PASSENGER AND BAGGAGE FLOW IN AN AIRPORT TERMINAL: A FLEXIBLE SIMULATION MODEL Lorenzo Brunetta Giorgio Romanin-Jacur University of Padova University of Padova Via Gradenigo 6/A Stradella San Nicola 3 35131 Padova PD 36100 Vicenza VI Italy E-mail: brunetta@dei.unipd.it Italy E-mail: romjac@dei.unipd.it KEYWORDS Airport terminal, Discrete simulation, Flexible model, User friendly interface. ABSTRACT The lack of a friendly and flexible operational model of landside operations motivated the creation of a new simulation model adaptable to various airport configurations for estimating the time behavior of passenger and baggage flows, the elements capacities and the delays in a generic airport terminal. The validation of the model has been conducted by comparison with the results of previous research about the average behavior of the future Athens airport. In the mean time the proposed model provided interesting dynamical results about both passenger and baggage movements in the system. INTRODUCTION The modeling of airport terminal operations has advanced significantly over the last 15 years; see for instance (Tosic, 1992) and (De Neufville and Odoni, 2003). Available models have improved in detail and reliability, as well as user friendliness. As a result, their use as decision support aids in management or design tools in terminal development projects has been steadily increasing. Some existing models are strategic in nature, as they sacrifice level of detail in exchange for speed and flexibility, and are employed for long-medium term decisions, while others are primarily tactical, incorporating high level of detail in data and system definition, and are employed for medium-short term decisions. Recent tactical simulation models evidence two main defects. Some of them, like for instance (Gatersleben and Van Der Wej, 1999) and (Joustra and Van Dijk, 2001) either model one airport but cannot be extended to other ones, or model one part of the airport. Other models, like for instance (Brunetta and Romanin-Jacur, 1999) get wide flexibility, i.e., ability of describing different airports in detail by means of limited adjustments, but lack user friendliness. The lack of a satisfactory tactical simulation model of landside operations motivated the creation of our new flexible simulation model for estimating the time behavior of passenger and baggage flows, the capacity and the delays in a generic airport terminal. Our model overcomes, on a lowcost platform, typical problems of the old packages, such as large data requirements and lack of flexibility. In fact it includes the common features of all airport terminals and, after an appropriate definition of the key parameters, is capable of describing any terminal configuration. All input data concerning airport structures and related operation times are inputed by means of guided menus, while flight plans are inputed by a text file. Any modification or adaptation is very easy, also for non expert users. ENVIRONMENT In recent years the amount of passengers simultaneously present in airports has dramatically increased because of many different causes. On one side the risk of attempts does not reduce significantly the amount of travellers by plane, but at the same time it imposes strong actions against terrorism taking place in the airports or originating from them; such actions require strict checks on passengers, baggage and goods to be boarded, which lower all movements inside the terminal, and consequently oblige passengers to extend their stay before boarding. On the other side significant changes in air travel management were imposed by the necessity of strong savings for airline companies. Many companies adopted the hub and spoke system, which replaces point-to point direct flights with a series of two or more indirect ones, linked together at the company hub airport (or airports); such a system plans several flights converging to the hub airport, linked to other flights originating from it, with a short time between arriving and starting ones; as a consequence large flows of transit passengers affect the hub terminal in some well defined intervals of the day, especially in the early morning and in the late afternoon; moreover the system permits to correctly dimension employed aircraft and therefore to obtain a high utilisation index. Code sharing between airlines, as well as charter and low cost companies, succeed in filling aircrafts, and often encourage the use of larger aircraft. All abovementioned changes contribute in increasing the amount of passengers simultaneously asking for the same facilities (ticket counters, check-in desks, security checks, baggage conveyors, baggage sorting stations, baggage inspection station, etc.), especially during peak intervals of the day. It is obvious that in this hard situation all (technological and human) resources are particularly stressed. In order to contain costs and yield efficiency (avoid congestion, i.e., delays, queues and bottlenecks) a clever utilisation of structures and facilities (and related servers) is required, possibly sharing some of them among different users. 361
MODELING APPROACHES A correct management of an airport terminal may be considered from two different points of view. A first approach concerns resource dimensioning based on the mean behaviour of the system: here we consider the typical days we foresee in the year (in particular the busiest day ), by taking into account the related peak intervals. Resource activation and flow ruling are statically decided, in order to obtain acceptable levels of service, based on the above estimates. Obviously we adopt a deterministic and aggregate model, which is generally utilised to determine the total amount of resources at disposition; the main advantage of such a model is its quick response and its wide flexibility, as it may be easily adapted to different realities; on the contrary, as it solves static dimensioning problems during predetermined intervals, it is able neither to describe transient evolutions of the system between adjacent (quasi) equilibrium states, nor to reproduce the system behaviour consequent to any alteration of the inputs (see for instance delays and similar). A second alternative (and sometime complementary) approach is addressed towards a detailed description of the system behaviour, consequent both to normal and to exceptional situations. As an example, we may want to examine the system evolution in time, arising from a temporary interruption of air services due to fog, from a congestion due to delay of one or more flights, etc., and thus to plan in advance, by accurate scheduling, all suitable reactions to reduce disease and to recover effectiveness and efficiency. As a second example, remember that the whole system simultaneously manages passengers, baggage and goods, to be loaded, unloaded or transferred from an aircraft to another; they generally follow different routes and utilise different facilities, but they meet in many common points and interact; possible delay or unsatisfactory behaviour of one of the subsystems may affect the whole system behaviour, causing further delays and deadlocks: we may want to foresee, monitor and rule all subsystem interactions to avoid delays and/or inefficiencies, or anyway to reduce system sensitivity to random accidents which may trouble it. In order to satisfy all abovementioned requests the only answer is supplied by discrete simulation. THE SUGGESTED MODEL A useful airport terminal simulation model should describe faithfully the particular airport under study; a model specifically built up and implemented only for that surely satisfies such a request, but it presents the limit of being hardly adaptable both to possible future transformations and expansion, and to other airports, due to all peculiar characteristics; in other words, it is a heavy job to obtain an instrument to be employed only a few times, to satisfy particular exigences. On the contrary, a generalised flexible model, describing the common structure of all airports, and built up to be easily parametrised in order to model different airports, requires about the same initial effort but permits to be employed many times, both for different airports and for different operating conditions. The aim of the paper is a flexible model implemented by a user friendly package. The model describes in detail all operations effected by both passengers and baggage; in particular we consider: passenger generation; departing, arriving and transit passenger movements in the terminal, with distinction between Schengen/domestic and extra Schengen/international passengers; departing, arriving and transit baggage movements by means of the baggage handling system. Flight departures and arrivals take place following the planned schedule (obviously delays may be considered): according to the flight type (arriving or departing flight, border crossing or non crossing, aircraft capacity and saturation, passenger characteristics), every flight generates passengers according to a known passenger presentation time distribution, generally a beta type one. Departing passengers may need to buy tickets, choose a check-in desk (according to the number of open desks and to given rules), ask for check-in and simultaneously may generate baggage, cross security control blocks, move towards the waiting lounge, are assigned a gate, move to the gate and are boarded; if extra Schengen/international they follow a different route after checking-in and have to cross passport control blocks too; late passenger may follow a special route, if possible, related to the flight departure time. Arriving passengers on their own enter the arrival gate and reach the baggage claim area; if extra Schengen/international they have to follow a different route crossing passport control and customs. Transit passengers constitute a fraction of arriving passengers: after entering the arrival gate they move towards the right waiting lounge to join departing ones. Obviously passengers may form queues whenever they ask for a scarce resource, like desks, security check blocks, passport control desks, etc.. Departing baggage, coming either from the check-in desks, cross x-ray machines (all baggage or a fraction, according to the airport rules) and possibly further inspection apparels and/or operators if necessary or convenient, then they are sorted and sent to the right pier to be loaded; particular (irregular, oversize, etc.) items are treated apart. Arriving baggage are unloaded, moved to the inbound area, sorted and then sent either to the baggage claim area (arriving baggage), or to the transit baggage area (transit baggage), to be possibly relabelled and finally moved to join departing baggage. Baggage may form queues whenever a scarce resource is to be employed. MODEL IMPLEMENTATION The proposed model is implemented by package MicroSaint 3.2 on a common use personal computer; MicroSaint showed to be a good compromise between abstraction and user friendliness. In particular it revealed to be easily readable also by non expert users. The model implementation include 51 tasks, as may be seen in Figures 1-4, where the main model network and three subnetworks are reported; it is comparatively simple and sythetic, but however it represents an actual airport in detail. There every task corresponds to a well defined operation, as evidenced by the related box name, and contains all information about: 1) possible queue for entering entities 362
(aircraft, passengers or entities) and related rules (e.g., priority); 2) beginning conditions (e.g., required resource at disposition) and beginning effects (e.g., resource sizing); 3) time to perform the operation (probability distribution); 4) ending effects (e.g., resource releasing); 5) decision about outgoing entities, which may be either multiple (i.e., the entity is multiplied and every new generated entity follows a different route) or probabilistic (a single route is chosen with given probability) or tactical (a single route is chosen according to a given deterministic rule); obviously information include the related parameters. Therefore all airport functional characteristics are stored in the related tasks, together with possible time and/or state depending operating rules. All input data concerning arriving and departing flights (type of aircraft, number of passenger and related baggages, aircraft arrival or departure time, distribution of the arrival time to the airport terminal for departing passenger, etc) are supplied by a text file, to be written in the Input data 1.2 box in Figure 2. Such a way all data which are necessary to define both the airport characteristics and the planned operations are easily and clearly set and may be written in plain language. Model outputs include: delays on departure, flight by flight, distinguished between those caused by passengers and those caused by baggage; last passenger check-in time, flight by flight (departing passengers); last passenger leaving the airport time, flight by flight (arriving passengers); queue length distribution, maximum queue length, queueing time distribution maximum queueing time, at every facility. Obviously simulation may be effected with different amount of open facilities and/or with different operating conditions and rules, both normal and exceptional (e.g., consequent to any alteration of flight schedule, due to bad weather conditions, strikes, etc). APPLICATION RESULTS The suggested simulation model was employed to describe Athens International Airport, newly built for the Olimpic Games, currently not yet completed. The airport model was tested under normal conditions, corresponding to 508 aircraft movements per day under the planned operating conditions. All average simulation results were positively checked by comparing them with the ones, reported in (Andreatta et al. 2001), supplied by aggregate deterministic model SLAM, described in (Brunetta et al. 1999). The results revealed that the airport is well dimensioned and possible departure delays or decrease from high values of service levels recommended by International Air Transport Association (IATA) are rare and last for very short time intervals. However it was possible to suggest some small changes in the number of open check-in desks during well determined intervals of the day, by which a strong reduction in the departure delays and in the passenger queueing times can be obtained. The model will be employed soon for other applications. The case of Athens International Airport. In Proceedings of Odysseus 2003. CD, Palermo, Italy. Brunetta, L., L. Righi and G. Andreatta. 1999. An Operations Research Model For The Evaluation Of An Airport Terminal: SLAM (Simple Landside Aggregate Model). Journal of Air Traffic Management, 5, 161-175. Brunetta L. and G. Romanin-Jacur. 1999. A flexible model for the evaluation of an airport terminal. In ESS 99 Simulation in Industry, G. Norton, D. Moller and U.Rude eds. SCS Delft, The Netherlands. De Neufville R. and A. Odoni. 2003. Airport system planning, design and management, Mac Graw Hill, Columbus, OH, USA. Gatersleben M. R. and S. Van Der Wej, Analysis and simulation of passenger flows in an airport terminal. In 1999 Winter Simulation Conference, P. A. Farrington, H. B. Nembhard, D. T. Sturrock and G. W. Evans, eds. IEEE Piscataway, NJ, USA. Joustra P. E. andn. M. Van Dijk. Simulation of check-in in airports. In 2001 Winter Simulation Conference, Peters B. A., Smiths J. S., Madeiros D. J. and Rohrer M. V., eds. IEEE Piscataway, NJ, USA. Tosic V. (1992) A review of airport passenger terminal operations analysis and modeling. Transportation Research 26A, 1, 3-26. BIOGRAPHY LORENZO BRUNETTA got a Laurea degree (equivalent to Master of Science) in Mathematics in 1991 and a PHD in Computational Mathematics and Computer Science in 1995 both from the University of Padova. From 1996 to 2001 he was Assistant Professor at the Engineering School of the Polytechnic of Milan. He is currently Associate Professor of Operations Research at the University of Padova. He collaborated at several EU, NATO, and MIUR Research Projects. He is coauthor of over 40 publications including fourteen in International journals. His main research interests are Combinatorial Optimization and Integer Linear Programming and their applications to Air Traffic and Telecommunication Systems. GIORGIO ROMANIN-JACUR got a Laurea degree (equivalent to Master of Science) in Electrical Engineering in 1970 from the University of Padova. He was fellow of the Italian Council of Researches in 1970-71; after military service in 1971-72 he was Research Associate of the same Council until 1980; in 1980-83 he was lecturer of Operations Research at the School of Engineering of the University of Padova; in 1983-2000 he was Associate Professor and since 2000 he is Full Professor of Operations Research at the same School. He currently teaches to the class of Management Engineering and to some Master and Doctorate classes in the Schools of Engineering, Medicine and Agriculture. He collaborated at several EU, CNR, and MIUR Research Projects. He is coauthor of over 100 publications including 28 in International journals. His current researches concern production scheduling in the fields of Industry and Services. REFERENCES Andreatta G., L. Brunetta, L. Righi and G. Romanin-Jacur. 2001. Evaluating Terminal Management Performance Using SLAM: 363
Figure 1: Model network
Figure 2: Passenger generator subnetwork Figure 3: Baggage Handling System subnetwork for departing baggage Figure 4: Baggage Handling System subnetwork for arriving baggage