ArticlePDF Available

Incorporating tolerances of customers’ requirements for customized products

Authors:
  • The Hang Seng University of Hong Kong

Abstract and Figures

Tolerancing conventionally deals with the variation of manufacturing processes to meet the requirements of product quality. With the development in product customization, it has been generally accepted that customers’ requirements also have acceptable tolerance range. This flexibility provides an opportunity to better match customer requirements with richer product offerings through customization. A probabilistic model is presented in this paper to incorporate the tolerance of functional requirements into customized product design. By leveraging on the requirements tolerance, customers are more likely to get their desired products. Customer requirements tolerance can thus open up a new dimension for product customization.
Content may be subject to copyright.
Incorporating
tolerances
of
customers’
requirements
for
customized
products
Yue
Wang,
Mitchell
Tseng
(1)*
Advanced
Manufacturing
Institute,
The
Hong
Kong
University
of
Science
and
Technology,
Hong
Kong
1.
Introduction
Tolerancing
is
an
important
research
topic
in
mechanical
engineering
dealing
with
accumulated
variations
in
manufacturing
processes
and
production
systems
for
meeting
the
requirements
of
product
quality.
The
study
of
tolerancing
aims
to
enable
machine
parts
or
components
to
be
precisely
made
and
interchangeably
assembled.
In
this
way
mass
production
efficiency
can
be
improved
and
at
the
same
time
manufacturing
cost
and
product
quality
can
be
controlled.
The
scope
of
tolerancing
has
been
focused
on
tolerance
analysis
and
the
tolerance
synthesis.
Basically,
tolerance
analysis
is
concerned
with
the
tolerance
stack-up
given
each
individual
component
tolerance.
The
tolerance
synthesis
tries
to
identify
each
individual
component
tolerance
given
the
tolerance
of
the
final
product.
Therefore
traditional
tolerancing
analysis
mainly
addresses
the
issues
of
variations
in
product
manufacturing
and
production.
With
recent
development
of
product
customization
and
person-
alization,
it
has
been
generally
accepted
that
customers’
functional
requirements
have
certain
acceptance
range
or
tolerance
range
[1].
Great
potential
can
be
gained
by
extending
the
scope
of
tolerance
analysis
in
production
and
manufacturing
to
product
requirements
definition
for
customized
product
development.
Product
development
usually
starts
with
the
process
of
capturing
customer
needs
and
then
transforms
them
into
tangible
product
variants
or
configurations.
It
attempts
to
meet
the
needs
of
customers
economically
and
timely.
The
key
to
product
success
relies
on
better
understanding
of
the
customer
needs
and
their
functional
requirements
with
respect
to
the
product
[2].
Research
in
mass
customization
shows
that
customers
are
often
willing
to
compromise
among
certain
product
attributes
[3].
Namely,
they
are
indifferent
or
insensitive
to
some
ranges
of
attributes
and
certain
product
features
[4].
In
this
paper,
this
behaviour
is
considered
as
customer
functional
requirements
tolerance.
Actu-
ally
the
tolerance
of
customer
needs
or
requirements
is
an
inherent
property
of
customer’s
decision
making
process.
Customers
often
adopt
a
‘‘satisficing
(satisfy
+
suffice)’’
strategy
instead
of
trying
to
find
the
optimal
product
variation
[4].
Thus
they
may
appreciate
a
set
of
product
attribute
alternatives,
exhibiting
the
tolerance
and
adaptability
in
the
product
requirements.
For
example,
it
is
reported
in
a
survey
that
22%
of
new
car
buyers
in
the
UK
accepted
changes
from
their
original
functional
requirements
and
detailed
specifications
in
year
2000
and
2001
[5].
However,
quite
a
few
challenges
exist
to
integrate
functional
requirements
tolerance
to
product
design,
particularly
custom-
ized
product
design.
From
customers’
side,
it
is
difficult
to
characterize
and
quantify
the
tolerance
of
functional
require-
ments.
Functional
requirements
are
often
vague
and
subjective.
They
are
hard
to
express
in
a
quantitative
form
objectively.
Traditional
methodologies
on
customer
needs
expression
do
not
consider
the
tolerances.
In
addition,
the
requirements
tolerance
is
subjective
by
nature.
It
depends
on
the
users
who
examine
the
product.
The
range
and
response
of
tolerance
degree
may
differ
greatly
from
person
to
person
[6].
All
these
factors
obstruct
the
integration
of
customer
requirement
tolerance
into
product
customization
practice.
To
address
the
challenges,
this
paper
attempts
to
discover
new
ways
to
characterize
and
incorporate
functional
requirements
tolerance
in
early
product
design
stage
and
to
find
the
mapping
between
customer
functional
requirements
tolerance
and
the
corresponding
product
configurations
in
a
product
family.
2.
Characterizing
the
tolerance
of
functional
requirements
Design
is
concerned
with
how
the
design
parameters
(DPs)
satisfy
a
customer’s
functional
requirements
(FRs)
[7].
Design
can
thus
be
restated
as
how
FRs
map
to
DPs
as
shown
in
Fig.
1.
CIRP
Annals
-
Manufacturing
Technology
63
(2014)
129–132
A
R
T
I
C
L
E
I
N
F
O
Keywords:
Design
Customization
Tolerance
A
B
S
T
R
A
C
T
Tolerancing
conventionally
deals
with
the
variation
of
manufacturing
processes
to
meet
the
requirements
of
product
quality.
With
the
development
in
product
customization,
it
has
been
generally
accepted
that
customers’
requirements
also
have
acceptable
tolerance
range.
This
flexibility
provides
an
opportunity
to
better
match
customer
requirements
with
richer
product
offerings
through
customization.
A
probabilistic
model
is
presented
in
this
paper
to
incorporate
the
tolerance
of
functional
requirements
into
customized
product
design.
By
leveraging
on
the
requirements
tolerance,
customers
are
more
likely
to
get
their
desired
products.
Customer
requirements
tolerance
can
thus
open
up
a
new
dimension
for
product
customization.
ß
2014
CIRP.
*
Corresponding
author.
Contents
lists
available
at
ScienceDirect
CIRP
Annals
-
Manufacturing
Technology
journal
homepage:
http://ees.elsevier.com/cirp/default.asp
http://dx.doi.org/10.1016/j.cirp.2014.03.091
0007-8506/ß
2014
CIRP.
Without
considering
FR
tolerance,
the
mapping
is
from
a
point
in
FR
space
to
a
small
DP
region.
With
FR
tolerance,
the
corresponding
DP
region
in
DP
space
can
be
much
larger,
meaning
more
flexibility
and
benefit
can
be
achieved
by
the
reduction
of
product
development
cost
and
the
utilization
of
manufacturing
resources
and
supply
chain.
From
this
perspective,
functional
requirements
tolerance
is
considered
as
the
customer’s
indifference
or
flexibility
to
certain
product
attributes
such
that
any
product
configuration
(combina-
tion
of
different
product
attribute
levels)
within
the
customer
acceptable
range
would
have
little
impact
on
customer
satisfaction.
Following
this
definition,
customer
requirement
tolerance
can
be
characterized
through
two
dimensions:
range
and
response
[1].
The
range
of
customer
requirements
tolerance
indicates
the
total
range
of
the
customer’s
desired
or
satisfactory
choices.
This
mainly
includes
an
acceptable
range
of
product
variety
or
configurations
offered
by
the
manufacturer.
Any
choices
beyond
this
range
are
unacceptable.
For
example,
considering
a
case
of
buying
a
laptop
with
requirement
for
Hard
Disk
being
at
least
500
GB,
then
any
laptop
with
larger
than
500
GB
disk
is
within
the
functional
requirement
tolerance
range.
Other
laptop
models
with
less
than
500
GB
Hard
Disk
are
not
acceptable.
The
response
of
customer
requirement
tolerance
measures
how
much
a
customer
values
different
attribute
choices
within
the
acceptable
range
of
the
attribute.
The
customer
may
perceive
different
values
or
satisfaction
levels
to
different
product
attribute
choices.
Some
choices
are
highly
preferred
or
desired.
Others
may
be
just
satisfactory.
For
the
abovementioned
laptop
example,
500
GB
and
1TB
Hard
Disk
are
both
in
the
tolerance
range.
But
1TB
hard
disk
is
more
preferred
by
the
customer.
The
response
of
requirement
tolerance
means
much
to
manufacturers
because
it
helps
manufacturers
capture
customers’
satisfaction
level
to
each
product
variant.
Most
importantly,
the
response
makes
it
easier
for
manufacturing
companies
to
evaluate
the
overall
satisfaction
level
for
a
product
configuration
or
variant
by
combining
the
satisfaction
to
each
individual
attribute.
3.
Problem
definition
and
preliminaries
The
context
of
this
paper
is
product
customization
via
product
configuration
searching
in
the
whole
solution
space
[9].
It
means
that
the
product
consists
of
a
set
of
predefined
attributes
and
the
corresponding
attribute
choices.
The
customization
task
is
to
explore
the
whole
configuration
space
and
find
the
optimal
combination
of
attribute
alternatives
to
satisfy
the
customer’s
requirement
given
the
customer’s
requirements.
This
mechanism
has
been
applied
widely
in
industry.
Many
current
online
configuration
systems
have
adopted
this
idea
to
realize
product
customization,
like
Dell
computer’s
online
component
selection
system.
Basically,
the
problem
can
be
modelled
as
a
combinatorial
optimization
problem,
namely,
finding
a
product
configuration
from
the
whole
products
space
to
satisfy
the
customer’s
requirement
by
leveraging
on
customer
requirements
tolerance.
However,
customer
requirements
are
usually
not
complete
and
contain
a
lot
of
uncertainty
during
requirements
elicitation
stage
of
product
design.
We
need
to
present
the
product
configuration
to
meet
the
customer’s
needs
without
knowing
his/her
entire
functional
requirements
and
the
corresponding
tolerance.
We
apply
a
data-driven
method
to
characterize
and
quantify
customer
requirement
tolerance.
We
assume
a
collection
of
previous
customers’
product
selection
data
is
available.
When
a
new
customer
inputs
certain
number
of
requirement
specifications
of
the
product,
we
try
to
calculate
the
likelihood
that
each
end
product
will
meet
the
customer’s
requirements.
The
likelihood
is
called
as
likelihood
of
acceptance
throughout
this
paper.
A
critical
point
is
that
although
one
attribute
is
not
specified
by
the
customer,
it
does
not
mean
that
the
attribute
is
unsatisfactory
due
to
requirement
tolerance.
The
satisfaction
degree
will
be
quanti-
fied
by
the
likelihood
of
acceptance.
It
is
a
subjective
measure
and
is
dependent
on
customers.
For
example,
two
customers
give
the
same
specification
of
4
GB
RAM
to
a
PC.
However
they
may
have
different
tolerance
degrees
to
the
alternative
of
two
combined
2
GB
RAMs.
By
calculating
likelihood
of
acceptance,
the
subjective
tolerance
is
quantified
and
integrated
in
the
configuration
searching
process.
Definition
1.
Let
C
¼
ðA
1
!
;
A
2
!
;
.
.
.
;
A
n
!
Þ
represent
a
product
variant
or
configuration
where
A
i
!
¼
ða
i1
;
a
i2
;
.
.
.
;
a
in
i
Þ
is
the
ith
attribute
choices
vector.
a
i
j
is
an
indicator
function
for
the
ith
attribute.
a
i
j
¼
1
indicates
the
existence
of
the
jth
alternative
in
the
product
variant
and
value
0
otherwise.
In
this
way,
there
is
one
and
only
one
alternative
of
an
attribute
having
value
a
i
j
¼
1
in
one
product
configuration,
i.e., P
j
a
i
j
¼
1
for
any
i.
Therefore
each
product
configuration
can
be
represented
by
a
vector
of
binary
variables
in
the
form
of
C
¼
ða
11
;
.
.
.
;
a
1n
1
;
.
.
.
;
a
n1
;
.
.
.
;
a
nn
n
Þ
with
constraint P
j
a
i
j
¼
1
for
any
i.
Similarly,
the
specification
can
also
be
represented
in
the
same
way.
Definition
2.
Let
I
R
ðC;
SÞ
be
a
relevance
indicator
function
where
S
is
a
customer’s
specification
of
the
product,
i.e.
the
customer’s
requirement.
If
a
product
configuration
C
can
satisfy
the
customer’s
requirement,
I
R
ðC;
SÞ
¼
1.
Otherwise,
I
R
ðC;
SÞ
¼
0.
It
should
be
noted
that
designers
know
the
value
of
I
R
ðC;
SÞ
only
if
the
customer’s
direct
feedback
of
satisfaction
to
the
product
configuration
S
is
obtained.
However,
designers
may
have
certain
belief
or
estimation
about
the
value
of
I
R
ðC;
SÞ
based
on
prior
knowledge
before
the
customer
gives
explicit
response
to
the
product
variant.
The
belief
function
will
be
represented
by
the
following
definition.
Definition
3.
Let
LðC;
SÞ
be
the
likelihood
of
acceptance
function
from
designers’
point
of
view.
It
represents
designers’
belief
or
estimation
that
a
product
configuration
C
can
satisfy
a
customer’s
requirement
S
without
knowing
the
customer’s
explicit
feedbacks
to
the
product
configuration.
4.
Incorporating
the
tolerance
of
requirements
in
product
customization
To
implement
the
data-driven
method,
we
need
previous
customers’
preferences
and
final
selection
data.
Here
we
assume
customer
preferences
are
expressed
in
detailed
product
specifica-
tion
level,
i.e.,
the
specification
of
each
attribute
or
component
choices.
But
customers
are
not
required
to
specify
the
tolerance
to
Fig.
1.
The
mapping
from
FR
to
DP:
(a)
without
requirement
tolerance;
(b)
with
requirement
tolerance.
Y.
Wang,
M.
Tseng
/
CIRP
Annals
-
Manufacturing
Technology
63
(2014)
129–132
130
attributes
explicitly.
We
try
to
estimate
the
tolerance
in
the
form
of
likelihood
of
acceptance
and
integrate
it
into
design.
Specifically,
a
specification–configuration
(SC)
data
pair
is
associated
with
each
customer
where
S
and
C
are
defined
in
Definitions
1
and
2.
The
specification
is
assumed
to
be
incomplete
in
this
paper
because
presenting
product
configuration
recommendation
to
a
customer
given
his/her
complete
product
specifications
is
a
trivial
and
easy
to
solve.
Example:
A
product
consists
of
three
components
X,
Y
and
Z
with
the
choice
set
ffX1;
X2g;
fY1;
Y2g;
fZ1;
Z2gg
for
each
compo-
nent.
The
choice
set
defines
the
whole
solution
space
of
product
customization.
Suppose
a
customer’s
requirement
corresponds
to
the
specification
fX1;
Y2g.
Then
the
configuration
searching
will
find
the
satisfactory
product
variant
by
calculating
the
likelihood
of
acceptance
function
LðC;
fX1;
Y2
where
C
can
be
any
product
configuration
such
as
fX2;
Y2;
Z2g
or
others.
Although
X2
is
not
specified
by
the
customer
in
this
example,
it
is
still
potentially
acceptable
due
to
the
tolerance
of
customer
requirements.
We
cannot
rule
out
the
possibility
that
the
configuration
of
fX2;
Y2;
Z2g
is
an
acceptable
product.
The
situation
in
the
example
is
quite
common
especially
when
a
customer
does
not
have
enough
expertise
to
select
product
attributes.
The
specifications
may
not
be
self-consistent.
If
the
tolerance
of
customer
requirements
is
not
considered
and
the
final
product
variant
must
consist
of
the
specified
attributes,
it
is
very
likely
that
there
is
no
satisfactory
end
product
existing
in
the
choice
set.
To
solve
the
issue,
we
need
to
integrate
customer
requirements
tolerance
into
the
customization
process.
Therefore
we
apply
likelihood
of
acceptance
to
quantify
the
possibility
that
unspecified
attribute
alternatives
are
also
acceptable.
The
method
calculates
the
likelihood
or
probability
that
a
product
will
meet
an
active
customer’s
specification
based
on
the
incomplete
or
even
contradicted
specifications
from
customers.
To
get
the
likelihood
of
acceptance,
let
us
start
with
the
probability
that
a
configuration
can
satisfy
a
customer’s
require-
ment,
i.e.
PðI
R
jC;
SÞ.
Applying
Bayes’
rule,
PðI
R
jC;
SÞ
¼
PðI
R
jSÞPðCjI
R
;SÞ
PðCjSÞ
.
To
simplify
the
analysis,
we
estimate
the
odds
instead
of
the
probabilities
directly,
i.e.,
OðxÞ
¼
PðxÞ
PðxÞ
¼
PðxÞ
1PðxÞ
.
It
is
easy
to
see
that
OðxÞ
is
strictly
monotonic
with
respect
to
PðxÞ.
Therefore
it
is
equivalent
to
present
the
satisfactory
product
based
on
odds
function.
Then
we
have
OðI
R
¼
1jC;
SÞ
¼PðI
R
¼
1jC;
SÞ
PðI
R
¼
0jC;
SÞ¼PðI
R
¼
1jSÞPðCjI
R
¼
1;
SÞ
PðI
R
¼
0jSÞPðCjI
R
¼
0;
SÞ(1)
Here
we
assume
the
linked
independence
to
simplify
the
calculation.
It
states
that
occurrence
of
different
components
are
independent,
i.e.,
PðCjI
R
¼
1;
SÞ
PðCjI
R
¼
0;
SÞ¼Y
n
i¼1
Pða
i
jI
R
¼
1;
SÞ
Pða
i
jI
R
¼
0;
SÞ(2)
This
assumption
has
been
widely
used
in
marketing
literature
to
strike
a
balance
between
computational
efficiency
and
the
effectiveness
of
algorithms
[8,9].
Let
p
i
¼
Pða
i
¼
1jI
R
¼
1;
SÞ
and
q
i
¼
Pða
i
¼
1jI
R
¼
0;
SÞ.
Then
we
get
PðCjI
R
¼
1;
SÞ
¼Y
n
i¼1
p
a
i
i
ð1
p
i
Þ
1a
i
and
PðCjI
R
¼
0;
SÞ
¼Y
n
i¼1
q
a
i
i
ð1
q
i
Þ
1a
i
.
Combining
them
with
Eq.
(1),
we
have
OðI
R
¼
1jC;
SÞ
¼
OðI
¼
1jSÞQ
i
p
ai
i
ð1
p
i
Þ
1ai
q
ai
i
ð1q
i
Þ
1ai
.
By
taking
loga-
rithm
on
both
sides
of
this
equation,
the
following
can
be
achieved
log
OðI
R
¼
1jC;
SÞ
¼X
i
log p
i
ð1
q
i
Þ
q
i
ð1
p
i
Þ
a
i
þX
i
log 1
p
i
1
q
i
þ
log
OðI
R
jSÞ
Since
the
second
and
the
third
terms
do
not
involve
a
i
,
they
are
identical
for
any
product
configuration.
As
a
result
they
do
not
have
any
effect
on
the
likelihood
of
acceptance.
We
only
consider
the
first
term
when
presenting
customization
choices
by
leveraging
on
customer
functional
requirements
tolerance.
Thus
the
likelihood
of
acceptance
can
be
expressed
as
a
linear
function
of
a
i
,
LðC;
SÞ
¼X
i
log p
i
ð1
q
i
Þ
q
i
ð1
p
i
Þ
a
i
(3)
The
parameter
of
p
i
and
q
i
can
be
estimated
by
applying
maximum
likelihood
estimation
(MLE).
Suppose
there
are
N
product
specifications
and
n
products
specifications
with
the
same
partial
specification
S
in
the
data
set.
Let
N
i
stand
for
the
number
of
final
product
specifications
offering
attribute
a
i
.
Among
the
N
i
final
product
configurations,
n
i
include
the
specification
S.
Then
we
can
yield
p
i
¼
n
i
n
and
q
i
¼
N
i
n
i
Nn
according
to
MLE
[10,11].
5.
Case
study
Here
we
use
a
simplified
personal
computer
example
to
test
the
viability
of
the
proposed
approach.
Table
1
lists
part
of
the
components
of
the
PC
and
their
alternatives.
The
data
used
in
this
example
were
obtained
from
a
survey
conducted
to
69
university
students.
Each
respondent
was
required
to
give
his/her
specifications
to
a
PC
and
then
selected
a
desired
one
from
a
predefined
product
offerings
list.
These
data
will
be
used
to
estimate
the
parameters
in
configuration
searching
process
by
applying
Eq.
(3).
To
handle
the
data
sparsity
issue,
we
generate
1380
sample
data
by
perturbative
bootstrap
method
[12].
Bootstrap
is
a
powerful
tool
for
data
resampling
in
statistics.
The
method
generates
new
data
samples
based
on
an
existing
data
set.
Each
new
sample
is
obtained
by
random
sampling
with
replacement
from
the
original
data
set.
Perturbative
bootstrap
allows
deviation
from
original
data
during
resampling
process
and
can
thus
introduce
variety
to
the
data.
For
each
component
alternative,
we
identify
its
potential
substitutable
alternatives
according
to
the
performance,
price,
and
other
properties.
For
example,
A2
(2.66G
processor)
and
A3
(2.8G
processor)
are
considered
as
substitutable
in
this
case.
A
pre-set
threshold
h
belonging
to
[0,1]
is
used
to
control
the
data
generation
for
requirement
tolerance.
To
generate
the
first
component
in
the
new
product
configuration,
a
random
number
u
in
[0,1]
is
generated
first.
If
h
>
u,
the
alternative
in
the
new
data
will
change
to
its
substitute
with
a
predefined
probability
p
which
is
set
to
be
0.5
in
this
paper.
Otherwise,
the
alternative
will
remain
unchanged
in
the
new
data.
The
same
process
is
applied
to
the
remaining
components
one
by
one
until
the
whole
new
specification
data
is
generated.
Thus
h
can
be
used
to
control
the
requirement
tolerance
degree
in
the
generated
data.
Bigger
h
means
the
generated
‘‘customer
requirement’’
is
more
tolerant.
By
applying
this
procedure,
sufficient
training
and
testing
data
can
be
obtained.
The
newly
generated
data
are
used
as
training
data
to
estimate
the
parameters
of
p
i
and
q
i
according
to
MLE
described
above.
Table
1
Component
list
of
a
PC.
Component
Code
Descriptions
Processor
(A)
A1
Intel(R)
Core(TM)2
Duo
3.16G
A2
Intel(R)
Core(TM)2
Duo
2.66G
A3
Intel(R)
Core(TM)2
Duo
2.8G
.
.
.
.
.
.
.
.
.
Memory
(B)
B1
4GB*DDR2
dual
channel
.
.
.
.
.
.
.
.
.
Display
Card
(F)
F1
Intel
1
Graphics
Accelerator
3100
F2
512MB
GeForce(R)
9800GT
.
.
.
.
.
.
.
.
.
Y.
Wang,
M.
Tseng
/
CIRP
Annals
-
Manufacturing
Technology
63
(2014)
129–132
131
In
the
testing,
we
set
h
to
be
from
0.05
to
0.85
with
step
size
0.2
and
generated
5
sets
of
testing
data.
Each
set
contains
207
testing
data
and
corresponds
to
different
degree
of
requirement
tolerance.
By
following
the
procedure
depicted
in
Fig.
2,
each
time
a
new
requirement
is
captured,
the
corresponding
likelihood
of
accep-
tance
is
calculated
based
on
Eq.
(3).
Some
potentially
accepted
product
configurations
are
presented
for
the
customer
to
screen.
The
average
number
of
specifications
that
customers
need
to
input
to
get
their
satisfactory
product
configurations
is
used
as
the
measure
of
performance.
Fig.
3
shows
the
experiment
result
from
different
testing
sets
under
the
condition
that
one
or
three
recommendations
are
presented.
Since
there
are
six
components
in
a
PC
as
shown
in
Table
1,
the
average
number
of
specifications
needed
from
customers
is
upper
bounded
by
six.
From
the
experiment
result
we
found
that
larger
degree
of
customer
requirements
tolerance
corresponds
to
fewer
inputted
specifica-
tions
from
customers.
It
means
customers
are
less
burdened
in
the
product
configuration
searching
procedure.
As
a
result,
customers
are
more
comfortable
in
the
customization
process
[12].
In
addition,
if
more
product
configurations
are
presented
for
the
customer
to
screen,
the
required
input
from
customers
also
decreases.
The
proposed
method
is
a
data-driven
approach.
To
make
it
work
better,
sufficient
amount
of
training
data
are
needed
to
estimate
the
required
parameters.
It
is
still
a
challenge
to
determine
the
number
of
training
data
for
each
customized
design
problem.
In
the
next
step,
we
will
explore
the
relationship
between
the
data
size
and
the
complexity
of
the
product
customization
problem.
6.
Summary
Customer
requirements
exhibit
certain
degree
of
tolerance,
meaning
that
customers
are
usually
indifferent
to
certain
product
attribute
choices.
This
flexibility
provides
an
opportunity
to
better
match
customer
requirements
with
richer
product
offerings
through
customization.
This
paper
presents
a
probabilistic
framework
to
leverage
on
the
tolerance
to
facilitate
product
customization.
It
operates
in
a
product
configuration
context
in
which
tolerance
of
customer
requirements
are
mapped
into
the
combinations
of
attribute
choices.
Likelihood
of
acceptance
for
each
end
product
is
calculated
to
select
the
product
configuration
that
is
within
the
range
of
customer
acceptance.
Case
study
shows
that
by
considering
the
tolerance
of
customer
preferences,
customers
only
need
to
input
part
of
the
specifications
to
get
the
satisfactory
product.
It
means
they
are
less
burdened
in
product
customization
process.
References
[1]
Zhang
Q,
Tseng
MM
(2009)
Modeling
and
Integration
of
Customer
Flexibility
in
Order
Commitment
Process
for
High
Mix
Low
Volume
Production.
Interna-
tional
Journal
of
Production
Research
47(22):6397–6416.
[2]
Daaboul
J,
Da
Cunha
C,
Bernard
A,
Laroche
F
(2011)
Design
for
Mass
Customi-
zation:
Product
Variety
vs.
Process
Variety.
CIRP
Annals
Manufacturing
Technology
60(1):169–174.
[3]
Wang
Y,
Tseng
MM
(2011)
Integrating
Comprehensive
Customer
Require-
ments
into
Product
Design.
CIRP
Annals
Manufacturing
Technology
60(1):175–178.
[4]
Simon
HA
(1996)
The
Sciences
of
the
Artificial,
MIT
Press,
Cambridge,
MA.
[5]
Elias
S
(2002)
New
Car
Buyer
Behaviour.
3DayCar
Research
Report,
Cardiff
Business
School.
[6]
Wang
Y,
Tseng
MM
(2013)
Customized
Products
Recommendation
Based
on
Probabilistic
Relevance
Model.
Journal
of
Intelligent
Manufacturing
24(5):951–960.
[7]
Suh
NP
(2001)
Axiomatic
Design:
Advances
and
Applications,
Oxford
University
Press,
New
York.
[8]
Kamakura
WA,
Wedel
M
(1995)
Life-Style
Segmentation
with
Tailored
Inter-
viewing.
Journal
of
Marketing
Research
32(August):308–317.
[9]
Wang
Y,
Tseng
MM
(2014)
Identifying
Emerging
Customer
Requirements
in
Early
Design
Stage
by
Applying
Bayes
Factor
Based
Sequential
Analysis.
IEEE
Transactions
on
Engineering
Management
61(1):129–137.
[10]
van
Rijsbergen
CJ
(1979)
Information
Retrieval,
Butterworths
Press,
London.
[11]
Crestani
F,
Lalmas
M,
van
Rijsbergen
CJ,
Campbell
I
(1998)
‘Is
this
document
relevant?.
.
.probably’:
A
Survey
of
Probabilistic
Models
in
Information
Retrieval.
ACM
Computing
Surveys
30(4):528–552.
[12]
Wang
Y,
Tseng
MM
(2011)
Adaptive
Attribute
Selection
for
Configurator
Design
via
Shapley
Value.
Artificial
Intelligence
for
Engineering
Design
Analysis
and
Manufacturing
25(1):189–199.
Expli
cate
functional
requirement
Confirm t
he
specifications
Provide cu
stomized
product
End product
Customer Product develo
pme
nt team
Capture customer
functional
requirement
Satisfi
ed with the
configurat
ion?
Y
N
Start
Process flow
Info
rmation flow
Knowledge base of
customer
functional
requirement
tolerance
Calculate the
likelihood of
acceptance for each
end product
Present product
configurat
ions
Fig.
2.
Process
of
product
customization
by
leveraging
on
tolerance
of
functional
requirements.
Fig.
3.
Experiment
result
with
different
testing
data
sets.
Y.
Wang,
M.
Tseng
/
CIRP
Annals
-
Manufacturing
Technology
63
(2014)
129–132
132
... Characterizing the tolerance of customer's requirements starts with understanding customers' behaviour to extract their preferences. In view of the prior work related to customer tolerance in product design, functional requirements tolerance is considered as the customer's indifference or flexibility for certain product attributes such that product configuration (combination of different product attribute levels) within the customer acceptable range that would have little impact on customer satisfaction [11] or the customer's willingness to make trade-offs in buying decisions on these attributes [6]. The alternative view in the supply chain is that a customer is willing to compromise his favourite item of product specifications [12] or often compromise the price and delivery date [13,14]. ...
... The basic concept of QFD is to translate the customers' needs (CNs) into design parameters (DPs) or engineering characteristics, and subsequently into production requirements related to its manufacture. Thus, the relationship between the customer's requirements and companies' capacities is often represented as a diagram called House of Quality [11]. The specific customer requirements are weighed in terms of importance, the intensity of the interactions is input into the matrix, and the result is a rating of importance and target specification for the product requirements. ...
... Wang and Tseng presented a probabilistic framework to incorporate the tolerance of functional requirements into customized product design [11]. It operates in a product configuration context in which tolerance of customer requirements is mapped into the combinations of attribute choices. ...
Article
Full-text available
The tolerance of design specification has been well accepted as an established quantity in design and manufacturing. However, the tolerance of customers’ requirements has not received comparable amount of attention in academic and industry. The possibility of providing different alternatives through customers’ tolerance creates economic value. With the emerging abundance of data from customers and providers, applying tolerances for economic advantage becomes possible. This paper presents a survey of recent works in describing requirement tolerances. Comparing the pros and cons of different approaches is also reported.
... However, research evaluating the effectiveness of configuration systems is still in its infancy. Work in this area has focused mainly on helping company managers to handle inhouse tasks such as reducing the costs of design and production, increasing sales or reducing the demand for product exchange [7]. So far, few efforts have been made to address the effectiveness of configurators from the customers' point of view, or to measure the customers' satisfaction with the configuring process or with the configured products. ...
Conference Paper
Full-text available
Product configurators have been important enablers of customised product design. In a configuration process, customers need to specify the attribute choices that best satisfy their needs. A combination of customer decisions then informs the desired product. Psychology and information retrieval researchers have acknowledged that time is an important factor in measuring customer satisfaction with decision-making process. Some recent studies have also shown that the subjective perception of time is an even more relevant measure of customer satisfaction. However, it remains unclear what factors are most significant in affecting a person's subjective perception of time. This study investigates the relationship between people's experience with the on-line configuration process and their subjective perceptions of the time involved. Empirical experiments are designed to answer the relevant research questions. The experiment's results show that two factors, namely the difficulty of the task and the customers' motivation to process information, both significantly affect customers' perceptions of time. We also find that customers' perceptions of time are significantly correlated with their satisfaction with the configured products, and that perceived time is moderately correlated with satisfaction with the configuration process.
... Thus better understanding of customer needs and the efficient mapping from customer needs to product design specifications can greatly contribute to shortening product development duration. In a typical design process, the recipients of products are invited to express their needs, and the designers seek to transfer those needs into tangible product specifications [5]. Extensive research has been conducted to characterize, understand, and elicit customer needs for product design. ...
Article
Full-text available
Although product design has been considered as a collaborative activity, progress in developing methods to facilitate the design process has been hampered by difficulties in building common bases among various stakeholders. Customers may only have general needs of the product in layman's terms instead of the sufficient domain knowledge to identify the product specifications. Thus, there is a great need in design research to translate the expressed needs in natural language to design specifications and bridge this semantic gap. By leveraging the massive amount of online user-generated content, we develop a deep learning-based method to automatically identify product design parameters or product specifications. Specifically, we crawl product review data and the corresponding product metadata from e-commerce websites. A convolutional neural network based solution is provided to map product reviews to product specifications. Experimental results show that the method can be well adapted to the mapping from customer needs in the natural language to product specifications. The method facilitates product design by addressing the semantic gap between general customer needs and detailed product specifications. In addition, the automated mapping greatly improves the efficiency and reduces labor in the design process.
... Li et al. [7] propose an approach unifying rough set methodology with QFD for modeling relationship measures in the process of product planning. Many variants of CR modeling methodologies have been proposed, such as QFD [8], probability analysis [9,10], and cluster analysis [11]. ...
... About qualitative method, Chen et al. [6] proposed an ontology learning CRs representation system, the system which preprocessed customer statements by language processing tools. Wang and Tseng [7]; Wang and Tseng [8] put forward the concept of customer demand bias and employed probability analysis methods to analyze CRs. Liu et al. [9] proposed a system management approach for demand management in industrial design. ...
Article
Full-text available
The precisely perception of key customer requirements (CRs) is critically important for customer collaborative product innovation (CCPI) design. A novel approach is proposed based on the Kano model, interval 2-tuple linguistic representation model, and prospect theory. First of all, a Kano model is constructed to preliminarily screen the relatively important product function attributes. For the uncertain and vague information of CRs, an interval 2-tuple linguistic representation model is proposed to determine the weight of CRs. Then, the comprehensive prospects value is utilized for sorting the innovative programs based on the prospect theory. Finally, a numerical example is given to verify the scientific and validity of the proposed method.
... Chen et al. proposed an ontology learning customer requirements representation system, which pre-processes customer statements using language processing tools [28]. Wang and Tseng proposed the concept of customer requirements bias and used probability analysis methods to analyze customer requirements [29,30]. Liu et al. proposed a system management approach to manage requirements in industrial design [31]. ...
Article
Full-text available
Low-carbon product design is an important way to reduce greenhouse gas emission. Customer collaborative product innovation (CCPI) has become a new worldwide product design trend. Based on this popularity, we introduced CCPI into the low-carbon product design process. An essential step for implementing low carbon CCPI is to clarify key low carbon requirements of customers. This study tested a novel method for perceiving key requirements of customer collaboration low-carbon product design based on fuzzy grey relational analysis and genetic algorithm. Firstly, the study considered consumer heterogeneity, allowing different types of customers to evaluate low carbon requirements in appropriate formats that reflected their degrees of uncertainty. Then, a nonlinear optimization model was proposed to establish the information aggregation factor of customers based on the genetic algorithm. The weight of customers was obtained simultaneously. Next, the key low carbon requirements of customer were identified. Finally, the effectiveness of the proposed method was illustrated with a case related to a low carbon liquid crystal display.
... Customisation has become popular because it provides flexibility via options which enable customers to make choices to tailor products according to their preferences. Buyers can easily avail of a multitude of different products with their favoured attributes to resonate with their own identities [1]. With the proliferation of Internet technologies, companies use different toolkits to facilitate the customisation process, particularly to elicit customer needs and transform them into tangible product configurations. ...
Preprint
Full-text available
Product configurators have been acknowledged as an important enabling toolkit to realise mass customisation. During the configuration process, customers are presented with an operational list of predefined attributes and their variants. They can choose between different variants in accordance with their preferences. The combination of their choices constitutes their desired specification and thus the customised product. This configure-to-order process has been implemented in various industries. Currently, some on-line configurators provide extra information on attribute variants to improve the information available to customers, such as popularity or bestseller ratings. This information is derived from the purchase behaviours of other customers. However, it is not clear how such information affects consumer decision making. Drawing on the notion of observational learning, it could be posited that customers tend to follow the choices made by others, particularly when they lack adequate prior knowledge about products. However, one of the advantages of customisation is that it enables customers to harness and reveal their own identities by configuring and purchasing a bespoke unique product that is tailored to their own desires. Thus, customers may not be willing to select the popular variants during configuration process. To solve this paradox, we empirically find that observational learning is significant in the product configuration process. Customers' abilities and motivations to process information mediate this observational learning effect.
... Product configuration systems serve as an important enabling toolkit to realize mass customization. Basically, a product configuration system attempts to elicit customer needs and map the needs to design parameters (Piller and Tseng 2010; Wang and Tseng 2014c). A product configurator consists of a set of predefined components or attributes, along with their constraints, to regulate the combination of components. ...
Chapter
Full-text available
Mass customization aims to deliver products and services that best meet individual customers’ needs with near mass production efficiency. In this paradigm, it is critical to provide individually designed products and services by considering every customer as an individual through process agility, flexibility, and integration of product life cycle. This entry is identical to the book chapter of “Mass Customization” in CIRP Encyclopedia of Production Engineering (2017 edition). Due to copyright concern, the publisher’s pdf version is not shared. Happy reading.
Conference Paper
Full-text available
The elicitation of customer needs (CNs) is a critical step in product development. However, these needs are often expressed in ambiguous, simple language and not in the form of well-defined specifications, causing a semantic gap in the product development process. Traditional methods to bridge the gap rely heavily on human action. Product development teams need to manually link CNs to product specifications in an ad hoc manner. This may be infeasible for large product variant spaces or evolving product families. We propose a machine learning mechanism to automatically bridge the semantic gap. This task is considered as a classification problem, with CNs being the class. The mapping function from product specifications to CNs is learned from training data by using a support vector machine and decision tree classifier. Given a new product variant, the learnt classifier can determine the needs that the product variant can satisfy. Numerical experiments show that the proposed method can achieve very high mapping accuracy. It can also shield product development teams from the tedious labour of linking CNs to product variants, and thus improve the efficiency of needs elicitation.
Article
Full-text available
Rapid changes of new technologies, market dynamics, and swift fluctuation of customer tastes acerbate the needs for companies to identifying the emerging customer requirements and incorporate them in the conceptual design stage. Discovering emerging customer needs has great potential to create new product opportunities for the success of business. However, identifying customers’ requirements in an early design stage has not been well addressed in the traditional design methodology. Because emerging needs are usually not obvious at the budding stage, the related observations are rare in dataset. Traditional methods fall short of providing enough support of eliciting the below-the-radar needs. This paper reports a new approach to identify emerging customer needs in the growing popular online interactive environment with customer participation. The Bayes factor, a methodology to quantify the occurrence possibility of a certain event, is used to calculate the likelihood that current offering cannot meet the customer's requirements whenever a new specification is incorporated. With the sequential input from customers, a series of the Bayes factor value can then be calculated as the weight of evidence that emerging customer needs appear. We show that the decreased value will be the potentially emerging needs which cannot be satisfied by the current product family. Numerical and analytical results are derived to demonstrate the viability and effectiveness of the proposed approach.
Article
Full-text available
The authors present a tailored interviewing procedure for life-style segmentation. The procedure assumes that a life-style measurement instrument has been designed. A classification of a sample of consumers into life-style segments is obtained using a latent-class model. With these segments, the tailored interview procedure classifies consumers into the same segments in future studies. The procedure minimizes the error of misclassification and decreases the interview costs by using only a fraction of the items for each respondent. In an empirical application to a questionnaire designed to assess lifestyle related to the consumption of fashion, the authors obtain a classification that agrees with the full interview (i.e., using all scale items) for 73% of the respondents while using only 22% of all the questions in a traditional life-style battery.
Article
Full-text available
Configurators have been generally accepted as important tools to elicit customers' needs and find the matches between customers' requirements and company's offerings. With product configurators, product design is reduced to a series of selections of attribute values. However, it has been acknowledged that customers are not patient enough to configure a long list of attributes. Therefore, making every round of configuring process productive and hence reducing the number of inputs from customers are of substantial interest to academic and industry alike. In this paper, we present an efficient product configuration approach by incorporating Shapley value, which is a concept used in game theory, to estimate the usefulness of each attribute in the configurator design. This new method iteratively selects the most relevant attribute that can contribute most in terms of information content from the remaining pool of unspecified attributes. As a result from product providers' perspective, each round of configuration can best narrow down the choices with given amount of time. The selection of the next round query is based on the customer's decision on the previous rounds. The interactive process thus runs in an adaptive manner that different customers will have different query sequences. The probability ranking principle is also exploited to give product recommendation to truncate the configuration process so that customers will not be burdened with trivial selection of attributes. Analytical results and numerical examples are also used to exemplify and demonstrate the viability of the method.
Conference Paper
The ability to model and simulate polymer electrolyte fuel cells (PEFCs) is of paramount importance to the fundamental study and advancement of the technology as well as engineering design and optimization of PEFCs. Despite the great efforts made in the past years, a challenge still remains in the area of developing a suitable sub-model that can simulate two-phase transport in PEFC channels and can be incorporated to a full PEFC fuel cell model. This is mainly due to the lack of efficient numerical methods to track the two-phase interface. The objective of this work is to make one of the first attempts at developing a sub-model of channel flows and couple it with a PEFC model that simulate other phenomena occurring in a PEM fuel cell, including those in the anode side. The boundary conditions at the channel and gas diffusion layer interface are briefly discussed. Comparison is also made between the single-phase channel sub-model and two-phase one, as well as between different model parameters in terms of liquid water saturation distributions.
Article
Product customization is attracting more attentions in industry as a viable strategy to better meet customer requirements and gain more profit. However the vast number of product variants in product customization process often makes it difficult for consumers to make purchase decisions, a phenomenon referred to as information overload. In this paper we take a two-prong approach to tackle the issue of information overload in customized products recommendation. Basically, the method answers two questions, namely, which products to recommend and in what order to present the recommendations. Firstly, a probability relevance model is deployed to calculate the probability of relevance for each end product. Then a probability ranking principle is exploited to present the recommendations. The approach also takes customer flexibility into consideration and thus mitigates the effect of inconsistent specifications from customers. It does not require any prior knowledge about an active customer’s preference and can accommodate the new customers challenge facing by recommendation approaches. Analytical results show that the method is optimal in terms of customer’s utility and product recommendation efficiency. Numerical experiments are also conducted to test the presented approach.
Article
Identification of customer needs is the starting point of design process. Most of design methodologies focus on technical domains to define customer requirements. However, the success of product design nowadays goes beyond technical features; it often depends heavily on multi-facets of customers’ needs including various business parameters. Moreover, the qualitative and subjective factors such as affection, aesthetic appearance, user friendliness and brand loyalty are essential to the ultimate acceptance of a new product. This paper presents a probabilistic approach to elicit, characterize the qualitative customers’ latent and subjective preferences and incorporate them into product design.
Article
With increasing product variety and dynamic demand fluctuation, manufacturing industry is moving towards a high product mix and low order volume production environment. Consequently, the order commitment process is becoming one of the most important processes for manufacturing firms to meet individual customer's needs with limited resources. However, demands for shortened delivery lead time, diverse customer requirements and more frequent customer orders have made the order commitment task more challenging. This paper attempts to tackle these new challenges by incorporating not only manufacturing flexibility but also flexibility from the demand side. Customer flexibility is characterised by customer indifference to certain product attributes and/or delivery schedules. Intuitively, with the consideration of customer flexibility, both manufacturers’ and customers’ interests can be better served since the solution space of matching demand and supply can be extended beyond the traditional domain purely from a manufacturing perspective. To this end, a systematic approach is developed to characterise and model customer flexibility. A mixed-integer-programming model is formulated to provide optimal order commitment decisions.
Article
Mass customization (MC) has become a reality and cannot be neglected as one of the leading strategies in satisfying customers and assuring companies survival in today’s markets. MC can be offered either via product variability or process variability. How this is achieved using both approaches is the main core of this article. After an overview of related works, the influence of design variability on the overall generated value of the product is studied via a metric approach. A case study of the shoe industry is also presented , based on the experience of a European project called Dorothy.
Article
Continuing his exploration of the organization of complexity and the science of design, this new edition of Herbert Simon's classic work on artificial intelligence adds a chapter that sorts out the current themes and tools -- chaos, adaptive systems, genetic algorithms -- for analyzing complexity and complex systems. There are updates throughout the book as well. These take into account important advances in cognitive psychology and the science of design while confirming and extending the book's basic thesis: that a physical symbol system has the necessary and sufficient means for intelligent action. The chapter "Economic Reality" has also been revised to reflect a change in emphasis in Simon's thinking about the respective roles of organizations and markets in economic systems.