Sridhar Kota A Metric for Evaluating Design

Associate Professor,
Mechanical Engineering and Applied Mechanics
The University of Michigan,
Commonality in Product Families
Ann Arbor, MI 48109
Many companies develop a market strategy built around a family of products. These
e-mail: kota@engin.umich.edu
companies regularly add new product variations to the family in order to meet changing
market needs or to attract a broader customer base. Although the core functionality
Kannan Sethuraman remains essentially unchanged across the products within a family, new functions, feature
Assistant Professor,
combinations and technologies are incorporated into each new product. If allowed to
School of Business Administration
grow unchecked, these component variations, commonly referred to as ‘‘complexity’’, can
The University of Michigan,
result in a loss of productivity or quality. The challenge lies in an effective management
Ann Arbor, MI 48109
of product variations in the design studio and on the manufacturing floor. The key is to
minimize non-value added variations across models within a product family without lim-
Raymond Miller iting customer choices. In this paper we discuss the factors that contribute to product
Research Assistant,
complexity in general, and present an objective measure, called the Product Line Com-
Mechanical Engineering and Applied Mechanics
monality Index, to capture the level of component commonality in a product family.
The University of Michigan,
Through our Walkman case study, we present a simple yet powerful method of bench-
Ann Arbor, MI 48109
marking product families1. This method gauges the family’s ability to share parts effec-
e-mail: ramill@engin.umich.edu
tively (modularity) and to reduce the total number of parts (multi-functionality).
关S1050-0472共00兲02704-5兴

Introduction and Motivation Related Research

In today’s competitive market environment, companies can no
longer achieve success through the mass production of a limited
number of products. Companies competing in every segment of
consumer goods and services are pursuing aggressive product ex-
pansion strategies—in particular, line extensions 关1兴. The U.S.
automobile industry nicely captures this phenomenon of ever in-
creasing product variety 关2,3兴. Carmakers using a platform ap-
proach design the essential features of a family of cars at the same
time, instead of one model at a time. Our experience suggests that
in many instances, although the system functionality remains es-
sentially the same, the individual components share minimal de-
sign and assembly features. The benefits of standardization are
widely known and most companies do standardize stock compo-
nents such as fasteners. However, many companies are not yet
standardizing their product-specific core components and thus fail
to reap significant benefits in quality and cost. We believe that by
standardizing the functional features, regardless of size variations,
and by developing a concurrent reconfigurable tooling design,
much of the non-value added complexity could be minimized. The
first step is to identify a set of functional building blocks across
the family of products in a manner that clearly identifies a core
design. The strategy for compiling a core set of design building
blocks is to analyze the existing designs, understanding the pack-
aging constraints, material and manufacturing constraints and er-
gonomic constraints. Then seek common functional design fea-
tures based on configuration similarities 共geometric shapes and
feature layout兲, kinematic similarities 共joint types and motion兲,
actuation and sensing similarities, etc.
In this paper we analyze the factors that contribute to product
complexity in general and suggest robust design/manufacturing
strategies that can help minimize non value-added variation across
models within a product family without limiting customer
choices. We then propose an objective measure to capture the
level of part commonality in a product family.
1
Many researchers in the past, including Sanderson and Uzumeri 关4兴, have used
the term ‘‘product family’’ to refer to a set of related products that are characterized
by a high level of similarity in their functionality and in the potential to share
components, subassemblies, and production processes.
Contributed by the Design Theory and methodology Committee for publication in
the Journal of Mechanical Design. Manuscript received Jun. 1998. Associate Tech-
nical Editor: J. Cagan.
The need for product variety in the market place has increased
substantially over the past decade. As such, considerable research
attention has been paid to the costs and benefits of offering high
product variety 关1,5兴. The impact of design decisions that are
made during the product development phase on a firm’s ability to
offer high product variety has also attracted much research inter-
est 关6,7兴.
Previous research 关8兴 observe that the unit costs for multi-item
lines can be 25% to 45% higher than the theoretical cost of pro-
ducing only the most popular item in the line. Two concepts that
can potentially benefit companies in economically providing the
variety desired by the market are modularity and commonality.
Pine 关9兴 emphasizes that the best method to achieve mass cus-
tomization 共that is, to minimize costs while maximizing individual
customization兲 is through creating modular components that can
be configured into a wide variety of end products or services. In
their discussion of modular design, Ulrich and Tung 关10兴 recom-
mend the use of common design modules and packages. This
would help companies to achieve faster completion of design
projects and reduce costs. Companies such as Toyota and Sony
have successfully been applying the principles of modularity to
help keep their design and build complexity to a minimum 关11兴.
Standardizing components is expected to save development
costs, tooling costs, administrative expenses and development
time. Ulrich 关7兴 presents an excellent analysis of the potential
linkages between the architecture of a product and product vari-
ety. He also notes that sometimes, in an effort to standardize,
firms will use a component with excess capability for a particular
application. For example, a standard power supply may provide
slightly more power than is strictly necessary in a particular ap-
plication. However, a firm may choose to adopt the standard com-
ponent even though the unit cost is higher than that of a compo-
nent more closely matched to the application.
To exploit the benefits of commonality while generating highly
differentiated end products, many firms are moving towards a
family approach to product development. Many researchers
关12–19兴 have examined the benefits of component commonality
interms of reduced component inventory and improved service
levels.2
2
Vakharia et al. 关20兴 provide a detailed overview of the existing literature on parts
commonality.

Journal of Mechanical Design Copyright © 2000 by ASME DECEMBER 2000, Vol. 122 Õ 403
Methodology Table 1 Features across various models of Sony Walkman

Product Differentiating Components—Definition. In a

given product family, each product will have a unique set of func-
tions to appeal to the targeted market segments. The set of func-
tions may include certain unique functions and/or a unique com-
bination of functions that are different from those offered by its
siblings. However, there will be certain ‘‘basic functions’’ that all
products within a product family share 共these are the very func-
tions that justify grouping the products as a ‘‘family’’兲. All physi-
cal components and sub-assemblies that serve the basic functions
which are common to two or more products within a family are
expected to have the same physical characteristics. On the other
hand, physical components and sub-assemblies that contribute to a
product differentiating function may be different in their physical
character when compared to their counterparts in other siblings. In
fact, ‘‘counterparts’’ do not even exist if the function in question
is new and unique to that particular product within the family of
products. Such components are termed product differentiating.
Factors Influencing the Proliferation of Non-Product Differ-
entiating Components. We believe that non-value added de-
sign variations arise primarily due to one or more of the following
reasons:
• Geometric variations to accommodate packaging constraints
that are specific to a given model.
• Introduction of different technologies across models.
• Idiosyncrasies of different product teams.
Of the three listed above, the first two can be somewhat justi-
fied, although to a lesser extent than is usually claimed. The third tures, some may have additional product differentiating features.
reason cannot be justified at all and therefore every effort should These features may be related to the primary features, such as
be made to eliminate the non-value adding design variations aris- Auto-Reverse; or unrelated, such as a clock. Table 1 summarizes
ing due to product development team idiosyncrasies. Other pres- various features offered on some of the Sony Walkmans. Figure 1
sures on designers, such as evershortening product development illustrates the function diagram for a typical walkman that we
lead times, may influence a design to stray from the optimal. examined. Figure 2 shows the parts inside the Sony product fam-
While efforts should still be made, amidst these pressures,3 to ily that we studied.
eliminate non-value added variation, such pressures are admit- A designer developing a new product family 共or adding a new
tedly a fact of today’s competitive product development product to an existing family兲 must first begin with a detailed
environment. function diagram of each model in the product family. The func-
tional configuration of a new product must be developed from the
Design Methodology for Component Standardization. needs identified through market research. When all functions are
What seems to be lacking in this struggle to gain control of prod- described at the same high level of abstraction, it provides a com-
uct proliferation is a methodology to design a family of products. mon ground for comparing alternate designs 关21–23兴. It is then
Even more fundamentally, a way to measure commonality across the responsibility of the product development teams to identify
models is required to capture the extent to which a company is physical embodiments for each function. If adding a new product
eliminating non-value-adding proliferation. Such a metric would to an existing family, the team must not reinvent these embodi-
also aid in benchmarking a firm’s initiative towards this goal with ments, but should pull from the existing designs. Only in the case
the best in the industry. Before we present our model for objective of new functions, distinct from all pre-existing functions, should
evaluation of commonality, we propose a methodology for new parts be designed. Additionally, these parts must be devel-
standardization.
The first step towards standardization is to develop a complete
function diagram. A function diagram captures the various func-
tions that a system has to perform. A function just describes
‘‘what to accomplish’’ without any reference to ‘‘how to accom-
plish’’ the task. Naturally, functions of a system can be classified
as primary and auxiliary 共secondary and tertiary兲 functions. Each
function can be divided into various sub-functions and so on,
hierarchically, until a physical component is identified with each
sub-function. In other words, a network of functions required to
fulfill a given task would serve as a skeletal design, representing
all that needs to be accomplished. We can also think of a product
on the level of its features, which directly map to a set of func-
tions. For instance, if we consider a Walkman radio/cassette
player, one can readily identify some of the primary features as i兲
Play, ii兲 Stop, iii兲 Fast-Forward, iv兲 Eject, v兲 AM Radio, and vi兲
FM Radio. While most Walkmans share the same primary fea-

3
In fact, modular principles that tend to decrease non-value added variety 共such as
the reuse of common design modules兲, have become an enabler of reduced product Fig. 1 Function diagram of non-product differentiating func-
development cycles. tions of a walman cassette player

404 Õ Vol. 122, DECEMBER 2000 Transactions of the ASME

 Fig. 2 Four of many members of the Sony Walkman family of products demonstrating 87% part commonality
in non-product differentiating functions. FX-221 costs about twice as much as EX-122 in retail market. „See
Tables 1 and 2…
oped to conform to the framework of existing functional and
physical configurations.
While this procedure may seem to stifle creativity on the level
of component design, creativity is now required to integrate dif-
ferentiating components with common architecture. It is not ben-
eficial to simply be creative in generating new design concepts.
New designs must occur with minimum changes to the existing
configuration. More importantly, if some component or subassem-
bly does not help in differentiating a product then it requires no
reinventing. A design already exists for this component within the
same product family.
As we have stated above, one of the vital performance mea-
sures of the product development effort is resource productivity.
This can be defined as the level of resources that are required to
take the project from concept to market. In order to increase re-
source productivity, it is imperative that firms conserve their ef-
forts by seeking ingenious ways to share designs of non-
differentiating features across models.
When designing a new family, the product development team
should study each function across the entire product family. The
goal is to standardize as many components as possible for those
functions that are common to the entire family. The design team
should evaluate each function in terms of:
1 ‘‘How’’ the function is implemented, namely the working
principle,
2 The number of components required to perform the function,
and,
3 The different types of components that are used in each de-
sign
• Geometry 共size and shape variations兲
• Material differences across models
• Manufacturing processes by which the parts are going to be
made
• Fastening and joining methods 共type and number of fasten-
ers, methods such as soldering, welding and adhesives etc.兲
• Mechanisms
• Actuation methods 共electro-mechanical, pneumatic, and hy-
draulic etc.兲
• Sensors 共Type of sensors: optical, pressure etc.兲
4 Vendor base
5 Number and type of secondary operations required.
Cost of Complexity. The use of standard components typi-
cally helps in lowering the complexity, cost and lead-time of
product development. From the product design and development
point of view, any variation in design leads to increased cost due
to an increase in quality assurance testing, uncertainties associated
Journal of Mechanical Design
with development, a lack of economies of scale, etc. However, not
all design variations are treated equally by the manufacturer. De-
pending on the production capabilities of the manufacturer, cer-
tain types of product variation lead to a greater increase in costs
than the others. Therefore, when we speak of the cost of product
design complexity, we must do so in the context of a manufactur-
ing facility. For instance, let us consider two different parts be-
longing to the same product family, produced on the same pro-
duction line but differing in one of the size dimensions. If the
manufacturing method used were stamping, a change-over would
involve only a simple indexing of the right size tool in real time to
accommodate the different part sizes. Whereas, if the parts were
injection molded, the same change in dimension from one part to
another may require changing dies. This would be a much more
time-consuming task when compared to automatically indexing
a stamping tool. This serves to illustrate that it is difficult to
map variations in product features to increases in manufacturing
cost in the absence of information about the capabilities of the
manufacturing facility and the process by which the parts are
manufactured.
Some of the major factors which influence the ease or difficulty
with which a certain product design change can be handled in a
given manufacturing facility are:
• Material handling system,
• Level of reconfigurability of tooling
• Part size-dunnage and space requirements
• Level of automation
• Universal grippers
• Worker training and skill levels
• Equipment maintenance schedules
• Compatibility of newer operations with existing operations
etc.
In fact, the product must be designed in concert with the
capabilities/limitations of the selected manufacturing process.
Therefore in order to reap maximum benefits from a manufac-
turing system, we stress the importance of the following attributes
being common across all products within a family:
• Manufacturing methods
• Part locators and reference surfaces
• Manufacturing and assembly related features
• Fastening and joining schemes
• General spatial orientation of the product features
Definition of Complete Commonality Across Models. As-
pects that differentiate members of a product family may include
a set of components and/or characteristics that influence the per-
DECEMBER 2000, Vol. 122 Õ 405
formance. They may also include augmenting features, safety fea-
tures, aesthetic appeal, and reliability as reflected in the warranty
information.
Conformance quality is the traditional definition of quality such
as the one given by Taguchi: Every product delivers the target
performance each time the product is used, under all intended
operating conditions, and throughout its intended life with no
harmful side effects.
Firms should strive to standardize all component characteristics
that neither help in differentiating a product nor influence the
conformance quality of a product. The component characteristics
that we consider here are its geometry, size, manufacturing
method, and fastening and joining methods. In other words, we
say that complete commonality across products within a product
family is achieved if all component characteristics that are 1兲 non-
product differentiating, and 2兲 non-influencers of conformance
quality, have been standardized.
Product Line Commonality Index „PCI…. As we strive to
compare different manufacturers on their efforts towards stan-
dardizing components across models, we would like to have an
objective measure to capture the existing differences in product
design strategies. One simple metric would be the percentage of
components that are common across a product family. However,
such a metric would over-look the degree of differentiation in a
product family, penalizing those families that have a broader fea-
ture mix. The degree to which products in a family are differen-
tiated is a marketing strategy decision, and thus beyond the scope
of our metric. Rather, we seek a metric that only penalizes those
differences that should ideally be common, given the product mix.
In developing this measure of commonality, we considered the
following factors:
• The number of different types of components that can be
ideally standardized across models 共i.e., those that are non-
product differentiating and non-influencers of conformance
quality兲,
• Geometric features of components in terms of their sizes and
shapes,
• Materials used across these components,
• Manufacturing processes that were used for their production,
and,
• Part assembly and fastening schemes that were used.
We define Product Line Commonality Index 共PCI兲 as follows:
P P
兺
i⫽1
CCIi⫺ 兺 MinCCI
i⫽1
i
PCI⫽ P P ⫻100 (1)
兺 MaxCCI ⫺ 兺 MinCCI
i⫽1
i
i⫽1
i
where,
CCIi⫽Component Commonality Index for component i
⫽ni⫻f1i⫻f2i⫻f3i
MaxCCIi⫽Maximum possible Component Commonality Index
for component i ⫽ N
MinCCIi⫽Minimum possible Component Commonality Index
for component i ⫽
1 1 1 1
n i⫻ ⫻ ⫻ ⫽ 2
n i ni n i n i
P⫽Total number of non-differentiating components that can
potentially be standardized across models
N⫽Number of products in the product family
ni⫽Number of products in the product family that have com-
ponent i
f1i⫽Size and shape factor for component i
406 Õ Vol. 122, DECEMBER 2000
f2i⫽Materials and manufacturing processes factor for compo-
nent i
f3i⫽Assembly and fastening schemes factor for component i
Substituting the values of CCIi , MaxCCIi and MinCCIi in the
expression for PCI, we obtain the following formula for product
line commonality index:
P P
1
兺 n ⫻f
i⫽1
i 1i ⫻ f 2i ⫻ f 3i ⫺ 兺n
i⫽1
2
i
PCI⫽ P ⫻100 (2)
1
共 P⫻N 兲 ⫺ 兺n
i⫽1
2
i
The three factors f1i , f2i , and f3i are assumed to be independent
of one another and the PCI score is expressed as a percentage and
takes values between 0 and 100. A PCI score of 0 indicates that
none of the non-differentiating parts are shared across models, or
if they are shared, their sizes/shapes, materials/manufacturing pro-
cesses, and assembly processes are all different. A perfect score of
100 would indicate that all the non-differentiating parts are shared
across models and that they are of identical size and shape, made
using the same material and manufacturing process, and the fas-
tening methods used are identical.
When computing the PCI, the first step is to sort through all of
the components and determine which are non-product differenti-
ating. These are the only components considered for the PCI.
Tables 2, 3, and 4 show calculations of the PCI for three Walkman
product families. Each row in these tables represents one compo-
nent, i, that is non-product differentiating, and may be shared
across different members of the family. A component is termed,
‘‘shared’’ if it is present in more than one model, and it performs
the same core function. Such a component is not necessarily iden-
tical from model to model. It can potentially vary in size, shape,
material, or fastening method, as described below.
For each component i, we determine the component common-
ality index 共CCI兲. This measure captures two important pieces of
information:
1 The extent to which the family shares architecture for the
functions represented by component i, and
2 The extent to which the size, shape, materials, manufacturing
process and assembly methods are shared across the product
family.
When ni⫽N, the function that maps to component i is accom-
plished using the same architecture in all models. However, if ni
⬍N, this implies that the function 共that component i maps to兲 is
performed with a different set of components in at least one
model, and thus the architecture is different. In order to capture
the size, material and assembly variations of the component across
models, we have introduced the three factors f1i , f2i , and f3i . The
size factor, f1i , is computed as the ratio of the greatest number of
models that share component i with identical size and shape to the
greatest possible number of models that could have shared the
component i with identical size and shape 共namely, ni兲. Ideally, if
in all models that share component i, the component is identical in
size and shape, then the size factor f1i would equal 1. On the other
hand, if no two models in the family have this component identi-
cal in size and shape, then the size factor f1i would take its lowest
value 共namely, 1/ni兲. The other two factors, f2i and f3i , are deter-
mined in a similar manner.
When calculating the PCI, in order to gain values between 0
and 100 we subtract the minimum possible value of the CCI and
normalize by the maximum possible value of the CCI. The mini-
mum value occurs when each of the size/shape, materials, and
assembly factors take on their lowest value. In this case, CCIi
⫽ni⫻1/ni⫻1/ni⫻1/ni⫽1/ni2. The maximum value occurs when
f1i , f2i , and f3i have a value of 1, and the component is present in
Transactions of the ASME
 Table 2 PCI table for Sony Walkman family „including FX-101, EX-122, FX-221,
FX-321…
each of the family members (ni⫽N). In this case, CCIi⫽N.
Tables 2, 3, and 4 show tabulated calculations of the PCI for each
of three Walkman product families.
The closer the PCI index is to 100, the better are the firm’s
efforts towards a value optimizing design. In the case of Sony’s
family of Walkmans 共not including Sports models兲, the PCI was
determined to be 87.0 共see Table 2兲, for the RCA family of prod-
ucts, it was determined to be 38.4 共see Table 3兲, and for Radio
Shack, it was 43.5 共see Table 4兲. To give the PCI an intuitive
appeal, we can think of it as the percentage of non-product differ-
entiating components that are identical. We can say that Sony
achieves 87% commonality in non-product differentiating compo-
nents, while Radio Shack and RCA achieve roughly 40% com-
monality for non-product differentiating components. If we weigh
this information together with the total number of components in
the product family, we can begin to assess the relative impact that
a change in the PCI value would have in terms of manufacturing
and development costs.
We make no assumption or precondition that the PCI be applied
only to families with uniform architecture. The core function
which comes from the function diagram describe ‘‘what to ac-
complish’’ without reference to ‘‘how to accomplish.’’ Thus, the
core function is independent of architecture. If the architecture of
Journal of Mechanical Design
each member of the family were the same, then ni 共in the 3rd
column of the PCI Table兲 would always be equal to N.
This analysis of product line commonality using the PCI cap-
tures the differences in component level commonality observed in
the same product family across three different manufacturers. The
differences in the level of commonality of components across
these product families should account for some very significant
differences in the product development costs and quality of these
products. In the case of Sony, sharing as many as 87% of the
non-product differentiating components between the low-end
共$27兲 and the high-end 共$90兲 models should ensure not only
higher quality across products, but also enable them to lower their
overall development costs.
Validation. The objective of the PCI is to provide a percent
common of non-differentiating components. It is tempting to do
away with PCI, a seemingly complex metric, and employ a
straight forward measure of percent of components that are com-
mon across a product line. The difficulty with such oversimplifi-
cation lies in defining what is ‘‘common’’? Is a component ‘‘com-
mon’’ if it is present in two of the five family members? Or does
it have to be present in all five? An added level of complexity in
DECEMBER 2000, Vol. 122 Õ 407
 Table 3 PCI table for RCA Walkman family „including RP-1800, RP-1810, RP-1820, RP-1875,
RP-1880…
the metric is required to ‘‘scale’’ the level of commonality,
depending on how many members of the family share the
component.
The PCI table groups information that is more valuable than
simple quoting of a percentage. This accounts for the inclusion of
the factors f1 , f2 , and f3 . In the most straight forward approach,
two parts would be considered common if they had identical size,
geometry, and material. Otherwise, they would be uncommon.
However, by splitting up the ‘‘common-ness’’ of parts into three
categories, it gives more information to the user. A glance at the
PCI table will not only tell the user that, for example, their PCI
ranks low in comparison to a competitor’s, but also that they can
target improvement activity on a specific aspect such as fastening
and assembly scheme, or architecture.
Benchmarking Through Normalized Product Line Com-
monality Index. In order to benchmark one company’s efforts
towards an increased sharing of components across products
within a product family with the efforts of another company, we
need to consider both the product line commonality index 共PCI兲
and also the total number of components that are present in each
of the products under consideration. It is important to remember
that it is difficult to identify a generic strategy that can adequately
408 Õ Vol. 122, DECEMBER 2000
address the tradeoffs between commonality and the total number
of parts. The issues are complex and necessarily specific to the
particular product and manufacturing system. However, the nor-
malized PCI presented here can give an indication as to how well
a particular firm’s strategy is working when compared to compet-
ing firms.
Generally, a modular design tends to have more components
than a multi-functional design performing the same overall func-
tion. The modular design will also tend to be more suited for
sharing parts across products. On the other hand, a product family
designed on the basis of multi-functional design principles will
tend to have fewer common parts across models that are less
suited for sharing. An optimum balance exists between the func-
tional content that each part should have and the level of com-
monality that should exist across products within a product fam-
ily. The ultimate goal of any product development team is not
only to maximize the level of shared parts across models but also
to effectively control the number of parts per product. Fewer parts
per product results in fewer part management related transactions
and thereby minimizes the level of complexity faced on the shop
floor. This is the rationale for normalizing the PCI index for the
number of parts that are used in each of the models.
Transactions of the ASME
Table 4 PCI table for Radio Shack Walkman family „including Table 5 Comparison of product line strategies for Sony, RCA,
14-1202, 14-1216, 14-1218, 14-1227… and Radio Shack

PCIj⫽Product line commonality index for the company j

Max共PCI兲⫽Maximum PCI value across all companies under
consideration
Numparts j ⫽Number of parts for the product family of com-
pany j
Min共Numparts兲⫽Minimum number of parts that are present
across all companies under consideration
From the above expression, we can see that a company would
have a high value for the normalized PCI if it has a high PCI and
a low number of parts. Table 5 shows values for the normalized
PCI for Sony, RCA, and Radio Shack.
Like the PCI, the normalized PCI is expressed as a percentage
and will take on values up to 100. We can say that a high value of
the normalized PCI means that the firm is doing well in its efforts
towards standardization and reduction of part numbers. Sony has
done the best across the three companies under consideration for
the specific set of products that we had studied. Radio Shack had
the worst score for the normalized PCI and it shows that they not
only had more part numbers than their counterparts, but also did a
poor job in standardizing these parts across models. There is an
intuitive feel to the normalized PCI value as well. For both the
PCI and the number of parts Radio Shack performs about twice as
‘‘poorly’’ as the best company, among those under consideration.
This compounds and gives a normalized PCI value of roughly 25
percent. These results are specific to the product families under
investigation and does not reflect the overall effectiveness of these
firms.

Closure
In this paper, we have analyzed the factors that contribute to
product complexity in general, and developed an objective mea-
sure to capture the level to which firms successfully share non-
product differentiating parts across products in a given product
family. The PCI measure provides designers with a method by
which they can benchmark their performance with the best in the
industry and it also highlights potential opportunities for improve-
ment in design efficiencies.

Acknowledgments
This study was supported by Ford Motor Company, Dearborn,
Michigan under the University Grants Program 共1997兲. The au-
thors would like to thank undergraduate seniors, Ms. Michelle
Lornz and Ms. Tara Greene, for their research assistance in diss-
section of several walkmans and preparation of PCI tables.

References
关1兴 MacDuffie, J. P., Sethuraman, K., and Fisher, M. L., 1996, ‘‘Product Variety
and Manufacturing Performance: Evidence from the International Automotive
Assembly Plant Study,’’ Manage. Sci.
关2兴 Clark, K., and Fujimoto, T., 1991, Product Development Performance: Strat-
In Table 5 we show the PCI values, the number of parts under egy, Organization and Management in the World Auto Industry, Harvard Busi-
consideration, and the normalized value of PCI. The normalized ness School Press, Boston.
value of PCI is computed as follows: 关3兴 Womack, J. P., Jones D., and Roos, D., 1990, ‘‘The Machine that Changed the
World,’’ Rawson Associates, New York.
For each company under consideration, 关4兴 Sanderzon, S. W., and Uzumeri, V., 1997, ‘‘Managing Product Families,’’
Normalized PCI for company j⫽ Irwin.
关5兴 Kekre, S., and Srinivasan, K., 1990, ‘‘Broader product Line: A Necessity to
Achieve Success?,’’ Manage. Sci., 36, No. 10, pp. 1216–1231.
PCI j Min共 Numparts 兲 关6兴 Whitney, D. E., 1993, ‘‘Nippendenso Co. Ltd: A Case study of Strategic
⫽ ⫻ ⫻100 (3) Product Design,’’ Res. Eng. Des., 5, pp. 1–20.
Max共 PCI 兲 Numparts j 关7兴 Ulrich, K., 1995, ‘‘The Role of Product Architecture in the Manufacturing
Firm,’’ Res. Policy, 24, pp. 419–440.
关8兴 Sethuraman, K., 1994, ‘‘The Impact of Product Variety on Manufacturing
where, Performance: An Empirical Investigation of the World Automobile Industry,’’

Journal of Mechanical Design DECEMBER 2000, Vol. 122 Õ 409

 Ph.D. thesis, Wharton School, University of Pennsylvania.
关9兴 Pine, J., 1992, Mass Customization: The New Frontier in Business Competi-
tion, Harvard Business School Press.
关10兴 Ulrich, K., and Tung, K., 1991, ‘‘Fundamental of Product Modularity,’’
ASME Winter Annual Meeting Symposium on Issues in Design/
Manufacturing Integration, Atlanta.
关11兴 Brooke, L., Nov., 1994, ‘‘Coping with Complexity,’’ Automotive Industries.
关12兴 Baker, K. R., Magazine, M. J., and Nuttle, H. L. W., 1986, ‘‘The Effect of
Commonality on Safety Stock in a Simple Inventory Model,’’ Manage. Sci.,
31, pp. 982–988.
关13兴 Gerchak, Y., Magazine, M. J., and Gamble, A. B., 1988, ‘‘Component Com-
monality with Service Level Requirements,’’ Mandge. Sci., 34, pp. 753–760.
关14兴 Dertouzos, L. M., Lester, R., and Solow, R., 1989, Made in America: Regain-
ing the Productive Edge, The MIT Press, Cambridge.
关15兴 Farrell, J., and Saloner, G., 1985, ‘‘Standardization, Compatibility, and Inno-
vation,’’ Rand J. Econ., 16, No. 1, pp. 70–83.
关16兴 Farrell, J., and Saloner, G., 1986, ‘‘Installed Base and Compatibility: In, Prod-
uct Preannouncements, and predation,’’ Am. Econ. Rev., 76, No. 5, pp. 940–
955.
关17兴 Fonte, W. G., 1994, ‘‘A deproliferation Methodology for the Automotive In-
410 Õ Vol. 122, DECEMBER 2000
dustry,’’ Masters of Science Thesis, Massachusetts Institute of Technology,
Leaders for Manufacturing Program.
关18兴 Gupta, S., and Krishnan, V. K., 1997, ‘‘A Product Family-Based Assembly
Sequence Design Methodology for the Economic Attainment of Product Va-
riety,’’ Working Paper University of Texas, Austin.
关19兴 Lee, H., and Billington, C., September, 1992, ‘‘Designing Products and Pro-
cesses for Postponement,’’ presented at the conference on Design Manage-
ment, Anderson Graduate School of Management, UCLA.
关20兴 Vakharia, Asoo, Parmenter, David, and Sanchez, Susan, 1996, ‘‘The Operating
Impact of Parts Commonality,’’ Journal of Operations Management, 14, No-
vember.
关21兴 Iyengar, G., Lee, C-L, and Kota, S., 1994, ‘‘Towards an Objective Evaluation
of Alternate Designs,’’ ASME J. Mech. Des., 116, pp. 487–492.
关22兴 Kota, S., and Chiou, S. J., 1992, ‘‘Conceptual Design of Mechanisms Based on
Computational Synthesis and Simulation of Kinematic Building Blocks,’’ J.
Res. Eng. Des. No. 4, pp. 75–87.
关23兴 Kota, S., and Lee, C. L., 1993, ‘‘A General Framework for Configuration
Design Part 1: Methodology, Part 2: Application to Hydraulic Systems Syn-
thesis,’’ J. Eng. Des., 4, No. 4, pp. 277–303.
Transactions of the ASME