Beruflich Dokumente
Kultur Dokumente
net/publication/233116968
CITATIONS READS
15 169
3 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Stephen Thomas Newman on 29 November 2015.
SUMMARYThe foundation for sculptured feature-based shoe last design, using extended form
(EF) feature method, is presented. The E F approach is particularly suited to sculptured
surface products that exhibit parameoic variation throughout a set or family, such as shoe
lasts. Some underlying principles and issues relevent to shoe lasts and common m sculptured
products that can be usefully modelled by the E F feature approach are also discussed. A n initial
feature anatomy for a common shoe style is proposed to support the need for localized shape
control for current and future design requirements. The following conclusions are reached: it is
possible to decompose a shoe last into a feature anatomy for EF feature-based design; there are
issues common to sculptured products, including shoe lasts, that are specifically addressed and
dealt with effectively within the E F method; an EF jeature-based approach shows significant
potential to benefit lasr design eficiency, enhance size grading, and improve manufacturing
processes; more work need to be done to evaluate the eficacy of a system based on the EF
feature methods and the proposed last anatomy.
1. Introduction
transverse 'v' groove and a sprung hinge roughly in the middle, to allow easy removal
from the finished shoe.
During the upper, sole and heel assembly process (known as 'attaching'), the last
supports the insole from inside the shoe, to provide the clamping pressure distribution
necessary for the adhesion of the sole to the insole, and that of the heel to the sole. T o
achieve this, the sole of the last is flatter and more uniform than the human foot, and
also has a sharper profile to distribute the load to the edges of the insole.
T o make a shoe style available to a variety of people, it must be made in a range of
sizes. This requires production sets of last pairs manufactured to form the different shoe
sizes. Initially, the last is designed by a craftsman modelling a wooden last for a single
size, typically to produce a size 4 (women's) or 7 (men's) shoe. The designer seldom
starts from scratch. Usually, helshe will start with at least a pan-machined block that
has a standard heel already copy turned from a previous design, or will start by adapting
a previous last model. The customer's design specification is usually a combination of
drawings, key measurements and often a sample shoe. Sometimes a cast from the inside
of the shoe will form the basis for the last shape. Once the initial model is accepted,
intermediate size variations are produced by 'grading' and subsequent 'coordination'.
The grading process progressively enlarges or reduces the last dimensions, and
commonly involves a combination of copy turning with a magnification factor and
manual adjustment. There are three approaches to grading I:1]:
Arithmetic: the increment for a specific dimension between sizes is specified as a
constant value.
Geometric the increment for a specific dimension between sizes is specified as a
constant percentage of that dimension.
Proponional: the increment for all dimensions between sizes is specified as the same
percentage applied to each dimension.
Proportional grading is little used today, even though it maintains the proportions of
the last, and so its shape and style, through the size range. Figure 1 shows that there
is little practical difference between the last dimensions produced by geometric or
arithmetic grading. Both approaches allow for the length of the last to increase or
decrease proportionately more than the width or girth. This produces a better fit.
T o maintain acceptable comfort levels for all shoe sizes, or to reduce the cost by
sharing 'heel units', for example, it is often necessary manually to alter the different
lasts, so that all sizes share key dimensions ('coordinating'). Typical adjustments
i
involve making transverse cuts and inserting wedges to keep the 'toe spring' and 'heel
pitch' constant through a coordinated set of lasts (Fig.2).
With any approach to grading, if a size 5; women's last is modelled, for example, then
the size 4 and 7 'submodel' lasts will be copy turned from the size 5; and then
coordinated. The size 8 last will then be copied from the coordinated 7. The smaller
coordination errors produced by grading submodels are usually tolerated. All other
intermediate sizes are produced in a similar manner, as necessary. As well as the normal
range of last sizes, there may also be special sizes produced for wide and narrow feet.
Additional grading rules are used to generate lasts for these fits.
Apart from a small number of developments in grading systems [l-31, there have
been few advances in last design methods over recent years. Although computer-aided
design (CAD) methods have been applied to many aspects of shoe design, the last still
tends to be made by traditional methods, with a model-maker developing the design
model by hand. For CAD of the shoe, the last is digitized to produce a CAD model on
which to design the upper. The last designs captured for CAD systems have generally
Downloaded By: [University of Bath Library] At: 11:05 28 November 2007
FIG. 1 . Arithmetic and geometric last grading sample comparison: (a) length dimension
comparison; (b) width dimension comparison; (c) girth dimension comparison.
Downloaded By: [University of Bath Library] At: 11:05 28 November 2007
been in the form of individual single-surface models. These models are difficult to
manipulate if changes to the last are required. Shoe design tends to be a process of
product variation rather than design from scratch, so the opportunity exists to use
standard lasts and vary only those features that require modification. Typically, the heel
section will have a standard shape, whereas the toe will be varied much more, subject
to the whims of fashion.
common product-specific shape elements and behaviours. This provides the foundation
for a generic approach to a product family's shape definition and control.
A specific product's anatomy is established by identifying three classes of sculptured
feature:
0 primary EF features;
0 secondary blend features;
tertiary ornamental marking features.
A sculptured product's shape is normally governed by a number of dominant elements,
and these can be extrapolated as the basis for primary E F features. An E F feature's
shape contribution within a product is limited to the boundary defined by its relation-
ships with neighbouring features. Typically, the boundary will comprise intersections
with other features andlor the tangency curves defined by secondary blend features to
other parts of the product. The secondary blend features generally do more to control
the aesthetic quality of the product, whereas the primary E F features define its
underlying shape.
The EF feature method is particularly suited to sculptured surface products that
exhibit parametric variation throughout a set or family. The following sections describe
in more detail its application to shoe last design, and identify some underlying
principles and issues common to sculptured products that can be usefully dealt with by
the E F feature approach.
the EFs and aesthetic blends indicated by the variation of highlights on the last
surface;
the parametric control complexity for individual features;
the total number of features;
the potential for shoe industry specification parameters to be used for feature
shape control;
the suggestion that a feature or group exists from industry terminology, such as
'heel', 'sole' or 'toe';
the potential requirement for localized conrrol of the last design.
This requires experience, consultation with existing and prospective designers, and
several iterations.
Where possible, the features are given names that relate to industry terminology.
Where suitable terms are unavailable or inappropriate, names have been formulated to
make as much sense as possible. For example, we have the following.
'Out-step' is used to describe the corresponding feature on the opposite side to the
'in-step'.
0 The 'tongue' feature might have been termed the 'front cone' by the shoe industry
but this would establish shape preconceptions; because the feature is seldom truly
conical.
The 'shank' region (roughly corresponding to the sloping region of the sole
between the ball of the foot and heel) has not been identified as a separate feature.
At present, this is considered to be an unnecessary complication of the sole group,
although it could be included if necessary.
T o establish feature bounding relationships, it is not necessary to have a 'vamp'
feature or group for this particular last type. However, a suitable group could be
included if this aids last manipulation, such as when interchanging whole vamps
between lasts. This is also true for heel, toe and waist regions. The result would
be an alternative view of feature groupings for design manipulation, as shown in
Fig. 6. It is useful to note that there is some overlap between the additional groups.
Obviously, there is the potential for feature synonyms; for example, the sole-heel and
sole-toe could be called the sole-forepart and -backpart respectively. There is no reason
why particular users cannot adapt or implement particular terms to suit their own
preferences.
Some features have been omitted from the initial anatomy. In particular, the 'v' cut
and hinge details have been omitted at this stage. This is because they are usually
produced by a dedicated machining process with little variation. They are generally
purely functional, so contribute little to the style of the last or of the shoe. However, it
is a simple matter to add a suitable feature to the last upper anatomy, should this be
needed.
Shoe lasts are mostly free from tertiary ornamental marking features, except for the
raised 'pips' used to indicate the position for key dimensions, such as the last girth
Downloaded By: [University of Bath Library] At: 11:05 28 November 2007
158
S. R. Mitchell et al.
Downloaded By: [University of Bath Library] At: 11:05 28 November 2007
-b slick Length I
(Fig. 7). Because these markings are both simple in shape and easily added to the
model, they have also been omitted, to concentrate initially on successful general shape
design.
Current last designers may consider the proposed anatomy overly complex and might
argue that blended heel, waist and toe features would be sufficient. Certainly, these
groups can be treated as single entities for the purpose of combining whole sections
from different last designs using the E F method. However, experience suggests that
shape control for the toe, for example, is best achieved by further subdividing the
region. This gives localized shape control rather than an excessively complex set of
shape control parameters or routines. Such complexity generally makes it almost
impossible to eliminate or control the side effects of making design changes to one
portion of a complex feature. For example, producing a square toe and smoothly
varying the severity of the internal blend regions in three dimensions using a single
feature would be parricularly difficult. Considering the more extreme case, these are
precisely the reasons why a decomposed last feature anatomy approach is recom-
mended instead of single-surface manipulation.
It could be further argued that, because the heel design changes little between last
types, it is unnecessary to subdivide it. Certainly, a standard heel group can be
introduced as one entity using the proposed features and, in most instances, this will
require no Further change. However, it is still possible that there will be a requirement
for change in the future, possibly to suit a new heel style or a new customer or
population group. In this case, the arguments for subdividing the toe will apply. This
highlights the need to determine carefully a product's features, to allow for both future
and current design requirement or activity levels.
fi n Ternaly Blend
Figures 8 and 9 indicate the relationships necessary between the identified features to
establish the bounded geometry. E F features are shown by a single-border 'cloud' and
feature groups are shown with a double-border cloud. Binary and ternary blends are
also suitably represented. It is interesting to note that the anatomy for the last upper is
symmetrical, as might be expected.
Downloaded By: [University of Bath Library] At: 11:05 28 November 2007
0 anatomy specification and manipulation elements, to cater for more than one last
style;
0 feature, anatomy and actual design library access;
0 suitable parameter specification means, and sensible limits and increments;
0 geometry engine initiation and three-dimensional model viewing;
0 set parameter variation specification;
0 derived characteristic measurement and shape optimization functions for girth,
length, etc.
This final point is particularly important, because the shoe industry currently uses
several measurements to quantlfy last specifications that can only be derived from the
E F design model (Fig. 7 shows some important examples). It is not practical, in terms
of computing time and complexity, to attempt to control the last's basic shape directly
from these dimensions. However, it is possible to automate optimization for selected
feature shape and position parameters, to achieve the desired result.
These optimization facilities will play an important role in achieving automatic set
grading based on the 'set parameter variation specification'. Where possible, set grading
will be achieved by direct control over feature (such as heel elevation or
pitch, toe spring, stick length or bottom width). For the derived characteristics (such as
joint, waist, instep and long heel girths), an optimization goal will be specified through
the set, together with feature parameters that can be refined to achieve the desired
dimension.
A customized version of the interface previously used for golf club design [6] (based
on Delcam .International's D U C T surface modelling software), that supports the above
facilities, is suitable. Further details for a shoe-last-specific system implementation will
be the subject of a future publication.
3. Discussion
3.1 General
Determining the optimum feature anatomy for a given product is fundamental to
developing a successful EF feature-based CAD system. The anatomy proposed in
Section 2 is the first attempt to achieve the subjective criteria balance described in
Section 2.2 and, as such, may need future modification in response to implementation
trial results, or even to suit a particular organization. Obviously, variations of this
anatomy will be required for markedly different last styles and a variety of feature
shapes will need to be available for each feature type in the anatomy. However, the
proposed anatomy is considered to be a good basis for further research.
the method shows some promise as a generic approach to sculptured product feature-
based design, as well as significant potential for application to shoe last design.
In particular, shoe lasts share the following characteristics with other sculptured
products (such as golf club heads, consumer electronics casings, ceramic tableware or
sanitary ware) considered suitable for E F feature-based design.
0 The shoe last is a fully sculptured product. The design prototype is usually
produced manually by craftsmen and the designs exhibit virtually no prismatic
engineering features.
Different lasts have similar anatomies and terminology. However, a typical last's
sole or heel feature is not equivalent in shape to those of a golf club or steam iron,
for example. This confirms the need for product-specific anatomies and features
to support efficient sculptured product CAD.
0 There are accepted parameters, properties and notional data used to specify the
last design characteristics. These are common to most manufacturers in the shoe
industry and show some promise as control parameters for individual last features
and size variation.
0 The notional data are often vague and open to interpretation. (For example, the
girth is measured in three positions. These locations are determined by each
craftsmen and are difficult to reproduce independently.) This causes some cultural
and implementation conflicts when applying CAD techniques, because a more
rigorous design specification is necessary.
0 Shoe lasts require typical design processes (such as hybrid design based on
previous lasts that require localized feature editing and swapping; automatic
anisotropic set generation; and optimization to achieve derived properties such as
the girth). This confirms the need for a generic sculptured feature architecture to
manipulate the product-specific anatomies and features.
0 Design characteristic optimization and set grading are partly based on properties
that are more easily derived or measured from the design rather than used to
manipulate directly the shape. (For example, the girth property is more easily
treated as a measured characteristic than as a feature control parameter, as are the
mass of a golf club head or a teacup's capacity.)
0 The main design goals are difficult to quantify (i.e. the last's effect on 'shoe fit' and
the golf club 'feel' are similarly enigmatic), so physical prototypes are required for
performance assessment (a last needs leather stretching trials and a golf club needs
play testing).
0 There is an equivalent abundance of previous last designs, all cross-referenced by
the design guru, and used as the basis for current designs. This creates a common
design capture, storage, classification and retrieval problem. Current research
indicates that approximating coordinate measuring machine (CMM) digitized
data with suitable E F fearures and blends produces acceptable CAD model
replicas for existing designs.
0 Design activity levels are high to keep pace with fashion changes. Thus, the
investment required to develop anatomies and features is warranted. The need for
economic limited-volume last customization for orthopaedic purposes also sup-
ports the need for feature-based CAD.
0 The current designers and craftsmen exhibit low computer literacy levels.
Shoe lasts also exhibit some idiosyncrasies that raise uncommon issues.
(1) The shoe last is a manufacturing process form tool. It is not the final commercial
Downloaded By: [University of Bath Library] At: 11:05 28 November 2007
Finally, using CAD tools for last designs has some implications for last manufacture.
Once the base model has been graded by the CAD system, it is relatively easy to
produce computer numerical control (CNC) machine code to manufacture a pair of
lasts for each size in the set. Production last sets for each size can then be copy turned
from these masters.
Alternatively, automatic C N C code generation may make it cost effective to pro-
duce low production volume injection-mould tooling for each last size. This approach
would remove much of the manual work in finishing the last heel and toe profiles.
Currently, these are used as holding points and the final shape is produced by manual
grinding and shaping the residual lugs. The 'v' groove and hinge recess could also be
cast at the same time, for further reduction of manual operations. Furthermore, it may
be possible for shoes to share the same heel moulds, where they use a standard heel
shape, and even for shoe manufacturers to use one set of heels (foreparts) with several
alternative sets of toes (backparts).
0 Training times: Existing designers require as little as 1 day of training for them to
use the system to develop valid designs. Normally, it would take a training course
lasting 2 weeks and a month's subsequent use to gain enough experience to
produce useful designs with normal surface modelling software.
0 Base model design: It is possible to produce the base design for a new set in a matter
of minutes, depending on the level of variation from an existing design. Usually,
this would take the best part of a day to do manually using normal surface
modelling software, and at least a day to craft by hand.
0 Base model refinement: A design can be modified in a matter of minutes, depending
on the degree of change. For a major alteration, it would normally be necessaty to
rebuild completely a regular CAD model, so it may take up to another day to do
so without feature-based tools.
0 Set generation: Because developing individual designs is faster, producing the
designs in a set is proportionally faster-except that the designer can do something
else while the 'donkey work' is done for him. For example, a full set of 11 golf
irons can be developed automatically within 1 h using E F feature methods,
whereas it takes 2 weeks using standard CAD software.
Design quality: Because the design process is faster, more effort can be spent
refining the shape. In addition, the inherent design modelling consistency results
in fewer problems for automatic C N C code generation.
Because they share the same underlying sculptured product characteristics as golf club
heads, it is anticipated that similar results will be experienced in using E F feature
methods for shoe last design.
4. Conclusions
An initial feature anatomy for a common shoe style is presented, to support the need
for localized shape control for current and future design requirements. T h e following
conclusions can be made based on this work.
(1) It is possible to decompose a shoe last into feature anatomy for E F feature-based
design.
(2) There are issues common to sculptured products, including shoe lasts, that are
specifically addressed and dealt with effectively within the E F method, such as
the following:
0 CAD tool implementation based on product-specific terminology;
0 localized and simplified shape control for sculptured product design;
0 hybrid and variational product design;
0 anisotropic grading of product families.
(3) An E F feature-based approach shows significant potential to benefit last design
efficiency, enhance size grading and improve manufacturing processes.
(4) More work needs to be done to evaluate the efficacy of a system based on the E F
feature methods and the proposed last anatomy.
Downloaded By: [University of Bath Library] At: 11:05 28 November 2007
Acknowledgements
The authors would like to thank The EPSRC, Delcam Int. Ltd, Dunlop Slazenger Int.
and the Department of Manufacturing Engineering at Loughborough University for
their support. They would also like to thank BUSM and Clarks Shoes for their help in
understanding the shoe and shoe last industries, as well as Sean Parsons, Jennifer
Rankin and David Venables for their valuable contributions.
REFERENCES
[I] MILLER,R.G. (Ed.) (1989) Manual of Shoemaking, 6th edn (Street, UK, C. &
J. Clark).
[2] THORNTON, J.H. (1970) Textbook on Footwear Manufacture, 3rd edn (London,
Butterworth, Heywood Books).
[3] HEATH,A. (1967) Some basic considerations in shoe last sizing systems, Leather
and Shoe, August.
[4] SALOMONS, O.W.,VAN HOUTEN,F.J.A.M. & KALS, H.J.J. (1993) Review of
research in feature based design, Journal of Manufacturing Systems, 12, pp. 113-32.
[5] JONES,R., MITCHELL,S. & NEWMAN,S.T. (1993) Feature based systems for the
design and manufacture of sculptured products, International Journal of Production
Research, 31, pp. 1441-1452.
[6] MITCHELL,S., JONES,R., NEWMAN, S.T. & HINDE,C.J. (1994) A design system
for iron golf clubs. In: A.J. COCHRAN & M.R. FA- (Eds), Science and GolfII,
Proceedings of 1994 World Scientific Congress of Golf, pp. 390-395.
[ 7 ] Rosst, W.A. (1988) The futile search for the perfect shoe fit, Journal of Testing and
Evaluation of ASTM, 16, pp. 393403.