Available online at www.sciencedirect.

com

ScienceDirect
IFAC PapersOnLine 55-2 (2022) 48–53
Development
Development of an advanced condition-based maintenance system for high-
Development of of an
an advanced
advanced
critical
condition-based
condition-based
industrial fans in a
maintenance
maintenance
foundry
system
system for
for high-
high-
Development of an critical
advanced industrial fans
condition-based
critical industrial
Development of an advanced in a foundry
maintenance
fans in amaintenance
condition-based foundry system for high-
system for high-
A.
A. Polenghi*,
Polenghi*, L.
L. critical
Cattaneo**, M.
critical
Cattaneo**, M. industrial
Macchi*, D.
industrial
Macchi*, D. fans in a
Pasanisi***,
Pasanisi***, foundry
V.
V. Pesenti***,
fans in a foundry A. Borgonovo****
A. Polenghi*, L. Cattaneo**, M. Macchi*, D. Pasanisi***, V. Pesenti***, A. Borgonovo****
Pesenti***, A. Borgonovo****
A. Polenghi*, L. Cattaneo**, M. Macchi*, D. Pasanisi***, V. Pesenti***, A. Borgonovo****
*Department
A. Polenghi*,of
*Department of
ofL.Management,
Cattaneo**,Economics
Management,
Management, Economics
M. Macchi*,
Economics and
and
andD. Industrial
Industrial
Pasanisi***,
Industrial Engineering,
Engineering, Politecnico
Politecnico
V. Pesenti***,
Engineering, Politecnico di
di Milano,
di Milano, P.zza
Milano,
A. Borgonovo**** P.zza
P.zza
*Department
A. Polenghi*, of Management,
L. Cattaneo**, Economics
M. Macchi*, and Industrial
D. Pasanisi***,Engineering, Politecnico
V. Pesenti***, di Milano,
A. Borgonovo**** P.zza
*Department Leonardo
of
Leonardo
Leonardo da da
da Vinci
Management,
Vinci
da Vinci
Vinci 32, 32,
32, 20133,
Economics
20133,
32, 20133, Milan,
and
Milan,
20133, Milan, Italy
Italy
Milan, Italy (e-mail:
Industrial
(e-mail:
Italy (e-mail: adalberto.polenghi@polimi.it,
(e-mail: adalberto.polenghi@polimi.it, P.zza
Engineering, Politecnico
adalberto.polenghi@polimi.it, di Milano,
*Department Leonardo
of Management,
Leonardo da Vinci 32,Economics 20133, and Italy
Industrial
marco.macchi@polimi.it)
Milan, (e-mail: Engineering, Politecnico di Milano, P.zza
adalberto.polenghi@polimi.it,
*Department of Management, Economics marco.macchi@polimi.it)
marco.macchi@polimi.it)
and Industrial
marco.macchi@polimi.it) Engineering, Politecnico di Milano, P.zza
Leonardo da Vinci 32, 20133, Milan, Italy (e-mail: adalberto.polenghi@polimi.it,
marco.macchi@polimi.it)
**
** School
School Leonardo
of
of da Vinci
Industrial
Industrial 32, 20133,Università
Engineering,
Engineering, Milan, Italy
Università (e-mail:
Carlo
Carlo adalberto.polenghi@polimi.it,
Cattaneo
Cattaneo -
- LIUC,
LIUC, C.so Matteotti
C.so Matteotti
Matteotti 22, 22, 21053
22, 21053
21053
** School of Industrial Engineering,marco.macchi@polimi.it) Università Carlo Cattaneo -- LIUC, C.so
** School of Industrial Engineering, Castellanza,
Castellanza, marco.macchi@polimi.it)
Italy (e-mail:
Università
Italy (e-mail: lcattaneo@liuc.it)
Carlolcattaneo@liuc.it)
Cattaneo - LIUC, C.so Matteotti 22, 21053
** School of Industrial Engineering, Castellanza, Italy (e-mail:
Università Carlolcattaneo@liuc.it)
Cattaneo - LIUC, C.so Matteotti 22, 21053
** School of Industrial Castellanza,
Engineering, Italy (e-mail:
Università Carlo lcattaneo@liuc.it)
Cattaneo - LIUC, C.so Matteotti 22, 21053
*** Consorzio
*** Consorzio Intellimech,
Consorzio Intellimech,
Intellimech,
Castellanza,via via
via Stezzano
(e-mail:87,
Stezzano
Italy 87,
87, 24126,
24126,
24126, Bergamo,
Bergamo,
lcattaneo@liuc.it) Italy
Italy
Italy (e-mail:
(e-mail:
(e-mail:
*** Stezzano 87, 24126, Bergamo, Italy (e-mail:
Castellanza,via
valerio.pesenti@intellimech.it,
*** Consorzio Intellimech,
valerio.pesenti@intellimech.it, Italy (e-mail:
Stezzano lcattaneo@liuc.it)
davide.pasanisi@intellimech.it)
87, 24126, Bergamo,
davide.pasanisi@intellimech.it)
davide.pasanisi@intellimech.it) Italy (e-mail:
valerio.pesenti@intellimech.it,
*** Consorzio Intellimech, via Stezzano davide.pasanisi@intellimech.it)
87, 24126, Bergamo, Italy (e-mail:
*** valerio.pesenti@intellimech.it,
Consorzio Intellimech, via Stezzano davide.pasanisi@intellimech.it)
87,66,24126, Bergamo, Italy (e-mail:
****
****
**** Brembo
Brembo
Brembo S.p.A.,
S.p.A.,
S.p.A., Via
Via
Via
valerio.pesenti@intellimech.it, Gian
Gian
Gian Maria
Maria
Maria Scotti,
Scotti,
Scotti, 66,
66, 24030
24030
24030 Mapello,
Mapello,
Mapello,
davide.pasanisi@intellimech.it) Italy
Italy
Italy (e-mail:
(e-mail:
(e-mail:
****
**** Brembo S.p.A.,
S.p.A., Via
valerio.pesenti@intellimech.it,
Brembo Via Gian
Gian Maria
Maria Scotti, 66, 24030 Mapello,
davide.pasanisi@intellimech.it)
angelo_borgonovo@brembo.it)
Scotti, 66, 24030 Mapello, Italy
Italy (e-mail:
(e-mail:
angelo_borgonovo@brembo.it)
angelo_borgonovo@brembo.it)
angelo_borgonovo@brembo.it)
**** Brembo S.p.A., Via Gian Maria Scotti, 66, 24030 Mapello, Italy (e-mail:
angelo_borgonovo@brembo.it)
**** Brembo S.p.A., Via Gian Maria Scotti, 66, 24030 Mapello, Italy (e-mail:
Abstract: In the
the recent years, angelo_borgonovo@brembo.it)
manufacturing companies are investing in sensors and information systems
Abstract:
Abstract:
Abstract: In
In
In the
the recent
recent
recent years,
years,
years, manufacturing
manufacturing companies
angelo_borgonovo@brembo.it)
manufacturing companies
companies are
are
are investing
investing
investing in
in
in sensors
sensors
sensors and
and
and information
information
information systems
systems
systems
to implement
Abstract:
to
to implement
implement In condition-based
the recent
condition-based
condition-basedyears, maintenance
manufacturing
maintenance
maintenance (CBM),
companies
(CBM),
(CBM), thus
thus
thus pursuing
are investing
pursuing
pursuing the
the
the inbenefits
sensors
benefits
benefits of
of
ofanddigital
digital
digital transformation.
information systems
transformation.
transformation.
to implement
Abstract:
Nevertheless,
to implement In condition-based
the
to recent
implement
condition-basedyears, CBM maintenance
manufacturing
as
maintenance advanced (CBM),
companies
(CBM),digital thus pursuing
are
system,
thus investing the
significant
pursuing the inbenefits
sensors of
investment
benefits ofanddigital
should
digital transformation.
information be systems
made
transformation. to
Nevertheless,
Nevertheless,
Abstract:
Nevertheless, In the to
to
to implement
implement
recent years,CBM
implement CBM
CBM as
as advanced
advanced
manufacturing
as advanced digital
digital
companies
digital system,
system, significant
significant
are investing
system, significant investment
investment
in sensors
investment should
should
and information
should be
be
be made
made
systems
made to
to
to
to implement
gather
gather and
Nevertheless, manage
and manage
managecondition-based
to all needed
implement
all needed
needed data CBM maintenance
data
data from
as advanced
from different(CBM),
different
different digital thus
sources;
sources; pursuing
besides,
system, the
significant
besides, qualified benefits
qualified
qualified human of
investment
human digital
resources
should
resources transformation.
are
be required
made
are required
required to
to implement
gather and condition-based
all maintenance
from (CBM), thus pursuing
sources; besides, the benefits human of digital
resources transformation.
are
Nevertheless,
for
gather
for data
data andanalytics.
manage
analytics.to implement
Given
all
Givenneeded this
this CBMpremise,
data
premise, as advanced
from the present
different
the present digital
paper
sources;
paper system,
aims atsignificant
besides,
aims at describing
qualified
describing investment
an industrial
human
an should
resources
industrial project
project bewhere
are made an
required
where to
an
Nevertheless,
for dataand analytics.to implement
Given thisCBM premise, as advanced digital
papersystem, atsignificant investment should bewhere
made an to
gather
advanced
for data
advanced manage
CBM
analytics.
CBM all needed
system
Given
system for
forthis data fromthe
high-critical
premise,
high-critical the
present
different
industrial
present
industrial sources;
fans
paper
fans is
is
aims
besides,
aims at
describing
implemented qualified
describing
implemented in
in
an
an
industrial
human
aa foundry. resources
industrial
foundry. Indeed,
Indeed,
project
project aregoal
the
the required
where
goal is
is to
an
to
gather andCBM
advanced manage all needed
system for data from industrial
high-critical different sources;
fans is besides, qualified
implemented in a human resources
foundry. Indeed, aregoal
the required is to
for
use data
already
advanced
use alreadyanalytics.
CBM Given
available
system
available data
for
data this premise,
from
high-critical
from the
the the
extantpresent
industrial
extant paperisaims
automation
fans
automation and at describing
additional
implemented
and additional in a an industrial
vibration
foundry.
vibration data
Indeed,
data project
to develop
to develop
the where
goal is an
state
stateto
for data
use alreadyanalytics. Given
available data thisfrom premise,the the present
extant paper aims
automation and at describing
additional an industrial
vibration data project
to developwhere an
state
advanced
detection
use already
detection CBM
and
and to system
identify
available
to identify for
dataany
any high-critical
abnormal
from
abnormal the industrial
behaviour
extant
behaviour offans
automation
of the
the isassets.
implemented
and
assets. The
additional
The in avibration
evidence
evidence foundry.
from
from the
the Indeed,
project
data
projectto the
is
develop
is goali)
that:
that: isthe
state
i) to
the
advanced and
detection CBM to system
identifydatafor
anyhigh-critical
abnormal industrialof
behaviour fans
theisassets.
implemented
The evidence in a foundry.
from the Indeed,
project the goali)isthe
is that: to
use already
vibration
detection
vibration and available
analysis
analysisto remains
identify
remains anyan
an from
easy
abnormal
easy the
and
and extant
behaviour automation
cost-effective,
cost-effective, of yet
the
yet and
assets. additional
well-performing
The
well-performingevidence vibration
way,
way, fromto
to thedata
monitor
monitorprojectto
the
the develop
state
is that:
state and
and state
i) the
the
vibration
use already
vibration analysis
analysis remains
available
remains dataan an easy
from
easy and
the
and cost-effective,
extant automation
cost-effective, yet
yet well-performing
and additional way,
well-performing to
vibration
way, to monitor
data tothe
monitor the state
develop
state and
and the
state
the
detection
health
vibration
health of
of and
analysisto identify
machines
machines with
remains
with anyan abnormal
rotating
rotatingeasycomponents; behaviour
and cost-effective,
components; ii)
ii) ofautomatic
the
yetassets.
automatic The evidence
regulation
well-performing
regulation system
way,
system fromtomaythemask
monitor
may project
mask the
the
the is that:
underlying
state and
underlyingi) the
health
detection
health of
of machines
and
machines with
to identify
with rotating
any abnormal
rotating components;
behaviour
components; ii)
ii) ofautomatic
the assets.
automatic regulation
The evidence
regulation system
system from may
maythemask
project
mask the
the is underlying
that:
underlyingi) the
vibration
behaviour
health
behaviour of analysis
and
machines
and remains
degradation
with
degradation an
of
rotating
of easy
complex
complex and cost-effective,
assets;
components;
assets; iii)
ii)
iii) yet
already
automatic
already well-performing
gathered data
regulation
gathered data fromway,
system
from to
extant monitor
may
extant automation
mask
automation the
the state
are
are and
mainly
underlying
mainly the
behaviour
vibration
behaviour and
analysis
and degradation
remains an
degradation of
of complex
easy
complex assets;
assets; iii)
and cost-effective,
iii) already gathered
yet well-performing
already gathered data
data from
fromway, extant automation
to monitor
extant automation the stateare
are mainly
and
mainly the
health
focused
behaviour
focusedofonof onmachines
the
and
onmachines with
process
degradation
the process
process rotating
parameters
of
parameters complexcomponents;
and provides
assets;
and provides ii)
iii)
providesii) an automatic
aid
already to
anautomatic
aid to regulation
describe
gathered
to describe
describe the
data system
working
from
the working
working may
state
extant
state ofmask
the
automation
ofmask the
assets,
the assets,
assets, underlying
arebut have
mainly
but havehave
health
focused the with rotating components;
parameters and an aid regulation
the systemstate mayof the the underlying
but
behaviour
limited
focused
limited onandthe degradation
potentialities
potentialitiesprocessfor
for of complex
novelty
parameters
novelty and assets;
detection.
provides
detection. iii)analready
Eventually,
Eventually, aid tothe gathered
paper
describe
the paper dataworking
envisions
the fromfutureextant automation
statedevelopment
of the assets, are
of aamainly
but more
have
behaviour
limited
focused
and
onapproach
degradation
potentialities
the process
of complex
forparameters
novelty assets;
detection.
and provides
iii) already
Eventually,
an aid tothe paper envisions
gathered
describe
data fromfuture
envisions
theextant
working
extant
future development
automation
statedevelopment
of the assets,
of
are
ofbut more
amainly
more
have
integrated
limited
integrated potentialities
approach aimed
for
aimed at
novelty
at aa combined
detection.
combined elaboration
Eventually,
elaboration of
of data
the
data from
paper
from the
envisions
the extant automation
future
automation and
development
and vibration
of
vibration a data.
more
data.
focused onapproach
integrated the process aimed parameters
at a combined and provides
elaborationan aid ofto describe
data from the theextant
working state of the
automation andassets,
vibrationbut data.
have
limited
The
integrated
The potentialities
integrated
integratedapproachapproachfor
aimed
approach novelty
is atunder detection.
development,
a combined Eventually,
hence
elaboration the
of the paper
paper
data from envisions
provides
the future
insights
extant development
on the
automation on-going
and vibration of a more
analyses.data.
limited
The potentialities
integrated approachfor is is under
novelty
under development,
detection.
development, hence
Eventually,
hence the
the paper
the paper
paper provides
envisions
provides insights
futureon
insights on the on-going
development
the on-going analyses.
of a more
analyses.
integrated
The integrated
Copyright
Keywords: approach
© 2022 aimed
approach
The
condition-based is at
Authors.undera combined is anelaboration
development,
This
maintenance, open access
CBM, hence of
thedata
article
prognostics from
under
paper
and the
the extant
provides
health CC automation
BY-NC-ND
insights
management, on thePHM, andstate
license
on-goingvibration
analyses.
detection,data.
Keywords:
integrated
Keywords: condition-based
approach aimed
condition-based atmaintenance,
a combined
maintenance, CBM,
CBM, prognostics
elaboration of
prognostics data and
from
and health
the
health management,
extant automation
management, PHM,
PHM, state
andstate detection,
vibration
detection,data.
The
health
The
integrated
integrated
approach
(https://creativecommons.org
health
Keywords: assessment, vibration,
condition-based
assessment, vibration,
approach
is under/licenses/by-nc-nd/4.0/)
warningdevelopment,
maintenance,
is under warning system,
development,CBM,
system,
hence
industrial the
prognostics
industrial
hence
paper
thefan,
fan, and
paper
provides
manufacturing. insights
health management,
manufacturing.
provides
on
insights on the
the
PHM, on-going analyses.
state detection,
on-going analyses.
health
Keywords: assessment, vibration,
condition-based warning
maintenance, system,
CBM, industrial
prognostics fan, manufacturing.
and health management, PHM, state detection,
health
Keywords: assessment, vibration,maintenance,
condition-based warning system, CBM, industrial
prognostics fan,and
manufacturing.
health management, PHM, state detection,
health assessment, vibration, warning system, industrial fan, manufacturing.
health assessment, vibration, warning system, industrial fan, manufacturing.
insights
insights
insights on
on
on machine
machine
machine health
health
health states
states
states through
through
through the
the
the application
application
application of
of
of
1. INTRODUCTION
1. INTRODUCTION insights on machine health states through the application of
1. INTRODUCTION advanced
insights
advanced onsolutions
machine
solutions for
for condition-based
health states
condition-based through maintenance
the
maintenance (CBM),
application
(CBM), of
1. INTRODUCTION advanced
insights onsolutions
machine for condition-based
health states through maintenance
the (CBM),
application of
Manufacturing
Manufacturing companiescompanies
companies are undergoing
are undergoing
undergoing radical radical changes
radical changes
changes collectingcollecting
advanced
collecting real-time
solutions
real-time data
for
data from the
condition-based
from the industrial
industrial assets
maintenance
assets as
as well
(CBM),
well as
as
Manufacturing 1. INTRODUCTION
are insights onreal-time
machinedata healthfrom states through the
the industrial application
assets as (CBM),
well asof
dictated by by the
Manufacturing
dictated by the digital
the digital1. transformation.
companies
digital INTRODUCTION
transformation.
are undergoing
transformation. It
It offers
offers to companies
radical
to changes
companies a
a advanced
elaborating
collecting
elaborating solutions
data
real-time
data for
through
throughdata condition-based
adequate
from
adequate the analytics,
industrial
analytics, maintenance
embedded
assets
embedded as in
well
in CPS
CPSas
dictated by It offers to companies a advanced
elaborating solutions
data for
through condition-based
adequate analytics, maintenance
embedded (CBM),
in CPS
wide
dictated
wide by the
Manufacturing
spectrum
spectrum
Manufacturing the digital
companies
of new
digital
of new
companies
transformation.
are undergoing
possibilities,
transformation.
possibilities,
are undergoing
It
It offers
like
like the
offers
the
to
to companies
radical
developmentchanges
companies
development of
of
aa collecting
(Cyber
elaborating
(Cyber real-time
Physical
Physical
(Cyber Physical
collecting Physical
real-time
data adequate
System),
data through
System),
System),
from
data from
to
to the
to the
industrial
support
analytics,
support
support reactive assets
embedded
reactive
reactive
industrial assets
or
or as
or as
well
proactive
in CPS
proactive
proactive
well
as
as
wide spectrum
dictated by the of newtransformation.
digital possibilities, like
It the radical
offers development
to
changes
companies ofa (Cyber
elaborating data System),
through to
adequate support
analytics,reactive
embedded or proactive
in CPS
new business
wide
new business
spectrum models that better fit
of newtransformation.
possibilities, fit like
the the requirement
development of the
the actions
ofa (Cyber
actions (Gao
Physical
(Gao et
et al., 2021).
System),
al., 2021). to support reactive or proactive
dictated
new by the models
business digital
models that
that better
better fit the
It offers
the requirement
to companies
requirement of
of the actions (Gao
elaborating
actions (Gao et al.,
data
et al., 2021).adequate
through
2021). analytics, embedded in CPS
wide
market
new spectrum
(Ulas, of new that
2019). possibilities,
This fit like
transformation the the isdevelopment
based on
of theof (Cyberactions Physical
(Gao et al.,System),
2021). to support reactive or proactive
wide business
market
market (Ulas,
spectrum models
2019).
of new This
(Ulas, models
2019). This
better
transformation
possibilities,
transformationlike requirement
the is based
based on
isdevelopment on the
of The
the (Cyber
The
actions data
dataPhysical
(Gao sources
sources
et
System),
al., are
are
2021).
to support
different
different and
and reactive
could
could or proactive
include
include sensors
sensors
new
marketbusiness
availability
(Ulas,
availability of more
2019). that
and
This better
more fit
data
transformation the requirement
thanks is to the
based of the
digital
on The data sources are different and could include sensors
businessof
availability
new
availability of more
more and
of models
more that more
and
and better data
more
more data
fit the
data thanks
thanks
thanks to
to the
requirement
to the digital
of the The
the digital
digital actions data
installed
The (Gao
data
installed on
on
sources
et
the
sources
the al., are
2021).
assets,
are
assets,
different
controllers
different
controllers
and
andlike
like
could
PLC
could
PLC
include sensors
(Programmable
include sensors
(Programmable
market (Ulas,
technologies
availability
technologies of 2019).
that
thatmoreare
are This
pervading
and transformation
more
pervading the
data
the manufacturing
thanks
manufacturingis based
to the on the
industry
digital
industry installed
installed on
on the
the assets,
assets, controllers
controllers like
like PLC
PLC (Programmable
(Programmable
technologies
market
technologies that
(Ulas,that are
2019).
are pervading
pervading the
This transformation
the manufacturing
manufacturing industry
is basedindustry
on the The Logic data
installed sources
Controller)
on the andare
assets, CNCdifferent
(Computer
controllers andlike could
Numerical
PLC include sensors
Control),
(Programmable as
availability
(Borangiu ofal.,moreare and more the data thanks to the digital Logic Controller) and CNC (Computer Numerical Control), as
(Borangiu et
technologies
(Borangiu
availability
(Borangiu etofal.,
et
et
that
al.,
al.,
2019).
2019).
2019).
more
2019). and
Consequently,
pervading
more datamanufacturing
Consequently,
Consequently,
Consequently,
the
the greater
the
thanks
the greater
greater
to the
greater
amount
industry
amount
amount
amount
of
of The
of
digital
of
Logic
Logic
installed
well
Logic
Controller)
data
as on
sourcesand
Controller)
the
enterprise
Controller) andareCNC
assets,
and CNC (Computer
different
(Computer
controllers
information
CNC (Computer like
systems
Numerical
and could
Numerical
PLCsuch
Numerical
includeControl),
sensors
Control),
(Programmable
as the
Control),MES
as
as
as
technologies
data available
(Borangiu
data availableet
availablethatthat
and
al.,
and areare
the
2019). pervading
higher
the higher
higher the
computational
Consequently,
computational manufacturing
the power
greater
power the industry
current
amount
the currentof well
installed
well as
as enterprise
on the
enterprise information
assets, controllers
information systems
like
systems PLCsuch
such as the
(Programmable
as the MES
MES
technologies
data and the pervadingcomputational
the
computational power
manufacturing
power the
the current
industry
current Logic Controller)
(Manufacturing
well as enterprise and CNC
Execution
information(Computer
System) or
systems Numerical
thesuch ERP asControl),
(Enterprise
the MES as
(Borangiu
devices
data et al.,
have
available
devices have and 2019).
allow
allow theto Consequently,
tohigher
better monitor the
bettercomputational
monitor andgreater
and control
power
control amount
the what
current
what of
is (Manufacturing
is (Manufacturing
Logic Controller)ExecutionExecution
and CNC (ComputerSystem)
System) or the
orNumerical
thesuch ERP (Enterprise
ERPasControl),
(Enterprise as
(Borangiu
devices et al.,
have 2019).
allow Consequently,
tohigher monitor the
bettercomputational andgreater
control amount
what of
is Resource
well
Resource as enterprise
(Manufacturing Planning)
Planning) information
(Tao
Execution
(Tao et
et al.,
System)
al., systems
2018).
2018). or Additionally,
the ERP
Additionally, the MES
relevant
(Enterprise
relevant
data available
happening
devices
happening at
have
at and
the
allow
the the
shopfloor
to
shopfloor betterlevel
monitor
level of
of companies
and
companies power
control the
(Lee
(Lee current
et
what
et al.,
is
al., well
Resource as enterprise
Planning) information
(Tao et al., systems
2018). such
Additionally, as the MES
relevant
data available
happening at and
the the higher level
shopfloor computational
of companies power(Lee the current
et al., (Manufacturing
data for
for CBM
Resource CBM Execution
purposes
Planning) (Tao may et System)
also2018).
al., come orfromthe third
from ERP parties,
third
Additionally, (Enterprise
parties, like
relevant
devices
2016). have allow to betterlevel monitor and control (Leewhat is data purposes may also come like
2016). This
happening
2016).
devices
2016). This
This atenables
have
This
the
enables
enables
allow
enables
production
shopfloor
production
to better and
production
production
and
and
monitor
and
ofmaintenance
companies
maintenance
maintenance
and control
maintenance
management
management
managementet al.,
what
management
(Manufacturing
data
is outsourcing for CBM
Resource
outsourcing
data for CBMPlanning)
Execution
purposes
companies
purposes (Tao may
that
may et
System)
also2018).
al.,
are
also in
come orfrom
charge
come
the third
ERP parties,
Additionally,
fromof the
third
(Enterprise
like
relevant
maintenance
parties, like
happening
to
to perform
2016). This
perform at the
better
enables
better shopfloor
than in
production
than in level
the
the past
and
past of bycompanies
responding
maintenance
by responding (Lee
managementet
faster
faster al.,
to
to Resource
outsourcing companies
Planning)
companies (Tao that
thatet are
al.,
are in charge
2018).
in charge of the
Additionally,
of the maintenance
relevant
maintenance
happening
to perform atbetter
the shopfloor
than in thelevel
pastof bycompanies
responding (Lee et al.,
faster to activities
data for CBM
activities
outsourcing (Murthypurposes
companies
(Murthy et
et may
al.,
that
al., alsoincome
2015).
are
2015). from
Overall,
charge
Overall, of third
the
the
the parties, like
collection
maintenance
collection of
of
2016).
unforeseen
to This
perform enables
events
better production
that could and slowmaintenance
down management
or stops the activities (Murthy et al., 2015). Overall, the collection of
unforeseen
unforeseen
2016).
unforeseen events than
events
This enables
events thatincould
that
production
that
the and
could
could
pastslow
slow
slow
by down responding
down
maintenance
down or
or
faster the
stops
ormanagement
stops
stops
to data
the
the
for CBM
activities
outsourcing
relevant
activities (Murthypurposes
companies
variables
(Murthy et
for
et
may2015).
al.,
that
CBM
al.,
also come
are
could
2015). in charge
be
from the
Overall, of
expensive
Overall,
thirdcollection
the
the
parties, like
maintenance
for various
collection of
of
to perform
production
unforeseen
production better
(Negri
events
(Negri than
et
et al.,
that
al., in the
2019).
could
2019). past
It
It slowis
isby responding
especially
down
especially or faster
maintenance
stops
maintenance to
the relevant
relevant
outsourcing
relevant variables
variables
companies
variables for
for
for CBM
CBM
that
CBM could
could
are
couldin be
be
charge
be expensive
expensive
of
expensive the for
for various
various
maintenance
for various
production
production (Negri
to perform (Negribetter et et
than al.,
al., 2019).
in2019). It
the past
It is
isbyespecially
responding
especially maintenance
faster to reasons,
maintenance activities
reasons,
relevant (Murthy
either
variables et
technical
for al., 2015).
(specific
CBM could Overall,
sensors
be to be
expensivethe collection
installed,
for of
cloud
various
unforeseen
that is finding
production events
its
(Negri etthat
“golden could
al., age”
2019).as aaaslow
progressive down evolution
ormaintenance
stopsfrom the activities
reasons, either
either technical
technical (specific
2015).sensors
(specific sensors to
to be
be theinstalled,
installed, cloud
cloud
that
that
that is
is
is finding
finding
unforeseen
finding its
its
events
its “golden
“golden
“goldenthat age”
age”
could
age” as It
as
as a slowis especially
progressive
progressive
progressive down evolution
or stopsfrom
evolution
evolution from reasons,
the computing
from relevant
computing
reasons,
(Murthy
either
variables
power
either
et al.,
technical
for
to
technical be (specific
CBMpaid)
(specificcould
and
Overall,
sensors
be to be
expensive
organizational
sensors to be
collection
installed,
for various
(workforce
installed,
of
cloud
cloud
production
the
that (Negri
E-maintenance
is finding
the E-maintenance
E-maintenance its et
of
“golden
of al.,
the
the 2019).
first
age”
first as It
decade
a
decade is especially
of
progressive
of 21th
21th maintenance
century
evolution
century (Guillén
from
(Guillén relevant
computing power
variables
power to
for
to be
beCBMpaid)
paid) and
could
and organizational
be expensive
organizational (workforce
for various
(workforce
production
the (Negri et of the
al., first
2019).decade
It is of 21th
especially century (Guillén
maintenance reasons, either
thatal.,
et
the
et is finding
2016)
E-maintenance its
to of
of the
“golden
the new
the first
age” decade of
as a progressive
concept
first decade of 21th
Smart
21th century
evolution (Guillén
Maintenance
century (Guillén allocated
computing
from allocated to
reasons, either
allocated to thetechnical
the
topower
the
to be (specific
monitoring
monitoring
technical
monitoring
paid)
(specific andsensors
activities
activities
activities
or
sensors
or
to be installed,
or persons
organizational
persons to
to train
to be installed,
persons to train
train
cloud
for
(workforce
for the
the
cloud
for the
thatal.,
et 2016)
is finding
al., 2016) to
to
its
to the
the
“golden
the new
new
new age”concept
concept
as
concepta of
progressive
of Smart
Smart
Smart Maintenance
Maintenance
evolution
Maintenance from data computing
data
allocatedanalytics
to
analyticspower
the to be paid)
activity).
monitoring
activity). and organizational
Therefore,
activities
Therefore, orthe implementation
persons
the to
implementation (workforce
train for of
the
of
the
et E-maintenance
(Bokrantz
al., et al.,
2016) al., of
2020).
to 2020).
the the
new first decade
Currently,
concept of
it is 21th
isSmart century
possible (Guillén
to gather
gather data
Maintenance computinganalyticspower to be paid)
activity). and organizational
Therefore, (workforce
(Bokrantz
(Bokrantz
the
et al.,
et
et
E-maintenance
(Bokrantz et
2016)
al.,
al.,
to
2020).
of the
2020).
the new
Currently,
Currently,
first decade it
Currently,
concept
it
of is
it
of
21th
is possible
possible
century
possible
Smart
to
to
to gather
(Guillén
gather
Maintenance allocated
advanced
data
advanced to
CBM
analytics
CBMthe monitoring
capabilities
activity).
capabilities activities
should
Therefore,
should be
beorthe
tested
the
tested
implementation
persons to train forwith
beforehand
implementation
beforehand
of
the
with
of
(Bokrantz et al., 2020). Currently,
et al., 2016) to the new concept of Smart Maintenance data it is possible to gather allocated
advanced to
CBMthe monitoring
capabilities activities
should beor persons
tested to train
beforehand for the
with
(Bokrantz et al., 2020). Currently, it is possible to gather advanced analytics
CBM activity).
capabilities Therefore,
should be the
testedimplementation
beforehand of
with
2405-8963
(BokrantzCopyright © 2022 The
et al., 2020). Authors. it
Currently, This is ispossible
an open access datathe
article under
to gather analytics
CCCBMBY-NC-NDactivity).licenseTherefore,
. should bethe implementation of
advanced capabilities tested beforehand with
Peer review under responsibility of International Federation of Automatic Control. advanced CBM capabilities should be tested beforehand with
10.1016/j.ifacol.2022.04.168
 A. Polenghi et al. / IFAC PapersOnLine 55-2 (2022) 48–53 49

available data within the company or with the introduction of aramid type and a power of 110 kW. Figure 2 shows a schema
general-purpose sensors in order to guarantee cost-effective illustrating the structure of an industrial fan.
proof of concepts (Ahmad & Kamaruddin, 2012).
This work describes the development of an advanced CBM
system for high-critical industrial fans. This multi-year project
is performed in collaboration with a foundry realizing semi-
finished products for the automotive market. In this project,
the developed solutions rely both on already available data
from the SCADA (Supervisory Control And Data Acquisition)
and acceleration data due to a newly installed accelerometers
system. The two solutions are initially developed separately,
to understand which one better fits for the project goal. Based
on the evidence gathered from the two solutions in a separate
way, some more insights on the possible integration of both Figure 2. Functioning schema of the industrial fan.
data, and related analytics, are proposed; it envisions the future
development of a more integrated warning system. The goal of the project is to develop an advanced CBM system
capable to promptly notify about an abnormal behaviour of the
The paper is structured as follows. Section 2 describes the case industrial fans. As the industrial fans guarantee a safe working
study, with details on the industrial fans (target of the CBM), environment, the prompt notification is essential. If they fail,
the process they must serve, and the adopted methodology. the entire plant should be stopped in order to restore adequate
Section 3 illustrates the FMECA (Failure Modes, Effects and working conditions. Hence, the industrial fans are high-critical
Criticality Analysis) identifying the most critical failure assets since they are operationally risky and do not have spare
modes. Section 4 deals with the vibration analysis performed assets in stand-by that could operate in case of failure.
on data coming from the accelerometers. Section 5 proposes
the exploratory analysis of the already available data from 2.1 Project methodology
SCADA. Section 6 describes the ongoing analyses, in order to The methodology followed during the project is inspired by
envision the future development of an integrated warning (Cattaneo et al., 2021), based on PHM (Prognostics and Health
system. Eventually, Section 7 summarises the results and Management), and is reported in Figure 3.
draws some conclusions on the project experience.
2. DESCRIPTION OF THE CASE STUDY
The case under analysis is the one of a foundry in the northern
part of Italy. The foundry produces semi-finished products for
the automotive sector. Figure 1 shows the phases of the process
according to the ASME (American Standards of Mechanical
Engineers) graphic symbols for process flow diagrams.

Figure 1. Process flow of the foundry.

The process represented in Figure 1 is the primary process of

the foundry, but there are ancillary processes, with relative
systems/assets, that enable the correct functioning of the
primary process. Amongst those ancillary processes, there is
the flue gases filtering to remove contaminants from output air
in order to guarantee safe workplace on the shopfloor as well
as compliance with environmental regulations. The flue gases
filtering relates to two main areas:
 Hot flue gases filtering for the die casting;
 Cold flue gases filtering for the gravity casting.
The two filtering processes are served by two assets. Their
characteristics are the following, respectively: an industrial fan
of 170000 m3/h, 600 bags of polyester type and a power of 315
kW, and an industrial fan of capacity 90000 m3/h, 350 bags of
Figure 3. PHM-inspired methodology in the project.
In addition to the traditional functional blocks considered in
the ISO 13374-1 (equivalently, in the OSA-CBM, Open
System Architecture for Condition Based Maintenance, by
MIMOSA – https://www.mimosa.org/mimosa-osa-cbm/), the
methodology includes a first functional block called Asset
Analysis that focuses on the prioritization of assets and their
failure modes and promotes a better understanding of the asset
functioning and characteristics. The Asset Analysis is
described in Section 3.
50 A. Polenghi et al. / IFAC PapersOnLine 55-2 (2022) 48–53

3. ANALYSIS OF THE INDUSTRIAL FANS Priority

Rank Sys. Component Failure mode
value
To prioritise the failure modes, it is important to understand
Supports
the physical decomposition of the industrial fan. Figure 4
4 CF (ring, Overheating 10.30%
reports the physical structure of the asset under analysis,
spacer, seal)
limited to the first level for the sake of visualisation.
Impeller
5 CF Overheating 9.76%
shaft
BF = bag filter, CF = centrifugal fan
The decision is due to a trade-off between the costs needed to
implement the solution and the opportunities subsequently
reachable. Specifically, the engine, coupling and supports are
Figure 4. Physical decomposition of the industrial fan.
really close each other; thus, the new monitoring system, to be
To prioritise the failure modes, the MAFMA (Multi-Attribute designed, could be composed by only three accelerometers and
Failure Modes Analysis) by (Braglia, 2000) is employed, one central unit. Thus, this solution allows to monitor the
given that the critical asset is already identified. MAFMA is evolution of three different failure modes without much
based on AHP (Analytic Hierarchy Process). In the project, the spending with respect to each of them. This was identified as
AHP criteria are the parameters of the FMECA, namely cost-effective solution to further develop. Section 4 describes
occurrence, severity, and detectability, while the alternatives the vibration analysis subsequently carried out.
are the eleven identified failure modes. The structure is
4. VIBRATION ANALYSIS OF CENTRIFUGAL FAN
represented in Figure 5.
The vibration signals are collected thanks to a new monitoring
system installed on the centrifugal fan as depicted in Figure 6.

Figure 5. MAFMA applied to the project.

The three criteria (occurrence, severity, detectability) are

equally weighted (0.33 each). For the alternatives evaluation,
after the definition of a scale for each of the three criteria, the
MAFMA was firstly filled in with data coming from the
maintenance reports of the centrifugal fan and of the bag filter, Figure 6. Vibration acquisition system.
later validated through an interview with three maintenance The accelerometers are characterised by a sampling frequency
technicians and the maintenance manager. The advantages to of 50 kHz and are installed one per bearing.
use MAFMA, built on AHP, is that it can manage both
quantitative and qualitative inputs. This is particularly useful The vibration analysis is carried out considering the ISO
for the detectability that is hardly measurable objectively. 10816-3 – Mechanical vibration - Evaluation of machine
Table 1 shows the top-five highly ranked failure modes with vibration by measurements on non-rotating parts to implement
the associated component and system (bag filter or centrifugal the state detection. The standard identifies specific HI (health
fan) of the industrial fan and their global priority value. index) thresholds, based on RMS (Root Mean Square), for
machines of certain size with respect to their power. The
Despite the failure mode of the solenoid valve is the one with industrial fan for the cold flue gases, whose power is of 110
the highest priority, together with the maintenance manager, it kW, is considered of medium size by the standard (from 15 to
was decided to monitor the engine, coupling and supports. 300 kW). In the remainder, the fan for the cold flue gases is
Table 1. Results of the application of MAFMA. used as exemplar to show the developed solution. The same
has been done for the hot flue gases industrial fan. Medium-
Priority sized machines have specific values of the RMS as reported in
Rank Sys. Component Failure mode
value Figure 7.
Solenoid Ageing of the
1 BF 12.38%
valve valve spool
Engine
2 CF (shaft and Overheating 10.57%
bearings)
Deterioration
3 CF Coupling of coupling 10.30%
slats Figure 7. HI thresholds according to ISO 10816-3.
 A. Polenghi et al. / IFAC PapersOnLine 55-2 (2022) 48–53
In Figure 8, it is observable the continuous monitoring of the
RMS during the last week of March 2021. Each point
represents the RMS value, and it is coloured to trace the
Figure 8. Continuous monitoring of RMS during the last week of March 2021; during the weekend the industrial fan is kept in active standby.
Continuing the methodology in Figure 3, the health assessment
functional block could be performed by looking at the ISO
13373 – Condition Monitoring and Diagnostics of Machines –
Vibration condition monitoring (specifically, part 3 and 5),
integrated by the tutorial published by (Randall & Antoni,
2011). The analysis includes both frequency spectrum and
envelope spectrum. However, the former suffers of some
pitfalls in industrial applications since random noise and
resonances could hinder the peaks due to fault, when at its
early stage. Therefore, the envelope spectrum is adopted, and
it was possible to identify that there was a problem, possibly a
misalignment, on the outer race of the motor bearings, given
Figure 9. Analysis of the envelope spectrum on March 4th, 2021, at 15:29:59; the ratio of amplitudes at BPFO and BPFI is almost 10.
Overall, by combining the analysis of vibration signals and the
predetermined HI defined by the ISO 10816-3 (Figure 7), it is
possible to conclude that the vibration analysis technique
provides enough information content for decision making
based, at least at first sight. Nonetheless, this analysis does not
provide useful insights on the overall health state of the
industrial fan, but only for some specific components (bearings
of the centrifugal fan) as it was any expected by design. In the
next section 5, the already available data collected by the
SCADA are analysed for possible further insights.
5. ANALYSIS OF DATA FROM SCADA
The data already available from the SCADA are reported in
Table 2, where fs is the sampling frequency.
Table 2. Already available data from SCADA.
Variable Location Unit fs [kHz]
Pressure Inlet pipe mmH2O 1
Temperature Inlet pipe °C 5
P Bag mmH2O 10
Motor RPM Inverter RPM 10
Current Inverter A 1
Particulate Chimney mg/nm3 5 Further analyses through data processing on the other variables
51
rotational speed of the shaft. It is worth noticing that the RMS
barely fell over 4 that is the limit for HI of type B “Unlimited
long-term operation allowable” according to Figure 7.
the peaks at 1x and 2x of the BPFO (Ball Pass Frequency of
the Outer Race), as shown in Figure 9, with an amplitude from
2 to 10 times the one of BPFI (Ball Pass Frequency of the Inner
Race.
Even though the envelope spectrum analysis suggested a
damage of the outer race of the bearings on the motor side, the
RMS values were still in the acceptable range. Therefore, the
maintenance manager decided not to act immediately on the
bearings, instead engaging the outsourcing company in charge
of the maintenance service to better investigate and define the
better way to proceed.
The available data underwent firstly a pre-processing phase
that was required since the sampling frequency differ between
variables, as notable from the rightmost column in Table 2.
The choice was to select the highest sampling frequency and
re-build the signals of variables with lower fs by adopting the
LOCF (Last Observation Carried Forward) imputation.
Secondly, an explorative analysis was performed to get a first
impression on the insights the data can produce. The result for
the P is shown in Figure 10 where the different states of the
industrial fans were highlighted (on, off, transient on, transient
off).
Figure 10. State analysis for the industrial fans.
did not show useful insights on the health state of the asset.
52 A. Polenghi et al. / IFAC PapersOnLine 55-2 (2022) 48–53
On the whole, the information content given by the data from
the SCADA is exclusively related to the working state of the
industrial fan. In fact, the collected data are mainly focused on
the process parameters and not on the asset side (like it
happens for vibrations). This results from different operating
conditions, as discussed with the maintenance manager:
1. the workload to which the industrial fan is subjected
to different changes during the week and even the day
depending on the number of casting operations;
2. the industrial fan has an automatic regulation system
in order to keep the same reference value for P by
adjusting all the other ones; however, the variation of
the other signals is very limited (plots are not shown
here for brevity); also, this inner control may “mask”
the behaviour of the industrial fan.
Concluding, it is apparent that the analysis of process-driven
variables did not show relevant results to be used for CBM
purposes since the relevant information useful to the manager
to take reactive actions is hidden and biased by other factors.
6. ONGOING ANALYSES FOR A FUTURE
INTEGRATED ADVANCED CBM SYSTEM
Currently, several analyses are on the way, to enable building
an integrated approach from SCADA and vibration data. This
is done to empower the data analytics, thus, to extend the
capability of data analytics from single component to asset
behaviour.
Firstly, from the vibration analysis, a strange behaviour of the
industrial fan, particularly, of the centrifugal fan, is detected,
even if the maintenance record does not show maintenance
interventions. In Figure 11, the plot of RMS versus RPM is
reported. It is worth noting that the trend of the RMS may be
modelled as a quadratic function with respect to the RPM with
a high coefficient of determination R2 for low RPM. However,
as the RPM increases, the R2 decreases, meaning that the
function is no more able to predict accurately the RMS based
on RPM. Moreover, the RMS values above RPM equal to 1240
are really scattered and present a U-shaped trend with high
variability. This may be induced by some resonance effects of
the centrifugal fan.
Figure 11. Scatter plot of RMS vs RPM on the motor side.
This evidence opens an intense discussion in the maintenance
department, to understand why this trend shows up and what
is the effect on the health state of the centrifugal fan. However,
this is just an initial integration step, solely building on a few
characteristics measurable from vibration and SCADA data
(respectively, RMS and RPM). Also, the research team aims
to extract additional information from the available data,
broadening the integration of vibration and SCADA-sourced
data. Currently, the selection of the best AI (Artificial
Intelligence) method is undergoing, and Neural Networks
(NNs) are under consideration given the supervised problem.
The selected method will take as input the available data to
predict the value of the Ppred (predicted P) and then, by
measuring the prediction error, notify a possible deviation
from normal behaviour of the industrial fan in relation to bag
filters states, based on the current working regime of the fan
itself. The solution should be verified and assessed so to
comply with technical constraints and barriers related to
algorithm deployment, computational power, response time
and others. Finally, the developed algorithms should be able to
scale between different indenture levels of the asset, providing
information on the health state and possible diagnostics and
prognostics of faults at both component and asset levels.
Figure 12. Future analysis with AI-based system.
7. RESULTS AND CONCLUSIONS
This work describes the experience of an ongoing project in a
foundry. The project aims at developing an advanced CBM
system for high-critical industrial fans that are used to remove
hot and cold flue gases from the shopfloor so to guarantee a
safe working environment. The criticality of the assets thus
relates to their functions in guaranteeing a safe workplace and
efficient operations.
The CBM system has been built considering the availability of
data from SCADA as well as additional sensors, namely
accelerometers, installed on the centrifugal fan. The vibration
signals are elaborated according to the guidelines provided by
the ISO 10816-3 (up to state detection) and ISO 13373-3/5 (for
health assessment). The results show that it is possible to use
the RMS as relevant feature for both state detection and health
assessment of bearings. The SCADA-sourced data are used,
but the retrievable information content is related to the
identification of the working states of the industrial fan (if on,
off, transient off, transient on) rather than on novelty detection.
At the moment, the CBM system is deployed as proof-of-
concept and before the final deployment additional data
analytics will be introduced.
This project experience has shown interesting evidence for
companies willing to pursue the benefits of CBM:
1. Vibration remains a useful signal to understand the
health state of rotating machines. This is not
 A. Polenghi et al. / IFAC PapersOnLine 55-2 (2022) 48–53
unexpected since from the basics of the P-F curve it
is already known that vibration is one of the main
early symptom of a potential failure (Moubray,
1997). Available international standards, as the ISO
10816 and 13373 support the development of CBM
systems based on vibration analysis.
2. Complex assets usually have automatic regulation
system, which guarantees to keep the P fixed, in
case of industrial fans. However, the inner control
may mask the behaviour and degradation of the asset.
Correctly considering this phenomenon is vital to
optimise algorithm performance, while providing
useful information to the maintenance engineer.
3. Despite the digitalisation, current assets are
monitored so to gather process-related data.
However, these data are useful to describe the
working states of the asset rather than its health state.
This should not be seen as a limit, but rather as an
opportunity: current CBM approaches are focused on
the introduction of working states as relevant
knowledge so to normalise state detection results and
take suitable maintenance actions.
As an additional takeaway, the maintenance manager
recognises the relevance to have a reference model, referring
to the PHM as in the ISO 13374, so to promote the
understanding of the entire program to maintenance engineers
and data scientists as well as to plant and company managers
to guarantee their commitment. Concluding, the maintenance
manager has already envisioned, and he is already working on,
a further enhancement in the overall management of the
industrial fans. Indeed, the better knowledge of the assets
allows to re-evaluate and improve the service contract with the
outsourcing company in charge of fans maintenance. Hence,
maintenance interventions could be better defined according
to current industrial fans health states.
The envisioned approach, based on a further integration by
means of proper AI methods, will be expected as a future step
to increase the capabilities of advanced CBM solutions for the
industrial fans under study. Especially, the integration of
information about the scheduling of the die casting operations
is expected to improve CBM performance given that the load
on the industrial fan highly depends on the flow rate of flue
gases to remove. As such, AI methods such as NNs could rely
on more input features, which are significant to predict the P
according to maintenance operators’ knowledge.
ACKNOWLEDGEMENT
This work is the result of a multi-year project (2020-2021)
framed within the MEGMI executive master (Master
Executive in Industrial Asset Management and Maintenance,
link) held by MIP - Politecnico di Milano Graduate School of
Business and by SdM – Scuola di Alta Formazione, University
of Bergamo. In 2021, the MEGMI master ranks in the first
place in the EdUniversal BestMasters Ranking in “Industrial
and Operations Management” in Western Europe.
REFERENCES
Ahmad, R., & Kamaruddin, S. (2012). An overview of time-
53
based and condition-based maintenance in industrial
application. Computers & Industrial Engineering,
63(1), 135–149.
https://doi.org/10.1016/j.cie.2012.02.002
Bokrantz, J., Skoogh, A., Berlin, C., Wuest, T., & Stahre, J.
(2020). Smart Maintenance: An empirically grounded
conceptualization. International Journal of
Production Economics, 223, 107534.
https://doi.org/10.1016/j.ijpe.2019.107534
Borangiu, T., Trentesaux, D., Thomas, A., Leitão, P., &
Barata, J. (2019). Digital transformation of
manufacturing through cloud services and resource
virtualization. Computers in Industry, 108, 150–162.
https://doi.org/10.1016/j.compind.2019.01.006
Braglia, M. (2000). MAFMA: multi‐attribute failure mode
analysis. International Journal of Quality &
Reliability Management.
Cattaneo, L., Polenghi, A., & Macchi, M. (2021). A
framework to integrate novelty detection and
remaining useful life prediction in Industry 4.0-based
manufacturing systems. International Journal of
Computer Integrated Manufacturing, 1–21.
https://doi.org/10.1080/0951192X.2021.1885062
Gao, G., Zhou, D., Tang, H., & Hu, X. (2021). An Intelligent
Health diagnosis and Maintenance Decision-making
approach in Smart Manufacturing. Reliability
Engineering & System Safety, 216, 107965.
https://doi.org/10.1016/j.ress.2021.107965
Guillén, A. J., Crespo, A., Macchi, M., & Gómez, J. (2016).
On the role of Prognostics and Health Management in
advanced maintenance systems. Production Planning
and Control, 27(12), 991–1004.
https://doi.org/10.1080/09537287.2016.1171920
Lee, J., Bagheri, B., & Jin, C. (2016). Introduction to cyber
manufacturing. Manufacturing Letters, 8, 11–15.
https://doi.org/10.1016/j.mfglet.2016.05.002
Moubray, J. (1997). Reliability-centered maintenance.
Industrial Press Inc.
Murthy, D. N. P., Karim, M. R., & Ahmadi, A. (2015). Data
management in maintenance outsourcing. Reliability
Engineering & System Safety, 142, 100–110.
https://doi.org/10.1016/j.ress.2015.05.002
Negri, E., Ardakani, H. D., Cattaneo, L., Singh, J., MacChi,
M., & Lee, J. (2019). A Digital Twin-based
scheduling framework including Equipment Health
Index and Genetic Algorithms. IFAC-PapersOnLine,
52(10), 43–48.
https://doi.org/10.1016/j.ifacol.2019.10.024
Randall, R. B., & Antoni, J. (2011). Rolling element bearing
diagnostics—A tutorial. Mechanical Systems and
Signal Processing, 25(2), 485–520.
https://doi.org/10.1016/j.ymssp.2010.07.017
Tao, F., Qi, Q., Liu, A., & Kusiak, A. (2018). Data-driven
smart manufacturing. Journal of Manufacturing
Systems, 48(January), 157–169.
https://doi.org/10.1016/j.jmsy.2018.01.006
Ulas, D. (2019). Digital Transformation Process and SMEs.
Procedia Computer Science, 158, 662–671.
https://doi.org/10.1016/j.procs.2019.09.101