You are on page 1of 10

925

The performance envelope of forming shoulders and


implications for design and manufacture

C J McPherson, G Mullineux*, C Berry, B J Hicks and A J Medland


Innovative Manufacturing Research Centre, Department of Mechanical Engineering, University of Bath, Bath, UK

Abstract: The production of packaging machinery is a highly competitive global market driven by the
ever-increasing demands of customers and legislation. The fundamental design principles of many
packaging machines are the result of incremental improvements made over the last few decades.
This paper looks at the underlying theory for forming shoulders and, starting with previously
published results, determines the performance envelope relating to certain critical parameters. The
findings are discussed in the light of their relevance for the creation of new designs.

Keywords: geometric modelling, differential geometry, machine–material interaction

NOTATION aim of increasing the production capabilities, improving


reliability and reducing down time and production costs
c0 , c2 , c3 , c4 , c5 coefficients in the normalized bending [2]. To accomplish this it is not only necessary to improve
curve the machines themselves, but also to understand the
eðÞ error in the bending curve interaction between the machines and the packaging
f ðÞ normalized bending curve materials [3]. Environmental legislation such as the
h height of the bending curve European Packaging Waste Directive aims to reduce
R radius of the tube the amount of material being used in packaging. This
v circumferential parameter around the means that the producers of fast-moving consumer
tube goods (FMCGs) are driven to use thinner, lighter-
zðvÞ bending curve weight and recycled packaging materials to convey
their goods to the consumer. The changing specification
 opening angle of packaging materials and the need to satisfy the new
 angle between surface normals on the requirements that they bring put additional pressure on
bending curve packaging machine manufacturers to update and
0 back angle improve their machines.
1 front angle The designs of many modern packaging machines are
 curvature of the wrapped bending the result of incremental improvements driven largely by
curve empirical experimentation [4]. Over the years, designers
 curvature of the planar bending curve have developed rules of thumb relating to the way that
 angular parameter around the tube particular materials interact with specific types of machine.
The fundamental understanding of the machine–material
interaction is largely absent. Computer-based modelling
1 INTRODUCTION of the process and machine components can be used to
improve the understanding of the complex processes and
The production of packaging machinery is a highly com- the physical interactions occurring in packaging processes.
petitive global industry. In the year 2000, US shipments This modelling aids the understanding but can also be used
of packaging machinery were valued at just under US$5 to extract or generate design rules to assist the develop-
billion [1]. Packaging manufacturers seek to improve ment of the next generation of machines.
continually the performance of their machines with the This paper concerns the design of the crucial compo-
nent of one particular type of machine. This is the
The MS was received on 4 July 2003 and was accepted after revision for forming shoulder (sometimes called a flow former) in a
publication on 19 April 2004.
*Corresponding author: Innovative Manufacturing Research Centre, vertical form, fill and seal packaging machine. This
University of Bath, Bath BA2 7AY, UK. type of machine creates bags or pouches from a flat
B12803 # IMechE 2004 Proc. Instn Mech. Engrs Vol. 218 Part B: J. Engineering Manufacture

Downloaded from pib.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on April 7, 2015


926 C J McPHERSON, G MULLINEUX, C BERRY, B J HICKS AND A J MEDLAND

reel of packaging material. Bags produced this way are


commonly used to package particulate products such
as flour, snacks, pasta, cereal and hardware.
Section 2 describes the current design practice for
forming shoulders. This is usually based on heuristics.
A number of mathematical approaches for describing a
forming shoulder have been proposed. Section 3 dis-
cusses one of these [5]. Four basic design parameters
and their associated limiting constraints are identified.
It can be seen that the shape of the shoulder is very
sensitive to changes in these parameters. Indeed, some
changes result in shoulders which violate some of the
required constraints. The effects of small changes are
examined in more detail in section 4 and this leads on
to the identification of some design implications in
section 5.

2 CURRENT DESIGN PRACTICE FOR FORMING


SHOULDERS Fig. 1 Schematic diagram of a packing system using a form-
ing shoulder
Bags or pouches made from film or paper are a familiar
form of protective packaging. The paper or film from
which these packages are made is manufactured as a These include packs with flat bottoms and tetrahedral
flat, pre-printed sheet of uniform width and stored on a tea bags [2].
roll. For dry free-flowing materials the vertical form, fill Although some attempts have been made at casting or
and seal packaging machine is the most commonly used moulding forming shoulders, the majority of formers are
for transforming the flat sheet into bags. Figure 1 shows still constructed from a flat metal sheet cut into two
a schematic view of such a machine. The single web of parts. For the purpose of this work, these sections are
film or paper is drawn from a roll and fed over a forming referred to as the tube and the collar (Fig. 2). The tube
shoulder that guides the material from the flat to a tubular portion of the forming shoulder has a circumference
shape around the product feed tube. The edges of the film equal to twice the full bag width.
are brought together. Either they are overlapped (with The process of making a shoulder starts with the bend-
opposite sides of the film meeting) to produce a lap seal, ing curve which ultimately forms the edge over which the
or they are formed as a fin seal (with the internal sides film passes. Initially the bending curve is regarded as a
of the film meeting). The seal is formed by heating or planar curve as in Fig. 2a. Suppose that the part below
glueing. In this way a tube is produced into which the the curve is cut from flat sheet. Wrapping this round
product to be packaged can be dropped in measured into a circular cylinder gives the vertical tube (Fig. 2c).
quantities. The tube of packaging material is then cross- Similarly, if the part above the planar bending curve is
sealed to form a sealed bag. The final seal also forms the cut from flat sheet and wrapped around, then it forms
bottom seal of the next bag. This basic approach has the shoulder surface (Fig. 2b). The assembly is shown
been adapted to produce a variety of different packs. in Fig. 2d. The shape that the forming shoulder assumes

Fig. 2 (a) Bending curve, (b) collar, (c) tube and (d) complete shoulder

Proc. Instn Mech. Engrs Vol. 218 Part B: J. Engineering Manufacture B12803 # IMechE 2004

Downloaded from pib.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on April 7, 2015


PERFORMANCE ENVELOPE OF FORMING SHOULDERS AND IMPLICATIONS FOR DESIGN AND MANUFACTURE 927

depends upon the bending curve and the radius of the in Fig. 2d. A model based on two conic surfaces
tube. separated by a planar triangle can be used [6] and the
The methods of forming shoulder construction have bending curve again taken as the intersection with a
an inherent degree of imprecision. This is compensated circular cylinder. This does indeed generate a collar
for by the experience of those personnel who design surface containing a triangular plane at the highest
and make them. A shoulder has to be matched to the point to permit the transition of the film from a flat
material that is to be run over it. The aim is to have plane into the tube. This approach results in an equation
the film track smoothly across the shoulder with no for the bending curve with four design parameters: the
permanent stretching and no tearing of the packaging height, the radius, the angle between the tube and
material. Minor problems can be compensated for by collar at its highest point, and the included angle of the
hand tuning of the shoulder but for major problems a planar triangle as variables.
completely new former has to be made. In some cases Although successful, the resultant family of practical
as many as four or five modifications may be made to shoulders is somewhat limited. There are potentially
produce a suitable former. This can be a costly and more versatile approaches available for finding suitable
time-consuming approach. This reliance on experience bending curves [5, 7]. It is the latter approach that is
and trial-and-error testing has led to shoulder manu- adopted here and the supporting theory is now reviewed.
facture being described as a ‘black art’. It is worth One of the requirements for the shoulder is that there
noting, however, that packaging machinery manufac- should be no stretching or tearing of the film as it passes
turers have been building reliable forming shoulders for across it. This requires that the collar surface needs to be
decades. isometric to the plane. This in turn means that the collar
has to be a developable surface [8]. Such surfaces are
formed from generators which are a single-parameter
3 MODELLING THE FORMING SHOULDER family of straight lines.
As noted in reference [5], it follows from a result in
A schematic view of the typical form, fill and seal reference [9] that, for a given bending curve wrapped
machine is shown in Fig. 1. Here the collar is that surface around a cylinder, there are two developable surfaces
over which the film passes before being taken over the containing the curve. One of these is the cylinder itself
edge of the shoulder and passing into the main circular and the other surface, once determined, can be used to
filling tube. This edge is referred to as the ‘bending form the collar.
curve’. If the filling tube were cut along a generator at Let  denote the curvature of the bending curve when
its lowest point and unrolled, the result is a flat sheet of wrapped around the cylinder and let  be the curvature
width equal to the circumference of the tube with the for the corresponding planar curve. Then  is certainly
bending curve defining its top side. Indeed one way to smaller than . Let  denote the angle between the sur-
form the tube is to take such a sheet and to roll it into face normals to the cylinder and the collar at the typical
a tube of circular cross-section. point on the bending curve. This is, of course, also the
In its flattened state, the bending curve is roughly angle between the tangent planes to the two surfaces.
parabolic. One design method is to assume that it is Then, as in reference [5], this angle is related to the
exactly parabolic and to define it by curvature by
 2
zðvÞ ¼ h½1  ðv=pRÞ2  for  pR 4 v 4 pR 
cos  ¼ 1  2

where R is the radius of the tube and h is the height of the
top of the bending curve above its lowest point. The If zðvÞ denotes the planar bending curve with
parameter v effectively represents the angular position pR 4 v 4 pR, then the curvatures can be found [5]
around the circumference of the tube. and the above relation becomes
One of the key design parameters is the height :radius
ratio h:R. In practice these normally range between 2 R2 z2vv  z2v  1
cos  ¼  ð1Þ
and 5 with values in the lower part of this range being R2 z2vv þ z2v þ 1
the most common. An alternative means for determining
the bending curve is to regard the collar as being part of where the subscript v denotes partial differentiation with
the surface of a cone and then to take the curve as the respect to v.
intersection of a cone and a circular cylinder [6]. The collar surface, if extended infinitely, has an edge of
In practice, packaging film is fed from a roll. This regression along which it turns back on itself. This clearly
means that material starts as a flat plane and there is a represents a singularity to be avoided on that portion of
need for a smooth transition between the plane and the the surface used for the collar itself. This can be ensured
collar. To this end a triangular planar surface needs to by arranging that the edge of regression lies within the
be included in the collar at its highest point, as illustrated region occupied by the circular cylinder. From reference
B12803 # IMechE 2004 Proc. Instn Mech. Engrs Vol. 218 Part B: J. Engineering Manufacture

Downloaded from pib.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on April 7, 2015


928 C J McPHERSON, G MULLINEUX, C BERRY, B J HICKS AND A J MEDLAND

Fig. 3 Main shoulder parameters

[5], the condition for this is and f ðÞ is the even function given by
8
R3 zvvvv þ Rzvv < 0 ð2Þ >
> c0 þ c 2  2 þ c 3  3
>
>
< þc ðcos   1 þ 1 2 Þ
The collar surface is formed from straight-line genera- f ðÞ ¼
4 2
ð4Þ
tors emanating from the bending curve. If the curve is > þc5 ðsin    þ 1 3 Þ
> for  5 0
>
> 6
smooth, then the generators emanate at smoothly vary- :
f ðÞ for  < 0
ing angles and the surface is continuous. However, if
the curve has a discontinuity at its highest point, then This involves trigonometric terms. Expanding these as
the generators on either side are in different directions power series gives the following polynomial which
and a planar triangle can be inserted into the collar. approximates f ðÞ for small values of :
Suppose that the bending curve zðvÞ in its planar version
is an even function so that zðvÞ ¼ zðvÞ and the surface f ðÞ ¼ c0 þ c2 2 þ c3 3 þ 24
1
c4 4 þ 120
1
c5  5 for  5 0
has symmetry. The appropriate discontinuity [5] is one in
which the third derivative of z is everywhere continuous ð5Þ
except at v ¼ 0.
Four design parameters are considered, as in Fig. 3:
In practice, the difference between the bending curve
itself with and without a continuous third derivative is h :R ¼ ratio of the overall height h of the wrapped
likely to be small and indeed smaller than the errors bending curve to the radius R of the cylindrical
introduced by the manufacturing process. Nevertheless, tube
the discontinuity is important from the geometric point 0 ¼ back angle, i.e. the angle between the normal to
of view of allowing the correct generators to be produced the surface and the normal to the tube at the
with the triangular insert. highest point of the bending curve, which is then
Let  be the angle between the generators at the also the angle between the triangular insert and
highest point of the bending curve and then this is also the vertical generator of the cylinder
the angle at the apex of the inserted triangle. It is given 1 ¼ front angle, i.e. the angle between the normal to
by the following, where it is assumed that the sign the surface and the normal to the tube at the
convention and choice of bending curve is such that lowest point of the bending curve, which is also
zvvv ð0þÞ is negative: the angle between the tangent plane to the surface
and the vertical tangent plane to the tube
2R2 zvvv ð0þÞ  ¼ opening angle, i.e. the angle at the apex of the
tanð12 Þ ¼  ð3Þ
R2 z2vv ð0Þ þ 1 triangular insert
A purely parabolic planar bending curve does not give There are conditions that define or restrict the choices
sufficient freedom to allow the required discontinuity to for the coefficients in equation (4). Firstly, there are the
be achieved. Therefore a modified curve is introduced. end conditions at  ¼ 0 and  ¼ p:
The form given in reference [5] is
h:R ¼ f ð0Þ ¼ c0 ð6Þ
z ¼ R f ðÞ
0 ¼ f ðpÞ
where
¼ c0 þ c2 p2 þ c3 p3 þ c4 ð2 þ 12 p2 Þ þ c5 ðp þ 16 p3 Þ
v
¼ for  p 4  4 p ð7Þ
R
Proc. Instn Mech. Engrs Vol. 218 Part B: J. Engineering Manufacture B12803 # IMechE 2004

Downloaded from pib.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on April 7, 2015


PERFORMANCE ENVELOPE OF FORMING SHOULDERS AND IMPLICATIONS FOR DESIGN AND MANUFACTURE 929

If equation (1) is applied to determine the back angle 0 , be solved to find two values for c4 and hence two values
then the following equation for coefficient c2 results: for c5 . Of these pairs, only one satisfies equation (11) for
 ¼ p. Once the coefficients are found, equations (11) and
c2 ¼  12 tanð12 0 Þ ð8Þ (12) need to be verified over the entire curve.
Clearly, the additional inequality constraints mean
Similarly, equation (3) relates the opening angle  to the
that some sets of design parameters generate coefficients
coefficients:
that are not acceptable. Thus the class of shoulders that
12c3 can be generated is restricted. The purpose of this paper
tanð12 Þ ¼  ð9Þ is to investigate what the limitations are.
4c22 þ 1

Using equation (1) at the lowest point of the bending


4 ENVELOPE OF FORMING SHOULDER
curve relates the coefficients to the front angle 1 :

½ f 00 ðpÞ2 ¼ f½ f 0 ðpÞ2 þ 1g tan2 ð12 1 Þ ð10Þ One of the aspects that emerges from the theoretical
model of the shoulder is the extent to which it relies
Further conditions, all inequalities, are imposed. The very precisely upon the shape of the bending curve.
first pair ensure that the planar bending curve has its For example, inequality (12) involves the fourth deriva-
turning point at  ¼ 0 uppermost and is concave: tive of the curve. Very slight deviations in the curve
due to manufacturing errors or to wear easily cause
f 0 ðÞ < 0, f 00 ðÞ < 0 for 0 4  4 p ð11Þ
this condition to be violated. In practice, the compliance
The third inequality follows from equation (2), which is of the film being drawn over the shoulder is sufficient to
the condition that the collar surface has no singularities: mean that some deviations do not have a significant
effect even if a theoretical constraint does not hold.
f 0000 ðÞ þ f 00 ðÞ < 0 for 0 4  4 p ð12Þ Nonetheless it is important to have some understanding
of what limitations on the choice of the design variables
For the given form of the bending curve, the left-hand are implied by the various conditions that exist, particu-
side of the inequality is found to be linear in  since the larly equations (11) and (12). Some of these limitations
trigonometric terms involved in f ðÞ in equation (4) are investigated in this section.
cancel: Consider first the back angle 0 . This is one of the four
f 0000 ðÞ þ f 00 ðÞ ¼ ð2c2 þ c4 Þ þ ð6c3 þ c5 Þ ð13Þ design parameters. For given values of the other three,
there is only a small interval of allowable values for 0 .
Hence inequality (12) holds throughout the range if and The graphs in Fig. 4 show the lower and upper bounds
only if it holds at both  ¼ 0 and  ¼ p. for 0 in a number of cases. In each case, the front
The strategy for finding a bending curve given the four angle 1 is 108. The four graphs are for different values
design parameters is firstly to use equations (6), (8) and of the opening angle . Each graph gives plots of the
(9) to determine c0 , c2 and c3 . Taken together, equations lower and upper limits of 0 against the height :radius
(7) and (10) lead to a quadratic equation for c4 which can ratio.

Fig. 4 Graphs of upper and lower bounds of back angle 0 against height : radius ratio for front angle 1 at 108
and opening angle  set to 308, 608, 908 and 1208

B12803 # IMechE 2004 Proc. Instn Mech. Engrs Vol. 218 Part B: J. Engineering Manufacture

Downloaded from pib.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on April 7, 2015


930 C J McPHERSON, G MULLINEUX, C BERRY, B J HICKS AND A J MEDLAND

Fig. 5 Graphs of upper and lower bounds of back angle 0 against height : radius ratio for front angle 1 at 308
and opening angle  set to 308, 608, 908 and 1208

The graphs show that there is a lower bound for the curves for the bounds on 0 have moved apart. The
height :radius ratio below which no acceptable values lower bound curves have moved significantly to the
of the back angle exist so that a shoulder is not possible. right and there is a smaller move of the upper bound
Just above this lower bound, the range of choices for 0 is curves. This means that there is now a greater range of
small. It increases as the ratio increases and then choice for the back angle 0 for any given value of the
becomes roughly constant at around 178. height:radius ratio.
Figure 5 shows the same set of graphs but now taking The effects of changes in the bending curve on the
the front angle 1 to be 308. The same form of curves is collar surface are of interest. Initially this is considered
seen and again there is a lower bound on the height:radius ‘in reverse’. Shoulders with front angle 1 at 108 and
ratio. This lower bound has increased slightly from the opening angle  at 908 (a value commonly used in
previous case. The most obvious difference is that the practice) are taken for a number of height :radius

Fig. 6 Sensitivity of the bending curve to changes in the back angle in the case when the front angle 1 is 108
and the opening angle  is 908

Proc. Instn Mech. Engrs Vol. 218 Part B: J. Engineering Manufacture B12803 # IMechE 2004

Downloaded from pib.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on April 7, 2015


PERFORMANCE ENVELOPE OF FORMING SHOULDERS AND IMPLICATIONS FOR DESIGN AND MANUFACTURE 931

ratios. For each, the bending curve is found when are interrelated but do not seem, as yet, to be fully under-
the back angle 0 is in the middle of its allowable stood. Figure 8 shows what is commonly believed to be
range. The value of 0 is disturbed by a small amount their relationship to the height :radius ratio. As the
up to 58 either side of this mid-curve. The plots on the ratio increases, so the stress in the film decreases and
left of Fig. 6 show the resultant bending curves. Indivi- its ability to track becomes poorer. If this is the
dual curves are not distinguishable, but what is seen case, then for good tracking a small ratio is required.
is that, for a given height :radius ratio, the curves agree However, this increases stress and for some materials
at the top and ends, as indeed is ensured by their may result in permanent distortion.
construction. The largest changes are in between these
extremes.
The largest deviation from the mid-curve is identified 5.1 Compactness
and plots of these against the change in the back angle
Usually there is a requirement for the forming system
are shown on the right of Fig. 6. The gradients of these
to occupy as little space as possible. For the tube itself,
curves give a measure of the sensitivity of the bending
this implies that the height :radius ratio should be low,
curve to changes in the back angle. The sensitivity
and this in turn suggests that the opening angle 
increases with increasing height : radius ratio. Arguing
should be reduced (say, less than 908), and the back
conversely, this suggests that small changes in the bend-
and front angles 0 and 1 should be small. However,
ing curve (which leave it smooth) have the greatest effect
for overall compactness, the opening angle  should
upon the surface again when the height : radius ratio is
be large so that the triangular insert in the collar
large.
opens out more quickly, meaning that the reels of film
Consider now the effects of small localized changes to
can be brought close to the shoulder. Overall com-
the bending curve such as might arise from wear or
pactness also suggests that the back angle 0 should be
damage. These may mean that inequality (12) is violated
around 908 so that the reels lie at the same height as
and the film may distort or buckle. Suppose the bending
the shoulder.
curve f ðÞ is changed to f ðÞ þ eðÞ where e is an error
The results in Figs 4 and 5 indicate that the front angle
term (which may be zero for most values of ). Then
1 needs to be chosen to be small, say, around 108, as this
the required inequality becomes
enables smaller height :radius ratios to be used. Increas-
ing the opening angle  has the effect of increasing the
½ f 0000 ðÞ þ f 00 ðÞ þ ½e0000 ðÞ þ e00 ðÞ < 0
smallest allowable value of the ratio, and a value for 
and by equation (13) this is the same as around 908 seems to be a reasonable compromise.
However, a small height : radius ratio means that the
½ð2c2 þ c4 Þ þ ð6c3 þ c5 Þ þ ½e0000 ðÞ þ e00 ðÞ < 0 back angle 0 needs to be small and so the film has to
be brought on to the collar from below. This can be
Figure 7 shows plots of the first term in square brackets, avoided by decreasing the value of  to around 308 and
the unaltered value of f 0000 þ f 00 for  between 08 and 1808 moving the reels away from the shoulder.
in various cases. These are all when the front angle 1 is set
to 108 and the opening angle  is 908. The cases are when
5.2 Sensitivity
the height:radius ratios are 2, 4, 6, 8 and 10, and for each
the front angle is taken at its lower bound, its upper bound At a theoretical level, small changes in the geometry of
and the mean of these. In each case, the value becomes the bending curve can mean large changes in the shoulder
more negative as the ratio increases. This means that, and potential violations of some of the necessary
for higher ratios, small (local) deviations in the bending constraints. Such changes may result in practice from
curve are more likely to be accommodated. errors in the manufacturing process and from wear
during use. The results associated with Fig. 6 suggest
that, if an error occurs over a wide range of the bending
5 DESIGN IMPLICATIONS curve, then a low height : radius ratio is desirable.
Current practice is often to form the collar by smoothly
In this section, some of the design considerations arising wrapping a flat sheet around the bending curve; therefore
from the previous results on the interdependence of the manufacturing errors are likely to be of this form and a
four main design parameters are discussed and summar- low ratio should be considered. It also means that the
ized in Table 1. This is carried out at essentially the level length of the bending curve is reduced and hence so is
of pure geometric considerations. There are other issues the opportunity for error.
that designers need to bear in mind and these include the However, a low ratio also increases material stress and
amount of stress in the material as it is pulled over the so is likely to increase wear. This may result in much
shoulder and the ability of the material to track well more localized damage to the bending curve. In this
and not tend to drift away to one side. These two ideas case the results of Fig. 7 apply and indicate that an
B12803 # IMechE 2004 Proc. Instn Mech. Engrs Vol. 218 Part B: J. Engineering Manufacture

Downloaded from pib.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on April 7, 2015


932 C J McPHERSON, G MULLINEUX, C BERRY, B J HICKS AND A J MEDLAND

Fig. 7 Graphs of f 0000 ðÞ þ f 00 ðÞ against  for shoulders with front angle 1 at 108 and opening angle  at 908
for various height : radius ratios and values of the back angle 0

increased ratio is preferable to ensure that the function of industrial practice is based on only a limited understand-
the shoulder is not impaired. ing of the process and often many forming shoulders
have to be designed and refined before a suitable
shoulder is configured for the particular application.
6 CONCLUSIONS This process can be time consuming and costly.
To address some of these issues, one theoretical model
This paper deals with the complex issues involved in of a forming shoulder has been examined. The model
the design of vertical form, fill and seal machines and ultimately depends upon the definition of the bending
in particular of the forming shoulder itself. Current curve in mathematical terms and leads to a definition
Proc. Instn Mech. Engrs Vol. 218 Part B: J. Engineering Manufacture B12803 # IMechE 2004

Downloaded from pib.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on April 7, 2015


PERFORMANCE ENVELOPE OF FORMING SHOULDERS AND IMPLICATIONS FOR DESIGN AND MANUFACTURE 933

Table 1 Design implications

Area Implication Parameters

Machine Tube Reduce height of tube h : R low


0 low
1 low
 reduced
Horizontal compactness Film feed close to shoulder  high
Minimum h : R increased
Vertical compactness Film feed same height as shoulder 0 around 908
Minimum h : R increased
Sensitivity Manufacture Reduce effects of general errors h : R low
Material Low stress in film h : R high
Operational Reduce effects of localized wear h : R high

Fig. 8 Heuristic relationships of tracking and material stress to the height : radius ratio

of the collar surface including a planar triangular region the compromises required at the design stage for a new
that allows film to be taken from a cylindrical roll. Four shoulder.
main design parameters are identified and constraints
upon the choices of these exist which ensure that the
collar surface is well defined. It is noted that very small ACKNOWLEDGEMENTS
changes in these may corrupt the good surface definition
at a theoretical level at least; in practice the film used may The work reported in this paper arose from a research
deform to alleviate some of the errors but this is consid- project led by Pira International and incorporated into
ered undesirable. the Engineering and Physical Sciences Research Council
The relationships between the design parameters and Innovative Manufacturing Research Centre at the
some of the limitations on their choices have been University of Bath (Grant GR/R67507/01). Additional
examined. This has led to the identification of some of support has been provided by the Process and Packaging
B12803 # IMechE 2004 Proc. Instn Mech. Engrs Vol. 218 Part B: J. Engineering Manufacture

Downloaded from pib.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on April 7, 2015


934 C J McPHERSON, G MULLINEUX, C BERRY, B J HICKS AND A J MEDLAND

Machinery Association and other collaborators. The 4 Hicks, B. J., Medland, A. J. and Mullineux, G. A constraint
authors gratefully express their thanks for the advice based approach to the modelling and analysis of packaging
and support of all concerned. machines. J. Packag. Technol., 2001, 14(5), 209–226.
5 Boersma, J. and Molenaar, J. Geometry of the shoulder
of a packaging machine. SIAM Rev., 1995, 37(3), 406–
REFERENCES 422.
6 Mot, E. The forming shoulder problem on pouch form,
1 PMMI 7th Annual Packaging Machinery Shipments and Out- fill and seal machinery. Verpackungs Rundschau, 1973, 5,
look Study, the U.S. Packaging Machinery Industry Executive 35–39.
Summary, 2001 (Packaging Machinery and Manufacturers 7 Culpin, D. A metal-bending problem. Math. Scientist, 1980,
Institute, Arlington, Virginia). 5, 121–127.
2 Hanlon, J. F., Kelsey, R. J. and Forcinio, H. E. Handbook of 8 Lipschultz, M. Theory and Problems of Differential Geometry,
Package Engineering, 1998 (Technomic Publishing Com- Schaum’s Outline Series, 1969 (McGraw-Hill, New York).
pany, Lancaster, Pennsylvania). 9 Forsyth, A. R. Lectures on the Differential Geometry of
3 Twede, D. and Goddard, R. Packaging Materials, 1998 (Pira Curves and Surfaces, 1912 (Cambridge University Press,
International, Leatherhead, Surrey). Cambridge).

Proc. Instn Mech. Engrs Vol. 218 Part B: J. Engineering Manufacture B12803 # IMechE 2004

Downloaded from pib.sagepub.com at UNIV CALIFORNIA SANTA BARBARA on April 7, 2015

You might also like