Proceedigs of the 15th IFAC Symposium on

Proceedigs
the
IFAC
on
Information
Control
Problems
in Manufacturing
Proceedigs of
of
the 15th
15th
IFAC Symposium
Symposium
on
Proceedigs
of
the
15th
IFAC
Symposium
on
Information
Control
Problems
in
Manufacturing
Available
online at www.sciencedirect.com
May
11-13,
2015.
Ottawa,
Canada
Information
Control
Problems
in
Manufacturing
Information
Control
Problems
in
Manufacturing
May
11-13, 2015.
Ottawa, Canada
May
May 11-13,
11-13, 2015.
2015. Ottawa,
Ottawa, Canada
Canada

ScienceDirect

IFAC-PapersOnLine 48-3 (2015) 727732

Lot-sizing
in
Lot-sizing
in
Lot-sizingfor
in
consideration
consideration
for
consideration for

flow-shop
with
energy
flow-shop
with
energy
flow-shop
with
energy
sustainable
manufacturing
sustainable
manufacturing
sustainable manufacturing
systems
systems
systems

Masmoudi Alice Yalaoui Yassine Ouazene

Alice Yalaoui Yassine Ouazene
Masmoudi
AliceChehade
Masmoudi
Hicham
Masmoudi
Alice Yalaoui
Yalaoui Yassine
Yassine Ouazene
Ouazene
Hicham
Hicham Chehade
Chehade
Hicham
Chehade

ICD, LOSI, Troyes University of Technology, UMR 6281, CNRS,

LOSI,
Troyes
University
of
Technology, UMR
6281,
CNRS,
ICD,
ICD,
LOSI,
Troyes
University
of
6281,
Troyes,
oussama.masmoudi,
alice.yalaoui,
ICD,
LOSI,France
Troyes(e-mail:
University
of Technology,
Technology, UMR
UMR
6281, CNRS,
CNRS,
Troyes,
France
(e-mail:
oussama.masmoudi,
alice.yalaoui,
Troyes,
France
(e-mail:
oussama.masmoudi,
alice.yalaoui,
yassine.ouazene,
hicham.chehade)@utt.fr
Troyes, France
(e-mail:
oussama.masmoudi,
alice.yalaoui,
yassine.ouazene,
hicham.chehade)@utt.fr
yassine.ouazene,
yassine.ouazene, hicham.chehade)@utt.fr
hicham.chehade)@utt.fr
Abstract: This paper presents a capacitated lot-sizing problem in flow-shop system with energy
Abstract:
This
paper
presents
a
capacitated
in
flow-shop
system
with
energy
Abstract:
This
paper
presents
a
lot-sizing
problem
in
system
with
consideration.
The
planning
horizon
is definedlot-sizing
by a setproblem
of periods.
Each period
is characterized
Abstract:
This
paper
presents
a capacitated
capacitated
lot-sizing
problem
in flow-shop
flow-shop
system
with energy
energy
consideration.
The
planning
horizon
is
defined
by
a
set
of
periods.
Each
period
is
characterized
consideration.
The
planning
horizon
is
defined
by
a
set
of
periods.
Each
period
is
characterized
by
demand, duration,
electricity
costis and
maximum
power.Each
Bothperiod
non-linear
and linear
consideration.
The planning
horizon
defined
by a setpeak
of periods.
is characterized
by
demand,
duration,
electricity
cost
and
maximum
peak
power.
Both
non-linear
and
linear
by
demand,
electricity
cost
maximum
peak
non-linear
and
mixed
integerduration,
programming
are proposed
solve the problem
with Both
the objective
of minimizing
by
demand,
duration,
electricity
cost and
andto
maximum
peak power.
power.
Both
non-linear
and linear
linear
mixed
integer
programming
are
proposed
to
solve
the
problem
with
the
objective
of
minimizing
mixed
integer
programming
are
proposed
to
solve
the
problem
with
the
objective
of
minimizing
the
production
costs. The costs
are
considered
as the
sum
of electrical,
holding,
setup
and power
mixed
integer
programming
are
proposed
to
solve
the
problem
with
the
objective
of
minimizing
the
production
costs.
The costs
are
considered
the
sumand
of
electrical,
holding,
setup
power
the
production
costs.
are
as
the
of
holding,
setup
and
power
demand
costs. Computational
are as
presented
the numerical
results
areand
discussed
the
production
costs. The
The costs
costsexperiments
are considered
considered
as
the sum
sumand
of electrical,
electrical,
holding,
setup
and
power
demand
costs.
Computational
experiments
are
presented
the
numerical
results
are
discussed
demand
costs.
Computational
experiments
are
presented
and
the
numerical
results
are
discussed
and
analyzed
to
evaluate the efficiency
of those
methods.and the numerical results are discussed
demand
costs.
Computational
experiments
are
presented
and
analyzed
to
evaluate
the
efficiency
of
those
methods.
and
analyzed
to
efficiency
of
methods.
and
analyzed
to evaluate
evaluate the
the
efficiency
of those
those Control)
methods.
2015,
IFAC (International
Federation
of Automatic
Hosting by Elsevier Ltd. All rights reserved.
Keywords: Lot-sizing, flow-shop, sustainable manufacturing, peak power, production planning
Keywords: Lot-sizing,
Lot-sizing, flow-shop, sustainable
sustainable manufacturing, peak
peak power, production
production planning
Keywords:
Keywords: Lot-sizing, flow-shop,
flow-shop, sustainable manufacturing,
manufacturing, peak power,
power, production planning
planning
1. INTRODUCTION
a just for peak buffer inventory to reduce power demand
1. INTRODUCTION
INTRODUCTION
aduring
for
peak
buffer
inventory
to
reducethe
power
demand
1.
aa just
just
for
buffer
to
power
demand
peak periods
without affecting
throughput
1. INTRODUCTION
just the
for peak
peak
buffer inventory
inventory
to reduce
reducethe
power
demand
during
the
peak
periods
without
affecting
throughput
during
the
peak
periods
without
affecting
the
throughput
of the manufacturing
systems.
Luo
et al. (2013)
and Bego
The aim of a lot-sizing problem is to determine the amount during
the
peak
periods
without
affecting
the
throughput
of
the
manufacturing
systems.
Luo
et
al.
(2013)
and
Bego
The
aim of
of a lot-sizing
lot-sizing
problem
is to
to determine
the
amount
of
the
systems.
al.
and
Bego
al. manufacturing
(2014) also considered
the et
energy
aspect
The
aim
problem
is
amount
of
products
realize at
each period
of a giventhe
horizon
to et
of
the
manufacturing
systems. Luo
Luo
et
al. (2013)
(2013)
andduring
Bego
The
aim of aato
lot-sizing
problem
is to determine
determine
the
amount
et
al.
(2014)
also
considered
the
energy
aspect
during
of
products
to
realize
at
each
period
of
a
given
horizon
to
et
al.
(2014)
also
considered
the
energy
aspect
during
the
production
by
taking
into
account
changes
in
the
of
products
to
realize
at
each
period
of
a
given
horizon
to
satisfy
demand
while
minimizing
total
costs.
In
this
paper,
et
al.
(2014)
also
considered
the
energy
aspect
during
of
products
to
realize
at
each
period
of
a
given
horizon
to
production
by
taking
into
account
changes
in
the
satisfy
demand while
while
minimizing
total costs.
costs.
In this
this
paper, the
the
production
by
taking
into
account
changes
in
the
cost
of
electricity
from
one
period
to
another.
A
multisatisfy
demand
minimizing
total
In
paper,
the
Capacitated
Single
Item
Lot-Sizing
Problem
(CSILSP)
the
production
by
taking
into
account
changes
in
the
satisfy
demand while
minimizing
total costs.
In this
paper, cost of electricity from one period to another. A multithe
Capacitated
Single
Item
Lot-Sizing
Problem
(CSILSP)
cost
of
electricity
from
one
period
to
another.
A
multiobjective
mixed-integer
programming
model
for
the
flowthe
Capacitated
Single
Item
Lot-Sizing
Problem
(CSILSP)
in
a
flow-shop
system
is
considered.
cost
of
electricity
from
one
period
to
another.
A
multithe
SingleisItem
Lot-Sizing Problem (CSILSP) objective mixed-integer programming model for the flowin aaCapacitated
flow-shop
system
considered.
objective
mixed-integer
programming
model
the
flowscheduling
problem
was proposed
by for
Fang
al.
in
is
Nowadays,
thesystem
consideration
of the ecological aspects in shop
objective
mixed-integer
programming
model
for
theet
flowin
a flow-shop
flow-shop
system
is considered.
considered.
shop
scheduling
problem
was
proposed
by
Fang
et
al.
Nowadays,
the
consideration
of the
the for
ecological
aspects in
in
shop
scheduling
problem
was
proposed
by
Fang
et
al.
(2011)
whose
objective
is
to
find
the
schedule
that
miniNowadays,
the
consideration
of
ecological
aspects
manufacturing
systems
is
essential
the
protection
of
shop
scheduling
problem
was
proposed
by Fang
et
al.
Nowadays,
the systems
consideration
of the for
ecological
aspects in
(2011)
whose
objective
is
to
find
the
schedule
that
minimanufacturing
is
essential
the
protection
of
(2011)
whose
objective
is
to
find
the
schedule
that
minimizes
the
makespan,
the
peak
power
consumption
and
the
manufacturing
systems
is
essential
for
the
protection
of
the
planet.
To
reduce
the
environmental
impacts,
several
(2011)
whose
objective
is
to
find
the
schedule
that
minimanufacturing
systems
is
essential
for
the
protection
of
makespan,
the
peak
power
consumption
and
the
the planet.
planet. To
To
reduce
the environmental
environmental
impacts,
several mizes
the
makespan,
the
peak
power
consumption
and
carbonthe
footprint.
Xu et
(2014)
considered
the reduction
the
reduce
the
impacts,
several
alternatives
could
be presented
as reduction
of greenhouse
mizes
the
makespan,
theal.
peak
power
consumption
and the
the
the
planet. To
reduce
the environmental
impacts,
several mizes
carbon
footprint.
Xu
et
al.
(2014)
considered
the
reduction
alternatives
could
be
presented
as
reduction
of
greenhouse
carbon
footprint.
Xu
et
al.
(2014)
considered
the
reduction
of
the
power
demand
for
the
scheduling
problem
in
a
alternatives
could
be
presented
as
reduction
of
greenhouse
gas
emissions,
energy
consumption
and water
consump- carbon footprint. Xu et al. (2014) considered the reduction
alternatives
could
be
presented
as
reduction
of
greenhouse
of
the
power
demand
for
the
scheduling
problem
in
a
gas emissions,
emissions,
energy
consumption
and the
water
consumpof
the
power
demand
for
the
scheduling
problem
in
hybrid
flow-shop
system.
gas
energy
consumption
and
water
consumption.
According
to
Mouzon
et
al.
(2007),
expansion
of
of
the flow-shop
power demand
for the scheduling problem in a
a
gas
emissions,
energy
consumption
and the
water
consumphybrid
system.
tion.
According
to
Mouzon
et
al.
(2007),
expansion
of
hybrid
flow-shop
system.
As
far flow-shop
as lot-sizing
problems are concerned, Absi et al.
tion.
According
to
Mouzon
et
al.
(2007),
the
expansion
of
research
to minimize
the energy
consumption
is due to the
hybrid
system.
tion.
According
to
Mouzon
et
al.
(2007),
the
expansion
of
far as
as
lot-sizingfour
problems
are concerned,
concerned,
Absicarbon
et al.
al.
research to
to
minimize
the
energy consumption
consumption
is due
due to the
the As
far
lot-sizing
problems
are
Absi
et
(2013)
suggested
alternatives
to introduce
research
energy
is
increase
of minimize
the price the
of electricity
and the intensification
As
far as
lot-sizingfour
problems
are concerned,
Absicarbon
et al.
research
to
minimize
the
energy consumption
is due to
to the As
(2013)
suggested
alternatives
to
introduce
increase
of
the
price
of
electricity
and
the
intensification
(2013)
suggested
four
alternatives
to
introduce
carbon
emission
constraint
in
the
single
item
lot-sizing
problem;
increase
of
the
price
of
electricity
and
the
intensification
of
the global
warming.
suggested four
alternatives
to lot-sizing
introduceproblem;
carbon
increase
of the
price of electricity and the intensification (2013)
emission
in
the
single
item
of the
the
global
warming.
emission
constraint
in
the
single
item
lot-sizing
problem;
areconstraint
periodic carbon
emission
constraint,
cumulative
of
warming.
As
far global
as scheduling
problems are concerned, Mouzon and which
emission
constraint
in
the
single
item
lot-sizing
problem;
of
the
global
warming.
periodic
carbon
emission
cumulative
As far
far as
as scheduling
scheduling
problemsand
areMouzon
concerned,
Mouzon
and which
are
periodic
carbon
constraint,
cumulative
carbonare
emission
global constraint,
carbon emission
conAs
problems
are
concerned,
Mouzon
and
Yildirim
and Yildirim
(2012)
proposed
which
are
periodicconstraint,
carbon emission
emission
constraint,
cumulative
As
far as (2008)
scheduling
problemsand
areMouzon
concerned,
Mouzon
and which
carbon
emission
constraint,
global
carbon
emission
Yildirim
(2008)
and
Yildirim
(2012)
proposed
carbon
emission
constraint,
global
carbon
emission
constraint
and
rolling
carbon
emission
constraint.
Yu etconal.
Yildirim
(2008)
and
Yildirim
Mouzon
(2012)
proposed
aYildirim
mathematical
model
whichand
minimizes
total
tardiness
of carbon emission constraint, global carbon emission
con(2008)
and
Yildirim
and
Mouzon
(2012)
proposed
straint
and
rolling
carbon
emission
constraint.
Yu
et
al.
a mathematical
mathematical
model
which
minimizes
total
tardiness
of
straint
and
rolling
carbon
emission
constraint.
Yu
et
(2013)
presented
a
lot-sizing
model
with
carbon
emission
ajobs
model
which
minimizes
total
tardiness
of
and total energy
on a single
machine. of
A straint and rolling carbon emission constraint. Yu et al.
al.
a
mathematical
modelconsumption
which minimizes
total tardiness
(2013)
presented
a
lot-sizing
model
with
carbon
emission
jobs
and
total
energy
consumption
on
a
single
machine.
A
(2013)
presented
lot-sizing
carbon
emission
constraint
in eachaa period.
Formodel
Heck with
and Schmidt
(2010),
jobs
and
total
energy
consumption
on
aaobjective,
single
machine.
A
generalized
case,
which
has the same
has been
(2013)
presented
lot-sizing
model
with
carbon
emission
jobs
and
total
energy
consumption
on
single
machine.
A
constraint
in carbon
each period.
period.
For
Heck and
and
Schmidt
(2010),
generalizedby
case,
which
has
the same
same
objective, has
has
been
in
each
Heck
(2010),
power usage,
dioxideFor
emission
and Schmidt
water consumpgeneralized
case,
the
objective,
been
developed
Liuwhich
et al.has
(2014)
who considered
a set
of constraint
constraint
in carbon
each period.
For
Heck and
Schmidt
(2010),
generalized
case,
which
has
the same
objective, has
been
power
usage,
dioxide
emission
and
water
consumpdeveloped
by
Liu
et
al.
(2014)
who
considered
a
set
of
power
usage,
carbon
dioxide
emission
and
water
consumption
are
considered
in
their
lot-size
study.
developed
by
Liu
et
al.
(2014)
who
considered
a
set
of
machines.
Two
zero-one
non-linear
programming
models
power
usage,
carbon
dioxide
emission
and
water
consumpdeveloped
by
Liu
et
al.
(2014)
who
considered
a
set
of
tion
are
considered
in
their
lot-size
study.
machines.
Two
zero-one were
non-linear
programming
models
tion
are
considered
in
lot-size
study.
this
lot-sizing
problem
flow-shop is considmachines.
Two
zero-one
non-linear
programming
for
a system
production
presented
by Wang models
and Li In
tion
are paper,
considered
in their
their
lot-sizein
study.
machines.
Two
zero-one were
non-linear
programming
models
In
this
paper,
lot-sizing
problem
in
flow-shop
considfor
a
system
production
presented
by
Wang
and
Li
In
this
paper,
lot-sizing
problem
in
flow-shop
is
considered.
Several
research
studies
have
dealt
with is
type
for
a
system
production
were
presented
by
Wang
and
Li
(2013)
for
the
total
electricity
consumption
minimization
thisSeveral
paper,research
lot-sizingstudies
problem
in dealt
flow-shop
isthis
considfor
a system
production
were presented
by Wang
and Li In
ered.
have
with
this
type
(2013)
for
the
total
electricity
consumption
minimization
ered.
Several
research
studies
have
dealt
with
this
type
of
system.
Babaei
et
al.
(2011)
and
Babaei
et
al.
(2014)
(2013)
for
the
total
electricity
consumption
minimization
and
for
the
total
electricity
cost
minimizations,
while
ered.
Several
research
studies
have
dealt
with
this
type
(2013)
forthe
thetotal
total electricity
electricity consumption
minimization
of
system.
Babaei
et
al.
(2011)
and
Babaei
et
al.
(2014)
and
for
cost
minimizations,
while
of
system.
Babaei
et
al.
(2011)
and
Babaei
et
al.
(2014)
presented
a
multi-level
and
multi-period
capacitated
lotand
for
the
total
electricity
cost
minimizations,
while
maintaining
an
amount
of average
cumulative
production
of
system.
Babaei
et
al.
(2011)
and
Babaei
et
al.
(2014)
and
for
the
total
electricity
cost
minimizations,
while
presented
aascheduling
multi-level
and
multi-period
capacitated
lotmaintaining
an amount
amount
of average
average
cumulative
production
presented
multi-level
and
multi-period
capacitated
sizing
and
problem
with
sequence-dependent
maintaining
an
of
cumulative
production
that
is
not
lower
than
the
required
level.
Starting
from
presented
ascheduling
multi-levelproblem
and multi-period
capacitated lotlotmaintaining
an amount
of average
cumulative
production
sizing
and
with
sequence-dependent
that
is
not
lower
than
the
required
level.
Starting
from
sizing and
and
scheduling
problem
withover
sequence-dependent
setups,
backlogging
andproblem
setup carry
in flow-shop systhat
is
not
lower
than
the
required
level.
Starting
from
an
advanced
planning
and
scheduling
system,
the energy
sizing
scheduling
with
sequence-dependent
that
is
not
lower
than
the
required
level.
Starting
from
backlogging
and
setup
carry
over
ina flow-shop
sysan advanced
advanced
planning
andmodifies
scheduling
system,
the
energy setups,
backlogging
and
setup
over
system. Ramezanian
al. (2013)
developed
mixed integer
an
planning
and
scheduling
energy
aware
scheduling
method
thesystem,
originalthe
timetable
setups,
backlogginget
and
setup carry
carry
over in
ina flow-shop
flow-shop
sysan
advanced
planning
andmodifies
scheduling
system,
the
energy setups,
tem.
Ramezanian
et
al.
(2013)
developed
mixed
integer
aware
scheduling
method
the
original
timetable
tem.
Ramezanian
et
al.
(2013)
developed
a
mixed
integer
programming
model
for
lot-sizing
and
scheduling
probaware
scheduling
method
modifies
the
original
timetable
to
reduce
the shopmethod
floor peaks
power.
This method
was tem. Ramezanian et al. (2013) developed a mixed integer
aware
scheduling
modifies
the
original
timetable
model
for
lot-sizing
and
scheduling
probto reduce
reduce by
theBruzzone
shop floor
floor
peaks
power.
This method
method was
was programming
programming
model
lot-sizing
and
scheduling
probwith availability
constraints
to minimize
to
the
shop
power.
This
proposed
et peaks
al. (2012).
A mixed-integer
programming
model for
for
lot-sizingthat
and aims
scheduling
probto
reduce by
theBruzzone
shop floor
peaks
power.
This method prowas lem
lem
with
availability
constraints
that
aims
to
minimize
proposed
et
al.
(2012).
A
mixed-integer
prolem
with
availability
constraints
that
aims
to
minimize
the
production,
holding
and
sequence-dependent
setup
proposed
by
Bruzzone
et
al.
(2012).
A
mixed-integer
programming
problem
aims
to
minimize
total
tardiness
and
lem
with
availability
constraints
that
aims
to
minimize
proposed
by
Bruzzone
et
al.
(2012).
A
mixed-integer
prothe
and
sequence-dependent
setup
gramming while
problem
aims to
to minimize
minimize
total demand
tardinessofand
and
production,
holding
and
sequence-dependent
setup
costs.production,
Mortezaei holding
and Zulkifli
developed a mixedgramming
problem
aims
total
tardiness
makespan
respecting
powers peak
the the
the
production,
holding
and (2013)
sequence-dependent
setup
gramming
problem
aims tothe
minimize
total demand
tardinessofand
costs.
Mortezaei
and
Zulkifli
(2013)
developed
a
mixedmakespan
while
respecting
the
powers
peak
the
costs.
Mortezaei
and
Zulkifli
(2013)
developed
a
mixedinteger
model
for
lot-sizing
problem
in
flow-shop
system.
makespan
while
respecting
the
powers
peak
demand
of
the
manufacturing
system.
Fernandez
et al.
(2013)
developed
costs.
Mortezaei
and
Zulkifli
(2013)
developed
a
mixedmakespan
while
respecting
the
powers
peak
demand
of
the
model
for
lot-sizing
in
flow-shop
system.
manufacturing system.
system. Fernandez et
et al. (2013)
(2013) developed integer
integer
model
for
lot-sizing
problem
in
system.
objective
minimizeproblem
the production,
storage
and
manufacturing
integer
model is
forto
lot-sizing
problem
in flow-shop
flow-shop
system.
manufacturing system. Fernandez
Fernandez et al.
al. (2013) developed
developed The

The
objective
is
to
minimize
the
production,
storage
The authors would like to thank the Champagne-Ardenne region.
The objective
objective is
is to
to minimize
minimize the
the production,
production, storage
storage and
and
The
and

Oussama
Oussama
Oussama
Oussama

authors would
like to
thank the
Champagne-Ardenne region.

The
 The
The authors
authors would
would like
like to
to thank
thank the
the Champagne-Ardenne
Champagne-Ardenne region.
region.
Copyright
2015 IFAC
760 Hosting by Elsevier Ltd. All rights reserved.
2405-8963 2015,
IFAC (International Federation of Automatic Control)
Copyright
2015
IFAC
760
Copyright

2015
IFAC
760
Peer review
of International Federation of Automatic
Copyright
under
2015 responsibility
IFAC
760Control.
10.1016/j.ifacol.2015.06.169

INCOM 2015
Oussama Masmoudi et al. / IFAC-PapersOnLine 48-3 (2015) 727732
May 11-13, 2015. Ottawa, Canada

728

makespan costs. Sahling et al. (2009) presented a mixedinteger model for multi-level capacitated lot-sizing problem with objective to minimize setup, holding and overtime costs. Mohammadi et al. (2010) proposed a multiproduct multi-capacitated lot-sizing and scheduling problem with sequence-dependent setups. The objective of this
model is to minimize the sequence-dependent setup costs,
holding and production costs. For this type of problem,
neither energy consumption nor environmental impacts
are taken into account.
Therefore, this paper proposes a Capacitated Single Item
Lot-Sizing Problem in flow-shop with energy consideration. According to Goldman (2010), there exist two types
of demand response programs: price driven and event
driven. For the first category, the price of electricity varies
over different time periods and thus leading to the existence of off-peak period and on-peak period. The price
during the off-peak period is lower than the on-periods
one. Consequently, the manufacturers planify and organize the activities in a way that minimizes the electricity
costs. Time Of Use (TOU), Critical Peak Pricing (CPP)
and Real-Time Pricing are some example of price driven
programs. In the event-driven program, rewards will be
allocated to customers who reduce their energy consumption in response to specific triggering events, depending
on several factors like weather conditions and systems
economics.
The outline of this paper is organized as follows. In section
2, a definition of the problem, assumptions and the models
are presented. Section 3 reports the numerical examples
obtained by exact methods of the proposed models. The
last section presents concluding remarks and future studies.
2. PROBLEM DESCRIPTION
A typical manufacturing system with N reliable machines
and N buffers with infinite capacity as illustrated in Fig. 1
is considered. Rectangles denote machines and circles
denote buffers.

Fig. 1. A typical manufacturing system with N machines

and N buffers

Fig. 2. A possible scenario

products start to be available at the output of machine
m1 .
2.2 Mathematical formulation
The parameters used in the model are:
T : Number of periods.
m : The power of the machine m.
Cot : The price of electricity during period t.
pm : Processing time for machine m.
h : Holding cost per unit.
wm,t : Setup cost of machine m in period t.
dt : External demand at period t.
Lt : Length of period t.
M : A large real number.
t : Price of power in period t.
t : The allowed maximal power in period t.
m,t = m pm Cot : Electrical consumption cost of
machine m at period t.
The decision variables used in the model are:

2.1 Assumptions
Principal decision variables

The model assumptions are as follows:

The horizon is split into T periods where each one is
characterized by its electricity price.
The demand must always be satisfied at each period.
For each period, the demand is known in advance.
The first machine is never starved.
The last machine is never blocked.
Only one product type is to produce.
Capacity of each machine is limited.
A machine m cannot begin producing xm,t products
in period t if this quantity is not available at the
output of the previous machine m 1.

This last assumption, translating handling equipments

constraints, results in production configuration in which
overlaps may be encountered in a given period.
In the example illustrated in Fig. 2, all the production
(covering demand for the two periods t1 and t2) is carried
out on machine m1 during t1 . The production of quantity
x2,1 on machine m2 at period t1 can only start once x2,1
761

xm,t : Quantity produced on machine m in period t.

Im,t : Inventory level downstream of machine m at the end
of period t.
Cm,t : Completion time of machine m in period t.
ym,t : A binary variable, equal to 1 if machine m is setup
in period t, 0 otherwise.
Ptmax : The maximum power demand during period t.
Complementary decision variables

vm,t : A binary variable, equal to 1 if the quantity xm,t is

available in buffer m 1 at the beginning of period t, 0
otherwise.
fm,r,t : A binary variable, equal to 1 if
Cr,t Cm,t xm,t pm , 0 otherwise.
gm,r,t : A binary variable, equal to 1 if Cm,t Cr,t , 0
otherwise.

INCOM 2015
Oussama Masmoudi et al. / IFAC-PapersOnLine 48-3 (2015) 727732
May 11-13, 2015. Ottawa, Canada

M inz =
+

T

t=1

T 
N


t=1 m=1

fm,m,t = gm,m,t = 0

(m,t xm,t + h Im,t + wm,t ym,t )

t Ptmax
(1)

xN,t + IN,t1 = dt + IN,t

t = 2, ..., T

xm,t + Im,t1 = Im,t + xm+1,t

m = 1, ..., N 1, t = 2, ..., T
xm,t ym,t

T


=t

m = 1, ..., N, t = 1, ..., T

(2)
(3)

(4)

Im1,t1 xm,t + M vm,t

m = 2, ..., N, t = 2, ..., T

(5)

xm,t Im1,t1 + M (1 vm,t )

m = 2, ..., N, t = 2, ..., T

(6)

Cm,t xm,t pm
(Cm1,t xm1,t pm1 + (xm,t Im1,t1 ) pm1 )
(1 vm,t ) m = 2, ..., N, t = 2, ..., T
(7)
Cm,1 xm,1 pm,1 Cm1,1 xm1,1 pm1,1
+xm,1 pm1 m = 2, ..., N
Cr,t Cm,t + xm,t pm M fm,r,t
m = 1, ..., N, r = 1, ..., N = m, t = 1, ..., T
Cm,t Cr,t M gm,r,t 1
m = 1, ..., N, r = 1, ..., N = m, t = 1, ..., T
Ptmax m ym,t +

N


r=1,r=m

(fm,r,t gm,r,t r )

Im,1 = xm,1 xm+1,1

(9)
(10)

(11)

(12)

m = 1, ..., N 1

(13)

IN,1 = xN,1 d1

(14)

C1,1 x1,1 p1 = 0

(15)

Cm,t xm,t pm Lt ym,t

Cm,t Lt

xm,t , Im,t , Cm,t N

(20)

ym,t , vm,t , fm,r,t , gm,r,t {0, 1}

(21)

The objective function (1) minimizes the production costs

composed of electrical consumption costs, storage costs,
setup costs and power required costs. Equations (2) and
(3) present the flow balance constraints. Constraint (4)
establishes the relation between the quantity produced
and the state of each machine m in each period t. The
existence of sufficient inventory at the beginning of each
period is given by constraints (5)-(8). They assure the
vertical interaction. In other words, the production on
each level can only be started when a sufficient amount of
product exists on the adjacent upstream buffer. Contraint
(9) compares the completion time of a machine with the
starting time of an-other machine whereas constraint (10)
compares the completion time of two different machines.
The calculation of the power used per period is given
by constraint (11). The maximum allowable power per
period is represented by constraint (12). Equations (13)(15) represent the initial conditions. The constraints (16)(18) force the starting and the completion time to be
in the same period. Equation (19) eliminates the case of
comparaison of the starting and completion time and the
comparaison of the completion times for the same machine.

Because of constraints (7) and (11), the previous mathematical model is non-linear. In order to assure the linearity, constraints (22) and (23) are proposed to replace
constraints (7) and (11) respectively.

Cm,t xm,t pm Cm1,t xm1,t pm1 +

(xm,t Im1,t1 ) pm1 M vm,t
m = 2, ..., N, t = 2, ..., T
Ptmax m ym,t +

N


r=1,r=m

(22)

(fm,r,t + gm,r,t 1) r

m = 1, ..., N, t = 1, ..., T

(23)

3. NUMERICAL EXPERIMENTS

(16)

m = 1, ..., N, t = 1, ..., T
(17)

m = 1, ..., N, t = 1, ..., T

(19)

2.3 Linearization of non-linear constraints

t = 1, ..., T

Cm,t xm,t pm 0 m = 1, ..., N, t = 1, ..., T

m = 1, ..., N, t = 1, ..., T

(8)

m = 1, ..., N, t = 1, ..., T
Ptmax t

729

(18)
762

In order to evaluate the behaviour of the proposed models, numerical experiments are presented. These problem
formulations are solved by LINGO 14.0 on an Intel Core
i7 with 3.0 GHz and 8 GB RAM. Inspired by the works
of Babaei et al. (2014) and Mohammadi et al. (2010),
the required parameters are extracted from the following
uniform distributions:
pm U(5,10), h U(1,6) and wm,t U(50,100). All these
parameters are considered to be entire and obtained by
rounding to the nearest integer real value.

INCOM 2015
Oussama Masmoudi et al. / IFAC-PapersOnLine 48-3 (2015) 727732
May 11-13, 2015. Ottawa, Canada

730

The price of electricity is inspired by the work of Wang

and Li (2013) (Table 1).
Table 1. TOU pricing profile
Season
June-Sep
Oct-May

Time of day
7pm-1pm
1pm-7pm
9pm-10am
10am-9pm

Cot ($/KWh)
0.08274
0.16790
0.08274
0.11224

t ($/KW)
0
18.8
0
8.12

We assume that the machine electric power m = 25KW

for each machine and the parameter t follows a stage
distribution. To assure feasible solutions d(t) U(0,100)
for N = 2 and N = 3, d(t) U(0,50) for N = 4, d(t)
U(0,20) for N = 5 and d(t) U(0,10) for N = 10.
For each problem size (N,T), 5 instances are generated.
The computational time (in seconds), for the linear formulation (L.F) and non-linear formulation (NL.F) are
illustrated in Tables 2, 3, 4 and 5. Here, we set a
completion time limit of 1 hour.
For N = 2, we have obtained the optimal solutions within
a short time for all scenarios. As it can be seen from Table
2, the average for the computational time of the linear
and non-linear formulations, expect the instances (2,10),
are close. For this instance, the average for computational
time of non-linear formulations is less than the linear ones.
Therefore, the non-linear formulation is recommended for
problems of this size.
For N = 3, we have obtained the optimal solutions for all
scenarios. As it can be seen from Table 3, the computational time averages of linear and non-linear formulations
are close for the instances (3,2) and (3,5), while it is more
interesting to resort to the non-linear formulation for (3,8)
instance.
For N = 4, we have obtained the optimal solutions
for all scenarios. The Table 4 shows that the linear
and non-linear formulations are equivalent in terms of
computational time for the instance (4,2), while there is a
huge difference between the computational times for the
instance (4,5). In this case, the non-linear formulation is
more efficient.
For N = 5, after one hour of resolution, some scenarios
do not obtain the optimal solution. For example, the
solution given by the scenario (5, 5) 2 of linear formulation is greater than the non-linear one, that obtains
the optimal solution within 87.47 seconds with a Gap of
0.56% (Gap = (f easable Solution Optimal Solution)
100/Optimal Solution). For the (5, 8) 2, the linear formulation does not obtain the optimal solution within one hour
of resolution, while, the non-linear one obtain the optimal
solution within 5.86 seconds. The same thing for (5, 8) 3
scenario. The Table 5 shows that it is more interesting
to resort to the non-linear formulation which guarantees
optimal solutions within a shorter time.
For N = 10, after one hour of resolution, the linear formulations do not find the optimal solution while the nonlinear formulations obtained the optimal solution within a
reasonable time. For example, the computational time of
the scenario (10,8) 1 is 26.32 seconds for the non-linear and
+3600 seconds for the linear one, with a Gap of 32.11%.
763

Table 2. Comparative results for the CPu time

of linear and non-linear formulations (N = 2)
Problem size (N,T)
(2, 2) 1
(2, 2) 2
(2, 2) 3
(2, 2) 4
(2, 2) 5
Average
(2, 5) 1
(2, 5) 2
(2, 5) 3
(2, 5) 4
(2, 5) 5
Average
(2, 8) 1
(2, 8) 2
(2, 8) 3
(2, 8) 4
(2, 8) 5
Average
(2, 10) 1
(2, 10) 2
(2, 10) 3
(2, 10) 4
(2, 10) 5
Average

CPu of L.F
0.09
0.08
0.06
0.08
0.09
0.08
0.69
0.23
0.48
0.08
0.51
0.4
2.12
2.86
0.34
2.87
3.10
2.26
4.94
3.14
0.33
73.51
10.79
18.55

CPu of NL.F
0.11
0.08
0.06
0.06
0.06
0.07
0.69
0.25
1.09
0.19
0.51
0.54
2.51
4.17
0.36
4.65
2.15
2.77
3.34
2.5
0.39
35.22
10.73
10.44

Table 3. Comparative results for the CPu time

of linear and non-linear formulations (N = 3)
Problem size (N,T)
(3, 2) 1
(3, 2) 2
(3, 2) 3
(3, 2) 4
(3, 2) 5
Average
(3, 5) 1
(3, 5) 2
(3, 5) 3
(3, 5) 4
(3, 5) 5
Average
(3, 8) 1
(3, 8) 2
(3, 8) 3
(3, 8) 4
(3, 8) 5
Average

CPu of L.F
0.27
0.33
0.05
0.09
0.08
0.17
0.23
3.23
0.61
0.63
4.53
1.85
15.27
1768.64
2.18
1396.77
3.17
637.21

CPu of NL.F
0.37
0.28
0.08
0.10
0.07
0.18
3.38
3.70
0.45
0.54
5.57
2.73
3.67
13.90
1.04
8.62
1.82
5.81

Table 4. Comparative results for the CPu time

of linear and non-linear formulations (N = 4)
Problem size (N,T)
(4, 2) 1
(4, 2) 2
(4, 2) 3
(4, 2) 4
(4, 2) 5
Average
(4, 5) 1
(4, 5) 2
(4, 5) 3
(4, 5) 4
(4, 5) 5
Average

CPu of L.F
0.11
0.16
0.12
0.12
0.09
0.12
49.05
+3600
2.42
+3600
51.69
730.34

CPu of NL.F
0.14
0.12
0.08
0.12
0.09
0.11
5.98
15.96
1.15
273.87
6.28
60.65

INCOM 2015
Oussama Masmoudi et al. / IFAC-PapersOnLine 48-3 (2015) 727732
May 11-13, 2015. Ottawa, Canada

731

Fig. 3. The Gantt chart of the production planning (3, 5) 4

Table 5. Comparative results for the CPu time
of linear and non-linear formulations (N = 5)
Problem size (N,T)
(5, 2) 1
(5, 2) 2
(5, 2) 3
(5, 2) 4
(5, 2) 5
Average
(5, 5) 1
(5, 5) 2
(5, 5) 3
(5, 5) 4
(5, 5) 5
Average
(5, 8) 1
(5, 8) 2
(5, 8) 3
(5, 8) 4
(5, 8) 5
Average

CPu of L.F
0.12
0.08
0.11
0.11
0.06
0.10
0.30
+3600
18.39
34.45
4.49
731.53
24.68
+3600
+3600
47.42
366.75
1527.77

Table 7. Data of instance (3, 5) 4

dt
w1,t
w2,t
w3,t
t

CPu of NL.F
0.12
0.11
0.11
0.09
0.12
0.11
0.50
87.47
0.56
1.19
0.84
18.12
2.33
5.86
5.62
7.12
34.71
11.13

CPu of L.F
0.10
0.11
0.12
0.09
0.08
0.10
0.23
+3600
781.65
453.67
+3600
1687.11
+3600
+3600
+3600
+3600
+3600
+3600

2
18
76
83
57
70

3
8
99
75
97
67

4
11
88
53
57
52

5
21
67
85
95
61

the first machine due to the vertical interaction condition.

In this case, the optimal solution is to produce all the
demand (d1 + d2 = 31) during the first period. This
solution is obtained without overlap.
The data of the second problem are illustrated in Table
7. Fig. 3 represents the optimal planning that minimizes
the objective function. In this case, during the first period,
which represents an off-peak period, it is more convenient
to produce the most products and put them in stock
(x1,1 = x2,1 = 47 and x3,1 = 38). Due to the capability
of the production system, machine m3 must produce
during the second period, which corresponds to an on-peak
period, to satisfy the demand (x3,2 = 9). The remaining
demand is produced during the third period rather than
producing during t4 and t5 .

Table 6. Comparative results for the CPu time

of linear and non-linear formulations (N = 10)
Problem size (N,T)
(10, 2) 1
(10, 2) 2
(10, 2) 3
(10, 2) 4
(10, 2) 5
Average
(10, 5) 1
(10, 5) 2
(10, 5) 3
(10, 5) 4
(10, 5) 5
Average
(10, 8) 1
(10, 8) 2
(10, 8) 3
(10, 8) 4
(10, 8) 5
Average

1
29
90
58
75
66

CPu of NL.F
0.20
0.17
0.17
0.17
0.13
0.17
0.41
25.50
35.80
8.83
10.60
16.23
26.32
333.26
311.73
+3600
239.79
842.22

Fig. 4. The Gantt chart of the production planning (2, 2) 2

4. CONCLUSION

To explain the obtained results, examples of instances

(2, 2) 2 and (3, 5) 4 are represented and interpreted. For
the first problem, d1 = 30 and d2 = 1, h = 4, w1,1 = 55,
w1,2 = 68, w2,1 = 57, w2,2 = 58, 1 = 34 and 2 = 25. In
this case the season (June-September) is considered. The
starting and completion time of each machine, the length
of each period are expressed in minutes. The starting time
of each period is equal to 0. Fig. 4 shows that the first and
the second machine produce only during the first period
which corresponds to the off-peak period while the second
one corresponds to the on-peak period. The starting time
of the second machine is equal to the completion time of
764

The contribution of this paper is to define a new lot-sizing

problem for flow-shop system with energy consideration.
To model the production planning, two mixed integer
programming formulations are proposed. The objective
of these mathematical formulations is to minimize the
total production costs that consists of electrical, holding,
setup and power demand costs. These models allow to
determine the optimal quantities per period where each
one is characterized by an electricity cost, duration and
maximum available power, using solver.
The computational study indicates that the non-linear
problem formulation obtains the optimal solution within
a shorter time than the linear problem formulation.
Other research will concentrate on developing theoretical
proprieties, heuristics and meta-heuristics to solve this
problem.

INCOM 2015
Oussama Masmoudi et al. / IFAC-PapersOnLine 48-3 (2015) 727732
May 11-13, 2015. Ottawa, Canada

732

REFERENCES
Absi, N., Dauz`ere-Per`es, S., Kedad-Sidhoum, S., Penz, B.,
and Rapine, C. (2013). Lot sizing with carbon emission
constraints. European Journal of Operational Research,
227(1), 5561.
Babaei, M., Mohammadi, M., and FatemiGhomi, S.
(2011). Lot sizing and scheduling in flow shop with
sequence-dependent setups and backlogging. International Journal of Computer Applications, 29(8), 5259.
Babaei, M., Mohammadi, M., and FatemiGhomi, S.
(2014). A genetic algorithm for the simultaneous lot
sizing and scheduling problem in capacitated flow shop
with complex setups and backlogging. The International
Journal of Advanced Manufacturing Technology, 70(14), 125134.
Bego, A., Li, L., and Sun, Z. (2014). Identification of
reservation capacity in critical peak pricing electricity
demand response program for sustainable manufacturing systems. International Journal of Energy Research,
38(6), 728736.
Bruzzone, A., Anghinolfi, D., Paolucci, M., and Tonelli, F.
(2012). Energy-aware scheduling for improving manufacturing process sustainability: A mathematical model
for flexible flow shops. CIRP Annals - Manufacturing
Technology, 61(1), 459462.
Fang, K., Uhan, N., Zhao, F., and Sutherland, J. (2011). A
new shop scheduling approach in support of sustainable
manufacturing. In Glocalized Solutions for Sustainability in Manufacturing, 305310.
Fernandez, M., Li, L., and Sun, Z. (2013). Just-for-peak
buffer inventory for peak electricity demand reduction
of manufacturing systems. International Journal of
Production Economics, 146(1), 178 184.
Goldman, C. (2010). Coordination of energy efficiency
and demand response. Lawrence Berkeley National
Laboratory.
Heck, M. and Schmidt, G. (2010). Lot-size planning
with non-linear cost functions supporting environmental sustainability. In Networked Digital Technologies,
volume 88, 16.
Liu, Y., Dong, H., Lohse, N., Petrovic, S., and Gindy, N.
(2014). An investigation into minimising total energy
consumption and total weighted tardiness in job shops.
Journal of Cleaner Production, 65, 87 96.
Luo, H., Du, B., Huang, G.Q., Chen, H., and Li, X. (2013).
Hybrid flow shop scheduling considering machine electricity consumption cost. International Journal of Production Economics, 146(2), 423439.
Mohammadi, M., Torabi, S., Fatemi Ghomi, S., and
Karimi, B. (2010).
A new algorithmic approach
for capacitated lot-sizing problem in flow shops with
sequence-dependent setups. The International Journal
of Advanced Manufacturing Technology, 49(1-4).
Mortezaei, N. and Zulkifli, N. (2013). Integration of lot
sizing and flow shop scheduling with lot streaming.
Journal of Applied Mathematics, 2013.
Mouzon, G. and Yildirim, M.B. (2008). A framework to
minimise total energy consumption and total tardiness
on a single machine. International Journal of Sustainable Engineering, 1(2), 105116.
Mouzon, G., Yildirim, M.B., and Twomey, J. (2007). Operational methods for minimization of energy consumption of manufacturing equipment. International Journal
765

of Production Research, 45(18-19), 42474271.

Ramezanian, R., Saidi-Mehrabad, M., and Fattahi, P.
(2013). Mip formulation and heuristics for multistage capacitated lot-sizing and scheduling problem with
availability constraints. Journal of Manufacturing Systems, 32(2), 392401.
Sahling, F., Buschkuhl, L., Tempelmeier, H., and Helbe,
S. (2009). Solving a multi-level capacitated lot sizing
problem with multi-period setup carry-over via a fixand-optimize heuristic. Computers and Operations Research, 36(9), 2546 2553.
Wang, Y. and Li, L. (2013). Time-of-use based electricity demand response for sustainable manufacturing
systems. Energy, 63, 233244.
Xu, F., Weng, W., and Fujimura, S. (2014). Energyefficiency scheduling for flexible flow-shops by using
mip. Proceeding of the 2014 Industrial and Systems
Engineering Research Conference.
Yildirim and Mouzon (2012). Single-machine sustainable
production planning to minimize total energy consumption and total completion time using a multiple objective
genetic algorithm. IEEE transactions on engineering
management, 59(4), 585597.
Yu, Y., Hong, Z., Chu, C., and Liang, L. (2013). A dynamic lot sizing model with carbon emission constraint
and multi-mode production: A general property and a
polynomially solvable case. In ICNSC13, 413417.