Page 1 Journal of Integrated Design and Process Science, Volume 3, Number 4, 1999. Modular Product Design: A Life-cycle View

Journal of Integrated Design and Process Science, Volume 3, Number 4, 1999.
Modular Product Design: A Life-cycle View
John K. Gershenson1, G. Jagannath Prasad2, and Srikanth Allamneni1
1Department of Mechanical and Aerospace Engineering
Utah State University
Logan, Utah 84322-4130
Phone: (435) 797-2834
E-mail: jkgershenson@mae.usu.edu
2Plumlee Associates, Inc.
Baton Rouge, Louisiana
ABSTRACT
1.1
Previous research into modularity
This paper discusses the incorporation of modularization
Most research into modularity originates from Suh’s
into mechanical designs. The research uses a definition
(1990) independence axiom that states, “[i]n good design
of modularity that incorporates the potential of
the independence of functional requirements is
modularity based not only on the form/function structure
maintained.” Therefore, if possible, each function that a
of a product but also life-cycle processes such as
product performs should be independent of all other
manufacture, assembly, service, and recycling.
functions the product performs. This axiom has led to a
search for a connection between physical independence
Modularization, due to the functional independence it
and functional independence. In one of the first works to
creates, has been called the goal of good design. Industry
discuss modular design theory, Ulrich and Tung (1991)
has made an effort to modularize products to be flexible
use product modularity as a design goal. They define
to the needs of end users. In addition, some modules are
modularity in terms of two characteristics of product
created with some aspects of assembly in mind. Life-
design: “1) Similarity between the physical and functional
cycle modularity entails maintaining independence
architecture of the design and 2) Minimization of
between components and all life-cycle processes in
incidental interactions between physical components.” In
different modules, encouraging similarity in all
an extension of this work, Ulrich (1995) states that a
components and processes in a module, and maintaining
modular product or subassembly has “a one-to-one
interchangeability between modules.
mapping from functional elements in the function
structure to the physical components of the product” and
In this paper, the definition of product modularity is
that all interfaces between the components of different
given. A measure of relative modularity and a modular
modules are decoupled.
design methodology are developed that encourage
modularity, prevent a cascade of product design changes
Newcomb, et al. (1996) discuss the role of product
due to changes in life-cycle processes, and support agile
architecture in modular design. They look at the effect of
reaction to changes in life-cycle processes. A short
modular architecture on the product life-cycle. While
example is used to clarify the work.
the authors do not discuss how to use modularity to
affect life-cycle cost, they do understand that modularity
1
INTRODUCTION
influences cost. While Chang and Ward (1995) use a
Modularity is a common but unexplored thread among all
more dynamic application of functional modules, Erixson
areas of life-cycle engineering. Modular products tend
(1996) and Kusiak (1996) detail design for modularity as
to have fewer components for assembly and are therefore
a tool to decrease assembly and manufacturing costs in
cheaper to assemble. Modularity allows for the
product families and manufacturing systems.
reduction of service costs by grouping components so
those less reliable components are easily accessed. In
Chen, et al. (1994) propose a measure of modularity
addition, grouping components into modules by how they
based upon the independence of functional requirements
are recycled can greatly reduce product retirement costs.
and their sensitivity to changes in design parameters.
However, their work could not account for the

---

interrelationship between the sensitivity and
components not in the module. These dependencies and
independence (Rosen, 1995). DiMarco, et al. (1994)
similarities include those that arise from the component
created a software tool that assesses a qualitative product
interactions and those which arise from the various
recycling cost. Their definition of modules, based on
processes the components undergo during their life-
physical characteristics and designer’s intent, does not
cycle. In an ideal module, each component is
capture all aspects of product modularity; neither do the
independent of all components not contained in that
authors develop a methodology for creating recycling
module throughout the entire product life-cycle
modules.
(independence). In addition, each component in the
module is processed in the same manner during each life-
In summary, work exists which defines modularity,
cycle stage (similarity) (Gershenson and Prasad, 1997).
develops methods of measuring modularity, and applies
This definition expands the form-function relationship to
modularity to product design. Some aspects missing in
a form-process relationship. Similarity is a new
the current works include a definition for modularity that
perspective on the separation of form and process. Each
takes into account aspects of a product other than its
part of the form (module) must undergo the same life-
function, a methodology for designing modular products,
cycle processes. Independence and similarity represent a
and an accompanying modularity measure.
significant increase in the rigor of defining product
modules versus past form/function independence.
1.2
Benefits of modularity
Modularity allows the designer to control the degree to
An important consideration when defining the relative
which changes in processes or requirements affect the
life-cycle modularity of a product is the level of detail
product and by promoting interchangeability, modularity
chosen when looking at the product structure. A product
gives designers more flexibility to meet these changing
may seem modular but, at some levels of detail, the
processes. This flexibility allows for delaying design
structure may not be modular. We use component trees
decisions until more information is available without
as a tool to describe the levels of detail of a product.
delaying the product development process. Another
Component trees show all of the components and
benefit is the ability of modularity to reduce life-cycle
subassemblies that make up the product. Components
costs by reducing the number of processes and reducing
can be examined down to their constitutive attributes
repetitive processes.
(material, geometry, tolerances, features,
etc.)
(Gershenson and Stauffer, 1995). The tree-like structure
Ulrich and Tung’s (1991) work details the costs and
is helpful in discerning levels of detail and showing
benefits of modular products. The benefits of modularity
subassembly interactions.
they discuss include 1) component economies of scale,
2) ease of product updating, 3) increased product variety,
Another important consideration when defining life-cycle
4) decreased order lead-time, and 5) ease of design and
modularity is the chosen level of abstraction of the life-
testing. The costs of modularity they discuss include 1)
cycle process itself. As an example, product retirement
static product architecture, 2) lack of performance
consists of many tasks (e.g., reuse, remanufacture,
optimization, 3) increased unit variable costs, and 4)
recycle). These tasks are made up of subtasks (e.g.,
excessive product similarity. Other works (Ishii, 1995;
collection, separation, and grinding.). A product may
Shah, 1996; Ulrich, 1991) concentrate on the benefits in
seem modular when examined from the standpoint of the
the design process.
overall retirement process but at some subtask level, the
product may not be modular. Process graphs are used to
2
DEFINITIONS
describe the levels of detail of life-cycle processes.
Life-cycle modularity is the foundation of this work.
Process graphs delineate each task and subtask of a
Life-cycle modularity is a relative property. Products
process.
possess a higher or lower degree of modularity. A
product with a higher degree of modularity either
Creating modular products involves comparing the
contains a larger percentage of components or
component tree and process graphs of a product and
subassemblies that are modular or contains components
making sure that, at each level of detail, the product’s
and subassemblies, which are, on average, more modular.
attributes are as independent from one another as
Subassemblies, which are relatively modular in nature,
possible for each level of detail of the life-cycle
are modules.
processes. If a dependency does occur, it should occur
within a module. In addition, within a module and at each
Modules contain a high number of components that have
level of detail, every process should be similar for every
minimal dependencies upon and similarities to other
component. Lastly, depending upon the product, the
2

---

connections between the modules should allow for the
3
MODULARITY MEASURE
interchangeability of modules.
The modularity measure, like most measures of
To increase independence and similarity, a product must
“goodness” or “X-ability,” is best used in comparing the
be designed with the following three facets of
relative modularity of two like products. There is quite a
modularity:
bit of initial work in calculating the measure. Analysis
with only a few life-cycle processes in life-cycle
Attribute Independence: Component attributes have
modularity may be more useful than total product
fewer dependencies on attributes of other modules,
modularity for a product, especially if a particular life-
called external attributes. If there are dependencies,
cycle process dominates the requirements of a product.
fewer attributes are dependent upon one another and
The four-step measure that follows relies heavily on
attributes that are related to external attributes are less
understanding the physical and process relationships
dependent. E.g., Lego® pieces which can be of any color,
among components. The example application used
size, shape, or material as long as they have the correct
throughout the next two sections of the paper is the
dot to attach to other pieces and an impression to accept
mechanical pencil in Figure 1.
other pieces. Attribute independence allows for the
redesign of a module with minimized effects on the rest
of the product. Attribute similarity is excluded because
having similar but unrelated components is not
detrimental as long as attribute independence is
Figure 1: Exploded view of the mechanical pencil
maintained.
highlighting the four modules: cone/tip, clutch/teeth, barrel,
and eraser.
Process Independence: Each task of each life-cycle
process of each component in a module has fewer
Step 1: Generating a Component Tree - A component
dependencies on the processes of external components.
tree details the physical relationships among components
This requires that the processes a module undergoes
at all levels of abstraction. To develop a component tree,
during its life-cycle are independent of the processes
the product is divided into its constitutive modules and
undergone by external modules. Any dependencies that
components. The modules are further classified into
do exist are minimized in number and criticality. E.g., in
subassemblies, then individual components, and lastly
separation for recycling, techniques that utilize grinding
product attributes that describe the components. A
and separation by material density are dependent upon the
partial component graph for a mechanical pencil
disassembly of all components containing materials that
highlighting the attributes of the cone/tip assembly is
are not compatible and are of a similar density. If the
shown in Figure 2. From Figure 2, it can be seen that the
disassembly process occurred later in the retirement
cone/tip assembly is comprised of components such as
process, grinding and density separation would not be
sleeve, rubber lead retainer, and cone/tip with similar
possible. Process independence allows for the reduced
geometric attributes (ID and OD for each) but very
cost in each life-cycle process and the redesign of a
different material attributes (steel, rubber, and plastic).
module in isolation if processes should change.
Mechanical Pencil
Process
Similarity: Group components and
subassemblies that undergo the same or compatible life-
Cone/Tip Assembly
Barrel Assembly
Eraser/ Refill Assembly
Spring/ Clutch/
Teeth Assembly
cycle processes into the same module. E.g., if a product
is being recycled through grinding, it would be least
Metallic pocket
Sleeve
Cone/ tip
Plastic barrel
clip
Eraser
Plastic cap
Rubber lead
Hollow plastic
expensive if all components undergoing this task were in
Rust proof Spring
retainer
Tube
Plastic lug/ unit
Brass Clutch unit
the same module therefore the entire module could be
M: Steel
SF: smooth
ground and then no other grinding would be necessary.
note: M: material, SF: Surface finish, T: Tolerance, G: geometry.
T: very close
ID: inner diameter
Brasscollet
Brass sleeve
OD: outer diameter
G: ID, OD, L
L: length
Process similarity minimizes the number of external
T: Thickness
Sid: Smaller ID
M: plastic
Sod: Smaller OD
M: rubber
Lid: Length of ID
components that undergo the same processes, creates a
L o d: Length of OD
SF: Smooth
W: width
SF: smooth
D: Depth
T: very close
strong differentiation between modules, reduces process
T: very close
G: cone angle, L,
OD, ID, O D & I D
G: ID, OD, L
of cylinder
repetition, and reduces process costs. Process similarity
Figure 2: A partial component tree of a mechanical pencil
also conserves redesign effort by insuring that changes to
cone/tip assembly.
individual life-cycle processes only affect one module of
the product.
Step 2: Generating Process Graphs - The various life-
cycle processes that each of the components in all of the
modules undergo are first jotted down and then all the
components which undergo each life-cycle process (e.g.,
3

---

manufacturing, assembly, function, service, retirement)
Component
Process
Modularity
are noted. For each process, a process graph must be
SubAssembly 1
Assy2
Pro2
Assy2
Process1
Evaluation
created that details each stage of the life-cycle, all of the
Component
Component
Component
Task1
Task2
Task
1
2
3
3
processes in each stage, and each of the tasks and
Matrix
Attrib.1 Attrib.2 Attrib .3 Attrib . 4 Attrib.5 Attrib .6 Subtask Subtask Subtask Subtask Subtask
1
2
3
4
5
.1
subtasks in each process. The manufacturing graph for a
1
Attrib
1
.2
mechanical pencil is shown in Figure 3. Components are
Component
Attrib
.3
grouped together according to the manufacturing process
Attrib
2
SubAssembly
.4
undergone and then each manufacturing process is
Component
Component
Attrib
.5
expanded to include the pertinent tasks and subtasks of
Attrib
3
.6
each process. The components cone/tip, hollow plastic
Assy2
Component
Attrib
tube, plastic cap, and plastic lug unit are all plastic
1
Subtask
injection molded components. The injection molding
Task1
2
Subtask
process is further classified into the tasks making the die,
Process1
3
Process
Subtask
locating the ejector pins, and melting the plastic in the
Task2
4
Subtask
auxiliary heating cylinder.
3
5
Pro2
Task
Subtask
Mechanical Pencil
Figure 4: A generalized modularity evaluation matrix. Each
subassembly and process is broken down into its constitutive
Components
Cone/ tip
Eraser
Brass collet/
Rubber lead
elements, attributes, and subtasks. The boxes contain the
Hollow plastic tube
sleeve
retainer
Plastic Barrel
Plastic cap
Plastic lug unit
Pocket clip
weights of the similarity and dependency relationships.
Spring
Metal sleeve
There are six possible relationships within both similarity
Processes
and dependency:
Rubber
Turning
Grinding
Special
Rubber
molding
purpose
molding
Injection
machinery
Finishing
molding
Milling
Component-Component Dependency occurs when two
Cutting
Shearing
Blanking
Hemming
components are reliant upon each other with respect to
Tube bending of
metal cap to fit
their physical design, specifically their attributes. An
barrel
The partition between the component, process and task levels is shown in italics. For space considerations,
example of this is a gear that fits on a shaft. While the
Locate ejector
M a k e d i e
only one task level for the injection molding process is shown by the arrow symbol.
Pins
gear and the shaft are two different components, the inner
Melt plastic in
Tasks
auxiliary heating
cylinder
diameter of the gear and the outer diameter of the shaft
Figure 3: A partial manufacturing process graph for the
are inextricably dependent upon each other.
mechanical pencil example.
Component-Component Similarity is not used because it
Step 3: Construction of the Matrices - Using the
does not tie the designs together so as to necessitate
component tree and process graphs, two modularity
changes in one due to changes in the other.
evaluation matrices are constructed, one to record
similarities and one to record dependencies. Figure 4
Component-Process Dependency details relationships in
shows the general form of the modularity evaluation
which product design is contingent upon the life-cycle
matrix. The square matrix has row and column headings
process a component undergoes, i.e., process drives
corresponding to the most specific levels of the
design. If the same process drives the designs of two
component tree and process graphs. The contents of the
different components, the components should be grouped
two modularity evaluation matrices, are the similarity and
in the same module so that they can evolve with the
dependency relationships among components and
process and minimize effects on other components. One
processes.
simple example is a tuner dial and a power switch on a
stereo, the two components are totally unrelated but
undergo the same manufacturing process. All such
plastic injection molded components could be combined
into one module so that they can be updated as one with
changes in the injection molding process.
Component-Process Similarity details relationships in
which a component uses or goes through the life-cycle
process. The logic is to group components that undergo
the same life-cycle processes in one module to minimize
the impact a change in process will have on the product.
As an example two fiber glass components of a
motorcycle such as the front and back mudguard which
4

---

are manufactured and retired by the same process or for
that matter any other fiber glass component, which is
A brief description of the calculation of the four prime
assembled in the same stage of assembly. These
parameters (Sin, Sout, Din, Dout) is given below:
components can be placed in the same module
Sin: Component similarities between each component
irrespective of their locations.
within a particular module.
Process-Process Dependency and Process-Process
M
s 1
s
T
S
S * S
in = ∑ ∑
− ∑∑
Similarity do not affect product design directly, due to
ik
jk
the exclusion of component interaction, and have been
m=1 i =r j=i+1 k=1
Where: m is a module, i, j are components in the same
excluded from both the relative modularity measure and
module, and k is a task.
design methodology.
M = # of modules in the product.
A set of ratings, the contents of the modularity evaluation
r = first component in module m or module n.
matrices, is shown in Figure 5. As an example, referring
s = last component in the module m or module n
T = # of processes under consideration
to Figure 6, a rating of 1 is given for the component-
S
component similarity between the rust proof spring and
ik is similarity between component i and task k
Sik is similarity between component j and task k
the sleeve whereas a rating of 5 is given for the
This value is the root mean square of the similarities
component-component similarity between the brass
between two components and a life-cycle process. Like
collet and the brass sleeve. This differential is due
all of the component-process measures to follow, it
mostly to a similarity in material attributes between the
allows a component-process relationship to be measured
collet and sleeve. In addition, a rating of 5 is given to the
in component-component terms. Sin is calculated for
component-process similarity between the component
only component-process interaction. Hence the
cone/tip and the injection molding process because the
calculation of Sin for a component of module A makes
entire cone/tip is injection molded.
use of the ratings of the component-process similarity
Similarity
Dependency
interactions for each component within module A. Sin
1: Not similar
1: Not dependent
has a positive effect on the measure as we are trying to
2: Slightly similar
3: Dependent
group components with similar life-cycles.
3: Similar
5: Highly dependent
Sout: Similarities between the components of a module
4: Very similar
and each component external to the module.
5: Extremely similar
M
s−1
M
s
T
Figure 5: Similarity and dependency ratings.
S
S * S
out = ∑ ∑ ∑ ∑ ∑
ik
jk
Step 4: Calculation of the Relative Modularity using
m=1 i=r n =m+1 j =r k =1
the Modularity Evaluation Matrix:
Where i, j are components not in the same module, and n is
For a high degree of modularity, it is important to have a
a module.
The calculation of S
high similarity between components within a module
out for a component of module A
makes use of the ratings of the component-process
(Sin), a low similarity between a component of a
similarity interactions for each component outside
concerned module and other components outside of the
module A. S
module (S
out has a negative effect on the modularity
out), a high dependency between components
measure, as we are interested in reducing process
within the module (Din), and a low dependency between a
similarities between components that are in two different
component within a module and a component outside of
modules.
the module (Dout).
Din: Dependencies between each component within a
The measure of relative modularity that was finally
particular module.
developed is:
Din = Component-Process interactions + Component-
Component interactions
Modularity = Sin / (Sin +Sout) +Din / (Din +Dout)
M
s 1
s
T
D
D * D
D
in = ∑ ∑
− ∑∑( ik jk + ij)
The above measure directly correlates to the definitions
of similarity and dependence put forth in the previous
m=1 i=r j=i+1k =1
sections. In this measure we find the ratio of the sum of
Where: i, j are components in the same module.
D
similarities inside the modules to the total similarity and
ik is the dependence between component i and task
k
sum it up with the ratio of the sum of dependencies
Djk is the dependence between component j and task
inside the modules to the total dependency. The values
k
that are obtained for the whole product show increasing
Dij is the dependence between component i and
modularity from 0 to 2.
component j
5

---

The calculation of Din for a component of module A
the cone/tip assembly, is calculated by applying the
makes use of the ratings of the component-component
formula for Sout from the previous section and summing
and component-process dependence interactions for
up the similarities for all the process. As an example, in
each component within module A. Component-
injection molding there are four external components
component dependencies are taken right out of the
that undergo the same process. Therefore, there are four
modularity evaluation matrix. Din has a positive effect on
sets of √(5*5) at the beginning of the calculation of Sout.
the measure as it is important to group dependent
D
components.
in=Din,c-c+Din,c-p
Din,c-c = 10+5=15
Din,c-p = 0
Dout: Dependencies between the components of a module
⇒ Din= 15
and each of the components that are external to the
Din includes both the component-component
module.
dependencies and component-process dependencies
Dout = Component-Process interactions + Component-
inside the module. Component-component dependencies
Component interactions
are depicted in the top part of the matrix whereas
M
s 1
M
s
T
component-process dependencies are computed from the
D
D * D
D
out = ∑ ∑
−
∑ ∑∑( ik jk + ij)
bottom part of the matrix. The method for calculating
m=1 i=r n =m+1 j=r k =1
both the dependencies is explained in the previous
Where: i, j are components not in the same module.
M = # of modules in the product.
section. D
D
in,c-c for the cone/tip assembly consists of two
ik is the dependence between component i and task
k
5-weight dependencies between two components, the
D
sleeve and the cone and the sleeve and the retainer. There
jk is the dependence between component j and task
k
is also a 5 weight relationship between the retainer and
Dij is the dependence between component i and
the cone. This leads to a 10 as the first part of Din,c-c and
component j
a 5 as the second. There are no common processes so
The calculation of Dout for a component of module A
Din,c-p = 0.
makes use of the ratings of the component-component
and component-process dependence interactions for
each component outside module A. Dout has a negative
impact on the total measure, as all external dependencies
Dout=Dout,c-c+Dout,c-p
must be minimized to have independent modules.
Dout,c-c = 9+13+6 = 28 Dout,c-p = 20+5+45 = 70
⇒ Dout= 98
Cursory Example of a Mechanical Pencil
Dout,c-c involves outside dependencies between, for
In this brief example, we have calculated the modularity
example, the sleeve and the brass collet, the brass sleeve,
of the cone/tip assembly only taking into account only
and the plastic barrel. Each dependency has a weight of 3
component, function, and manufacturing interactions.
leaving a 9 as the first value in Dout,c-c.
The relationships are shown in Figure 6 at the end of the
paper. The first step in measuring the modularity of the
Once the calculations for Sin, Sout, Din, and Dout are made
mechanical pencil is to calculate the values of S
for all four modules of the mechanical pencil, the relative
in, Sout,
D
measure for the product can be calculated by summing up
in, and Dout.
the relative measures of each of the modules as in Table
Sin = 0
1. The table displays the relative modularity values for
If we refer to the matrix in Figure 6, we see at the bottom
each of the four modules and the relative measure for the
part of the matrix for the cone/tip assembly that none of
mechanical pencil as a whole. Notice that the values in
the processes are common and hence there are no
the cone/tip row coincide with those described above.
similarities.
The cone/tip scored low because it consists of varied
S
components unlike the teeth/clutch, which has
out = Sinjection molding + Srubber molding + Sfunction 1 + Sfunction 2
components that are more congruous. The relative
= (√(5*5)+√(5*5)+√(5*5)+√(5*5)) + (√(5*5)) +
modularity for the pencil is 0.87 where the possible
(3*(√5*5 + (√5*5) + (√5*5)) + (√(5*5))
range of values is 0 to 2; the pencil scored low. This
+(√(5*5))
⇒
value has little meaning by itself but is useful to compare
Sout = 80
design options and to guide the redesign process.
In the manufacturing section of the matrix, there are
external similarities in the processes injection molding
Table 1: Total relative modularity of the mechanical pencil
and rubber molding and for the two functions lead slide
and each of its four modules.
and stopping of lead slide (not shown in the matrix). The
Module
Sin
Sout
Din
Dout
RM
value for Sout for each of the above processes, for only
Cone/Tip
0
80
15
98
0.13
6

---

Teeth/Clutch
45
75
131
101
0.94
module, a feasibility check is carried out to determine
Barrel
0
20
5
39
0.11
the practicality of shifting. If reconfiguration is
Eraser
5
35
20
40
0.46
impossible, the component attributes are redesigned. This
Total Relative Modularity
0.87
cycle is continued until the components are shifted to the
modules that yield the highest total relative modularity.
RM = modularity of the component; TRM ≠ ΣRM
Cursory Example of a Mechanical Pencil
4
MODULAR DESIGN METHODOLOGY
The redesign methodology was applied for three "rounds"
Using the definition of characteristic modularity, a
to the mechanical pencil. There were significant
specific methodology for designing products that are
improvements in the value of the total relative
modular in terms of their life-cycle processes was
modularity. First, we calculated the relative modularities
developed. The design methodology is a set of
shown in Table 1, leading us to approach the barrel
quantitative guidelines that direct product development
assembly first. Since elimination of the barrel module or
towards modular products with all of the benefits therein.
any other module was impossible, the next step was to
approach, in order of relative modularity, the components
The goal of the design methodology is to redesign a
of the modules. Again, no opportunity for elimination
product eliminating components or modules, rearranging
occurred. The next step is to look for possible
components or modules, or changing component
opportunities for reconfiguration. While there were
attributes. Elimination is the simplest process.
opportunities to increase modularity, they did not yield
Reconfiguration is the cost effective shifting of
feasible products. The last step was to attempt to
components to other modules to increase the total
redesign a component. Beginning with the lowest
relative modularity. Redesign is the changing of the
module, the barrel assembly, we approached the
component attributes to reduce outside similarities and
components in order of worst RM. The plastic barrel was
dependencies or increase inside similarities and
the first candidate but nothing could be done. Next came
dependencies. Redesign is more difficult than
the metallic clip, we redesigned the clip's material
reconfiguration because there is a need to redo the
attribute from metal to plastic. This increased the
engineering analysis. The logic of the design
modules relative modularity due to increased process
methodology, is as follows:
similarity within the module. Then, it was back to the
1. eliminate the modules if they are not necessary;
beginning of the methodology and, this time only
2. if the whole module cannot be eliminated, then
searching those modules and components that had
look to eliminate the components of these
experienced a change in the first round. After a
modules;
recalculation of the module and component RMs, we
3. if elimination is impossible, then try to shift the
again came all the way down to redesign before finding a
components to other modules or into new
possibility. Again, the barrel assembly was worst with the
modules to increase the overall value of product
plastic barrel component being worst in the module. Now
modularity;
that they were of the same material, the plastic barrel and
4. if reconfiguration is not possible, redesign the
the clip could be combined in a single component. Again,
attributes of the components to decrease or
the relative modularity of the modules and components
eliminate similarities or dependencies with
were recalculated and the barrel assembly was still
outside components or increase similarities with
lowest. After elimination was ruled out, reconfiguration
components of the same module.
was attempted. By comparing [Sin + D ] with [S
+
j
inj
outjk
D
] for each module, it was determined that the plastic
The complete algorithm for modular product design is
outjk
not included due to its length. In the algorithm, the
barrel (with integral clip) could be reconfigured and,
module with the lowest of the relative modularity is first
based on similar processes and that the threaded portion
taken up for analysis as long as its modularity has
of the barrel is assembled to the inner diameter of the
changed since it was last redesigned (in an iterative
cone/tip, the component plastic barrel/clip was moved to
effort). The relative modularity of all components in that
the cone/tip assembly. This eliminated the barrel
module is calculated and the component with the lowest
assembly since there were no components left in that
relative modularity is approached first. If the component
module. Note that for elimination and reconfiguration
cannot be eliminated, it is then taken up for
steps, some additional redesign is usually necessary. The
reconfiguration. For reconfiguration, it is necessary to
value of total relative modularity was increased from
determine into which other module the component can be
0.87 to 0.94. The new values of relative modularity are
moved. Once the component has been shifted to another
described in Table 2.
7

---

Table 2: Total relative modularity of the mechanical pencil
will lead to a better measure of the similarity and
after application of the modular design methodology.
dependence relationships between components.
Module
Sin
Sout
Din
Dout
RM
Cone/Tip
28
116
10
95
0.13
ACKNOWLEDGMENTS
Teeth/Clutch
45
75
131
109
0.94
The authors gratefully acknowledge The University of
Alabama Research Grants Committee, Alabama EPSCoR
Eraser
5
40
20
43
0.46
- NSF, and the National Science Foundation/Lucent
Total Relative Modularity
0.94
Technologies Industrial Ecology Fellowship program.
RM = modularity of the component; TRM ≠ ΣRM
REFERENCES
5
CONCLUSION AND FUTURE WORK
Chang, T.S. and A.C. Ward (1995), “Design-in-
In this paper, we have discussed the development of a set
Modularity with Conceptual Robustness,” Proceedings
of definitions that structure life-cycle product
of the 1995 ASME Design Technical Conferences -
modularity. We have also proposed a relative modularity
21st International Conference on Advances in Design
measure and design methodology. By using a structured
Automation, Boston, Massachusetts.
modularity measure built upon component independence,
Chen, W., et al. (1994), “Modularity and the
process similarity, and process independence, it is
Independence of Functional Requirements in Designing
possible to compare the degree to which a design is
Complex Systems,” Proceedings of the 1994 ASME
enjoying the benefits of modularity. Our approach
Design Technical Conferences - Conference on
focuses on independence and similarity across the life-
Concurrent Product Design, Minneapolis, Minnesota.
cycle and includes a step-wise redesign methodology to
DiMarco, P., C.F. Eubanks, and K. Ishii (1994),
guide designers towards modular products. It is
“Compatibility Analysis of Product Design for
important to view product modularity from the standpoint
Recyclability,” Proceedings of the 1994 ASME
of creating more modular products. This is quite
Design Technical Conferences - 14th International
different from designing products with interchangeable
Conference on Computers in Engineering,
or reconfigurable parts. It is also quite different from
Minneapolis, Minnesota.
maintaining form/function independence. It is the goal of
Gershenson, J.A. and L. Stauffer (1995), “The Creation of
modular design to group all attributes with like life-cycle
a Taxonomy for Manufacturability Design
processes into a single module and decouple them from
Requirements,” Proceedings of the 1995 ASME
all other attributes and life-cycle processes.
Design Technical Conferences - 7th International
Conference on Design Theory and Methodology,
The example product chosen for our analysis is a
September, 1995, Boston, Massachusetts.
mechanical pencil. We have deliberately chosen a simple
Gershenson, J.K. and G.J. Prasad (1997), “Modularity in
product. Our intention is to show the measure and apply
Product Design for Manufacturing,” International
the methodology before the final phase of testing on
Journal of Agile Manufacturing, Volume 1, Issue 1,
several comparison products and a more complex
Society of Agile Manufacturing, August, 1997.
product. In the example, we were able to use the measure
He, D.W. and A. Kusiak (1996), “Performance Analysis
and methodology to significantly increase the modularity
of Modular Products,” International Journal of
of the product. Through our cursory example, we have
Product Research, Vol. 34, No. 1, pp. 253-272.
seen that one shortcoming of the modular design
Ishii, K., C. Juengel, and D.F. Eubanks (1995), “Design
methodology is the work necessary to apply it. The
for Product Variety: Key to Product Line Structuring,”
matrices necessitate deep product knowledge and tedious
Proceedings of the 1995 ASME Design Technical
work. We are working to automate the evaluation and
Conferences - 9th International Conference on
reconfiguration.
Design Theory and Methodology, Boston,
Massachusetts.
Our next task will be to explore the ties between
Newcomb, P.J., B. Bras, and D.W. Rosen (1996),
modularity and life-cycle cost. Using the relative
“Implications of Modularity on Product Design for the
modularity measure and life-cycle cost calculations; we
Life-cycle,” Proceedings of the 1996 ASME Design
will research the point at which the wastes of redundancy
Technical Conferences - 10th International
and additional features outweigh the benefits of more
Conference on Design Theory and Methodology,
efficient product families and flexibility in meeting all
Irvine, California.
customer needs. In addition, we are moving towards
Rosen, D. (1995), personal conversation at ASME
improved development on the characteristics of
Design Technical Conferences, September, 1995.
similarity and dependence in each life-cycle area. This
8

---

Shah, et al. (1996), “Research Opportunities in
Ulrich, K. (1995), “The Role Of Product Architecture In
Engineering Design,” NSF Strategic Planning
The Manufacturing Firm,” Research Policy 24,
Workshop Final Report.
Elsevier Science B.V.
Suh, N.P. (1990), The Principles of Design, Oxford
Ulrich, K. and K. Tung (1991), “Fundamentals of Product
University Press.
Modularity,” Proceedings of the 1991 ASME Design
Technical Conferences - Conference on Design
Manufacture/Integration, Miami, Florida.
Similarity and dependenceTree
CG
CG
CG
CG
CG
CG
CG
CG
CG
CG
CG
CG
CG
evaluation matrix
Assy
cone/tip
cone/tip
cone/tip
Teeth/clt
Teeth/clt
Teeth/clt
Teeth/clt
Teeth/clt
Barrel
Barrel
Eraser
Eraser
Eraser
Comp/Assy
Sleeve
Cone
RL Ret.
H Pl. Tube RP Spring Clutch Un. Clutch Un. Clutch Un.
M Clip
Pl Brl.
Eraser
Pl. Cap
Pl. lug
Tree
Assy
Comp/Assy
Comp
Br. Collet Br. Sleeve Collet Cyl.
CG
cone/tip
Sleeve
0/0
3/5
2/5
3/0
1/0
5/3
5/3
2/0
3/0
2/3
2/0
3/0
4/0
CG
cone/tip
Cone
3/5
0/0
3/5
3/0
1/3
1/5
2/0
2/0
2/0
3/5
3/0
3/0
3/0
CG
cone/tip
RL Ret.
2/5
3/5
0/0
2/0
1/3
2/3
2/0
2/0
2/0
4/0
3/0
3/0
3/0
CG
Teeth/clt
H Pl. Tube
3/0
3/0
2/0
0/0
2/3
4/3
4/5
4/5
2/0
3/3
2/0
3/0
4/5
CG
Teeth/clt
RP Spring
1/0
1/3
1/3
2/3
0/0
4/0
4/5
4/5
5/0
4/2
4/0
4/0
4/0
CG
Teeth/clt
Clutch Un.
Br. Collet
5/3
1/5
2/3
4/3
4/0
0/0
5/5
5/5
5/0
2/0
1/0
2/0
2/0
CG
Teeth/clt
Clutch Un.
Br. Sleeve
5/3
2/0
2/0
4/5
4/5
5/5
0/0
5/5
5/0
2/2
4/0
2/0
2/0
CG
Teeth/clt
Clutch Un.
Collet Cyl.
2/0
2/0
2/0
4/5
4/5
5/5
5/5
0/0
5/0
2/3
1/0
2/0
2/0
CG
Barrel
M Clip
3/0
2/0
2/0
2/0
5/0
5/0
5/0
5/0
0/0
3/5
2/0
2/0
2/0
CG
Barrel
Pl Brl.
2/3
3/5
4/0
3/3
4/2
2/0
2/2
2/3
3/5
0/0
3/0
3/0
3/0
CG
Eraser
Eraser
2/0
3/0
3/0
2/0
4/0
1/0
4/0
1/0
2/0
3/0
0/0
4/5
4/5
CG
Eraser
Pl. Cap
3/0
3/0
3/0
3/0
4/0
2/0
2/0
2/0
2/0
3/0
4/5
0/0
5/5
CG
Eraser
Pl. lug
4/0
3/0
3/0
4/5
4/0
2/0
2/0
2/0
2/0
3/0
4/5
5/5
0/0
MG
MG
MG
MG
MG
MG
MG
MG
MG
MG
MG
Cone
Era. Uni
Pkt. Clip
Pkt. Clip
Pkt. Clip
Pkt. Clip
Collet
Collet
Collet
Spring
Metal Slv.
Inj. Mold Rub. Mold Shearing
Blanking Tube bend. Hemming
Turning
Grinding
Milling
SPM
Cutting
Tree
Assy
Comp/Assy
Comp
.
comp. Mold
Mak. Sh. Punch&die
Make blnk punch&cavity
make punch & die
make punch & die
cy time, collet, m
d/sn job template
d/sn milling cutter
wire drawing
tube cutting
CG
cone/tip
Sleeve
0/0
0/0
0/0
0/0
0/0
0/0
5/5
5/5
5/5
0/0
5/5
CG
cone/tip
Cone
5/5
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
CG
cone/tip
RL Ret.
0/0
5/5
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
CG
Teeth/clt
H Pl. Tube
5/5
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
CG
Teeth/clt
RP Spring
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
5/5
0/0
CG
Teeth/clt
Clutch Un.
Br. Collet
0/0
0/0
0/0
0/0
0/0
0/0
5/5
5/5
5/5
0/0
0/0
CG
Teeth/clt
Clutch Un.
Br. Sleeve
0/0
0/0
0/0
0/0
0/0
0/0
5/5
5/5
5/5
0/0
0/0
CG
Teeth/clt
Clutch Un.
Collet Cyl.
0/0
0/0
0/0
0/0
0/0
0/0
5/5
5/5
5/5
0/0
0/0
CG
Barrel
M Clip
0/0
0/0
5/5
5/5
5/5
5/5
0/0
0/0
0/0
0/0
0/0
CG
Barrel
Pl Brl.
5/5
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
CG
Eraser
Eraser
0/0
5/5
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
CG
Eraser
Pl. Cap
5/5
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
5/5
CG
Eraser
Pl. lug
5/5
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
0/0
Figure 6: Partial similarity and dependence modularity evaluation matrix for the mechanical pencil. The first value of each cell
represents similarity; the second value represents dependence. All process rows and the attribute levels of components are
eliminated to save space. The columns represent the elements of the component and the process graph.
9

---