twin roll creasers were studied at various control and noise parameter
settings. Results showed that board failures were primarily due to tearing of the
inner liner and false creasing. Tearing was further identified to have been
originated from the shoulder edge and the tip of the creaser; whereas insufficient
crease definition was ultimately responsible for false creasing.
Crease location was found to have a direct influence on the folding resistance, with
location 'B' (i.e. the peak of the corrugation) producing the most resistance.
Reducing the creaser roll gap or creasing with a sharper profile would, in general,
lower the folding resistance for the single-wall boards. Sharp creaser profiles also
had the effect of increasing the board's resistance to shear, so long as the inner
liner could withstand tearing.
The extreme board grades, namely the 250 TCH double-wall and the 230 CC
single-wall boards, were identified as the most problematic. The double-wall
148
board, as demonstrated in the experiment, needed to have its thickness reduced
substantially prior to creasing; its crease performance was most sensitive to the
twin roll gap setting. On the other hand, 230 CC boards, fabricated mostly from
recycled material, was shown to be more sensitive to moisture conditions.
The twin roll creaser was optimized to produce minimum shear in the board
samples. Further work, however, was required to investigate the geometrical
parameters of the creaser design and the rheological properties of paper in relation
to temperature and humidity.
c) Optimization Experiment
The objectives in the final experiment were to verify the operations of the twin roll
creaser in conjunction with the pre-crusher, and to investigate a combined profile
for creasing single- and double-wall boards. Only the 250 TCH double-wall boards
were considered.
Optimization was achieved for the rate of defects and the I st peak folding
resistance. The ID" creaser profIle was demonstrated ~s the most robust in
minimizing shear deformation and therefore fishtailing. Smallest twin roll gap was
also shown to give the least shear, but the largest gap setting would correspond to
the least number of defects. This was consistent with the result from the finite
element model, which indicated that the highest principal stress would appear in
that part of the inner liner lying within the twin gap. A full scale experiment
would be required so as to verify the true performance of the optimized creaser.
149
10.4 REMARKS
Some unique engineering techniques have been adopted in the research. For
example, in addition to optimizing the case conversion process, Taguchi method
was proved useful as an investigative tool. In the design of the coil cam, the
method of running a high-level software program in order to drive a proprietary
software by means of macro commands was demonstrated as a powerful technique
for iterative design. ESPI, a non-contact optical method for measuring
displacements was successfully applied on corrugated fibre boards.
10.5 RECOMMENDATIONS
i) Other means of folding which avoids any ink smearing and backward force
component as found in the folding rail arrangement. Any mechanical action
on the case panel will inevitably impart physical damage onto the case, be it
undesirable force component or print quality. The objective therefore will
be to minimize contact between the case and the conversion machine.
150
layered boards. A full scale experiment is proposed on the grounds that it
will verify the relative case strength with creases produced from the single
blade and the twin roll creasers, and any dynamic response on creasing at
high speed will be available for examination.
151
APPENDICES
152
Appendix Description Page No.
2a Tests for corrugated fibreboards and cases. 154
3a Case sample specification. 158
3b Folding rail calibration. 159
3c Glue dispenser calibration. 161
3d Tables of experimental data. 163
5a Determination of the third screw axis. 170
5b Cylindrical coordinate transformation. 178
5c CAD program listing. 181
6a Twin cam design optimization 186
8a Theoretical model on corrugation. 188
8b Determination of maximum twin roll gap. 190
8c Fourier harmonics for B-flute corrugation. 195
8d ESPI equation derivation. 196
9a Folding rig design specification. 202
9b LIS ortbogonal array. 211
9c Shear test 212
9d First experiment results. 216
ge Defective sample results. 226
9f SIN ratio ANOV A tables. 227
9g Optimum settings. 233
9h Confirmation run results. 235
9i Profile design for 2nd experiment. 236
9j Second experiment results. 238
9k Third experiment results. 242
153
Appendix 2a
Tests for corrugated tibreboards and cases
8S4817 Caliper test to determine the thickness of the liner or corrugated board.
8S2644 More generally known as the Cobb method, this is to determine the water
absorbancy of paper and board in a given time. The capacity of water
absorption is a function of the various characteristics of paper or board,
such as sizing, porosity, etc.
8S3137 Or known as the Mullen burst test, the method determines the bursting
strengths of paper and board. The test sample, placed in contact with a
circular elastic diaphragm, is rigidly clamped at the periphery but is free to
bulge with the diaphragm. Hydraulic fluid is pumped at a constant rate,
bulging the diaphragm until the test piece ruptures. The maximum value
of the applied hydraulic pressure will be the bursting strength of the test
sample. Usually, corrugated boards with kraft liners have higher bursting
strength than those with recycled material; this is because of the longer
fibres, and hence the higher fibre strength, found in the kraft liners.
154
subjected to puncture by a triangular pyramid puncture head attached to a
pendulum. The energy required to force the puncture head completely
through the test piece, i.e. to make the initial puncture and to tear and bend
open the board, is measured as the puncture resistance.
BS4826: VII Case compression test designed to measure the maximum force
sustainable by an erected case. The test differs slightly from the IS03037
[17].
IS03037: Determination of edgewise crush resistance. This forms the standard for
determining the edge-wise compressive strength of a case. A new method
for measuring the edge-wise compressive strength has also been proposed
by lackson et al. [18], through which the load and deformation
characteristics can be recorded simultaneously.
'A = 2..;3 -I
h
The slenderness ratio can then serve as a criterion for deciding what mode of
compressive failure will incur on the board sample of a specific size. For example,
a shon column sample with a slenderness ratio of less than 30 will provide the true
compressive strength. A medium column of slenderness ratio berween 30 and 100
will fail as a result of inelastic buckling. When the slenderness ratio is higher than
100, then long column elastic buckling will take place, giving rise to excessive out-
of-plane deformation.
155
Flexural stiffness is a strength propeny which refers to the combined board rather
than its individual component. Various methods such as the 3-point and 4-point
beam tests have been explored by McKee et al. [27] and Hohmann [13], who have
subsequently called for the standardization of such tests. Kellicutt [22] has
progressed further by examining the constituents of flexural stiffness, which
comprise the modulus of elasticity and the moment of inertia. Modulus of
elasticity is an inherent propeny that results from the fibre type. Moment of
inertia, however, varies with the dimensions and the arrangement of the individual
board components. Both properties can be affected by the type, thickness and
weight of the liners and the corrugating medium, the adhesive bond and the
manufacturing process.
Glue tests:
156
Other tests:
ISO R 287 A method for detennining the moisture content of paper boards. The
moisture content is expressed in tenns of the mass of water originally
present in the board.
157
Appendix 3a
Case Sample Specification
PANEL SCORE
28 273 499 273 497
158
Appendix 3b
Folding rail calibration
The objective is to adjust the position of the first set of folding rails (i.e folding
from 00 to 900).
The folding section is divided into the left-hand side (LHS) and the right-hand side
(RHS) in the direction of board travel. Each folding rail progresses from a
horizontal to a vertical position, and is held rigidly at three points. The first point
of contact, being a horizontal one, allows the rail clamp to slide into or out of the
folding section. The third point of contact enables the clamp to be raised or
lowered in the vertical direction. The middle point of contact can be varied in both
the vertical and horizontal directions.
Before conducting the experiment, the position of the rail was noted down in terms
of the displacement measured between the clamps and the end of the supporting
rods. It so happened that the rails were set close to the folding hinge. This was
subsequently used as the first rail setting. Afterwards, the rails were adjusted to
the other extreme positions (i.e away from the folding hinge) for the next set of
experiments. Finally, the third set of rail settings was chosen to be somewhere
between the previous two positions.
The actual experimental settings in mm taking the clamp position as origin were as
follows:
159
Point of contact
Setting First Second Third
x y
N.B With regard to A2 and A3, the second positions of the folding rails were set
in their natural position when the rails were suspended from the horizontal and
vertical end-points. During the confirmation run, it was impossible to return the
second position of Al to its original setting; for this reason, the rails were clamped
in their natural positions as in the case of A2 and A3.
160
Appendix 3c
Glue Dispenser Calibration
The first calibration test was performed on the glue flow adjust situated at the front
panel of the VC-352 unit. Strips of masking tapes were weighed prior to being
attached onto the glue flaps of the trial boards. The glue flow adjust was then set
at a maximum value. The trial boards were subsequently run through the machine
at a constant speed. The machine was stopped before the flaps were closed and
bonded together. At this stage, the masking tapes, which were carrying the glue,
would be carefully detached from the case's glue flaps and reweighed. The change
in mass would represent approximately the amount of glue applied at that particular
setting. The whole procedure was repeated, but with the glue flow adjust being set
at a minimum.
The results conclusively showed that the glue flow adjust would not be suitable for
providing three levels of glue amount settings. Incidentally, O.Olg of board ftbre
was recorded on the adhesive side of the masking tape after being removed from
the glue panel.
161
The second glue calibration test was performed using the gauge pressure valve
inside the EPC unit. The pressure line supplied air which would in turn eject the
glue onto the glue flaps. Three levels of pressure settings were chosen and the
same glue weighing technique was employed. The results were as follows:
The results from the second test showed that the pressure flow regulator would be
a better glue controller.
162
Appendix 3d
Tables of experimental data
1. Experimental Results
No 1 2 3 4 5 6 7 8 9 10
1 1 0.5 3 15 14 1 10.5 1 1 7
1 1 0 -1 -1 1 0.5 1 1 1.5
2 1 -0.5 0 0 0 0 0 1.5 4 2
0 -1 0 -1 -1 0 -1 0.5 3.5 1.5
3 0 2 2 1 0.5 0 0 0 0.5 0.5
3 1.5 2 1 1.5 1 0.5 2.5 2 1
4 5.5 1 3 1.5 3 10 2 5 3 10
-2.5 0 0 0 -0.5 0 -1 ·1 -1.5 -1
5 3 1.5 2 1 0.5 1 1.5 1.5 2 2
3 1 2 1 1 2 1.5 1.5 2 2
6 1.5 2 3 2.5 3 1.5 2 1 1 1
3 3 0 3.5 3 2 1.5 0 1 1
7 18 0.5 0 8 0 0 1 2 1 1.5
-1.5 0.5 0 2 1.5 1 2 1 0 1.5
8 2.5 4.5 1 3 0.5 3 1 0.5 0.5 3
1 0.5 1.5 -1 0.5 1 1 0 0.5 1
9 2 2 2 3 2 2 2 1 2 1
2 2 2 3.5 3.5 2 0 5 2.5 2
163
2. SIN values for the various control factors.
164
3. Combined SIN ratios at various control operation levels.
165
4. ANOVA table for SIN ratio
Combined Response
Factor DoF Sum of sq. Mean square F
A - Rail 2 19.2 9.6 7.8
B - Glue 2 14.1 70.7 57.9
C- Speed 2 2.4 1.2 -------
D - Crease 2 36.1 18.1 14.8
Total 8 199.1
(Error) (2) (2.4) (1.2)=S2
166
5. Alternative mean and variance analysis.
167
cont.!..
168
6. Confirmation Runs
CRI I-lOa 3.5 3.0 2.0 3.5 2.0 3.0 3.5 5.0 3.0 5.5 Ytot=34.0
SIN = -5.3
I-lOb 3.0 2.0 0.5 2.0 4.5 1.0 5.0 2.5 3.5 4.0 Y 101=28.0
S/N=-4.5
169
Appendix Sa
Determination of the Third Screw Axis
Referring to publications by Hunt [15] and Phillips [36], the following section
provides a summary on the three axes theorem and the derivation of the relevant
equations. Scaled parameters from the prototype model are then substituted into
the appropriate equations in order to determine the characteristics of the third screw
axis, ISA32, between the coil cam and the case panel.
Notation:
00 = angular velocity of a body in the right hand screw convention (0012 = angular
velocity of body 2 with respect to body 1).
h = pitch of screw (+ve if right-handed).
't = linear velocity along an instantaneous screw axis, ISA.
v = instantaneous linear velocity of a rigid body.
z = displacement along the cylindroid axis with respect to its origin.
\jI = angles as measured clockwise from a specified line to the relevant 00 vector
At an instant during the relative spatial motion of two rigid bodies 1 and 2, an
instantaneous screw axis ISA I2 can be uniquely defined, about or along which the
body 2 is said to be screwing with respect to body 1. All points I, in the body 2
instantaneously resident in ISA I2 have equal linear velocities 't 12 relative to 1 along :_\
the ISA, while all other points in the body 2 have velocities other than this. The
linear velocity with respect to 1 of body 2, 't 12, can be calculated as follows:
170
In general, for finite motion, the location in space and the h-value of an ISA
changes continuously.
Rigid body motion can be generalized further such that the instantaneous linear
velocity with respect to body 1 of any point R2 in body 2 is given by:
Figure Sa.! (with acknowledgement given to Hunt [15] for the reproduction of the
figure) shows two rigid bodies in relative spatial motion. co12 is a free vector,
therefore it may be drawn anywhere with significance; V IR2 and '1: 12 however are
bound vectors such that they must be drawn from the points in body 2 to which
they specifically refer.
When three bodies are in relative spatial motion, the vector polygon of the three
angular velocities will close; hence:
This vector polygon is in the form of a planar triangle, which means that the three
ISA's reside respectively in three planes all parallel to one another. Phillips and
171
Hunt [35] have also shown that the three ISA's have the same line as common
perpendicular. By taking the directed distance along the common perpendicular
from ISA 12 to ISA23 to be ZI2.23' it can be shown graphically that:
Hence, the three axes theorem states that, for three rigid bodies in spatial motion,
their instantaneous screw axes must reside in parallel planes and share a common
perpendicular. Kennedy-Aronhold's theorem [23] is a special case whereby all the
ISA's are orientated parallel to each other.
The three axes theorem is useful from the practical point of view. For example, if
properties about two ISA' s are known, then those of the third ISA can also be
identified. It is interesting to note that while the unique solution for '00' depends
upon the actual values assigned to the other two ' 00', the corresponding values of
the directed distance, pitch and axis orientation for the third ISA will depend solely
upon the co-ratio of the two known ISA's.
Body Description
a cylindroid system can be drawn with the corresponding ISA's as shown in Figure
5a.2.
172
(SA 13
~o
Fig. 5a.2 Three axes screw systein.
ISA 21
Figure 5a.3 shows the velocity polygon for the three angular velocities:
w
2l
173
-------
From the three axes theorem, the sum of the angular velocities can be expressed in
terms of equation (3). Also, from Figure 5a.3, the following expressions can be
derived geometrically:
hence,
sincj)
= - coscj) .........................(5)
tamlr32
_ sincj)
-. ..................... (6)
Slmjr32
Expressing 10013 x zl3.321 in equation (4) as 0013(Z,2 - Z,I - b), and resolving
perpendicularly to ISA 32 will give:
Solving equations (5), (6), (7) and (8) simultaneously for z", h 23 , "'23 and 00n, a
unique solution will be derived in terms of ro,l and 0012 for ISA23 • Dividing
equations (7) and (8) by 0012 and then substituting (5) and (6) will result in the
following equations:
174
which can be explicitly expressed as:
Also, substituting equations (13) and (14) back into equations (11) and (12) will
finally give:
During the initial design stage, a coil cam prototype, as shown in Figure 5a.4, has
been constructed in order to estimate some of the scaled design parameters.
175
Fig. 53.4 Coil earn prototype.
Based on the scaled prototype, the following design parameters have been
arbitrarily chosen:
(013= I rad / s
co" = 3 X ~, = 3 rad / s
h!3 = 100 mm / rad
h'2 =0
b = 50 mm
$ = 45 0
Substituting the above values into equations (15), (4), (6) and (2) will yield the
following results:
p = 79.06 mm
0/32 = 10.80
~2 = -3.77 rad / s
h32 = -29.36 mm
176
The velocity polygon is shown in Figure 5a.5:
/>/'",12
.. , ('--- 1(11 ,2 J
"V 21
Finally, from Sine Rule, the angle '1'32,21 can be derived as follows:
W
= sin-l[~ sin(135)]
w32
= 10.810
177
Appendix Sb
Cylindrical Coordinate Transformation
The three angles ($,9,,,,) shown in Figures 5b.l and 5b.2 are those of roll, pitch
and yaw (refer to Paul (32)).
x z y
'2 'y
r:/
~ 0 i'
z y y x x
178
Transfonning the coordinates (x,'!',Tx) from a cylindrical frame about the x-axis to
one about the z-axis will require the following equations:
z = r...sinlJr
r .COS'"
4> = atan2( z 'I' )
X
x = r.cos4>
IJr = atan2( ~ )
r.sm4>
i) x-axis to y-axis
y = rpnlJr
e = atan2( r;:os1Jr )
x
x = r,cose
IJr = atan2( ~ )
r;me
z = r;me
4> = atan2(y
rcose )
y
179
iv) z-axis to y-axis
y = r.cos6
6 = atan2( ~ )
r zSIDcj)
180
Appendix Se
CAD Program Listing
181
'include <math.h>
'include <stdio.h>
'include <assert.h>
'define FALSE 0
'define TRUE 1
main () (
pi-4*atan (1);
poqaO;
data_input ();
transform() ;
data_output():
poq-(-atan(3.0/4.0»*180/pi;
rx_a-sqrt(pow(rx_a,2)+900);
transforrn() ;
datayq() ;
void data_input() (
printf("\n3D CAM PROFILE OPTIMIZATION \n\n"):
printf("Enter values for the following:\n");
printf("Angular vel. ratio c ~);
scanf("\lf",&w_ratio);
printf("Offset distance, a · H);
scanf{"\lf",&a);
printf("Pitch of ISA 1, P - n);
scanf("\lf",&p);
printf("Radial distance of panel, rx - M);
scanf("\lf",&rx_a);
printf("Angle psi{ab) - H);
scanf(H\lf",&psi_ab);
printf("cylindroid axis distance, z32 ~ H);
scanf("\lf",&z32);
printf(HTwist angle alpha32 ~ H):
scanf(H\lf",&alpha32);
do (
printf{"Number of angular increments _ H);
scanf(H%ld",&delta);
I
while «(delta> 2) && (delta < 252») =a FALSE);
I
void transform() (
double xa[2S0],xb(250),psia[250],psib[2501,zb[2S0],rx_tmp;
double phi b(2501,rz b[250},zd(250),phi d[2S0],psi d(250J,rx d[250];
int count : I; - - - -
182
phi b[count]-atan2(rx a*cos(psib[count]),xb[countJ);
r~_b(COunt]-sqrt(pow(~(countJ,2)+pow(rx_a,2)*pow{cOS(pSib[countJ),2»);
void data_output() I
FILE *datal, *data2,.*data3;
datalafopen("datal.com",·w+");
fprintf(datal,"INPUT LIST @Ol \d \n", (delta+I)*4):
for (count-I; count<-delta+l; count++ ) {
if (count .... 1) (fprintf(datal,nAPPEND LIST eOI \f lif lif lid \n",ylcount],
z{count),-(psi_dinc[count]-90),count);)
else (if (count !- delta+l) fprintf(datal,"APPEND LIST eOI \f \ f \f lid \n",
y{count],z(count),-psi_dinc[count],count);
else fprint£(datal,"APPEND LIST eOl lif li£ \f \d",y{count],z[count],
-psi_dinc[countl,count);
fclose(datal);
data2-fopen("data2.com","w+");
fprintf(data2,"INPUT LIST e02 \d \n", (delta+l) *2);
for (count-I; count<-delta+l; count++ ) I
if (count !- delta+l)
fprintf(data2,"APPEND LIST e02 \d lif \n",count,x[count]l;
else fprintf{data2,"APPEND LIST e02 \d lif",count,x[count]);)
fclose(data2);
I
void datayqll (
FILE: *data3;
data3~fopen("data3.com","w+");
fprintf(data3,"INPUT LIST e03 \d \0", (delta+l) *6);
for (count-I; count<-delta+l; couot++ ) (
if (count !- delta+l)
fprintf(data3,"APPEND LIST @03 \f \f lif \f 'H \f\n",xp{countl,
yp(countl,zplcountl,xq[count]-xplcount},yq!count}-yp[count),
~qlcount]-zp(countl);
else fprintf{data3,"APPENO LIST @03 \f \f %f lif \f \f",xp[count),yp[countl
~p{count],xq[countl-xp[count),yq{count)-yp!countl,zqlcount}-zp[count]
fclose(data3);
I
183
cont.!..
2. Input lists generated the optimization program.
184
cont.!..
3. Doer Macro commands for imponing input lists in order to generate coil
cam proftle.
185
Appendix 6a
Twin cam design optimization
Factors \ Levels I 2 3
Since there were four factors in an ~ orthogonal array. and only three were needed
to be optimized, one factor column was therefore left empty as dummy.
Two criteria. designed to minimize the fmal cam radius and the cam velocity ratio.
were studied in the optimization process. The signal-to-noise ratios were as
follows:
186
Results from the spreadsheet program were as follows:
AI -40.5 -11.66
A2 -39.41 -11.84
A3 -38.52 -11.87
BI -35.32 -9.5
B2 -39.72 -11.97
B3 43.39 -13.89
Cl -38.56 -14.17
C2 -39.44 -11.47
C3 -40.44 -9.73
Optimized settings were: A3. BI. Cl (min. r) & Al. BI. C3 (min. 'V ratio)
187
Appendix 8a
Theoretical Model on Corrugation
Figure 8a.1 shows a point load acting at location 'A' as it penetrates deep into the
corrugation.
The flute can be modelled as a simple sinusoidal structure whose function is:
21tX
Y = Y cos- (1)
max P
As the inner liner curves over the flute, it will tangent off at some distance X. and a
depth d, such that: .
21tx,
d Y
(l-cos-)
=
'mu p
2y -d
y-(Ymax -d,) = mu '(x-x,) (2)
x- P
, 2
Differentiating eq(l) and equating with the line tangent from eq(2) will give:
2ymax -d,
x-I!..
, 2
188
substituting d, into the above expression and converting it into an iterative function:
(3)
p -x
2 '
x, can be detennined by iteration with a seed value of x equal to p/4 (see Figure
8a.l for approximation), putting each successive X. into the right-hand side of the
equation.
The next step is to consider the total length of the inner liner which has been
stretched between the two peaks of the corrugation (Le one flute pitch). The length
's' for a curve with function f(x) can be expressed as follows:
X·X, , - - - - - -
tip = 2 J p
P
2 "
2
mu
21U
1 + B(sin 21tXidx + 2 (- -x \ +[y (cos-' +1)]2-p
p
x-o
where B = (y 21ti
.... p
189
Appendix 8b
Determination of Maximum Twin Roll Gap
The idea for adopting a variable gap twin roll design is to avoid a roll change
when it comes to creasing boards of different calipers. Currently there are three
major design variables that need to be established; they are the twin roll gap, the
cone angle '9' and the creaser width, as shown in Figure 8b.1. This appendix aims
to estimate the three parameters which will serve as seed values for the finite
element model and experimentation.
profile width
shoulder radius
creaser lip
Theoretical derivation
At present, there are two propositions for setting the twin roll gap. One way is by
relating the gap width to the thickness of the board; but this may result in a crease
width larger than required and therefore poor appearance at the case corners.
Another way is to correlate the twin roll gap setting with the flute geometry. To
be more precise, a twin roll gap is chosen in such a way that the least damage is
imparted to that part of the corrugated board which will be most susceptible to
tearing, i.e the inner liner.
Figure 8b.2 models a flute structure subject to a point load at location 'C' -
somewhere between the peak and the trough of the flute.
190
, 0.,.", . ,
,••• Ir." •.
Placing a point load at location 'C' will depress the liner by o. Assuming no
slippage between the creaser and the inner liner, the part of liner most likely to tear
will be that between the points 'c' and 'B'. If the liner has a maximum
permissible strain rate of E, then the condition for no tearing will be:
from which,
By moving the point 'C' from x=O, the point on the liner will eventually meet the
flank of the flute, after which the original load will be supponed by the flute rather
than the liner. For the present purpose, this point 'x' beyond which the liner is
off-loaded will be called Xml' Using the harmonic components for a typical B-flute
corrugation (see Appendix 8c), y(x) in equation (5) can be expressed in terms of
the corrugation curve. From experimental data on load elongation, as quoted by
Thomas [50], most liners in the moisture range of dry to 10 % RH can stretch by
1.5 % to 2.5 % without breaking. Hence, by assuming a liner strain rate of 1.5%,
191
xcrit. can be derived from equation (5) and the corrugation curve, and is shown in
the following table:
If the twin roll creaser lands on the corrugation such that the twin rolls staddle
over a peak, then the tips of the creaser will behave like two point loads. The
maximum gap width for the twin roll, as shown in Figure 8b.3, can therefore be
determined from the previous X- values.
....
192
What happens if the twin roll gap lands at or in the vicinity of location 'A'? This
is where the twin roll shoulder effect comes into play. Figure Sb.4 shows one half
of the twin roll creaser, where the coordinate origin has been shifted along the
corrugation centreline to location 'A', such that the y-axis is now coincident with
the centreline of the twin roll.
% ',U?W
s: ZS: .
. . . . 1 •• 1'"11
The creaser profile gradient can be determined by examining the strain imposed on
the liner when the creaser begins to penetrate into the corrugation. It follows that:
e=tan-,_a_ (3)
l!. -x
2
where,
a=( P -x)/2€ +€2
2
In fact, e is independent of the flute pitch; its value is calculated to be 9.S6° for a
liner strain rate of 1.5 %. Using this value as our creaser shoulder angle, the twin
roll creaser width can be derived in terms of the minimum number of flutes
affected by the crushing action. For example, if the board is reduced to 'k' times
its original thickness, then the creaser shoulder angle will affect a certain width of
the flute. Referring to Figure Sb.4, half the creaser width can be determined by:
w = (l-k)h-a (4)
tane
193
When the middle of the twin roll gap falls in line with location 'A' (i.e x=O),
creasing of single-wall boards to a penetration of 'k' equal 20 % will imply 'w'
values of 8.52 mm and 12.76 mm for the B- and C-flute respectively. Doubling
these values will give the total twin roll creaser width.
With double wall boards, since the top layer is essentially a C-flute board, it may
seem logical to adopt a twin roll creaser whose width is designed for crushing C-
flute boards. However, interaction between the two layers prevents the board from
being treated like two sheets of independent lamina. Hence, it is not feasible to
derive a twin roll creaser width for the double-wall boards.
Finally, the assumption that there is no slipping between the creaser roll and the
inner liner is invalid (unless the friction coefficient between the two media is very
high - a situation made possible by increasing the moisture content of the liner).
Since the maximum permissible depression for a single-wall board is found to be
0.051 mm using equation (5). However, the present single blade design has an
effective blade penetration of 0.762 mm with no sign of liner tearing. If slipping is
accounted for, a value of 0.518 mm will be arrived at after a similar calculation.
Further penetration will have to be due to the compliance of the corrugation
structure.
194
Appendix Se
Fourier Harmonics for B-flute Corrugation
The following table provides the first ten Fourier components of the B-flute
corrugation waveform. Each harmonic component is represented by the sum of a
sine wave and a cosine wave. In the table below, the 'real' number is the
amplitude of the cosine wave, and the 'imaginary' number is that of the sine wave.
The magnimde of each component is the square root of the sum of the squares of
the real and imaginary values. The phase value is the number of degrees that the
harmonic component is shifted from a pure cosine wave. Harmonic 0 is the d.c
component
195
Appendix 8d
ESPI Equation Derivation
When the surface is displaced, the phase of all the scattered components changes
by the same amount ~c)l(r) such that the complex amplitudes of the object beam at
P(r) is now:
196
12 = u~ + u.; + 2uoUg::os[4>o - 4>R + a4>(r)] .••••••.••.(4)
Comparing the equations for 11 and 12 , the intensities will be the same when Llcp(r)
= 2n7t. This is also the case when the intensities of two sets of randomly
distributed speckle pattern are correlated using the correlation coefficient:
p = -;:=:=E[:::XYJ:::::::::;-;:::E[=X]:::E[:::YJ==:;:
JE[(X-E[X]~E[(Y-E[YJ~
Expressing intensities in terms of the two random variables 'X' and 'Y' will give:
1
p(a4» = -[1 + cos(a4>)]
2
where n = 1.2.3....
On comparing the two speckle patterns. fringes formed by lines of equal intensities
will provide information about the relative displacements of different pans of the
surface. In ESPI. the comparison of speckle patterns is achieved by means of a
TV system. In order to compensate for the low spatial resolution of a video
system. a small aperture viewing lens is needed for resolving the speckles (see
Wykes [57]).
The image plane is first of all positioned to coincide with the face plate of a TV
camera. In its undisplaced state. the object image is recorded on a video store.
The object is then displaced and the live image is subtracted electronically from the
stored picture. hence the term •electronic' in ESPI. Areas of zero intensity will be
those where Lle!> = 2nn. Conversely. the speckle pattern will have a maximum
contrast and mean intensity when LlCP = (2n+ l)n. It is the variation of mean
197
contrast and mean intensity when ~cp = (2n+ 1)1t. It is the variation of mean
intensity and the contrast of the speckle pattern across the subtracted speckled
images that constitute the fringe pattern. The fringes represent the correlation
between the two images and are therefore referred to as correlation fringes (which
must not be confused with interference fringes).
In an orthogonal coordinate system, if an object lies in the ~,X3 plane and viewing
is in the x, direction, then fringes can be obtained which represent contours of
constant out-of-plane displacement (d,) and in-plane displacement in a specific
direction ~ or d3 •
0,_
Do x2
u. V
/'
1/
~ OR
x,
_-+__ IrNJOO~
198
between the image plane speckle patem of Do and the 'dashed' position DR' For a
given point in the image plane, the intensity I, is given by equation (3).
4nd,
acj)(d,) ; -A- ....................(6)
which changes the intensity at the point to 12 as shown in equation (4). Combining
equations (5) and (6) will give the maximum correlation between I, and 12 , Hence,
1
d, ; -nA ..........................(7)
2
X2 o
o x,
L B
For a general form of A<I> such that either or both the viewing and illumination
directions are not parallel to the surface-normal,
2n
acj) ; -(no - n,).d ......................... (8)
A
199
The situation arises when divergent illumination is used or the object is non-planar.
B ut for any deviation of less than 150 from the normal, the following will be true
to a good approximation:
Figure 8d.3 shows the optical arrangement for an in-plane displacement sensitive
speckle pattern correlation interferometer.
x,
L
u,
L,
Fig. 8d.3 ESPI in-plane interferometer.
Wavefronts Uo' and Uo" are obtained by allowing half the incident plane wavefront
U o to illuminate the object D directly, while the other half is reflected onto the
surface by the mirror M. In this way, Uo' and Uo" will be inclined at equal and
200
opposite angles to the xI-axis surface-nonnal. Displacing the object D by a
distance d(dl'~,d,) will give:
Since the relative phase of V; and VD" is constant over planes lying parallel to the
x l x3-plane, the displacement components d l and ~ in the plane will not introduce a
relative phase change. From equation (5) and (9), the in-plane displacement can be
calculated as follows:
nA
~ = 2sin6 ......................(10)
Refer to Wykes [57] and Jones et al. [20] for further infonnation on ESPI.
201
Appendix 9a
Folding Rig Design Specification
BACKGROUND
Special folding rigs are known to have been used in paper research. The main
purpose of the rigs is to obtain the crease stiffness or the folding resistance of
corrugated fibreboards from its initial position of O" to the final folded position of
slightly less than 180". Though there is no detailed publication on the rig design,
Vogelpohl [51], Grebe [9] and McKee [26] have all reported substantial findings
from such apparatus.
The folding test rig used by Vogelpohl is unique in design, in that it consists of a
platform which is symmetrically divided along a folding hinge. One half of the
platform performs the folding action, while the other half remains static. Vacuum
pads on the platform enable the board samples to be held down, though it is not
clear how the force sensor is allowed to operate at the same time. Force sensing is
performed by means of strain gauges, and a rotational transducer measures the
angular displacement simultaneously. The signals, once digitized, can be displayed
and analysed on a dual channel plotter.
Other folding test rig design reported by Buchanan [3] and Grebe [9] are capable
of recording the force required to fold a 127 mm wide strip of creased sample.
The angle of fold at any instant can be displayed on an oscilloscope over a range
of folding speeds.
DESIGN SPECIFICATIONS
202
factors which must be built into the rig. The test rig must be robust enough to
process a range of board samples. Graphical representation of the folding
resistance is particularly informative, when different board samples can be
compared at a glance. The overall rig, including instrumentation and software,
must be economically viable and fit for the present purpose.
To summarize, the folding test rig design is expected to fulfil the following
criteria:
I. To measure the folding resistance relative to the angular displacement
when a board sample is subject to the folding action.
2. To give a graphical representation for instant interpretation.
3. Design must be robust enough to cater for corrugated fibreboards of
various thicknesses and liner grades.
4. Inexpensive to build.
Figure 9a.1 shows the folding test rig design. The basic mechanical features of the
folding rig consist of a counter-balanced folding leg the end of which is fitted with
a knife edge. A board sample is placed onto the principal support bar, whilst one
edge of the board is sitting on the tip of the cantilever. By latching the folding
leg, and hence the knife edge into position, the crease will lie immediately under
the knife edge. When the d.c motor is switched on, the centre roller on the folding
arm will act against the board panel. At this instant, the centre roller and the tip of
the cantilever act as though they were two point contacts, imposing a folding
couple on the board about the knife edge.
Once the board is folded to slightly over 90", the folding leg will be unlatched
automatically. This enables the folding leg to swing to its balanced position,
thereby clearing way for the board sample. The d.c motor is deactivated by means
of two limit switches at both the forward and the reverse strokes.
203
Fig. 9a.l Folding test rig design.
During the folding action, the strain and the encoder signals are sampled
continuously. The sampling rate corresponds to half a degree for every strain
sample reading.
Several minor problems were highlighted during the calibration of the folding test
rig. One problem occurred when the board sample momentarily left the support
edge as soon as the folding leg was unlatched. Together with the high sensitivity
of the strain gauges on the cantilever, the net effect was a sudden transition in
folding resistance as indicated by the kink in the middle of the folding resistance
204
curve. Side shims were therefore added to prevent the board from leaving the
suppon bar. In addition, a piece of spring steel was fitted onto the locking ring at
the top of the folding leg so as to dampen the swinging motion.
The rig was calibrated by adjusting the gain in the electronic strain gauge amplifier
for a known weight at the tip of the cantilever. A calibration factor was
subsequently included in the operation program FOLD3.BAS, a listing of which is
included in this appendix.
To operate the test rig, the board sample is first of all loaded with the folding leg
properly latched in position. After initializing the program 'foldrig2.exe', folding
is commenced by pressing the key '5' and switching on the motor simultaneously.
Upon saving the test data, the results can be displayed at a later stage by retrieving
the datafile via the display program 'fold3.exe'.
205
cont.!..
206
,**************************************************************************
,* Folding Test Rig Data Analysis Program *
,* (Adapted from D. Holman's program for flute height measurement) *
,* *
,* By A M Lau 10/8/93
,* *
*
,* Department of Mechanical Engineering, Loughborough University *
'**************************************************************************
DEFINT A-N
DEFSNG O-Z
'$DYNAMIC 'Allocate storage space for arrays only when the program is
'running and not during compilation
DIM SHARED profile(O TO Limit) AS INTEGER 'UPPER BOUND OF PROFILE & FILTER
CLS
SCREEN 9
CALL Startup Screen & graphics start up.
CALL Main
END
DEFINT A-N
SUB DrawProfile
*~**************************************************** ********
Procedure for plotting points in array "profile" onto screen
*~********************************************~******* ********
SHARED profile() AS INTEGER
count 1
FOR X = 0 TO XMax STEP 1
PSET (X, profi1e(X»
NEXT X
END SUB
DEFINT A-N
SUB Extract (Curve(»
,**-****************************************************
Algorithms for extracting the curve characteristics.
,*******************************************************
YAmax = 0
YBmax = 0
ScanWidth = 5 , Proportion of 1st peak curve to be sampled
for determining the gradient of curve.
FOR X = 0 TO (XMax / 2) STEP 1
IF Curve(X) > YAmax THEN
YAmax Curve (X)
MkA = X 1st peak marker.
207
END IF
IF Curve (XMax / 2 + X) > YBmax THEN
YBmax = Curve (XMax / 2 + X)
MkB = XMax + X , 2nd peak marker.
END IF
NEXT X
YC = YAmax
FOR X = (MkA + 1) TO MkB STEP 1
IF Curve (X) < YC THEN YC Curve (X)
NEXT X
DEFINT A-N
, ************************************************************
Returns a file name from user input
, ************************************************************
FUNCTION GetFileName$ STATIC .
INPUT "Enter data file to load (press ENTER to quit): ", FTemp$
GetFileName$ = FTemp$
END FUNCTION
DEFINT A-N
SUB Main
, ***************
Main Program
***************
DIM P! (0 TO 20)
TopValue = Limit
DO Program main DO loop.
OPEN "OUT.DTA" FOR OUTPUT AS #2
FileName$ = GetFileName$ , Input file data.
IF FileName$ = ., .. THEN , End if <ENTER> pressed.
END
ELSE OPEN FileName$ FOR BINARY AS #1
END IF
GET # I, 1, filesize
IF filesize > TopValue THEN
filesize = TopValue , Keep filesize within limits
END IF
208
count = count + 1
END IF
IF count = 6 THEN slope = TRUE
entry = entry + 2
WEND
count = 1
FoldE! = 0
DO WHILE entry <= filesize
GET i1, entry, number
Value = number / CalFac
WRITE 12, Value
profile (count) = CINT(Value)
FoldE! = FoldE! + Value
entry entry + 2
count = count + 1
LOOP
CALL DrawProfile
CLOSE n, #2
CLOSE OUT.DTA
END SUB
DEFINT A-N
SUB Startup
I ***********************************************
Set up graphics and screen for curve display.
***********************************************
209
LOCATE 23, 15
PRINT "30 11
LOCATE 23, 28
PRINT "60"
LOCATE 23, 40
PRINT "90"
LOCATE 23, 52
PRINT "120"
END SUB
210
Appendix 9b
LIB Orthogonal Array
Control Factors
Expt. A B C D E F G H
1 1 1 1 1 1 1 1 1
2 1 1 2 2 2 2 2 2
3 1 1 3 3 3 3 3 3
4 1 2 1 1 2 2 3 3
5 1 2 2 2 3 3 1 1
6 1 2 3 3 1 1 2 2
7 1 3 1 2 1 3 2 3
8 1 3 2 3 2 1 3 1
9 1 3 3 1 3 2 1 2
10 2 1 1 3 3 2 2 1
11 2 1 2 1 1 3 3 2
12 2 1 3 2 2 1 1 3
13 2 2 1 2 3 1 3 2
14 2 2 2 3 1 2 1 3
15 2 2 3 1 2 3 2 1
16 2 3 1 3 2 3 1 2
17 2 3 2 1 3 1 2 3
18 2 3 3 2 1 2 3 1
211
Appendix 9c
Shear Test
Procedures
The objective of the shear test is to provide a measure for fishtailing. The shear
test arrangement is as shown in Figure 9c.1.
Essentially, the folded board sample is placed onto the platform of the Fletcher's
trolley, with the edge of the base panel acting against the steel. By attaching a
spring clip onto the corner of the overlapping panel, and loading it with a pulley
mechanism, the shear effect can thus be simulated under experimental condition.
The fishtailing measure will be qeterrnined by the displacement between the edges
of the panels.
212
Calibration
A calibration test was conducted on the HH 300(B) board samples with a size of
lOO mm x 400 mm. Following creasing at a compression ratio of 90 % and
folding by means of the folding test rig, the fishtail result was measured before
loading the sample onto the Fletcher's trolley. The shear measure was then
recorded at each mass increment until the board could no longer withstand any
more shear loading. The shear radian could hence be derived by dividing the
difference in shear before and after loading by the edge length of the overlapping
panel. In order to differentiate between the direction in which shear could take
place, a positive shear was designated for the one with the panel displacing in the
direction of the applied load, and a negative shear for the contrary.
The results, including the first peak of the folding resistance, were as tabulated
below:
213
Table below showing the shear deformation (after) at different load increments:
1 35 6 18 23 25 --- --
2 0.5 2 4 32 -- --- ---
3 6.0 9 285 37 49 .--. ---
4 0.5 35 10 185 25 -- --
5 0 0.5 3 17.5 -- -- --
6 -7.0 -6 -45 -2.0 -- -- --
7 -3.0 -1 0 9.5 19 245 --
8 1.5 4 17 25 35 -- --
9 -6.0 --- -- -- -- -- --
10 0 3 55 31 --- -- --
214
0·1\
C·Il· .
SHEAR
/rad
) ,
,,
'"
LOA,D/g
Recommendation
From the calibration test, it was found that a mass of around 0.7 kg would be
suitable for loading the single-wall boards whose sample size is 200 mm x 400
mm. A mass of 1.4 kg would be recommended for loading the double-wall boards.
The masses must be kept constant throughout the experiment as evidence from the
shear radian plot suggested that sudden shear jumps might occur as a result of
increasing the load. This sudden transition in shear deformation could be explained
by the buckling of the flute structure.
215
Appendix 9d
First Experiment Results
The folding resistance first peak data for all the experiments have been stored in
the following computer fIle directories:
The sample number has been labelled in tenus of the experimental number, the
batch material number and the number within the test batch. For example,
'ONElll' corresponds to the frrst sample in the eleventh material batch type of
experiment ONE.
216
Expt. I: Lab. RH = 45%, Expt. 2: Lab. RH = 45%,
No. 1st peak Shear Shear Edge Shear Comment No. 1st peak Shear Shear Edge Shear Comment
before after length radian before after length radian
ONE I I 0.703 -I 17.5 195.5 0.094 MC TWOII 0.854 0 6 190 0.031 SFB
ONEI2 0.734 I 15 194.5 0.071 TWOl2 0,712 2 IO 195,5 0.040 SFB
ONEI3 0.73 I 17.5 196 0.084 TWOl3 0.792 0.5 6.5 193 0.031 SFB
ONEI4 0.818 -I 16.5 196.5 0.089 TW0I4 0.938 0.5 11.5 194.5 0.056 SFB
ONEI5 0.752 -I 17.5 197.5 0.093 SFB TWOl5 0.947 0.5 7.5 189.5 0.036 SFB
ONE41 1.279 -1.5 3.0 193.5 0.023 TW041 1.51 0.5 2.5 193.5 0.01
ONE42 1.504 -0.5 4.0 195.5 0.023 TW042 --- 0 1.0 191.5 0,005
ONE43 1.473 -1.5 1.5 198.5 0.015 TW043 1.76 2 6 191 0.Q2
ONE44 1.279 -1.5 5.5 197 0.035 SFB TW044 0.76 3 5 196.5 om SC
ONE45 1.283 -1.5 -1.0 197 0.002 TW045 1.75 2.5 3 190.5 0.002
ONE72 2.41 -0.5 0.5 197 0.005 TW072 2.80 3.5, 3.5 192.5 0
ONE73 2.39 -2.0 -1.0 198.5 0.005 MC TW073 3.04 2 2.5 194.5 0,002
ONE74 2.28 0.5 1.5 196.5 0.005 MC TW074 2.84 0.5 1.5 193.5 0.005
ONE75 2.62 -0.5 0.5 197.5 0.005 TW075 2.60 1.0 1.5 190.5 0.002
ONEIOI 2.61 -2.0 -1.0 197 0.005 TWOIOI 3.2 -11.5 -11 200 0.002
ONEI02 2.53 1.0 2.0 196 0.005 TWO 102 3.19 2 I 196.5 0.005
ONEI03 2.77 -6.5 16.0 195.5 0.115 SFB TWOI03 3.35 -9 -8.5 199.5 0.002
ONEI04 2.62 0,5 2.5 196.5 0.01 TWOI04 3.33 -935 -9 200.5 0.002
ONEI05 2.67 0 2.5 196 0.012 TWOI05 3.33 -7.5 -6 199 0.007
ONEIII --- --- --- --- --- CL TWOIII 3.18 3.0 7.5 187.5 0.024
ONEI12 --- --- --- --- --- CL TWOl12 ---- ---- "--- ---- .--- CL
ONEI13 --- --- --- --- --- CL TWOIl3 1.85 ---- ".". ---- .--- SFB
ONE I 14 2.36 2.5 7.5 190 0.026 TWO I 14 2.58 3.5 6 188 0.013
ONEI15 3.14 0.5 1.5 193 0.005 TWO I 15 3.02 ---- ---- ---- .---
Expt. 3: Lab. RH = 45%, Expt. 4: Lab. RH = 45%,
No. 1st peak Shear Shear Edge Shear Comment No. 1st peak Shear Shear Edge Shear Commenl
before after length radian before after length radian
THRII 0.77 0.5 10.5 195.5 0.051 SFB FOUll ---- ---- .-.- ---- ---- SFB
THRI2 0.73 7.5 6.5 195.5 0.005 MS FOUI2 1.0 0 7 195 0.035
THRl3 0.64 .. _- ---- ---- ---- SFB FOUl3 ---- ---- ---- ---- .--- SFB
THRI4 0.93 6.5 10.5 195.5 0.018 FOU\4 ---- ---- ---- ---- ---- SFB
THRI5 0.83 ---- ---- ---- ---- SFB FOUlS ---- ---- ---- ---. ---- SFB
THR41 1.49 ---- ---- ---- ---- SFB FOU41 2,32 ---- ---- ---- ---- SFB
THR42 1.78 ---- ---- ---- ---- SFB FOU42 ---- ---- ---- ---- ---- SFB
THR43 1.6 .-.- ---- ---- ---- SFB FOU43 ---- ---- ---- ---- ---- SFB
THR44 1.63 15 15 197.5 0 FOU44 ---- ---- ---- ---- ---- SFB
THR45 1.83 9.5 10.5 188.5 0.005 FOU45 2,16 -2.5 ·2.0 197.5 0.002
THR71 2.63 13.5 13.5 188.5 0 FOU71 3.S9 ·2 -1.5 198 0.002
THR72 ---- 16.5 16.5 186.5 0 FOU72 3.26 ·2 -1.5 196 0.002
-
N
00 THR73
THR74
----
2.68
----
14
----
14.5
----
186.5
----
0.002
MD FOU73
FOU74
3.57
3.74
-3
0
·2.5
0.5
195.5
195.5
0.002
0.002
-
THR75 2.53 18 18 187 0 FOU75 3.22 0.5 1.0 195 0.002
THRI02 ---- ---- ---- ---- ---- CL FOUI02 3.87 ·1.5 I 194.5 0.002
THRI03 ---- .--- ---- ---- ---- CL FOUI03 3.73 2.5 3 191.5 0.002
THRI04 ---- ---- ---- ---- ---- CL FOUI04 3.54 L5 I 192.5 0.002
THRill ---- ---- ---- ---- .--- CL FOUIII 2.97 I 2 191 0.005
THRI12 ---- ---- ---- ---- ---- CL FOUI12 2.46 3.5 5 189.5 0.007
THRI13 ---- ---- ---- ---- ---- CL FOUI13 3.31 0 7 188 0.037
THRI14 ---- .--. ---- ---- ---- CL FOUI14 3.4 0.5 2.5 190.5 0.01
THRIIS ---- ---- ---- ---- ---- CL FOUI15 2.9 ·1 3 191 0.01
Expt. 5: Lab. RH = 45%. Expt. 6: Lab. RH = 45%,
No. 1st peak Shear Shear Edge Shear Comment No. 1st peak Shear Shear Edge Shear Comment
before after length radian before after length radian
F1VII 0.57 2.5 6 193 0.018 SIXII ---- --.~ ---- ---- ---- SFB
F1V12 0.55 -2.5 I 195.5 0.007 SIXI2 ---- ---- ---- ---- ---- SFB
F1V13 0.65 6.5 15 195 0.004 SIXI3 1.04 0 1.5 187 0.008
FIVI4 0.65 10.5 13 195 0.015 SIXI4 0.8 -2.5 5 193.5 0.038
F1V15 0.63 2.5 7 193.5 0.022 SIXI5 ---- ---- .... ---- ---- SFB
FIV41 1.15 0.5 0.5 194 0 SIX41 1.86 3 3.5 193.5 0.002
FIV42 1.06 -I -0.5 194.5 0.002 SIX42 2.08 -0.5 0 193.5 0.002
F1V43 1.06 -2.5 -2 198 0.002 SIX43 1.75 3 3.5 192.5 0.002
F1V45 1.11 8 9.5 191.5 0.007 5IX45 1.81 -0.5 0.5 192.5 0
F1V71 1.6 -0.5 0 195 0.002 5IX71 2.89 11.5 11.5 198 0
FIVn 1.65 -2 -1.5 195.5 0.002 SIXn 2.85 -2 -1.5 196 0.002
FIVIOI 2.48 4.5 6 197 0.007 SIXIOI 3.47 -6.5 -6 198 0.002
F1V102 2.09 5 6 195 0.005 SIX 102 3.06 -2.5 -2 195.5 0.002
F1VI04 2.04 6.5 7.5 195 0.005 SIXI04 3.16 -5 -4.5 196 0.002
FIVI05 2.21 2.75 3.5 198 0.003 SIX 105 ---- ---- ---- ---- ---- CL
FIVIII .--- ---- ---- ---- ---- CL SIX11I ---- ---- ---- ---- _.-- CL
F1V112 ---- ---- ---- ---- ---- CL SIXI12 ---- ---- ---- ---- ---- CL
FIVI13 ---- ---- ---- ---- ---- CL SIXI13 ---- ._-- .--- ---- .. -- CL
FIVI14 ---- ---- ---- ---- ---- CL SIX114 .--- ---- ---- ---- ---- CL
F1V115 ---- a 1.5 191 0.007 SIX115 ---- ---- .--- ---- ---- CL
Expt. 7: Lab. RH = 47%.
Expt. 8: Lab. RH = 45%,
No. 1st peak Shear Shear Edge Shear Comment No. 1st peak Shear Shear Edge Shear Comment
before after length radian before after length radian
SEVII 0.96 ---- ---- ---- ---- SFB EIGII ---- ---- .--- ---- ---- SFB
SEVI2 0.77 .--- ---- 0--- ---- SFB EIGI2 1.08 ---- ---- ---- ---- SFB
SEVI3 0.87 ---- ---- 0--- ---- SFB EIGI3 I .--- ---- ---- ---- SFB
SEVI4 0.79 -I.S S 195 0.033 EIGI4 0.8 ---- ---- --- ."-- SFB
SEVIS 0.96 .--- ---- ---- ---- SFB EIGI5 1.16 ---- ---- .--- ---- SFB
SEV41 2.03 -3.5 -3 199 0.002 ---- .--- ---- SFB
EIG41 2 ----
No. 1st peak Shear Shear Edge Shear Comment No. 1st peak Shear Shear Edge Shear Comment
before after length radian before after length radian
NINII 0.51 I II 193 0.051 SFB TENI1 ---- ---- ---- ---- .--- SFB
N1N12 0.56 ._-- "--- ---- ---- SFB TENI2 0.376 4 8.5 187 0.024
N1N13 0.55 ---- ---- ---- ---- SFB TEN 13 0.358 ---- ---- ---- ---- SFB
N1N14 0.57 4 14.5 196 0.053 TENI4 0.81 -I 15 192.5 0.083
NINI5 0.56 -I 7 194.5 0.041 TEN15 1.07 ---- ---- ---- ---- SFB
N1N41 1.05 4 6 196.5 om TEN41 2.11 ---- ---- ---- .--- SFB
NIN42 -_.- ---- -._- ---- ---- MD TEN42 1.82 -I 3 193.5 0.02
NIN43 1.19 3.5 5 191 0.007 TEN43 1.85 2 5.5 190.5 0.018
NIN44 1.47 -1 0 192.5 0.005 TEN44 0.81 ---- ---- ---- ---- SFB
NIN45 1.58 1.5 0.5 189.5 0.005 TEN45 1.94 ---- ---- ---- ---- SFB
N1N71 1.21 3 3 190 0 TEN71 3.06 2 2.5 189.5 0.002
N1N72 2.16 0 0.5 189 0.002 TEN72 3.43 -1.5 -1.5 190 0
N1N73 1.95 6 6.5 190 0.002 TEN73 3.13 5 5.5 190.5 0.002
N1N74 1.95 3 3.5 191 0.002 TEN74 3.11 2.5 3 191 0.002
NIN75 2.33 0 0 192 0 TEN75 3.28 1.5 2 191.5 0.002
NINIOI ---- ---- "--- .-.- .--- CL TENtol 3.79 14.5 15.5 186 0.005
NINI04 2.35 3 5.5 192.5 0.012 TENI04 3.66 17.5 19 187.5 0.008
N1N105 ---- ---- ---- ---- ---- CL TENto5 3.79 13 14 189 0.005
NINIll ---- ---- ---- ---- ---- CL TEN 11 1 2.41 6 8.5 182 0.013
NINI12 ---- ---- -.-- ---- ---- CL TEN 112 2.13 0.5 6 190 0.005
NINl13 ---- ---- -_.- ---- ---- CL TENI13 1.82 4 6.5 190 0.013
NINI14 ---- ---- ---- ---- ---- CL TENI14 2.65 3 8 184 0.027
---- ---- CL
Expt. 11: Lab. RH = 45%,
Expt. 12: Lab. RH = 45%,
SFB
TWEII ---- ---- ---- CL
___ A
----
SFB
TWEI2 0.624 ---- ---- ---- ---- SFB
ELEl3 0.80 2 7 193.5 0.025
TWEI3 ---- ---- ---- .--- ---- CL
ELEI4 1.13 I 6 193.5 0.025 MS
TWEI4 ---- .--- ---- ---- ---- CL
ELEI5 1.15 0 5.5 194.5 0.028
TWEI5 0.522 ---- ---- ---- ---- SFB
ELE41 2.5 ---- ___ A
CL
ELEW3 3.73 -8 -6 193.5 0.01
TWEI03 ---- ---- ---- ---- ___ A
CL
ELEI04 4.26 -7.5 -5.5 193.5 0.01
TWElO4 ---- ---- ---- ---- ___ A
CL
ELEW5 3.43 0.5 1.5 196 0.005
TWEI05 ---- .--- ---- ---- ___ A
CL
ELEIII ---- ---- ___ A ___ A
---- CL
TWEIII ---- ---- ---- ---- ___ A
CL
ELEI12 ---- ---- ___ A ___ A
---- CL
TWEI12 ---- ---- ---- ---- .--- CL
ELEI13 ---- ---- ___ A
___ A
---- CL
TWEI13 ---- ---- ---- ---- ---- CL
ELEI14 2.79 0.5 1.5 193.5 0.005
TWEI14 ---- ---- ---- _._- ---- CL
ELEI15 2.69 ---- ---- .--. ---- SFB
TWEI15 ---- ---- ---- ---.
Expt. 13: Lab. RH = 45%. Expt. 14: Lab. RH = 45%.
No. 1st peak Shear Shear Edge Shear Comment No. 1st peak Shear Shear Edge Shear Commem
before after length radian before after length radian
THIlI 0.93 1.5 3.5 192.5 0.005 FORI I 0.6 -2 - 197.5 0.005
THIl2 ---- ---- ---- ---- ---. SFB FORI2 0.63 ---- ---- ---- --.- SFB
THIl3 ---- ---- ---- ---- .--. SFB FOR!3 0.5 -5 -4 197 0.005
THIl4 ---- .--- ---- ---- ---- SFB FORI4 0.58 9 10.5 192 0.007
THII5 ---- ---- ---- ---- ---. SFB FORI5 0.64 5 6.5 195 0.007 SC
THI41 ---- ---- ---- ---- ---. SFB FOR41 1.26 -0.5 I 196 0.007
TH142 .--- ---- ---- ---- ---. SFB FOR42 1.22 9 10 191 0.005
THl43 ---- ---- ---- ---- .--. SFB FOR43 1.09 ·4 ·3.5 198 0.002
THl44 ---- ---- ---- ---- .--. SFB FOR44 0.89 0 I 195.5 0.005
TH145 -_.- .--- ---- ---- ---. SFB FOR45 1.05 ·2.5 ·1.5 196 0.0005
THI71 3.35 I 1.5 192.5 0.002 FOR71 1.69 -2.5 -2 200 0.002
THl72 3.84 4.5 5 188 0.002 FOR72 1.66 -3.5 -3 198 0.002
THI73 4.07 8 8.5 194.5 0.002 FOR73 1.66 -5 -4.5 198 0.002
TH174 3.74 0.5 I 196 0.002 FOR74 1.91 -2.5 -2 197 0.002
THl75 4.14 5.5 6 193.5 0.002 FOR75 1.62 -2.5 -2 199.5 0.002
THII03 4.43 1.25 2.5 190 0.006 FOR 103 1.89 0.5 1.5 193.5 0.005
THII04 4.16 1.5 2.5 191 0.005 FORI04 1.96 4 4.5 192 0.002
THII13 3.24 0.5 1.5 185 0.005 FORI13 2.27 2.5 4 194 0.007
THII14 3.45 1.5 2 185 0.002 FORI 14 .--- ---- ---- -.-- ---. SFB
THll15 3.15 0 I 186 0.005 FORI 15 ---- ---- ---- ---- ---- CL
Expt. 15: Lab. RH = 45%,
Expt. 16: Lab. RH = 45%,
-_.- SFB
FIF12 ---- .--- ---- ___ A
---- SFB
SXTl2 0.75 ---- ---- .--. ---- SFB
FIFI3 ---- ---- ---- ___ A
---- SFB
SXTl3 0.64
___ A
---- ___ A
---- SFB
FIFI4 ---- ---- ---- ___ A
---- SFB
SXTl4 0.84 ---- ----
___ A
.--- SFB
FIFI5 0.97 6.5 9 194.5 0.012 SC
.. -
___ A
SFB
FIF41 1.72 ·2.5 ·2 197.5 0.002
SXT41 1.45
___ A
---- SFB
FIF43 1.19 ·3 ·2.5 196.5 0.002
---- ----
___ A
SXT43 1.33
___ A
SFB
FIF44 1.37 ·1 ·0.5 196.5 0.002
.---
___ A
SXT44
___ A
---- ___ A
.--- SFB
FIF71 2.04 1.5 1.5 195 0
SXT71 2.25 0.5 1.5 196.5 0.005
FIF72 2.21 0.5 0.5 195 0
SXT72 2.39 ·2.5 2 199.5 0.002
FIF73 2.32 2 2 196.5 0
SXT73 ---- ·8.5 7.5 198 0.005 MC
FIF74 2.16 ·0.5 0 198.5 0.002 SC
SXT74 ---- 1.5 2 196.5 0.002
4 4 198
FIF75 2.4 0
SXT75 -_.- ---- ---- ___ A
---- MD
FIF 10 1 1.91 0 0 192 0
SXTl01 2.31 ---- .--. ---- ---- SFB
FIF102 2.04 0.5 I 190 0.002
SXTl02 2.64 ·4.5 3 193.5 0.007
FIF103 2.02 0.5 1 193 0.002
SXT103 2.33 2 4 190 0.01
F1F104 1.9 0 0 193 0
SXT104 2.55 0 3 194 0.015
F1F105 ---- ---- ---- ___ A
.--- CL
SXTl05 2.63 2 3.5 197 0.007
FIFlll ---- ---- ----
___ A
---- CL
SXT111 _.-- -_.- ---- ---- .--- CL
F1Fl12 ---- ---- ---- ---- ---- CL
SXTl12 2.58 ·1 3 193 0.02
F1Fll3 ---- ---- ----
___ A
---- CL
SXTl13 3 5 11.5 188 0.034
FIF114 _.-- ---- ---- ___ A
.--- CL
SXTl14 ---- ---- ---- .-.- ---- CL
F1Fl15 ---- ---- .--- ___ A
---- CL
SXTl15 2.71 0.5 2 190.5 0.007
Expt. 17: Lab. RH = 45%. Expt. 18: Lab. RH = 45%.
No. 1st peak Shear Shear Edge Shear Comment No. 1st peak Shear Shear Edge Shear Comment
before after length radian before after length radian
SYTIl 0.89 ---- ---- -_0- ---- SFB EGTlI 0.38 .--- --.- ---- ---- SFS
SYTI2 0.73 ---- ---- "--- ---- SFS EGTl2 ---- ---- ---- ---- ---- MD
SYTI3 0.74 ---- ---- -._- ---- SFS EGTl3 0.87 ---- ---- ---- ---- SFB
SYTI4 0.87 ---- ---- ---- ---- SFS EGTl4 0.87 ---- ---- -.-- ---- SFB
SYTI5 0.79 ---- ---- ---- ---- SFB EGTl5 0.89 ---- -.-. ---- ---- SFS
SYT41 l.5 0 0.5 196.5 0.002 EGT41 l.97 -1.5 I 191.5 0.002
SYT42 1.5 ---- ---- ---- ---- SFS EGT42 l.92 -2.5' 2 194 0.002
SYT43 1.49 0 0.5 195 0.002 EGT43 0.87 2.5 2.5 192 0
SYT44 l.67 -I 0 192.5 0.002 EGT44 1.95 ---- .--. ---- ---- SFB
SYT45 1.59 ---- ___ A
Board Type
230 CC 250TH 400 KT 400KK 250TCH
226
Appendix 9f
SIN Ratio ANOV A Tables
Folding resistance
N.B 'F' ratios marked with *** indicate those control factors which have been
pooled:
230 CC
Overall mean = 2.128
Factor Sum of Sq DoF Ave. Sq F Rank
A 0.156 1 0.156 ***
B 0.457 2 0.229 ***
C 1.281 2 0.64 2.815 4
D 0.552 2 0.276 ***
E 2.097 2 1.049 4.61 2
F 1.376 2 0.688 3.024 3
G 13.7 2 8.352 36.72 1
H 0.427 2 0.214 ***
Total 23.05 15
Error 1.592 7 0.227
227
250 TCH
Overall mean = -4.04
Factor Sum of Sq DoF Ave. Sq F Rank
400 KT
Overall mean = -8.26
Factor Sum ofSq DoF Ave. Sq F Rank
A 0.007 1 0.007 ***
B 0.378 2 0.189 ***
C 7.629 2 3.814 36.41 2
D 0.148 2 0.074 ***
E 0.786 2 0.393 3.751 4
F 0.2 2 0.1 ***
G 31.18 2 15.59 148.8 1
H 2.929 2 1.465 13.98 3
Total 43.26 15
Error 0.733 7 0.105
228
400 KK
Overall mean = -9.09
Factor Sum of Sq DoF Ave. Sq F Rank
A 0.146 1 0.146 ***
B 4.933 2 2.467 6.472 4
C 12.58 2 6.291 16.51 2
D 0.856 2 0.428 ***
E 0.015 2 0.008 ***
F 1.65 2 0.825 ***
G 13.58 2 6.788 17.81 1
H 5.481 2 2.74 7.19 3
Total 39.24 15
Error 2.668 7 0.381
250 TCH
Overall mean = -7.82
Factor Sum of Sq DoF Ave. Sq F Rank
A 2.282 1 2.282 ***
B 11.81 2 5.907 1.472 3
C 7.975 2 3.987 ***
D 11.9 2 5.951 1.483 2
E 12.79 2 6.393 1.593 1
F 10.3 2 5.148 1.283 4
G 8.573 2 4.287 ***
H 9.261 2 4.631 ***
Total 74.89 15
Error 28.09 7 4.013
229
Lateral shear
230 CC
Overall mean = 31.5
Factor Sum of Sq DoF Ave. Sq F Rank
A 44.4 1 44.44 ***
B 170 2 84.8 4.82 1
C 55.8 2 27.9 1.59 3
D 79.5 2 39.77 2.26 2
E 16.7 2 8.368 ***
F 9.34 2 4.67 ***
G 44.2 2 22.08 ***
H 43.5 2 21.75 ***
Total 463 15
Error 158 9 17.58
250TH
Overall mean = 47.8
Factor Sum of Sq DoF Ave. Sq F Rank
230
400 KT
Overall mean = 54.1
Factor Sum of Sq DoF Ave. Sq F Rank
A 8.83 1 8.828 ***
B 20.6 2 10.31 3.58 3
C 129 2 64.63 22.4 1
D 0.65 2 0.327 ***
E 25.9 2 12.95 4.5 2
F 7.1 2 3.549 ***
G 2.67 2 1.336 ***
H 6.68 2 3.338 ***
Total 202 15
Error 25.9 9 2.881
400 KK
Overall mean = 45.1
Factor Sum of Sq DoF Ave. Sq F Rank
A 4.14 1 4.139 ***
B 99.2 2 49.59 7.68 3
C 130 2 64.98 10.1 1
D 47.2 2 23.61 3.66 4
E 26.7 2 13.35 ***
F 12.6 2 6.316 ***
G 119 2 59.33 9.19 2
H 1.72 2 0.858 ***
Total 440 15
Error 45.2 7 6.456
231
250 TeH
Overall mean = 40.5
Factor Sum of Sq DoF Ave. Sq F Rank
G 48 2 23.98 2.91 3
H 6.77 2 3.385 ***
Total 268 15
Error 57.7 7 8.245
232
Appendix 9g
Optimum Settings
230 CC 1 2 3 4 Comment
Folding resistance Gl - F2 C3
Shear - D2 C2 -
Optimum Gl D2 C2 F2
250TH 1 2 3 4 Comment
Folding resistance Gl - D2 C3
Shear C3 - - G3
Optimum G2 C3 D2 F2
400 KT I 2 3 4 Comment
Folding resistance G1 C3 - -
Shear C3 - - -
Optimum settings GI C3
233
Optimum set for 400 KK (Gemini).
400KK 1 2 3 4 Comment
Folding resistance G1 C3 - -
Shear C3 02 - D3
Optimum settings C2 02 D3 F2
Folding resistance - Dl - Fl
Shear F2 - Gl -
Optimum settings F2 G3 Cl D1
234
Appendix 9h
Confirmation Run Results
N.B Letter next to the 1st peak folding resistance indicates the position of crease
along the flute.
235
Appendix 9i
Profile Design for 2nd Experiment
The generation of the twin roll creaser profile was based on the geometrical
analysis of a double-flute structure. To replace the pre-crusher would mean the
incorporation of the crushing function into the creaser itself. Consequently, the
profile must be wide enough to reduce the section thickness of the corrugation to a
reasonable level, but not too wide to encourage shear deformation.
The depth of penetration for the creaser could be estimated from the thickness of
the buckled medium over which the creaser would define a score line. For
example, there would be at least five layers of buckled material in a creased
double-wall board. The width of the creaser could also be estimated from the pitch
of the C-flute board, since, being the top layer of the double-wall board, the C-flute
would be subject to the creasing and crushing forces on initial contact with the
creaser.
The twin roll creaser design consists of a pair of identical creaser rolls with
shoulder radius 'R' and tip radius 'r', as shown in Figure 9i.1.
236
where p = f x C-flute pitch.
f = fraction of the flute pitch
k = compression ratio as defined by Vogelpohl [51]
w = effective width over which creaser has contact with the board
H = total board caliper
h = creaser roll gap
(kHi + p2
R = -'-~-'-
2kH
Choosing the values 1.5, 1.75 and 2 for 'f', and a C-flute flute pitch of 7.547 mm
under an .80 % compression (i.e k=0.8), the following values were calculated and
were subsequently used in the experiment:
Creaser no. R h w
2 20.31 2 13.4
4 15.57 2.5 11
5 20.31 1.5 14
6 25.78 2 15.3
7 15.57 2 11.5
9 25.78 1.5 16
237
Appendix 9j
Second Experiment Results
238
B5 0.796 -1 5 193 0.031
VI B1 1.022 0 3.5 196 0.017 SFB
B2 1.208 5 6.5 193 0.007
B3 0.942 -1 3 196.5 0.02 SFB
B4 1.084 0 5 193.5 0.025 SFB
B5 0.92 1 3 197 0.01
VII B1 1.168 2.5 10 189 0.039 SFB
B2 1.124 -1.5 8 190 0.05 SFB
B3 1.155 -0.5 9 191 0.049 SFB
B4 1.199 -3.5 7 192 ·0.054 SFB
B5 0.911 -0.5 5 193 0.028 SFB
VIII B1 1.27 -1 1.5 193.5 0.012
B2 1.376 -1.5 4 194 0.028
B3 1.487 0 4.5 191 0.023 SFB
B4 1.327 0 2.5 193.5 0.012
B5 1.327 0 4 193 0.02
IX B1 0.792 0 5.5 195 0.028
B2 0.867 0 5.5 194 0.028
B3 0.978 -2.5 5 196 0.038 SFB
B4 0.902 8.5 11.5 190 0.015
B5 0.712 0 3.5 193 0.018 SFB
239
Percentage of defects and signal-to-noise ratios for both the ftrst peak folding
resistance and the shear deformation are tabulated as follows:
240
ANOV A tables for the folding resistance and the shear deformation are as follows:
Folding resistance
Shear deformation
241
Appendix 9k
Third Experiment Results
242
2.51 3.4 0.131 0 2.5 194.5 0.012
2.25 2.76 0.04 3 14 189 0.058
2.81 2.37 0.091 -4.5 4 192.5 0.044
2.58 3.49 0.033 1.5 11.5 191.5 0.052
2.47 2.86 0.081 4 10 188.5 0.031
2.05 2.82 ---- 0 10 190 0.052 NPA
2.88 2.37 0.124 1.5 10 189 0.044
VI 1.45 3.28 ---- -0.5 5 193.5 0.028 NPA
2.51 2.3 0.04 -9 8 192.5 0.088
2.4 2.75 0.012 0 14 196 0.074
2.19 2.95 0.023 3.5 11.5 191.5 0.041
1.91 2.88 0.013 0 5 190.5 0.026
VII 3.51 4.72 0.197 0 3.5 188.5 0.018
3.18 3.3 0.183 -1 1 193 0.01
3.2 3.3 0.076 -11 -7.5 199.5 0.017
VIII 2.66 3.64 0.18 6.5 11 193 0.023
2.59 3.62 0.098 -2.5 2 193 0.023
1.82 2.4 0.013 -5.5 9 188 0.077
* NPA - No clearly defined peak 'A', hence 1st peak gradient could not be
recognized by software.
243
Test No. No. and type of defects+
I 4pee
n 3 pee, 1 WC
1lI 2 pee
IV 1 pee, 6 CC
V 1 pee
VI 2 pee, 3 CC
VII 1 pee, 6 cc
VllI 6 cc
IX 1 pee, 1 CC
SIN ratios for the fraction of defective samples (with the two types of failure being
separated, i.e. those due to pre-crushing and creasing), the folding resistance peaks
, A' and 'B' and gradient 'D', and the shear defonnation are as shown below:
pee Defect Rate cc Defect Rate
Exp Orth. P l-p T]""f P I-p T]""f
No. Array
244
Shear Folding resistance
Exp Orth. l1'bear No. 11A 11B No. 110 No
no. Array
** 11rce deliberately left out because no pre-crushing failures could possibly be due
to these factors.
245
ANOV A Tables:
Total 19.13 8
Error 1.03 4 0.26
246
IV) Folding resistance llD' M = -28.21
Factor Sum of sq. DoF Ave. Sq. F Rank
A 150.28 2 75.14 2.02 2
B 83.5 2 41.75 ----
C 247.47 2 123.74 3.33 1
D 65.31 2 32.66 ----
Total 546.56 8
Error 148.82 4 37.2
Response\Rank 1 2 3
Defects % 01 B3 C2
Shear Bl/2 C3/2 02/3
Folding resistance 'D' C2 A3/2
Folding resistance 'A' Cl Al/2
Folding resistance 'B' 01/3 Cl/2
247
The optimized settings chosen were D1, B2, C2 and A2, and the experiment was
conducted at 53 % rel. humidity and room temperature of 24 ·C. Two samples
were torn during the experiment: one during pre-crushing and one during creasing.
The results were as follows:
248
REFERENCES
249
[1] Ball, R.S., "The theory of screws", Cambridge University Press, 1900.
[2] Benson, R.E., "Effects of relativity humidity and temperature on tensile stress-
strain properties of kraft linerboard", TAPPI 54(5), p699-703, 1971.
[3] Buchanan, 1.S., "The effect of crease form on the compressive strength of
corrugated cases", Packaging, vo!. 34, 1963.
[4] Chakraborry, J. & Dhande, S.G., "Kinematics and Geometry of Planar and
Spatial Cam Mechanisms", Wiley Eastern Ltd., 1977.
[5] Chen, F.Y., "Mechanics and Design of Cam Mechanisms", Pergamon Press,
1982.
[7] Edholm, B. & Eriksson, E.L., "Swedish study tells how flap folding affects
cartoning", Packaging, vo!.28, 1983.
[8] Goodman, 1.W., "Laser Speckle and Related Phenomenon", Chap2, ed. 1.C.
Dainty, Springer-Verlag, Berlin, 1975.
[9] Grebe, V.W. & Nebel, F., "Uber das Rillen von Wellpappe", Papier &
Kunststoff Verarbeiter, Teil 2: 8/80 und Teil 3: 2/81.
250
[13] Hohmann, RJ., "Biegesteifigkeit von Pappe und Kanon", Verpackungs-
Rundschau, nr.5, 1971.
[16] Ishikawa, K., "What is Total Quality Control? The Japanese Way", Prentice-
Hall, 1985.
[18] Jackson, C.A., Koning, J.W. and Gatz, W.A., "Edgewise compressive test of
paperboard by a new method", Pulp Paper Mag. Can., vo1.77, pan 10, p43-
46,Oct.1976.
[20] Jones, R., Wykes, C., "Holographic and speckle interferometry", Cambridge
University Press, 1989.
251
[23] Kennedy, A.B.W., "The Mechanics of Machinery", Macmillan, London, 1886.
[24] Lau, A.M. & Davies, T.H., "Case-making Process Optimization", TAPPI Conf.
Journal, Oct. 1992.
[25] Leung, F.B., "Dynamic of a spatial cam", BSc Project Repon, Mech. Eng.,
Loughborough University of Technology, 1993.
[26] McKee, R.C. & Altmann, F.J., "Comparative Evaluation of panel or body
creasing wheel contours", TAPPI vol. 39, pan 7, 1956.
[27] McKee, R.C., Gander, J.W. and Wachuta, J.R., "Flexural Stiffness of
Corrugated Board", Paperboard Packaging, Dec. 1962.
[28] McKee, R.C., Gander, J.W. and Wachuta, J.R., "Compression strength formula
for corrugated boxes", Paperboard Packaging, 48(8), p.149-159, 1963.
[29] Mechanica v.4: "Applied Structure", licensed under Rasna Corporation, 2590
Nonh First St., #200, San Jose, California 95131, U.S.A.
[30] Nordman, L., Kolhonen, E. and Toroi, M., "Investigation of the compression
of corrugated board", FEFCO, 15th Congress, Estoril, Ponugal.
[31] Olmstead, E.H., "Poppet valve dynamics", J. E.S. Taylor Aeronautical Sci.,
p.370-375, 6 Jul., 1939.
252
[34] Phillips, E.A, "Quick calculations for corrugation stiffness", published in
"Engineering Data for Product Design", edited by D C Greenwood,
McGraw-Hill, 1961.
[35] Phillips, J .R. & Hunt, K.H., "On the theorem of three axes in the spatial
motion of three bodies", Aust J. Applied Science, vol.15, 1964.
[36] Phillips, J., "Freedom in Machinery", vo!. 1 & 2, Cambridge University Press,
1984 & 1989.
[37] Plucker, J., "On a new geometry of space", Phi!. Trans., vol.155, 1865.
[38] Rothban, H.A., "Cams: Design, Dynamics and Accuracy", Wiley, NY, 1956.
[39] Rother, V., "Faltung im Luftstrom", Papier & Kunststoff Verarbeiter, Nr. 3,
1988.
[41] Ruvo, A., Fellers, C. and Engman, c., "The influence of raw material and
design on the mechanical performance of boxboard", Svensk Papperstidning,
nr.18, 1978.
[42] Seymour, W.B. & Terle, M., "Perfect Square Box Folding with the Multi-T
Belt Concept", TAPPI Journal, Jan. 1985.
253
[44] Smith, G.E., "Elastic buckling in shear of infinitely long corrugated plates
with clamped edges", thesis submined for the MAeroEng, Cornell Univ.,
USA,1957.
[45] Sprague, C.H. & Whitsin, W.J., "Medium fracture and strength losses in
fluting", TAPPI Journal, Oct. 1982.
[46] Sze, W.S. & Yu, K.M., "Rapid prototyping - a review", Hong Kong Engineer,
Dec. 1993.
[47] Taguchi, G., Elsayed, A. & Hsiang, T.C., "Quality Engineering in Production
Systems", McGraw-Hill Book Company, 1989.
[51] Vogelpohl, V.H. & Hohmann, H.J., "Versuche zur Optimierung des Rillens
von Wellpappen", Verpackung-Rundschau, Nr.8, 1987.
254
[54] Werner, A.W., "The Manufacture of Fibre Shipping Containers", Fibre
containers, part 12, p42, Dec. 1940.
[56] Whitsitt, W.J., "Papermaking Factors affecting Box Properties", TAPPI Proc.,
COIT. Containers Conf., 1988.
255