International Journal of Clothing Science and Technology

Improved Methods for Colour Inventory Management in the Apparel Industry

Dan R. Teel J. Richard Aspland James P. Jarvis Karen L. Dunlap
Article information:
To cite this document:
Dan R. Teel J. Richard Aspland James P. Jarvis Karen L. Dunlap, (1992),"Improved Methods for Colour Inventory
Management in the Apparel Industry", International Journal of Clothing Science and Technology, Vol. 4 Iss 2/3 pp. 66 - 70
Permanent link to this document:
http://dx.doi.org/10.1108/eb002995
Downloaded on: 12 April 2016, At: 00:26 (PT)
References: this document contains references to 0 other documents.
To copy this document: permissions@emeraldinsight.com
The fulltext of this document has been downloaded 124 times since 2006*
Users who downloaded this article also downloaded:
(1992),"Detection and Removal of Fabric Defects in Apparel Production", International Journal of Clothing Science and
Technology, Vol. 4 Iss 2/3 pp. 54-65 http://dx.doi.org/10.1108/eb002994
Downloaded by University of Wollongong At 00:26 12 April 2016 (PT)

(1978),"Inventory Management Techniques", International Journal of Physical Distribution & Materials Management, Vol.
8 Iss 8 pp. 385-428 http://dx.doi.org/10.1108/eb014432
(2001),"Inventory management in the women’s retail clothing industry", Management Research News, Vol. 24 Iss 3/4 pp.
40-44 http://dx.doi.org/10.1108/01409170110782595

Access to this document was granted through an Emerald subscription provided by emerald-srm:235887 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service
information about how to choose which publication to write for and submission guidelines are available for all. Please visit
www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company manages a portfolio of
more than 290 journals and over 2,350 books and book series volumes, as well as providing an extensive range of online
products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee on Publication
Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive preservation.

*Related content and download information correct at time of download.

 INTERNATIONAL JOURNAL
OF CLOTHING SCIENCE
AND TECHNOLOGY

Improved Methods for Colour

Inventory Management in the
Apparel Industry
Dan R. Teel
Tanner Companies, Rutherfordton, North Carolina, USA, and
Downloaded by University of Wollongong At 00:26 12 April 2016 (PT)

J. Richard Aspland, James P. Jarvis and Karen L Dunlap

Clemson University, Clemson, South Carolina, USA

INTRODUCTION between samples can be computed as a

The major objectives for a colour inventory
management system in an apparel-
manufacturing plant include better colour
consistency within shade groups, large and
fewer cutting lays, and efficient inventory
control. The Clemson Colour Clustering
(CCC) system has been shown to be an
effective method for numerical shade sorting.
With the co-operation of Tanner Companies,
Inc., the developers of CCC have
implemented a system for colour inventory
management which is based on CCC shade
sorting. This article includes a brief summary
of numerical shade sorting and the CCC
system, a description of the enhancements
made to the CCC technology to make it a
more effective tool for shade sorting, and
examples of the increased efficiencies in
inventory utilization which are possible
through the use of the CCC system.
Colour Measurement
Colour can be quantified using three
numbers in a variety of colour measurement
systems. Two widely used systems are L*a*b*
(light/dark, red/green, and yellow/blue
respectively) and L*C*h° (light/dark, chroma
or saturation, and hue). Colour differences
International Journal of Clothing Science and Technology, Vol. 4 No. 2/3.
1992, pp. 66-70, © MCB University Press, 0955-6222
66
weighted difference of the colour co­
ordinates in either of these colour systems.
Appropriate weights can be determined from
historical data or mathematically from
methods such as the CMC (2:1) colour
difference formula[l]. The latter has proved
particularly effective for measuring colour
differences in L*C*h° colour space. Colour
differences, referred to as ΔE, are scaled
depending on the standard, so that a value
of 1 would generally be judged perceptibly
different in colour[3]. The CMC (2:1)
formula is rapidly becoming an industry
standard for measuring colour difference.
Numerical Shade Sorting
Simply stated, numerical shade sorting is a
procedure which uses objective standards and
values obtained via colour measurement
equipment to group a population of similarly
coloured goods into a minimum number of
groups, none of which exceeds a specified
colour tolerance. A common method for
numerical shade sorting has been the 555
system [3].
The 555 method groups similarly coloured
samples by dividing three-dimensional colour
space into an array of cubes or boxes centred
on the standard of the population and
aligned with the co-ordinate axes. Shade
groups lie within a single "555 box". Edge
 VOLUME 4 NUMBER 2/3
1992

555. CCC typically produces two to three

times fewer groups at a given colour
tolerance than 555[6]. In addition, groups
produced by CCC tend to be more compact
(in colour) and more uniform (in size) than
those produced by 555 (or similar methods).
These observations are based on extensive
comparisons of CCC and 555 on real data
provided by a variety of textile and apparel
manufacturers (e.g.[7]). A representative
comparison is shown in Figure 2.
While CCC is demonstrably superior to
555 for numerical shade sorting, it has seen
little practical use in the apparel or textile
industry. The remaining portion of this
article describes the implementation of the
Downloaded by University of Wollongong At 00:26 12 April 2016 (PT)

CCC system at Tanner Companies, Inc.

sizes of the boxes are determined by (Rutherfordton, NC) and includes an
acceptable colour variation in each dimension assessment of the practical advantages of the
for each standard. The 555 method and CCC colour inventory management system
similar "fixed grid" schemes suffer from two for the apparel industry.
major problems: non-uniformity of the
shade-sorting boxes (opposite corners being
more widely separated than opposite faces), IMPLEMENTING CLEMSON COLOUR
and the division of naturally occurring CLUSTERING
groups of samples which happen to be The objectives of the joint project with
intersected by shade box boundaries[4]. These Tanner were threefold: to obtain practical
difficulties are illustrated in Figure 1. experience in using CCC in a production
environment; to integrate the shade-sorting
capabilities of CCC into an inventory
Clemson Colour Clustering management system; and to provide a useful
In contrast with the 555 method, Clemson methodology to Tanner and hence other
Colour Clustering (CCC) uses a manufacturers in the apparel industry. Tanner
computational technique, called hierarchical uses HunterLab equipment for colour
or agglomerative clustering, to find naturally measurement to assist in shade sorting and
occurring groups in the sample data[5]. utilizes a number of fabrics which are
Starting with each individual sample as a suitable for numerical shade sorting. Based
separate cluster, clusters are successively on initial conversations, the implementation
merged. The two clusters to be merged at of CCC shade sorting appeared to be
each step are chosen to minimize the straightforward.
diameter of the resulting cluster. The merge
step is repeated until the size of the next
merged cluster would exceed a specified
tolerance. This procedure guarantees that, on
termination, no cluster exceeds the tolerance
and no additional clusters can be combined
without exceeding the tolerance.
The groups or colour clusters produced by
CCC depend on the actual colour data being
analysed — natural groups of samples form
the colour clusters. The CCC method is fluid
in the sense that a change in the underlying
colour population will usually result in a
shift of the colour clusters. This apparent
draw-back is overshadowed by the increased
efficiency of the CCC method compared with

67
 INTERNATIONAL JOURNAL
OF CLOTHING SCIENCE
AND TECHNOLOGY

Multiple Measurements per Roll

To obtain better colour consistency in its
products, Tanner measures the colour of
incoming rolls of fabric three times — at the
right, centre, and left of each roll. When
CCC was applied to such measurements, a
serious problem became evident. Whether
CCC, 555, or any other method for
numerical shade sorting was used, there was
no way to assure that all measurements taken
from a single roll would be contained in the
same shade group. Even for shipments with
relatively little colour variation among or
between rolls, measurements from a single
roll were often associated with different
shade groups.
Downloaded by University of Wollongong At 00:26 12 April 2016 (PT)

This problem is illustrated by Figure 3. A

population of 80 rolls of a turquoise fabric
was measured and shade-sorted using CCC
and 555. At ΔE=0.8, CCC produced five
colour groups and 555 produced 18 groups.
As noted above, this was typical for CCC
versus 555. However, examination of the data
revealed that, of the 80 rolls of fabric, only
27 had all three measurements in single CCC
shade groups (similarly for 555). Clearly,
both CCC and 555 were of little use for
shade-sorting the 80 rolls. The procedure
actually used by Tanner for shade sorting
was to convert the measured L*a*b* values
into 555 values, which were used only as a
guide to assist in visual sorting.
□ applied to the 80 rolls of turquoise fabric
TANNER PERSONNEL USED THE
CCC SYSTEM TO SHADE-SORT A
VARIETY OF FABRICS
□ excessive side-centre-side colour variation.
Although the use of multiple measurements
per roll resulted in better colour consistency,
it initially appeared to present an
unsurmountable problem for numerical shade
sorting. However, it really represented the
opportunity for a major advancement in
shade sorting. Close examination of the CCC
method revealed that it could be modified to
incorporate directly multiple measurements
per roll. Instead of forming the initial
clusters from a single measurement, they
could be formed from those measurements
common to a roll. From that point, the
merging of clusters could proceed as in
ordinary agglomerative clustering. The
resulting clusters have the property that their
colour variation will not exceed the tolerance
unless that cluster consists of a single roll,
the internal colour difference of which was
initially greater than the tolerance. In other
words, in the modifed CCC procedure,
neither the intra- nor the inter-roll colour
difference for a cluster of more than one roll
will exceed the specified colour tolerance. In
addition, a roll with excessive internal colour
differences will never be merged into a cluster
with other rolls. To the authors' knowledge,
no other procedure for numerical shade
sorting can incorporate multiple
measurements per roll.
When the modified CCC procedure (which
will now be referred to simply as CCC) was
described above, seven groups were produced
at the same colour tolerance, which resulted
in 18 555 groups and five groups from the
original CCC method. One of those seven
groups consisted of a single roll with
Hence no great penalty was associated with
using the modified CCC procedure for numerical
shade sorting (irrespective of the fact that it
was actually the only procedure available).
Setting Colour Tolerances
Having established CCC as a viable
procedure for numerical shade sorting at
Tanner, the next stop was to establish
numerical colour tolerances which
corresponded well with the visual tolerances
which were previously in use. With
measurements taken with the HunterLab
equipment, Tanner personnel used the CCC
system to shade-sort a variety of fabrics and
colours over a range of colour tolerances.

68
 VOLUME 4 NUMBER 2/3
1992
These were visually inspected for colour
uniformity. It was determined that ΔE values
of 0.70 for single garment construction and
0.85 for co-ordinates corresponded closely to
the established visual tolerances. Thereafter
the CCC method was used exclusively for
instrumental shade sorting with decreasingly
frequent visual inspections as confidence in
CCC grew. The system has been in use for
several months with no complaints about
colour consistency.
It should be noted that the success in
setting a single colour tolerance applicable to
a wide range of materials is more a
recommendation for the CMC (2:1) colour
difference formula than CCC itself. The
CMC (2:1) formula produces a consistent
Downloaded by University of Wollongong At 00:26 12 April 2016 (PT)
numerical colour difference which
corresponds closely to visual perception. The
CCC method simply guarantees that no
cluster exceeds the colour tolerance and
groups a population into a relatively small
number of clusters. Any method for
numerically measuring colour differences can
be employed in the CCC method.
Choosing Rolls for Cutting
The goal of shade sorting in apparel
manufacturing is the grouping of coloured
inventory for subsequent cutting and garment
construction. A criticism of CCC which has
been made is that the colour clusters move
around as inventory is added (new shipments)
or removed (for cutting). The shade groups
produced by 555 remain fixed. As discussed
below, the dynamic nature of CCC is
actually an advantage. As inventory is added
or removed, the shade groups produced by
CCC change to reflect the current inventory.
A procedure was devised whereby the CCC
system could be used to determine rolls for
cutting to meet a specified yardage. The
current inventory is shade-sorted according to
the appropriate colour tolerance. Given a cut
yardage requirement, the smallest yardage
CCC cluster which contains sufficient
yardage to meet that requirement is chosen.
The rolls within that cluster are sorted in
order of decreasing distance from the
standard. Rolls with the greatest colour
difference compared with the standard are
successively chosen until sufficient yardage to
meet the cut requirement is achieved. These
rolls are removed from the inventory and the
remaining inventory is reclustered. (The
inventory is also reclustered after any new
shipment.)
69
Of course, if the entire cluster is required
for the specified yardage, then all the rolls in
the cluster are selected. However, if some
rolls remain, these tend to be close to the
standard and usually can be combined with
other inventory without increasing the
number of sort groups. The same procedure
could be used with 555 (if 555 could handle
multiple measurements per roll) but any
remaining rolls would remain in their
previous shade group. Over time, this results
in a large number of shade groups with
relatively few rolls per group. This
phenomenon is best illustrated by example.
Tanner received a shipment of 87 rolls of a
turquoise knit fabric. Each roll was
approximately 50 yards in length. Production
required that three cuts of 581, 484 and 676
yards (about 35 rolls) be made from this
inventory. Using 555-assisted visual sorting,
the 87 rolls were divided into eight shade
groups totalling 77 rolls. One roll was found
to have excessive side-to-side colour variation
(hence excluded) and the remaining nine rolls
could not be sorted with any other rolls and
were treated on a "piece-to-piece" basis
(requiring ply markers on cutting). After
making the three cuts, 50 rolls remained in
inventory. Of these, 36 rolls fell into three of
the original shade groups. The piece-to-piece
category now consisted of 14 rolls composed
of eight rolls initially in that category plus
six additional rolls which were remnants of
the original shade groups. These 14 rolls
represented approximately 30 per cent of the
remaining inventory.
The same scenario was analysed using the
CCC system. From the original 87 rolls,
seven rolls were excluded because L*a*b*
values were not available. The remaining 80
rolls were shade-sorted into six groups plus
one roll with excessive side-to-side variation
(excluded). For each cut in turn, the smallest
shade group with sufficient yardage was
chosen, rolls furthest from the standard were
selected from that group, and the remaining
rolls reclustered using CCC. This was
repeated for each of the three cuts. At that
time, 45 rolls remained in inventory and were
shade-sorted into four groups (the smallest of
which contained five rolls). Significantly, no
rolls fell into the piece-to-piece category.
When additional cuts were simulated from
the same inventory, the CCC system
consistently utilized the entire inventory. The
yardage for every cut was always chosen from
a single shade group and no rolls were ever
placed in the piece-to-piece category.
 INTERNATIONAL JOURNAL
OF CLOTHING SCIENCE
AND TECHNOLOGY
The same comparison was repeated with a
population of 24 rolls (approximately 100
yards each) of a rose fabric. Six cuts (about
12 rolls) were made. Using Tanner's visual
sorting, the inventory after the cuts consisted
of three shade groups totalling nine rolls and
three rolls in piece-to-piece. Using CCC, after
six cuts all the remaining 12 rolls of
inventory were included in a single shade
group.
These examples are repesentative of the
experience of Tanner in using the CCC
system. The combination of a numerical
shade-sorting system which incorporates
multiple measurements per roll and an
associated procedure for choosing rolls for
Downloaded by University of Wollongong At 00:26 12 April 2016 (PT)
cutting, which does not fragment the
remaining inventory, clearly holds promise
for both improved colour consistency and
efficient use of inventory.
CONCLUSIONS
The CCC method for numerical shade
sorting, in conjunction with the CMC (2:1)
colour difference formula, had been shown
previously to be considerably more effective
for numerical shade sorting than existing
procedures such as 555. A major
enhancement to the CCC method is the
capability of incorporating multiple
measurements per roll directly into the shade-
sorting process. This is expected to be useful
in detecting side-to-side colour variation as
well as for on-line colour measurement
systems which make colour measurements
end-to-end.
The shade-sorting capabilities of CCC have
been incorporated into a procedure for
choosing rolls for cutting. A single shade
group having sufficient yardage to meet the
cut requirement is chosen and rolls selected
from that group in order of their (decreasing)
distance from the standard. These rolls are
removed for cutting and all remaining rolls
shade-sorted again. Experience has shown
that the colour inventory is actually
consolidated with this procedure. Few, if any,
rolls must be treated on a piece-to-piece
basis. These developments in the CCC system
70
are being used to save money and improve
colour consistency at Tanner Companies.
Future effects in this project will include
the development of the CCC system into a
software package for general use in the
apparel industry. Capabilities for restricting
roll selection by width or other criteria will
be added. A particularly interesting
possibility is that of choosing rolls from
inventory to match the colour of previous
cuts. Also being considered is shade sorting
for both co-ordinate and single garment
construction.
References
1. Clark, F.J.J., McDonald, R. and Rigg, B.,
"Modification to the JPC79 Colour-difference
Formula", Journal of the Society of Dyers and
Colourists, Vol. 100 No. 4, 1984, pp. 128-32.
2. Aspland, J.R. and Jarvis, J.P., "Colour
Tolerances, Specifications, and Shade Sorting",
American Association of Textile Chemists and
Colorists Magazine, Vol. 18 No. 6, 1986, pp.
27-9.
3. Simon, F.T., "Shade Sorting by the 555
System", American Dyestuff Reporter, Vol. 73
No. 3, 1984, pp. 17-26.
4. Aspland, J.R., Jarvis, C.W. and Jarvis, J.P., "An
Improved Method for Numerical Shade
Sorting", American Association of Textile
Chemists and Colorists Magazine, Vol. 19 No.
5, 1987, pp. 21-5.
5. Kaufmann, L. and Rousseeuw, P.J., Finding
Groups in Data: An Introduction to Cluster
Analysis, John Wiley & Sons Inc., New York,
1990.
6. Aspland, J.R., Jarvis, C.W. and Jarvis, J.P., "A
Comparison of Four Approaches to Shade
Sorting", American Association of Textile
Chemists and Colorists Magazine, Vol. 19 No.
9, 1987, pp. 67-8.
7. Aspland, J.R. and Jarvis, J.P., "Color
Consistency and Shade Sorting", American
Association of Textile Chemists and Colorists
Magazine, Vol. 22 No. 10, 1990, pp. 17-18.
 This article has been cited by:

1. J. R. Aspland, K. D. Balasaygun, J. P. Jarvis, T. H. Whitaker. 2000. Alternative mathematical approaches to shade sorting.
Color Research & Application 25:10.1002/1520-6378(200010)25:5<>1.0.CO;2-G, 368-375. [CrossRef]
2. Y S W Li, C W M Yuen, K W Yeung, K M Sin. 1998. Instrumental shade sorting in the past three decades. Journal of the
Society of Dyers and Colourists 114, 203-209. [CrossRef]
Downloaded by University of Wollongong At 00:26 12 April 2016 (PT)