Amsterdam University of Applied Sciences
CPN-simulation methodology for the boarding process of aircraft
Mujica Mota, Miguel; Flores, Idalia; Guimarans, Daniel
Publication date 2014
Document Version Final published version
Link to publication
Citation for published version (APA):
Mujica Mota, M., Flores, I., & Guimarans, D. (2014). CPN-simulation methodology for the boarding process of aircraft. Hogeschool van Amsterdam.
General rights
It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).
Disclaimer/Complaints regulations
If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please contact the library:
https://www.amsterdamuas.com/library/contact/questions, or send a letter to: University Library (Library of the University of Amsterdam and Amsterdam University of Applied Sciences), Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.
Download date:27 Nov 2021
CPN-SIM
ABSTRACT The boarding identified bo aircraft arrive a methodolo boarding proc understand in and cause-eff boarding pro integrated at program to si combination such a way efficient for dynamics are approach can configuration process whil process.
Keywords: sim coloured petri 1. INTROD The turnarou between cons airport. Depe and space ava long (Bazarg depend upon flight schedu from business times of the difficult to ho influence the congestion at The step aircraft are:
Prior passe lugg
The
Bloc
The
MULATION
(a)
Aviat
(b)
Fac
(c)
Op
(a)
m.mu
T
g process of ottlenecks in es to an airpor ogy that mod
cess using co n detail what a ffect relationsh ocess. Then a different imulate in det of both techn
that the de r developing e difficult to n be used fo ns and improve
le reducing
mulation, disc i nets, aircraft DUCTION und of an airc
secutive flight ending on the ailable, the tur gan 2004). F the business ules are differ s model to bu ground han omogenise du e operations
the airside, et ps followed b
r preparation engers are o gage and docum
plane arrives ck-In is perfor passengers an
N METHOD
Miguel Mu
ion Academy cultad de Inge ptimisation Re ujica.mota@h
f an aircraft the turnaro rt. The present dels the key oloured Petri
are the main hips that hin the differen level using a tail the boardi
niques reinfo evelopment p g micro-mod o understand.
or testing dif e the efficienc
the cost ass
crete event sim t boarding pro
craft is the t ts when the a e type of com rnaround will Furthermore t model of the rent and the usiness model.
dling operati ue to the diffe such as ha tc.
by a typical
for boardin rganised and mentation is c at the parking rmed.
nd bags disem
DOLOGY
ujica Mota
(a),
, Amsterdam U eniería, Univer
esearch Group hva.nl,
(b)idalia
is one of t ound when t t article presen y aspects of
nets in order micro-dynam nder the smoo
nt models a a discrete-eve ing process. T
rce mutually process is ve dels when t . This kind fferent boardi cy of the studi sociated to t
mulation, ocess
emporary spa aircraft is in t mpany, the tim
be more or le this length w company as operations va In addition, t ons result ve erent factors th andling agen stopover of
ng: the lines d all their ha
checked.
g stand.
mbark.
FOR THE
, Idalia Flores
University of rsidad Nacion p, National IC a@unam.mx,
the the nts f a to mics oth are ent The in ery the of ing ied the
ace the me ess will all ary the ery hat nts, an
of and
fo to co op th tim Fi th
F
E BOARDIN
s
(b), Daniel G
f Applied Scie nal Autónoma CT Australia (N
(c)
daniel.guim
The pl
Wheth cleanin
When cleanin for t Simult loaded
During coordi docum
As soo and pa
The ch
The p corresp The busine orces them to o reduce turna ompetitive. Fo perations acq hose operation me when the igure 1 illustra he turnaround
Figure 1: The o
NG PROC
Guimarans
(c)nces, The Net de México, M NICTA), Aus marans@nicta.
lane is fuelled her there is ng team will p
the last p ng services ha the next fl
taneously the d for the new f
g the board inator shall mentation to th
on as the plan assengers, the hocks are rem plane perform
ponding runw ess model of take out som around times a
or this reaso quire more an
ns are the one e aircraft is ates the differ of an aircraft
outstation turn
ESS OF AI
therlands México
stralia com.au
d.
a scheduled proceed to cle passenger lea ave finished, t light shall e bags shall flight.
ding of pas deliver th he captain.
ne is loaded w doors are clos oved.
ms the taxiing way for the tak the low cost c e of the opera and thus maki n the boardin nd more imp es that consum performing rent processes
(Airbus 2014
naround time
IRCRAFT
cleaning, the an the plane.
aves and the the passenger
be boarded l start to be ssengers, the he necessary with fuel, bag
sed.
g towards the ke-off.
carriers (LCC ations in orde ing them more
ng/deboarding portance since me most of the the stopover s that compose
).
for the A-320 e e s d.
e e y s
e
C) er e g e e r.
e
0
From Figure 1 it can be noticed that the boarding and deboarding operations are currently in the critical path.
Therefore in order to make the operations more profitable it is necessary to reduce the turnaround operations.
The current paper presents a methodology that will be useful for improving the boarding process using a combination of coloured Petri nets with simulation.
1.1. Review on Boarding Operations
Different authors have put focus on ways to improve the boarding process. Most of them put their efforts in managing the seats and the schedule of the boarding process in the best possible way. Some of the initial efforts for testing the seat scheduling are the ones presented by Marelli, Mattocks, and Merry (1998).
These authors simulated the allocation of seats by two different states (On-Off). Using this approach they analysed the results when the boarding is performed using a window, middle aisle (WILMA) scheme. Their conclusion is that the boarding time can be decreased compared to a random assignment boarding. That experiment was followed by defining a different one for the WILMA BLOCK, which uses the same approach but enhanced with separating the boarding in blocks, first the ones at the back, then the ones in the middle, and then those in the front.
Another relevant study is the one presented by Van Landeghem and Beuselinck (2002), through the use of EXCEL and ARENA they simulated some scenarios and conclusions were drawn from the discrete event approach focusing mainly on the boarding time.
Other authors have put focus on policies that could help reduce the boarding time and they use different techniques to test their policies. One interesting example is the work done by Steffen (2008), in which he finds the optimal allocation of passengers using an approach that uses Markov Chains and Monte-Carlo Simulation. Furthermore Soolaki et al. (2012) present an approach that models the problem using linear programming and genetic algorithms. Finally, Steffen and Hotchkiss (2012) tested different configurations using a real-size fuselage of a Boeing 757 for improving the use of the aisle to reduce the boarding time.
The review performed reveals that most of the studies have used different modelling techniques ranging from abstract models developed using linear programming to real-size sets that try to copy the real boarding situations. In the previous years the studies focused mainly on the boarding strategies while in recent years scientific community is paying more attention to the main drivers that are behind the processes that could reduce the boarding times. These drivers are inherent to human nature such as age, companions, family relationships, passengers with bags, disabilities, and even psychological conditions. For these reasons the researchers are investigating with scenarios that are as close as possible to a real situation.
However real-size experiments cannot be performed
anytime the researcher wants because they are expensive and it would result difficult to coordinate all the needed elements. This situation puts in evidence the need for the use of digital models that allow the researchers to integrate more characteristics that play a role in the process and result close to real situations.
The paper will focus on a methodology that avoids using the traditional abstract approaches such as Monte Carlo simulation or linear programing. Instead, the approach goes one step further using coloured Petri nets (CPN) together with simulation for integrating in a straight forward way the characteristics that play an important role in the boarding process and at the same time allowing the modeller to integrate as much elements as needed.
2. CPN-SIMULATION APPROACH
In this paper CPN is used together with simulation for developing a methodological approach that allows modellers to cope with the inherent complexity present in systems whose performance depend on multiple factors. The advantage of using a modelling formalism with the simulation approach is that the different micro- dynamics can be easily understood and the simulation model can be developed in a structured fashion.
Different situations, such as the ones present when the passengers with bags must allocate them in the upper compartments of the cabin, or those that appear when the passengers already sit block the ones that need to sit in the same section, can be modelled in a structured way resulting in a more reliable and robust simulator.
2.1. Coloured Petri Nets
Coloured Petri Nets is a simple yet powerful modelling formalism that allows to properly modelling discrete- event dynamic systems that present a concurrent, asynchronous and parallel behaviour (Moore and Gupta 1996; Jensen 1997; Christensen, Jensen, Mailund, and Kristensen 2001). CPN can be graphically represented as a bipartite graph which is composed of two types of nodes: the place nodes and the transition nodes. The entities that flow in the model are known as tokens and they have attributes known as colours.
The formal definition is as follows (Jensen 1997):
CPN (,P,T, A, N,C,G, E, I) (1) where:
∑ = {C
1, C
2, … , C
nc} represents the finite and not-empty set of colours. They allow the attribute specification of each modelled entity.
P = {P
1, P
2, … , P
np} represents the finite set of place nodes.
T = {T
1, T
2, … , T
nt} represents the set of
transition nodes, such that P T , which
normally are associated to activities in the real
system.
A = arc s such
N i assoc is a trans
C is spec place C(P
i) C
G is node used trans proc
E is E : A the q selec node is de the v gene
I is the v the simu scen
EXP any etc.) The stat marking, wh associated to expressions i.
3. THE CP A discrete-e developed in identifying a backward, p compartment, type of passen CPN model developing a model is that depending o identified and be developed simulator alon and impleme Mujica and P simulation m software SIM
The CPN that a passeng
{A
1, A
2, … , set, which rela h as A P T
is the node ciated to the place node sition node an s the colour
ifies the com e node such as C
j: P
i P,C
jthe guard fun es, G(T
i), G d to inhibit th
sition upon essed entities the set of ar A EXPR . F quantity and cted among th e in order to e
ealing with a values of the erated when tr the initializat value specific place nodes ulation. It is th nario.
PR denotes lo inscription
te of every C hich is com o each place .e. they canno PN-SIMULAT
event system n which the person in the putting the , or defining nger that boar lling forma DES simulat t the differen on the type
d therefore a d, which woul ne. The CPN ented using Piera (2011). F
model has b MIO.
N model mod ger experienc
, A
na} represe ate transition T T P .
e function N(
input and out then the ot nd vice versa.
set function mbination of c s C : P :
j
nction, associa : T EXPR . he event asso the attribute .
c expressions For the input a type of entit he ones prese enable the tran an output plac e output token ransition fires.
tion function, ation for the i s at the beg
he initial state ogic expressio
language (lo
CPN model is mposed by t
P and they t have any fre TION APPRO m (DES) mo
e micro-opera e block seat, m luggage on g different att
rd the aircraft lism. The ion model ba nt micro-opera
of passeng robust simula ld be too com elements hav the technique Following the
been develop dels the differ e when they a
ents the direct and place nod (A
i), which tput arcs. If o ther must be n, C(P
i), whi colours for ea
( ated to transiti It is norma ociated with t values of t s, E(A
i), such arcs they spec
ties that can ent in the pla nsition. When ce, they spec
ns for the sta .
I(P
i). It allo initial entities ginning of t e of a particu ons provided gic, function
also called t the expressio must be clos ee variables.
OACH odel has be ations such moving forwar the overhe tributes for t are based on t advantage sed on the CP ations that va ger are clea
ation model c mplex for a DE
ve been mapp e presented e same logic, t ped using t rent interactio are boarding t
ted des is one e a
ich ach
(2) ion ally the the as ify be ace n it ify ate ws in the ular by nal,
the ons sed
een as rd- ead
the the of PN ary arly can ES ped by the the ons the
ai th re pr us in m th ha th str m Th of co re ca m
co de Co
rcraft. The in hat appear w espective seat reviously hav sing the CPN nteractions ar modify the resp
hen implement ave been iden hen the imple
raightforward As it ha modelled are th he relations w f the aircraft onstructed u eplicating the abin is ob methodological
Figure
The CPN omposed by efinition of co Table 1:
olour Defin X Inte
Y
{000 010, 011, 11 Z Inte R Inte W
{0 01 10 D {0
nitial interacti when a passe t and he is b ve arrived. Th N model and e identified pective transit t it in the simu tified and mo ementation in d.
as been me he ones at th will be implem and the total using the s seat models s btained. Fig l approach.
2: The metho N model of
16 transitions olours is presen
Colour Defin nition eger It de
the s , 001, 100, 101, 11}
It de by th
eger
It rep peop the p eger It rep pass 01, 10, 00}
It rep the p 0,1} It rep perso
ions modelled enger needs blocked by p hese relations they illustrate then is just tion in the CP ulator. After th odelled with th n the simulat entioned, the he seat-level i mented in a mo l simulation m simulation p so that a final gure 2 ill
dological App f the seat-i s and 3 plac nted in Table nition and Des Descrip escribes the ro
seat block escribes the se
he passengers presents the am ple waiting in passenger to s presents the ro enger is suppo presents the se passenger.
presents if the on is seated in
d are the one to sit in hi passenger tha
are modelled e that if more necessary to PN model and he interaction he CPN mode tion model i e interaction in the aircraft odel of the sea model will be program and l model of the lustrates the
proach interaction i ce nodes. The
1.
cription ption
ow number of
eats occupied
mount of the aisle for it.
ow where the osed to be sat eat location of e waiting n the middle.
s s at d e o d s el s s t.
at e d e e
s e
.
f
W
SIT1 S
P
1’(Z, 1’(X,Y) D)
1’(R,100)
1’(Z,D) 1’(X,100)
@t6[Z=2 or Z=0 or z=1 , Y=000 ]
The information related to the place nodes is presented in Table 2.
Table 2: Definition of Place Nodes Place Colour
Set Description
S Product X*Y
This place represents the information of the left block of seats of one row of the cabin.
W Z, D
This place holds information about the passengers standing up and waiting for the passenger to sit.
P Product R*W
This place holds the information of the passenger.
The following figures present examples of the different transitions that compose the model. These transitions are grouped in the transitions that model the different sitting situations that a passenger can face when boarding a plane. The remaining interactions are modelled taking advantage of the characteristics of the simulation program once the replication of the modules has been performed.
Figure 3: Transition for Sitting
Figure 3 illustrates the situation when a passenger has to sit at the window and the row of seats is empty.
This situation is modelled by the value of the colour W for the passenger (W=100). In this example 100 means that the passenger seat is the one in the window and the other two are not his (using a binary coding). The value of the variable Y is 000 which represents that the row is empty and the value of the colour Z can be either 2, 1 or 0. For this situation the result is the same, either no one is waiting for the passenger to sit or there are passengers that stood up for letting the passenger reach the window (Z=1 or Z=2) . In this example the corresponding time consumption can be associated to the value of the variable t6, but it would depend on the correspondent study and maybe on the characteristics of the passenger sitting. Once the passenger is sat, the new colour value for Y is assigned with the output arc to S (Y=100).
Another example is presented in Figure 4, which represents the situation in which a passenger must sit at a window seat (W=100) and the middle seat is occupied by another passenger (Y=010) who was previously sat and must walk out to let the former reach his window seat.
Figure 4: The Walking Out of a Passenger Seated in the Middle
In this model a unit is added to the colour Z of the token in the waiting place node and the variable D turns to 1; in this way track is kept for the number of passengers waiting until the correspondent passenger is seated. Once this event takes place the correspondent person has left the middle seat (Y=000) and the corresponding time consumption is modelled by the value of t5. For the remaining transitions the authors encourage the reader to contact them but using the same logic the model can be developed.
Using the CPN approach it is possible to identify clearly the cause-effect relationships that sometimes hinder the smooth flow of passengers inside the cabin during the boarding or deboarding process. Moreover the colours can also be used to model characteristics such as age, size, number of bags, disabilities etc. and those characteristics can be used to simulate in a more accurate way the micro-processes that generate emergent dynamics once the people is interacting with each other.
4. IMPLEMENTATION OF THE CPN MODEL The CPN models can be integrated with the software tool making a mapping or implementing some logic such as the one presented by Mujica and Piera (2011).
In this example the micro-simulation of the interaction present in the system is implemented following the methodology previously presented. Since the software uses an object-oriented modelling paradigm the simulation of the complete cabin is performed efficiently. As it has been mentioned, firstly, the different transitions of the CPN model are used to develop the modular object that represents a row inside the cabin. Secondly, advantage is taken from the use of a modular approach when the different rows of the cabin are put together in order to make a complete model for the cabin that takes into account not only the micro-interaction between passengers but also the interaction that occurs at higher levels e.g. aisle blocking, speed reduction duo to the passengers, etc.
Figure 5 illustrates the elements used in the module that simulates the interaction of people at seat-level.
The methodology proposed by Mujica and Piera (2011) is used to implement the different transitions that occur during the seating process. The transitions implemented in the model are evaluated using the Separator objects
W
SIT1 S
1’(Z,D) 1’(X,Y)
1’(R,100)
1’(Z+1,1) 1’(X,000)
@t5 [ Y=010, Z=0or z=1 ]
S
j H
D
from SIMIO.
using Connec object (CPN those transitio fired thus per In Figure 5 th 1 to 10 an implemented
Figure 5: T Some SI nodes and tra logic behind different step SIMIO and elements calle stations used just used to passengers sa HOLD and P W and P of th
Figure 7 transition cal DECIDE ev (ModelEntity (Binary_Occ zero, one or tw
ModelEntity.Seat==
ld.contents==1||Hol
If the co assigned (AS availability of send the entit consumed dep and the leng software itsel
. The elemen ctors so that transitions) is ons that satisfy forming the si he TransX obj nd the logic
using the proc
he Elements o IMIO element
ansition node d the transit ps within the
the place no ed Stations. F in the object o store the at in the cabin PAX, are used
he CPN model
Figure 6: Sta 7 illustrates lled SIT2 (G valuates if th y.Seat=100), cupiedL=000)
wo (Hold. Co
="100"&&Binary_O ld.contents==2)
ondition is ful SSIGN) to th
f the row. Th ty to the corre pending on th gth of the pa
f).
nts are configu the logic ass s evaluated at fy the differen
imulation with ects correspon behind eac cesses window
of the DES So ts are used to es. As mentio ion nodes i e Processes odes are mod Figure 6 prese t; some of th entities that n seats. Other s
to represent l, respectively
ation Element the impleme General Sit).
he passenger and if the sea ) and if the pe ontents):
OccupiedL==0&&(
lfilled then the he variable e remaining s esponding seat he characterist ath (this is m
ured in casca sociated to ea t once and on nt restrictions a h high accurac nd to transitio ch transition
w of SIMIO.
oftware Model model the pla oned earlier, t is coded usi windows fro delled using t ents the differe hese stations a t simulate th stations, name the place nod y.
ts
ented logic First the st r wants to at row is emp eople waiting
(Hold.contents==0||
e new values a that stores t steps are used t and the time tics of the ent managed by t
ade ach nly are cy.
ons is
l ace
the ing om the ent are hat ely des
for tep sit pty g is
|Ho