INTEGRATED OPTIMIZATION OF PRODUCT DESIGN
CONCEPT AND PRODUCT LIFECYCLE SCENARIO
BASED ON GENETIC ALGORITHM
Masakazu Kobayashi and Masatake Higashi
Toyota Technological Institute, 2-12-1 Hisakata, Tempaku, Nagoya, Japan
Keywords: Design optimization, Conceptual design, Lifecycle design, Lifecycle assessment, Functional optimization,
Layout optimization, Hierarchical optimization, Genetic algorithm.
Abstract: Due to rise of environmental awareness in recent years, companies are required to assess and reduce
environmental burdens of their products. However, in practical product development, since not only
environmental burdens but also product characteristics such as performance and cost need to be
simultaneously considered for creating attractive products, designers are forced to take a great deal of time
and effort to balance them at a higher level at every stage of product development. In response to this, this
paper proposes an integrated method for optimizing product design concept and product lifecycle scenario
for supporting conceptual design phase. The proposed method combines integrated optimization of
functional / layout design which we developed in the previous researches and lifecycle assessment (LCA).
Using the proposed method, optimal functional structure, components / parts layout and lifecycle scenario
that balance product characteristics and environmental burdens at a higher level can be obtained.
1 INTRODUCTION
Due to rise of environmental awareness in recent
years, companies are required to assess and reduce
environmental burdens of their products such as
carbon emissions. However, in practical product
development, since not only environmental burdens
but also product characteristics such as performance,
cost and size need to be simultaneously considered
for creating an attractive product, designers are
forced to take a great deal of time and effort to
balance them at a higher level.
Based on the above background, this paper
proposes a new integrated optimization method for
creating a product concept and its lifecycle scenario
that balance various criteria including lifecycle ones
at a higher level. To allow for such optimal design,
the proposed method is based on our integrated
optimization method (Kobayashi et al., 2009). This
method is an integration of functional / layout
optimizations, which are based on genetic algorithm
(GA), for supporting a conceptual design phase.
During a conceptual design phase, since there are
various decision-makings, designers are asked to
make optimal decisions to create great product
concepts by considering various product
characteristics such as performance, cost and size.
However, since functional / layout designs, which
are main two tasks of a conceptual design phase, are
very different tasks, their design problems are highly
hierarchized and their solution spaces are vast, it is
extremely difficult for designers to build up great
concepts only with their own decision makings. To
overcome such difficulty, in our previous method,
functional / layout optimizations are combined and
executed cooperatively by exchanging information.
Using this method, both a functional structure and a
parts layout that balances various criteria at a higher
level can be obtained. To allow for simultaneous
consideration of various criteria including lifecycle
ones during a conceptual design phase as described
above, this paper makes an attempt to combine this
integrated optimization method and LCA. As for
design variables, in addition to functional structure
and parts / components layout, decision makings
throughout product lifecycle are considered. As for
criteria, in addition to product characteristics,
environmental burdens are considered. Using the
proposed method, optimal product concept (a
functional structure and a components / parts layout)
and its lifecycle scenario that balance various criteria
208
Kobayashi M. and Higashi M..
INTEGRATED OPTIMIZATION OF PRODUCT DESIGN CONCEPT AND PRODUCT LIFECYCLE SCENARIO BASED ON GENETIC ALGORITHM.
DOI: 10.5220/0003671702080213
In Proceedings of the International Conference on Evolutionary Computation Theory and Applications (ECTA-2011), pages 208-213
ISBN: 978-989-8425-83-6
Copyright
c
 2011 SCITEPRESS (Science and Technology Publications, Lda.)

including lifecycle ones at a higher level can be
obtained.
2 INTEGRATED OPTIMIZATION
METHOD
2.1 Overview
This paper proposes an integrated method for
optimizing a product design concept (a functional
structure and a components / parts layout) and its
lifecycle scenario by considering various criteria
including lifecycle ones, based on our previous
method. The improved point is (1) To integrate LCA
in order to evaluate environmental burdens as
additional criteria of the integrated optimization and
(2) To handle decision makings throughout product
lifecycle as additional design variables of the
integrated optimization.
Figure 1 shows the overview of the proposed
integrated optimization method. This method
consists of functional / layout optimizations plus
LCA. Functional optimization is the main part of the
proposed method and executed just one time.
Functional optimization is based on the hierarchical
genetic algorithm (HGA) (Yoshimura and Izui,
2002) in order to consider hierarchical nature of a
functional structure. In the functional optimization,
selection of functions and parts (Parts correspond to
the functions at the bottom level of the functional
structure) from their alternatives and selection of
reuse / recycle / disposal scenario at the EOL stage
for each parts are considered as design variables. As
for criteria, performance, cost, total area and total
carbon emission are considered. Although only one
decision making throughout product lifecycle and
one index of environmental burden are considered in
this paper, the proposed method has a potential to
handle more design variables and criteria. Any of
them can be configured as an objective function and
the rest of them are configured as constraint
conditions. This paper assumes that performance and
cost can be calculated by simply summing up the
values associated with each part, whereas to
calculate total area and total carbon emission, layout
optimization and LCA need to be executed. So, they
are repeatedly invoked from the functional
optimization for evaluating generated design
proposals at its each iteration. In the layout
optimization, based on the information about size of
parts selected in the functional optimization, layout
with minimum area is calculated. Layout
optimization is based on the sequence-pair
representation and the traditional GA with the
special crossover operator (Murata et al., 1996);
(Nakaya et al., 2000). In lifecycle assessment, total
carbon emission is calculated from the selected parts
and lifecycle scenario.
Generation of solutions
Evaluation of solutions
Layout optimization
using GA
S
o
l
u
t
i
o
n
C
a
r
b
o
n
e
m
i
s
s
i
o
n
S
o
l
u
t
i
o
n
C
a
r
b
o
n
e
m
i
s
s
i
o
n
Iteration
•Performance
•Carbon emission
•Cost
•Area
(Selection, crossover, mutation)
Functional optimization using HGA
Optimal solution
Functional structure includin
g
alternatives
•Optimal functional structure
•Optimal layout
•Optimal lifecycle scenario
Lifecycle assessment
S
o
l
u
t
i
o
n
O
p
t
i
m
a
l
l
a
y
o
u
t
A
r
e
a
S
o
l
u
t
i
o
n
O
p
t
i
m
a
l
l
a
y
o
u
t
A
r
e
a
Area
Figure 1: Overview of the integrated optimization.
2.2 Preconditions of the Proposed
Method
Before explaining the details of the proposed method,
the following assumptions are made.
 A components / parts layout is limited to two-
dimensional plane. Shape of components and parts is
limited to rectangle.
 All components and parts can be placed freely
without any connection constraints.
 Total area of the product equals to the area of
rectangle that envelopes all parts placed in a
configuration plane.
 Each part has the values associated with
performance and cost. Their total values can be
calculated by simply summing up these values.
2.3 Functional Optimization using
Hierarchical Genetic Algorithm
In practical product designs, functional structure is
highly hierarchized and decision makings in the
upper level may affect lower functional structures
greatly. To optimize such hierarchical selection
problems, our method adopts HGA.
The most distinctive feature of HGA is
hierarchical genotype representations to exactly
describe hierarchical structures of mechanical
system designs and special operators of crossover
INTEGRATED OPTIMIZATION OF PRODUCT DESIGN CONCEPT AND PRODUCT LIFECYCLE SCENARIO
BASED ON GENETIC ALGORITHM
209

and mutation for manipulating them. Figure 2 shows
an example of a hierarchical design problem.
. . .BA . . .BA
cba ed
A-2A-1
cba ed
A-2A-1
a-2a-1 a-2a-1
b-2b-1 b-2b-1
e-2e-1 e-2e-1
B-3B-2B-1 B-3B-2B-1
Alternatives of A and B
Alternatives of a, b, c, d and e
Upper structure
Sub-structure of
A-1 and A-2
d-3d-2d-1 d-3d-2d-1
c-2c-1 c-2c-1
Figure 2: Hierarchical design problem.
In our method, HGA is adopted to optimize
product’s functional structure in the form of a
hierarchical tree structure. An individual of HGA
corresponds to one design proposal and its organism
strings show selections of functions / parts and
selection of reuse / recycle / disposal scenario at the
EOL stage in that design proposal. The lowest string
corresponds to the combinations of the part selection
and its lifecycle scenario selection. In other word,
the same parts with different lifecycle scenarios are
considered as different alternatives. For example,
part A with disposal scenario and part A with reuse
scenario are different alternatives. The other higher
strings correspond to selection of functions. Since
our method adopts single-objective HGA, any of
performance, cost, total area and total carbon
emission can be configured as an objective function
and the rest are configured as constraint conditions.
As described the previous section, performance and
cost can be calculated by simply summing up the
values associated with each part, whereas total area
and total carbon emissions can not be calculated by
simple summation. Therefore, layout optimization
and LCA, as described in the later sections, are
executed for evaluating each individual in each
generation.
2.4 Layout Optimization using GA and
Sequence-pair Representation
In our method, parts shape and a configuration space
are limited to rectangle and two dimensional, so
components / parts layout can be described by
sequence-pair representation and solved by GA.
Sequence-pair was originally developed for
VLSI layout design, which is the rectangle packing
problem. This method represents relative positions
of rectangles by using a pair of rectangle name
sequences, called 
+
and 
-
. 
+
and 
-
indicate the
rectangle sequences in diagonally right up and
diagonally right down respectively. Figure 3 shows a
layout example, its relative position and its sequence
pair.
b
d
a
e
c
f
a
b
d
f
c
e
0
1
2
3
4
5
0
1
2
3
4
5
(a) Layout (b) Relative position (c) 
-
and 
+

-
= [f, c, d, e, a, b]

+
= [e, c, a, b, f, d]

-

+
Figure 3: Example of layout and its sequence-pair.
When relative positions of rectangles are
described by 
+
and 
-
, the absolute positions of the
rectangles without overlap within minimum area can
be uniquely obtained by making horizontal and
vertical constraint graphs based on 
+
and 
-
and by
calculating longest paths in both graphs. Figure 4
shows their examples. See the reference (Murata et
al., 1996) for the details.
a
b
d
f
c
e
width
a
b
d
f
c
e
width
a
b
d
f
c
e
height
a
b
d
f
c
e
height
Figure 4: Horizontal / Vertical constraint graphs.
The original research (Murata et al., 1996) uses
simulated annealing for searching the optimal layout
with minimum area, whereas, our method uses GA
with the special crossover operator called PREX
(Placement-based Partially Exchanging Crossover)
(Nakaya et al., 2000).
Since components / parts layout of a practical
product is hierarchized, layout optimization is
repeatedly executed from a part level to a product
level to obtain hierarchical components / parts layout.
Figure 5 shows its concept.
A-1
b-2
b-1
a-1
a-2
a-3
A-2
A-3
B-1 B-2
B-3
Other
A
B
A-2
A-3
A B
Parts
Components
Product
Figure 5: Hierarchical layout optimization.
ECTA 2011 - International Conference on Evolutionary Computation Theory and Applications
210

2.5 Lifecycle Assessment
In the practical LCA, there are various indexes of
environmental burden such as emissions of CO
2
,
SO
x
and NO
x
throughout entire product lifecycle,
usage rate of renewable material and reuse / recycle
rate, which are determined by entire product
lifecycle scenario. However, because this research is
at an early stage, the proposed method handles only
selection of reuse / recycle / disposal scenario at the
EOL stage and carbon emission as a design variable
and an index of environmental burden respectively.
Total carbon emission of each design proposal is
calculated by the following concepts.
 Carbon emission is evaluated for each part and
sum of them is defined as total carbon emissions
GHG
total
. GHG is an acronym for Green House Gas.
 There are two types of parts. One has the fixed
value of carbon emission GHG
i
and the other has the
value of carbon emission per unit area uGHG
i
. Most
parts belong to the former type, whereas some parts
such as an electronic substrate belong to the latter
type. In the latter type, actual value of carbon
emission is calculated by multiplying uGHG
i
by the
area Area
i
calculated by layout optimization.
 The value of carbon emission varies with the
selection of reuse / recycle / disposal scenario, so the
value for each scenario needs to be assessed by LCA.
The selection also affects performance and cost of
the part, so the relationship among scenario selection,
carbon emission, cost and performance needs to be
assessed. For example, use of recycled material
increases manufacturing cost in exchange for lower
carbon emission. Reuse of used parts considerably
reduces both carbon emission and cost but
performance of such parts can not be expected. In
particular, in the field where rate of technological
evolution is high, old parts become rapidly obsolete,
so reuse of such parts equals use of low performance
parts. Thus the proposed method is useful in
balancing product eco-friendliness and other criteria
at higher level.
Finally, total carbon emission GHG
total
is calculated
by the below equation.



n
j
jj
m
i
itotal
uGHGAreaGHGGHG
11
(1)
Where GHG
i
is the fixed value of carbon emissions
of part i, whereas, uGHG
j
is the value of carbon
emissions per unit area of part j. Area
j
is the value of
area of part j.
3 CASE STUDY
3.1 Problem Description
In the case study, internal devices of a personal
computer are designed using the proposed method .
“Internal devices” means that input devices, a
display and an enclosure are not included. To make
the case study simpler, only reuse and disposal are
considered as alternatives of the lifecycle scenario at
the EOL stage. Reuse scenario means reuse of the
part used in the previous generation whereas
disposal one means use of the new part.
A computer consists of the following 5
components: motherboard, HDD, cooling system,
power supply and auxiliary storage. Motherboard,
cooling system and power supply can be
decomposed into more than one part, whereas HDD
and auxiliary storage can not be decomposed any
more. Figure 6 shows its functional structure. Note
that, due to space limitation, the lower functional
structure of Motherboard B is not described in this
figure. Motherboard B is similar to Motherboard A,
but has powerful CPU, more Memory and discrete
graphic card. (R) writen after the part name means
that the part is reused one. Table 1 shows examples
of their alternatives and specifications. As for the
new parts, prices and sizes are based on their retail
price surveys and size measurments. Performances
are subjectively and intuitively configured. Carbon
emission is based on the reference (Japan Environm
ental Management Association For Industry, 2007).
As for the reused parts, their specifications are
estimated from ones of the similar or same new parts.
Table 1: Part example (HDD).
HDD
Cost
(USD)
Dimension
(cm)
Perfor
mance
CO
2
(kg)
MK8009GAH 75 5.5*8.0 2 1.31
MK1214GAH 150 5.5*8.0 3 1.31
MK4007GAL (R) 30 5.5*8.0 1 0.78
WD1600BEVT 42 7.0*10.0 4 2.72
WD3200BEVT 66 7.0*10.0 5 2.72
MK8025GAS (R) 21 7.0*10.0 2 1.63
WD5000AAKB 160 10.0*14.5 7 10.88
WD1002FBYS 220 10.0*14.5 10 10.88
6L320R0 (R) 80 10.0*14.5 4 6.53
As for criteria, performance is handled as an
objective function, whereas cost, total area and total
carbon emission are handled as constraint conditions.
Table 2 shows parameters of HGA and GA.
INTEGRATED OPTIMIZATION OF PRODUCT DESIGN CONCEPT AND PRODUCT LIFECYCLE SCENARIO
BASED ON GENETIC ALGORITHM
211

Supply
Power
Read/Write
Outer information
Process
Information
Storag e
Inter information
Battery
Optical drive
Memory card
reader
HDD
Process
Information
Personal ComputerPersonal Computer
DVD combo
DVD burner
BD burner
DVD combo
DVD burner
BD burner
SD memory reader
CF memory reader
SD memory reader
CF memory reader
Motherboard B
Power supply
Discard
Heat
Cooler
Control
Power
Controller
Convert Electric
power to Torque
Motor
Generate
Wind
Fan
80mm Fan
120mm Fan
Process
Information
Store Data
Transfer
Data
Other
functions
Chipset
Atom Z530
Atom Z520
Atom Z540
Atom Z530
Atom Z520
Atom Z540
DDR2-533 1G
DDR2-533 2G A
DDR2-533 2G B
DDR2-533 1G
DDR2-533 2G A
DDR2-533 2G B
Support
communication
Motherboard A
WLAN module A
WLAN module B
Support Wireless
communication
WLAN module
TV tuner A
TV tuner B
Receive TV
TV tuner
Other
functions
Other
devices
Other
devices
Other
devices
Sub-boardDIMMCPU
MK8009GAH
WD1600BEVT
WD5000AAKB
MK8009GAH
WD1600BEVT
WD5000AAKB
MK1214GAH
WD3200BEVT
WD1002FBYS
MK1214GAH
WD3200BEVT
WD1002FBYS
Lithium polymer A,B
Lithium ion A,B
Discard
Heat
Cooling system
Remove
Heat
Convey
Heat
Cooler
Heat pipe
Radiation sheet
Generate
Wind
Motor
Motor FanFan
40mm Fan
60mm Fan
80mm Fan
Convert Electric
power to Torque
Sheet A
Sheet B
SD memory reader (R)
CF memory reader (R)
SD memory reader (R)
CF memory reader (R)
DVD combo (R)
80mm Fan (R)
120mm Fan (R)
Pentium M (R)
DDR2-533 512M (R)
80mm Fan
(
R
)
60mm Fan (R)
40mm Fan (R)
6L320R0 (R)MK8025GAS (R)
WLAN module (R)
TV tuner (R)
DVD burner (R)
DDR2-533 1GB (R)
MK4007GAL (R)
Figure 6: Functional structure designed in the case study.
Table 2: Parameters of HGA and GA.
HGA GA
Population 100 60
Crossover rate 1 1
Mutation rate 0.05 0.01
Generation gap 0.9 0.5
Terminal generation 200 50
3.2 Results
For comparison, both our previous method that
considers neither lifecycle scenario nor carbon
emission and the proposed method are executed here.
Figure 7 shows the results using our previous
method. In this case, 12 optimizations ars executed
under 12 various cost constraints from 550 USD to
2550 USD and constant area constraint (Area < 1200
cm
2
). Note that reused parts are not available in this
case. The optimal layouts of the design solutions
denoted by two stars in Figure 7 are shown in Figure
8.
0
20
40
60
80
100
120
500 1000 1500 2000 2500 3000
Cost
Performance
Figure 7: Relationships between performance and cost of
the obtained solutions using our previous method.
Battery
Cooling
system
HDD
Card reader
Motherboard
240mm
174mm
Battery
Cooling
system
HDD
Card reader
Motherboard
240mm
174mm
435mm
240mm
Cooling
system
Motherboard
HDD
Optical
drive
Power supply
435mm
240mm
Cooling
system
Motherboard
HDD
Optical
drive
Power supply
Figure 8: Optimal layouts (Left: star 1, Right: star 2).
Figure 9 shows the result using the proposed
method. In this case, 16 optimizations are executed
under 16 various constraints of carbon emission
from 7.5 kg to 45 kg and constant const and area
constraints (Cost < 3000 USD and Area < 1500 cm
2
).
This results shows that the constraint of carbon
emission makes it difficult to design high
performance PCs even if the constraints of cost and
area are sufficiently relaxed. The design solusions
with higher performance tend to consist of only new
parts whereas the design solutions with lower
performance tend to consists of many reused parts.
From the other point of view, the parts that are
making rapid progress such as CPUs and HDD are
infrequently reused, whereas the parts that are not
making progress such as auxiliary storages and
cooling fans are frequently reused. Use of reused
parts can reduce carbon emission, but at the same
time makes it difficult to achieve high performance.
Thus the proposed method is useful in balancing
product eco-friendliness and other criteria at a higher
level.
ECTA 2011 - International Conference on Evolutionary Computation Theory and Applications
212

0
20
40
60
80
100
120
0 1020304050
Carbon emission
Performance
Figure 9: Relationships between performance and cost of
the obtained solutions using the proposed method.
4 CONCLUSIONS
Due to rise of environmental awareness in recent
years, companies are required to assess and reduce
environmental burdens of their products. In response
to this, this paper proposes a new integrated method
for simultaneously optimizing a product concept and
its lifecycle scenario by evaluating product
characteristics such as performance, cost and size
and environmental burdens such as carbon emission
as criteria. Using the proposed method, optimal
product concept and its lifecycle scenario that
balance product characteristics and environmental
burdens at a higher level can be obtained. In the case
study, the proposed method is applied to a design of
a personal computer and the results show the needs
of consideration of lifecycle scenario and
environmental burdens during conceptual design
phase.
As for future works, we are planning to consider
modularization of parts and components. In recent
products, components and parts are modularized due
to various reasons. Modularization also affects
lifecycle characteristics of the product. Therefore, to
allow optimal modular design for product lifecycle,
we are planning to expand our method proposed in
this paper.
ACKNOWLEDGEMENTS
This study was supported in part by a grant of
Strategic Research Foundation Grant-aided Project
for Private Universities from Ministry of Education,
Culture, Sport, Science, and Technology, Japan
(MEXT), 2008-2012 (S0801058).
REFERENCES
Kobayashi, M., Suzuki, Y. and Higashi, M. (2009).
Integrated Optimization for Supporting Functional and
Layout Designs during Conceptual Design Phase.
Proceedings from IDETC/CIE 2009: ASME 2009
International Design Engineering Technical
Conferences & Computers and Information in
Engineering Conference, San Diego, CA.
Yoshimura, M. and Izui, K. (2002). Smart Optimization of
Machine Systems Using Hierarchical Genotype
Representations. Transaction of ASME, Journal of
Mechanical Design, 124, 375-384.
Murata, H., Fujiyoshi, K., Nakatake, S. and Kajitani, Y.
(1996). VLSI Module Placement Based on Rectangle-
Packing by the Sequence-Pair. IEEE Transactions on
Computer-Aided Design of Integrated Circuits and
Systems, 15(12), 1518-1524.
Nakaya, S., Wakabayashi, S. and Koide, T. (2000). An
adaptive genetic algorithm for VLSI floorplanning
based on sequence-pair. Proceedings from
ISCAS2000: 2000 IEEE International Symposium on
Circuits and Systems, Geneva, SWITZERLAND.
Japan Environmental Management Association for
Industry (2007). Implementation manual of product
LCA.
INTEGRATED OPTIMIZATION OF PRODUCT DESIGN CONCEPT AND PRODUCT LIFECYCLE SCENARIO
BASED ON GENETIC ALGORITHM
213